Hindawi Publishing Corporation

e Scientific World Journal

Volume 2016, Article ID 9858101, 14 pages
http://dx.doi.org/10.1155/2016/9858101

Research Article
Multiagent Systems Based Modeling and
Implementation of Dynamic Energy Management of
Smart Microgrid Using MACSimJX

Leo Raju, R. S. Milton, and Senthilkumaran Mahadevan

SSN College of Engineering, Chennai, Tamil Nadu 603110, India

Correspondence should be addressed to Leo Raju; [email protected]

Received 23 November 2015; Accepted 1 March 2016

Academic Editor: Chin-Chia Wu

Copyright © 2016 Leo Raju et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The objective of this paper is implementation of multiagent system (MAS) for the advanced distributed energy management and
demand side management of a solar microgrid. Initially, Java agent development environment (JADE) frame work is used to
implement MAS based dynamic energy management of solar microgrid. Due to unstable nature of MATLAB, when dealing with
multithreading environment, MAS operating in JADE is linked with the MATLAB using a middle ware called Multiagent Control
Using Simulink with Jade Extension (MACSimJX). MACSimJX allows the solar microgrid components designed with MATLAB to
be controlled by the corresponding agents of MAS. The microgrid environment variables are captured through sensors and given to
agents through MATLAB/Simulink and after the agent operations in JADE, the results are given to the actuators through MATLAB
for the implementation of dynamic operation in solar microgrid. MAS operating in JADE maximizes operational efficiency of solar
microgrid by decentralized approach and increase in runtime efficiency due to JADE. Autonomous demand side management
is implemented for optimizing the power exchange between main grid and microgrid with intermittent nature of solar power,
randomness of load, and variation of noncritical load and grid price. These dynamics are considered for every time step and complex
environment simulation is designed to emulate the distributed microgrid operations and evaluate the impact of agent operations.

1. Introduction reducing cost, and increasing degree of operational redun-

dancy.
Industrial automation is reaching a new level due to initia- The world today is witnessing a gradual transition in
tive taken towards industry 4.0. Conventional engineering the form of generation and distribution of electricity. The
approaches require every future interaction to be known in liberalization of the energy industry, together with decar-
design time, whereas the future approaches require adaptive bonization and inclusion of renewable energy sources, as well
systems capable of efficiently adapting to failures, component as ground-breaking Information and Communication Tech-
replacements, and changes in the environment with less nologies (ICT) developments, has huge consequences on the
human intervention. They should be scalable, adaptable to way the sector works: we are moving rapidly towards a more
unforeseen situations, and capable of handling evolution of decentralized, more sustainable, and smarter power system.
their complexity [1]. These characteristics can be realized The smart grid paradigm represents a transition towards
by enabling the system to dynamically reorganize by itself. intelligent, digitally enhanced, two-way power delivery grids
These types of systems are getting more complex, open, [2]. Microgrid is the building block of smart grid and is poised
and critical because they are adopted to remotely supervise to play a major role in enabling the widespread adoption of
and control most critical infrastructure utilities such as renewable, distributed energy resources.
power grids and water transportation. MAS has received However, renewable energy integration into the micro-
compelling attention in the recent times because of its capa- grid entails new challenges for control as the power generated
bility of significantly improving the operational efficiency, is intermittent in nature, which impacts the dynamics and
2 The Scientific World Journal

stability of the microgrid [3]. To meet these challenges, power management in microgrid with distributed agents is
microgrid monitoring should incorporate communication discussed with case studies in [16]. The various trends in
and control to achieve an optimal balance between gen- microgrid control are discussed in [17]. The various control
eration, energy storage, and load demand and also react strategies to improve stability of microgrid are discussed here.
quickly to changes. Dynamic energy management is a key The complete review of microgrids in MAS perspectives is
enabler for the integration of renewable power generation discussed in [18].
onto the microgrid. Smart energy management of microgrids Distributed online optimal energy management for smart
using genetic algorithm and neural networks is discussed grid is discussed in [19]. Here MAS is used for interactive
in [4]. Energy management of hybrid renewable energy operation of economic dispatch and demand response for
generation using fuzzy logic method and constrained opti- maximizing the social welfare. MAS simulations imple-
mization methods is discussed in [5, 6]. Other computational mented through JADE are discussed here. Multiagent system
intelligence methods and classical algorithms for energy implementation for home energy management is clearly
management of microgrids are discussed in [7]. explained with case studies in [20]. But the hardware imple-
All of these researches address microgrid operation mentation details are not given. A detailed review on agent
problems in a centralized manner. Traditional Supervisory concepts applied to intelligent energy systems is given in [21].
Control And Data Acquisition (SCADA) system has a central Real-time microgrid control strategies are discussed in
controller, which gathers all system knowledge and makes the very recent paper [22]. The control strategies to optimize
decision. Dynamic adaptation leads to complication in cen- the power exchange between microgrid and main grid are
tralized systems. Moreover this central coordinator will pro- discussed here. But facilitating MAS for real-time operations
voke considerations about scalability, computational com- of microgrid is not discussed in detail. In most of the
plexity, and communication overhead. But the MAS archi- references, MAS simulations of microgrid operations are
tecture can reduce these concerns by following decentralized implemented in JADE platform and the real-time operation
approach. A multiagent solution to energy management in of MAS is not adequately addressed. In order to practi-
hybrid renewable energy generation system is discussed in cally realize the MAS operations in the field, MAS has
[8]. Agent based control frameworks are discussed in [9]. An to be linked with MATLAB/Simulink model to sense the
agent based operation of microgrid is discussed in detail in environment variables and the outcome of the multiagent
[10]. Here the capabilities of MAS to improve the operational operations should be given back to the actuators through
efficiency are analyzed. Optimization of microgrid with inter- MATLAB/Simulink for physical action. But MAS operations
mittent renewable energy resources using MAS is discussed in JADE cannot be linked with MATLAB, as MATLAB
in [11]. Here, MAS approach is used in a dynamic environ- does not support parallel operations, which is essential for a
ment to distribute decisions related to microgrid operations decentralized approach such as MAS. So we consider a novel
and eventually to improve performance and stability. But approach, Multiagent Control Using Simulink with Jade
decentralized approach of MAS is not exploited here. MAS Extension (MACSimJX), to link MATLAB with MAS to allow
technology is used for controlling microgrid operations and system designed with Simulink to be controlled by agents
also for optimal energy exchange with the grid as discussed that operate in JADE, a powerful development environment
in [12]. Fully decentralized approach of MAS for microgrid for modeling MAS [23]. A decentralized multiagent system
operation and participation in energy market is discussed approach for service restoration using islanding of distributed
here. But each agent should have complete knowledge about generator is discussed in [24]. Only outage management is
the environment and decision making capabilities to be fully discussed in detail here. The other challenges of integration
autonomous. It is not economical to be fully autonomous. of renewable energy resources and the controllable loads
Also we want to use agent communication for coordinated are not considered. Latest survey on multiagent systems for
action. So some level of centralized control is necessary for microgrid control is given in [25]. The papers considered in
coordinated operation. So a hierarchical approach of MAS for the survey explain the idea of practical implementation of
microgrid operation is considered in [13]. Here, the design MAS but a hardware ready model is not clearly addressed
and implementation details of MAS in microgrid energy in any of the papers. Very recently the integration of JADE
management and MAS facilitating the seamless transition with Simulink using MACSimJX is discussed for energy
from grid connected to an island mode are also discussed. management in distributed energy resources (DER) of micro-
But linking MAS with MATLAB is not explained in detail. grid in [26]. However, the agent programming concepts,
Real-Time Digital Simulator (RTDS) is used for real-time linking MAS with MATLAB/Simulink model and strategies
operation of MAS on microgrid is discussed in [14]. Here for sensing and controlling the dynamic variations of the
MAS system was implemented in JADE platform and was environment, are not adequately addressed in this paper.
interfaced with RTDS via TCP/IP protocol. But sensing all
The shortcomings of the related works which bring the
the dynamic variations and the control strategies are not
motivations for the proposed approach are given below.
addressed here.
Multiagent based distributed energy management for
intelligent microgrid is discussed in [15]. Here MAS sim- (i) Previous research related to microgrid energy man-
ulated for market operations of integrated microgrid is agement using MAS has mainly focused on prototype
discussed in detail. Practical realization of MAS for market design and simulation of distributed control methods
operation is not attempted in this paper. A comprehensive but the need to test and validate the implemented
The Scientific World Journal 3

applications, in a coded form ready for deployment,

has not been adequately addressed.
(ii) MAS linking MATLAB for microgrid energy man-
agement is not discussed in detail. Time synchroniza- DC/AC
Switch
tion in cosimulation between MAS and MATLAB is
not explained in detail.
(iii) All the options available for effective energy manage-
ment of microgrid has not been considered. DC/DC
(iv) The impact of stochastic nature of the renewable
resources on the microgrid operations and the mit-
igation process are not adequately addressed. Also
problems of stability and reliability due to increased DC/DC DC/AC
penetration of renewable energy resources are not
considered.
(v) Microgrid is building block of smart grid and so all
the features of smart grid can be implemented in
microgrid. But only energy management is discussed
Figure 1: Solar microgrid.
in most of the works. Controllable loads, dynamic
pricing, and plug and play features are not discussed
so far in the microgrid environment.
to process the environmental information, received from
(vi) The process of development, verification, and deploy- Simulink, and optimize the microgrid operation through
ment of time-critical MAS applications has not been autonomous actions and also collaborating with other agents
adequately addressed to date. for strategic decision.
(vii) Multiagent system features such as runtime adaptive The rest of the paper is organized as follows. Solar
behavior, less communication overhead, and less microgrid is explained in Section 2. A detailed discussion
computational demands are not adequately exploited on multiagent system approach and multiagent platform
for microgrid energy management applications. is given in Section 3. Implementation of dynamic energy
management of solar microgrid in distributed environment
In the proposed approach, all the above-mentioned short- using JADE is given in Section 4. Section 5 deals with
comings are addressed adequately. Here a comprehensive agent formulations and Section 6 deals with implementation
approach is used, considering all the options available in a of control operations of solar microgrid for demand side
microgrid for improving the energy management efficiency management, using MACSimJX. Case study, simulation, and
in a dynamic, distributed environment. In this approach, validation are given in Section 7. Conclusion is given in
MAS is implemented in JADE and then it is linked with MAT- Section 8.
LAB/Simulink, using Multiagent Control Using Simulink
with Jade Extension (MACSimJX). This opens MATLAB to 2. Solar Microgrid
one of the most strongly supported open source agent devel-
opment environments, allowing real-time hardware systems The architecture of the considered solar microgrid is shown
to be modeled alongside and interacting with software model in Figure 1. The consumer can cover his demand partly by
such as multiagent systems, bringing MAS closer to practical using the electricity produced by the local renewable (solar)
implementation. Here, the solar microgrid system designed generator, partly by stored electricity in the battery when the
with Simulink is controlled by MAS agents that operate solar source is not available. The battery is charged when
in JADE. MAS in JADE is fully exploited for control and the solar source is available, and can discharge the storage
automation in distributed, dynamic energy management of when needed. The consumer has the possibility to control the
solar microgrid, considering the intermittent nature of solar storage and the solar power generator. The challenge in solar
power, randomness of load, and dynamic pricing of grid. New microgrid is that the power supply is intermittent in nature
control strategies are formulated through generating System and with the ultracapacitors and battery the randomness
State Number (SSN) and Control State Number (CSN) in the supply should be managed so that the load is given
to sense all the environment variable values dynamically constant supply all the time irrespective of the nature of the
every hour and reflect these values in JADE agents for solar power. The battery can be charged by the solar power or
taking strategic action to maintain the stability and improve from the grid.
reliability of the microgrid with stochastic nature of solar A solar PV module consists of a number of solar cells
power and load. Also demand side management strategies connected in series or parallel based on the requirement.
are implemented autonomously by controlling of noncritical There are various factors which affect the solar power like
Load (NCL). The automated application is represented as an solar irradiance, temperature, partial shading and cloud,
agent control application where each microgrid component arrangement of the cells, and the angle of tilt of the panel.
is represented by an agent and the task of the agent is The maximum power is found by Maximum Power Point
4 The Scientific World Journal

Tracking (MPPT) algorithm [27]. The MPPT refers to the

Agent
point with a maximum power output in the curve under
specific external temperature and solar irradiation.
Reasoning
3. Multiagent Systems Approach
3.1. Multiagent Systems. Autonomous components and coor- Proactive Optimizing Reactive
dination are the basic ingredients of any distributed system.
Distributed systems that involve many heterogeneous entities
Controlling
have some major limitations:
(i) Interactions among participating entities are fixed by
Learning
application developer while coding and hence they
lack runtime adaptive behavior.
(ii) Some applications have to operate in environments
where maintaining continuous communication is
expensive. So distributed systems with many ongoing Social
interactions are almost infeasible.
These considerations have motivated the development of Figure 2: Functionality of an agent.
approaches to distributed system based on agents which
provide ways for adaptation and ongoing interaction. A
multiagent system (MAS) is a distributed system consisting networks, may be inadequate to cope with the high
of multiple software agents which form “a loosely coupled penetration of DERs and complex control decisions due
network,” to work together to solve problems that are beyond to its centralized approach and lack of flexibility and
their individual capabilities or knowledge of each entity. extensibility. In the microgrid, uncertainty in SCADA
Multiagents overlay a way to elaborate systems that are systems arises when sensor data or inferred knowledge
decentralized rather than centralized, emergent rather than cannot be deemed accurate due to intermittent nature
planned, and concurrent rather than sequential. MAS has of DERs. Also SCADA has problem dealing with inherent
inherent benefits such as flexibility, scalability, autonomy, noise/error in sensor data or knowledge as well as uncertainty,
and reduction in problem complexity. Agents have certain incompleteness, and inconsistent or conflicting data from
behavior and tend to satisfy certain objectives using their multiple, heterogeneous sources. Humans have traditionally
resources, skills, and services. Ability to produce is a skill supervised such problems to reason and resolve issues. A
and selling power to grid is a service. The way that the agent multiagent Energy Management System (EMS) can cope
uses its resources, skills, and services defines its behavior with heterogeneity and give better, faster solution than
and the behavior of each agent is formed by its goals. The SCADA, taking the automation of microgrid to the next
MAS approach’s goal is to control a very complicated system level. With decentralized approach, a MAS has its own
with minimum data exchange and minimum computational perception of the environment, goal, and agenda and it tries
demands. to achieve the best for itself while behaving strategically.
In MAS, several autonomous and intelligent entities MAS can deal with disadvantages of SCADA and increase
called agents are working in collaboration to achieve the over- the operational efficiency of solar microgrid due to its inher-
all goal of a system. Agents have four behavioral attributes, ent characteristics and functionalities taking the automation
autonomy, social, proactive, and reactive. Autonomy refers to of microgrid to the next level. MAS are by nature distributed
the principle that agents can operate on their own to meet and concurrent; they are independent entities engaged in the
their goals without the need for human guidance. Agents system; microgrid is tightly associated with the communica-
are proactive, that is, the ability to take the initiative rather tions between stakeholders and entities (agents) to exchange
than acting simply in response to their environment. Agent information. Plug and play adaptability and connection to
can cooperate with other agents for coordinated action. In external grid are seamless in MAS based microgrid. By
order to cooperate, agents need to possess social ability, nature, MAS can be scaled up by adding other agents or
that is, the ability to interact with other agents with some by dispersing them in new environment with new resources
communication language like Agent Communication Lan- and capacities. MAS is particularly useful for designing
guage (ACL). Agents are reactive to changes in environment. distributed systems requiring autonomy of their entities.
Reasoning, optimizing, controlling and learning are the The development of smart grid and related technologies
inherent characteristics of an agent, as shown in Figure 2. For combines advances in distributed systems, artificial intelli-
agent systems to be truly smart, they would have to learn as gence, control, and information and communications tech-
they react and/or interact with their external environment. nologies. MAS in mirogrid leads to high level of autonomy,
self-healing, and reliability and leads to providing features
3.2. Multiagent Systems in Microgrid. Centralized SCADA, such as reconfiguration, protection, restoration, and interac-
which was originally designed for traditional passive tion with other users through demand response.
The Scientific World Journal 5

In MAS, set of economic and control mechanisms is used

for dynamic balance of supply and demand across the entire Agents Agent Management Directory
electrical infrastructure using value as a key operational System facilitator
parameter.

Message transport system

3.3. Multiagent Platform. JADE provides a convenient dis-
tributed platform for users to focus on developing agents for
control and monitoring of power balance during microgrid
operation. JADE is an open system which supports plug
and play capabilities and is also scalable without much
modification to the control scheme. In this paper, JADE Figure 3: Architecture of multiagent system.
framework that conforms to FIPA (Foundation of Intelligent
and Physical Agent) standard for intelligent agents is used.
Agent platform is a software environment, where software (PV) systems, one in the department of the college and the
agents run. Agent lives in a container, and a collection of other in a hostel with the capacity 1000 kW each, each of
containers make up a platform. Agent Management Service which contains a local consumer, a solar PV system, and a
(AMS) is responsible for managing the agent platform, which battery. Every hour solar power is taken from NREL (National
maintains a directory of Agent Identifiers (AIDs) and agent Renewable Energy Laboratory). All the electrical appliances
states. Each agent must register with an AMS in order to are considered for per hour load calculation based on their
get a valid Agent ID. Directory Facilitator (DF) provides the usage for the department and hostel. The solar power and the
default yellow page services in the platform which allows load graphs of department and hostel are shown in Figures 4
the agents to discover the other agents in the network based and 5.
on the services they wish to offer or to obtain. A MAS The solar power, load, state of charge (SOC) of the battery,
architecture is shown in Figure 3. noncritical loads, and dynamic pricing of grid are monitored
JADE is also used as the runtime environment in which hourly, and based on these data, the agent takes best possible
agents execute, thereby masking from the agents the under- actions autonomously for dynamic energy management of
lying complexity of the operating system or network. Agents the solar microgrid in a distributed environment. Consider-
can span multiple computers or be on one computer, yet for ing all the possible options available for the solar microgrid,
the implementation, the code for sending and receiving mes- a flow chart is drawn as shown in Figure 6. The proposed
sages is the same. A platform encompasses all the containers system has the following agents: solar power generator agent,
within an agent system. JADE contains main containers load agent, grid agent, diesel agent, and control agent. Each
and peripheral containers. When a platform is created, the PV system has all these agents. Multiagent programming is
main container is always the first container to be initialized done in JADE in Eclipse environment. The overall procedure
in JADE. Agent Management System (AMS) and Directory is the following.
Facilitator (DF) agents are also automatically created once the
main container is initialized. The directory services and other (1) Here first the department load agent (LDA) commu-
administration services are hosted on the main container, nicates the power demand through an ACL message
which is the first container launched in the platform, but with the available solar power in department solar
are duplicated on the other containers for robustness. JADE agent (SDA) in the department at that specific hour.
platform provides a set of functions and classes to implement If surplus power is available in SDA, SDA takes
agent functionality, such as agent management service, Direc- autonomous decision to give power to LDA. The
tory Facilitator, and message passing services [28]. remaining power is used to charge BDA and BHA
through control agent. After charging the batteries,
the excess power is given to the grid through control
4. Implementation of Multiagent agent. Thus, the load agent makes the local decision
System Based Energy Management of as well as communicates with other agents for global
Solar Microgrid with JADE decision.
The proposed multiagent system is implemented using the (2) If power available in SDA is not sufficient for covering
Java Agent Development Environment (JADE). JADE is a load, it checks with the availability of solar power
Java-based open source software framework for develop- in the hostel solar agent (SHA). If required power
ing multiagent systems. The JADE architecture is built on is not available then it looks into the battery of the
peer-to-peer modality. Intelligence, initiative, information, department (BDA). If required power is fully available
resources, and control can be fully distributed across a group it takes from BDA and if partially available, SDA
of heterogeneous hosts and devices, through wireless or checks with the BHA. If required power is available,
wired networks. Each agent can communicate and negotiate it is taken.
with its peers to reach mutually acceptable agreements for (3) If the power is still required after taking from solar
cooperative problem solving. We consider a grid connected unit and battery, SDA checks for noncritical load
solar microgrid system containing two solar photo voltaic shedding through control agent for implementing
6 The Scientific World Journal

1200 Thus, every hour the solar microgrid energy manage-

ment is done dynamically for distributed optimiza-
1000
tion of solar microgrid by using multiagent system in
800 JADE platform. Programming is done for every agent
in Java in JADE platform and executed in Eclipse
(kW)

600 environment. All the communication between the

agents is done through Agent Communication Lan-
400 guage (ACL). The complete interactions are shown
in the sniffer diagrams. The console output gives the
200
transaction report of a particular scenario.
0
1 3 5 7 9 11 1 3 5 7 9 11
5. Agents Formulations
(hr)
Load The proposed multiagent system comprises many intelligent
Solar power agents representing various components in a microgrid. Each
agent has a localized knowledge base, containing rules and
Figure 4: Solar power and load for department.
behaviors, which governs its decision making process.
The following agents are formed to simulate multiagent
700 system in JADE environment.

600 5.1. Generator Agent. Generation agent (GA) receives power

request from load agent. It allows owners of solar generators
500 to set selling price for trading. The initialization for all agents
begins with registering itself with the Directory Facilitator
(DF) by providing it with a set of service descriptions like
400
to setName and setType. For GA, setName and setType
(kW)

are chosen to be solar power source and power selling,

300 respectively. Department solar agent (SDA) and hostel solar
agent (SHA) are considered in our works.
200
5.2. Load Agent. Load agent allows customer to specify the
100 amount of power to purchase and communicate with the
generator agent. The load agent is taken as a buyer agent
0 because it searches the DF yellow pages for generation agents
1 3 5 7 9 11 1 3 5 7 9 11 offering power supply services. It first registers itself with
(hr)
DF by providing load and power buyer as input parameters
Load
to setName and setType, respectively. If there are no GA
Solar power in the DF yellow pages, it checks for batteries and finally
ends with control agent (CA) buying power from the grid to
Figure 5: Solar power and load for hostel. meet its load demand. Thus, load agent makes local decisions
and communicates and collaborate with other agents for
collective decision. Department load agent (LDA) and hostel
demand side management strategies. Load response load agent (LHA) are considered in our case.
strategies include both load shedding and load shift-
ing. Load shedding involves curtailing equipment
5.3. Grid Agent. Grid agent (GA) registers with DF by giving
that is not mission critical and load shifting is the
grid agent and grid status as inputs to setName and setType,
rescheduling of energy-intensive operations to a dif-
respectively. Grid agent collects real-time grid pricing and
ferent time period. Noncritical loads can have many
informs other agents about connection status of grid. Grid
priorities based on the requirement. Even after this,
price varies every hour leading to dynamic pricing concept
if load requires power, it check with the pricing of
of smart grid. The main set of behaviors for grid agent
the grid at that hour and the diesel power price and
is receiving request from CA. This behavior will wait for
chooses the least priced one.
messages from CA to determine whether the microgrid needs
(4) Every hour based on the load requirement and avail- to buy power or sell net surplus power to the grid during
ability of solar power the agent makes the best possi- grid connected mode. Then, it gives the power or accepts the
ble decision for economic operations in a distributed power according to control agent instructions.
environment.
(5) Similar steps are followed for the load agent in the 5.4. Diesel Agent. Diesel agent registers with DF by giving
hostel. All the communication is done through ACL. diesel agent and diesel status as input to setName and setType,
The Scientific World Journal 7

Start

T
SP > LR

T F T
Other loads Excess power in
other solar
requires
Get from
F other solar
Supply other Get from batteries
loads
Not fully
F drained
SP still Check SOC of
excess
batteries
Fully drained
T
NCL shedding
Full
Check SOC of T
the batteries
LR < SP
Grid
Not full F
F
Charge batteries
GP > DP

Diesel (till max) Grid

Grid

SP Solar Power NCL Non Critical Load

LP Load Power Requirement GP and Grid Unit Price
SOC State of Charge of Battery DP Diesel Unit Price

Figure 6: Flowchart.

respectively. It gives the price of per unit power. It receives with control agent and inform the requirements. For example,
request from the CA and sells power to load agent when the department solar agent gives power to the department load
diesel price is less than the grid price. requirement if it is having enough power. If there is a shortage
of power, it informs the control agent about the power
5.5. Control Agent. Control agent (CA) is responsible for requirements. The control agent gets the status of all the
monitoring, controlling, and negotiating power levels and agents from Directory Facilitator (DF). where all the agents
performing power exchange between the solar microgrid and are registered with their services. It validates the request made
main grid. Initially CA registers as control and microgrid by other agents and informs other agents about the best place
control management as inputs to setName and setType, to get the requirement for optimal energy management in the
respectively. The control agent (CA) displays the total micro- solar microgrid.
grid power generation and loading as well as computing The agents relationship diagram is shown in Figure 7.
the net microgrid power every hour. Then, the agent will Here, the load agents participate in the system as buyers of
determine whether to buy or sell power to the grid based on energy, while the solar generator agent participate as sellers
the value of net power. If the grid power unit price is greater of energy. Here the hierarchical approach, which combines
than diesel generator (DG) unit price then it prefers to buy centralized and decentralized approach, is used. Every agent
from DG. autonomously makes decision then through control agent
it communicates to other agents for strategic decision. We
5.6. Agent System and the Agent’s Relationship. Agents sense consider solar department agent (SDA), load department
the environment and make simple decisions which are within agent (LDA), battery department agent (BDA), solar hostel
their capacity. For complex decision, agents communicate agent (SHA), load hostel agent (LHA), battery hostel agent
8 The Scientific World Journal

SHA
load in the department and hostel is 500 kW and 200 kW,
SDA
respectively. The sequence of operations is given below.
(1) Department Load needs 500 kW but taps 200 kW
from the available department solar power as depart-
LDA Control agent (CA) LHA ment solar power available is only 200 kW. Still it
requires 300 kW.
(2) Department load checks with hostel solar power. But
in hostel no excess power is available as the load
BDA BHA requirement is equal to the available solar power
GA DA of 200 kW. Department load then checks with the
battery of department and hostel but no power is
LHA Load Hostel Agent SDA Solar Department Agent available in the battery.
GA Grid Agent LDA Load Department Agent (3) Department load then goes for noncritical load shed-
DA Diesel Agent BDA Battery Department Agent
ding before going to grid or diesel for demand side
BHA Battery Hostel Agent SHA Solar Hostel Agent
management. Noncritical load in that particular hour
Figure 7: Agents relationship diagram. is 200 kW. So it sheds 200 kW and still it requires
100 kW.
(4) Finally, it checks with the unit price of grid and diesel.
(BHA), grid agent (GA), diesel agent (DA), and control agent In the dynamic pricing of grid the unit price that hour
(CA). Every hour based on the net power availability and the is Indian Rupees 8/kWh and the diesel unit price is
load requirement, the transaction with the grid is made by the Indian Rupees 10/kWh. It chooses the least priced
CA. When LDA request power from SDA, SDA gives power one, which is grid, and so it gets the remaining power
to LDA autonomously. If surplus power is available, it is given of 100 kW from grid.
to BDA. Further excess power is given to BHA and finally to
the grid agent (GA). The validation is done through control The console outputs and sinffer agent diagram are shown
agent. If there is no enough power available in SDA then LDA in the following.
contacts SHA and receives the available power. If the power
is still required it contacts the BDA and then BHA and finally Output
it communicates with control agent to do the NCL shedding
Department
and finally the post-NCL shedding power is received from
grid or diesel agent based on the unit price at that point of Power: 200 kW
time.
Similarly, the LHA contacts SHA for power. SHA makes Load: 500 kW
the autonomous decision to give the power to LHA. If surplus Power tapped from local agent: 200 kW
power is available after supplying to LHA, SHA checks the Power tapped from the other agent: 0 kW
BHA and BDA for charging and further excess power is given
to grid. If SHA does not have enough power required by Power tapped by battery: 0 kW
LHA, LHA contacts SDA and gets the power available there. If Battery charge: 0.0%
power is required still, LHA contacts BHA and BDA to get the
Power remaining: 0 kW
power available in the battery. The further power required is
managed by contacting the control agent which does the NCL Power needed: 300 kW
shedding and finally the post NCL shedding power is received Noncritical load shed: 200 kW
from grid or diesel agent after comparing the unit prices.
Power needed after load shedding: 100 kW
5.7. Simulation Results and Case Study. One round of oper- Preferred nonrenewable power source: grid
ation of solar microgrid was simulated for 24 hours, con- Power tapped from grid: 100 kW grid price/kWh =
sidering all the possible scenarios happen in a day. All the Rs.8
operations are considered as shown in the flow chart and
Total power tapped from grid: 100 kW
for these scenarios, the Java programming is done for all the
eight agents of the solar microgrid in JADE framework and
executed in Eclipse Integrated Development Environment. Hostel
Various scenarios are considered and sniffer diagrams and the
console output representing the interaction of the agents and Power: 100 kW
transaction details are studied. Load: 500 kW
The solar power and the load for the department and
hostel at 8 am are considered for the case study. The solar Power tapped from local agent: 100 kW
power of department and hostel is 200 kW each and the Power tapped from other agent: 0 kW
The Scientific World Journal 9

Power tapped by battery: 0 kW Microgrid parameters from Simulink

Battery charge: 0.0% Microgrid Agents

MAS model
Power remaining: 0 kW
MACSimJX
Power needed: 400 kW TCP/IP
Layer
Noncritical load shed: 100 kW MATLAB JADE
Simulink platform
Power needed after load shedding: 300 kW
Preferred nonrenewable power source: grid Control signals from JADE
Power tapped from grid: 300 kW grid price/kWh =
Figure 8: Multiagent system with MATLAB.
Rs.8
Hostel load tapped 100 kW from hostel solar
Department load tapped 200 kW from department several processes run concurrently inside Simulink. To over-
solar come this problem, MACSimJX, which acts a middleware
Hostel load tapped 0 kW from department solar between Simulink models and the agents, is used to bring
MAS closer to the physical models as shown in Figure 8.
Department load tapped 0 kW from hostel solar MACSimJX has a client-server architecture, separating the
Hostel load tapped 0 kW from hostel battery MAS from Simulink. Client is in the Simulink and the server
is at the agent environment.
Department load tapped 0 kW from department bat-
Named pipes in windows are used for communication
tery
[29]. Two pipes are used, one for passing configuration infor-
Hostel load tapped 0 kW from department battery mation and the other for passing simulation information. The
Department load tapped 0 kW from hostel battery MAS simulation framework has the following characteristics:
Hostel load tapped 300 kW from grid (i) A fixed-rate, discrete time, simulation with no explicit
Department load tapped 100 kW from grid internal state.
(ii) True multithreading in parallel with Simulink’s simu-
After trading and negotiation are completed, the final lation cycle.
result is being reported by control agent, which tells users how
much power is being traded with the grid. (iii) The capability to synchronize with Simulink opera-
tions.
6. Practical Realization of MAS in Initially the environment variables of solar microgrid are
JADE for Dynamic Energy Management of sensed by the sensors installed in all the physical devices of
Solar Microgrid Using MACSimJX the solar microgrid. The complete model of MACSimJX is
shown in Figure 9, which shows how the signals are given
6.1. Multiagent Simulation Using MACSimJX. A microgrid from MACSim client to the MACSim server. Once simulation
using a MATLAB/Simulink structure can effectively achieve signals arrive at the MACSim server, they are passed on to
energy management and demand side management through the agent model, which is divided into two parts, agent envi-
JADE, allowing each component in the microgrid to com- ronment and agent task force (ATF). The agent environment
municate and obtain the status of the entire network at any provides essential facilities such as coordination and messag-
time. The component status of MATLAB/Simulink can be ing. The agent environment has an agent coordinator and an
sent to the JADE agent system. If the breaker needs to switch agent server. The agent coordinator posts messages to ATF
after judgment, the JADE agent system will send a signal which assigns the work to various agents, requesting them
to the corresponding breaker in MATLAB/Simulink. When to carry out any operations necessary to prepare outputs for
unexpected situations occur in the microgrid, JADE can the specified time step. The ATF consists of all the agents that
immediately respond and manage the situation; for example, jointly operate on the data arriving from Simulink in order to
it can rapidly disconnect the power source in island mode and accomplish the goal of optimizing the energy management of
integrate distributed energy into the grid to provide power solar microgrid. The decentralized approach of MAS with the
supply. Once power has been restored to the source, it can be runtime efficiency of JADE improves the performance and
reconnected with the system immediately to provide power. reduces the computational complexity. Then, the computed
The use of multiagent systems in microgrids is extremely outputs are passed back to the agent environment. Once the
flexible because the monitoring and management can be agent environment has the completion messages from all
adjusted to microgrids of various structures and demands. agents, the output values are passed to the pipe server to
The Simulink model is developed for the solar microgrid with be returned to Simulink. The messaging, therefore, provides
solar power, load, and battery. In Simulink, the S-functions synchronization with Simulink’s simulation cycle so that the
are unable to handle multiple threads of execution, which is messages can be sent as command signals to the actuators for
an essential characteristic of MAS: they become unstable if switching operation of microgrid devices.
10 The Scientific World Journal

value. The lesser the available power, the longer it takes for
the battery to fully charge, and vice versa. Similarly, if more
power is drawn, it discharges in short time and if less power
is drawn it discharges for long time. Like this model for solar
Simulink model Agent environment Agent task force power, load, grid, and diesel are developed. The abbreviations
used are shown in Table 1. The input port and output ports are
defined in simulation model of solar microgrid. Eight input
ports and 18 output ports are considered. These 18 output
ports are connected to the actuators for physical action.
MACSim communica- JADE communication All the components are considered for optimal energy
tion channel channel
management, such as solar power, load power, and the state
Figure 9: A complete model of MACSimJX. of charge of battery. Based on all possible environmental
parameter values of solar microgrid components, a unique
number is formed, which is broadcasted to all the agents.
By observing it, each agent decides the subset of switching
This paper aims at dynamic energy management by
operations it is expected to do since it is already supplied
distributed automation of solar microgrid in a distributed
with the consolidated state numbers while programming.
environment. The important components of solar microgrid
It observed the state number and if it is present in its
are programmed in agents so the component status of
consolidated list of state numbers it will do the switching
MATLAB/Simulink can be sent to the JADE agent system.
operations of the corresponding switches; else, it will not
If the breaker needs to switch after judgment, received after
respond.
operation of agents, the JADE agent system will send a
signal to the corresponding breaker in MATLAB/Simulink,
which in turn is given to the physical breaker for real- 7. Case Study
time action. The approach considers a two-layer framework
7.1. Case Study with Critical Load. The Simulink modeling is
comprising agent layer containing the agent based control
implemented for solar power unit, load, and batteries of the
and functional layer, where the physical infrastructure is
department and hostel and grid and diesel power. Here BD
simulated in MATLAB/Simulink. The behavior parts of the
and BH refer to department and hostel battery SOC. SDP,
agents (i.e., decisions and control strategies) and coordinating
LDP, SHP, and LHP refer to solar power and load values
functionalities are done in the agent layer aiming to optimize
of the department and hostel. DPR and GPR refer to the
microgrid operation. The functional layer includes the model
diesel price and grid price in INRS (Indian Rupees). The
of the real solar microgrid comprising solar units load, bat-
input values taken for the case study are shown in Table 2.
teries, diesel unit, and grid. MACSimJX acts as a middleware
All the agents programs are executed in JADE and then
linking these two. The functional layer gets the values of
MACSimJX is executed to link the agent environment with
the environmental variables and passes them to the agent
Simulink model. Then, the solar microgrid Simulink model
layer. The agent layer observes values and coordinates with
is made to run. The environment values are given to the
required agents using the decentralized approach of MAS
JADE agents through the Simulink and after the coordinated
operating in JADE and takes necessary actions for optimizing
operations of agents, the results are given as command signals
the solar microgrid under dynamic environment. The agent
to the Simulink for the switching operations. The agent senses
layer needs to interact with the function layer to have a
the environment through Simulink and takes the strategic
mean to test the control strategies. The agent layer consists
action for optimal energy management of the microgrid in
of department load (LDA) and solar (SDA) and battery
the dynamic environment. Every hour the environment is
agents (BDA) and hostel load (LHA) and solar (SHA) and
changing due to intermittent nature of solar power and the
battery agents (BHA) apart from grid (GA) and diesel agents
variation of load. The agent takes the best possible action
(DA). The control agent (CA) monitors and controls all the
every hour.
agents. The agent layer uses MATLAB/Simulink as gateways
Every hour based on the solar power and load values the
to translate semantics from agent world to services world,
optimal distributed energy management is implemented by
where the commands can be physically executed. The quality
the strategic action of agents. By using multiagent system
of service and the dynamic energy management are done by
implemented in JADE, the runtime efficiency of the solar
proper control and management strategies, which accommo-
microgrid is improved. Thus, dynamic energy management
date different heterogeneous entities and also remain secure,
and demand side management are done for all the possible
sustainable, and reliable. The MATLAB/Simulink model of
scenarios in this case study. The individual switching action
the solar microgrid is shown in Figure 10.
of all the agents is analyzed. For example, the switching action
of the solar hostel agent (SHA) in critical load case study
6.2. Simulink Model of Solar Microgrid. The Simulink model is shown in Figure 11. Here the switch SH-LH is always on
is developed for SD, LD, BD, SH, LH, BH, GRD, and DSL as the local load LHP is always supplied by SHP. Initially
individually. In the battery model, three types of state of from 0th to 1st hour, the department and hostel solar powers
charge (SOC) are considered. SOC is full, SOC is between (SDP and SHP) are 200 kW more than the corresponding
full (100%) and cut-off value (40%), and SOC is below cut-off load demand and so the department and hostel batteries
The Scientific World Journal 11

Solar_Dept Load_Dept DIESEL

[SD] [LD] [DSL]

Battery_Dept
[BD]

Controller

Battery_Hostel Scope
[BH]

Load_Hostel
[LH]

Solar_Hostel
[SH] GRID
[GRD]

Figure 10: Simulink model of solar microgrid.

are charged. The solar hostel agent charges its battery BH department battery by switching on SD-BD till it gets fully
by switching on SH-BH and after BH gets fully charged charged and then the power is given to grid by switching on
the power is given to grid by switching SH-GRD. From SD-GRD. It is again switched on from 6th hour till it gets fully
5th to 6th hours, department load requires 300 kW but the charged and then the power is given to grid by switching on
available solar power is 200 kW. So the 100 kW deficiency in SD-GRD. In the 9th hour, it is switched on again to charge as
the department is managed by 200 kW excess in the hostel there is excess power in the department solar unit and then
unit. The hostel unit gives 100 kW to the department load it is given to grid through SD-GRD. Simulation output of all
and the remaining 100 kW is used to charge the hostel battery the components of the solar microgrid is observed and the
first and then the department battery. Here SHA switches actions are verified. The Simulink model output can be given
on its local load (LHP) and also switches on SH-LD to give for real-time physical action through Intelligent Electronic
100 kW deficiency in the department load. The remaining Devices (IED) and Program Logic Controllers (PLC). Thus,
100 kW is used to charge its battery (BH) by switching on SH- microgrid quickly adapt to environment dynamics improving
BH. After hostel battery gets fully charged, the switch SH- stability and reliability. Multiagent system is exploited for
BH is switched off and SH-BD is switched on to charge the microgrid automation through cyber physical system and
department battery (BD). Similarly, the switching actions of scaled up using cloud computing concept for larger benefits.
all the agents are observed. The switching operation of the Advanced SCADA uses MAS to meet the challenges due to
department solar agent is shown in Figure 12. Here SD-LD increased penetration of renewable energy resources.
switch is always on as department load is always connected to
the department solar power. From 2nd to 3rd hours, there is 8. Conclusion
deficiency of 100 kW in the hostel solar in supplying to hostel
load. So it receives 100 kW from department solar through MAS is implemented in JADE framework for optimal energy
SD-LH switch. In the 6th hour, it is again switched on to management and effective demand side management of a
manage the deficiency in hostel solar unit. In the first hour, solar microgrid. Solar microgrid model is developed in
the excess power in the department solar is used to charge MATLAB/Simulink platform and it is linked with JADE using
12 The Scientific World Journal

Table 1: Device description.

Number Abbreviations Description

1 SDP and LDP Department solar power and load
2 SHP and LHP Hostel solar power and load
3 BD and BH Department battery and hostel battery
4 GRD and DSL Grid and diesel
5 GPR and DPR Grid and diesel unit price
6 CL and NCL Critical load and noncritical load
7 LCDP and LCHP Critical load in department and hostel
8 LNCDP and LNCHP Noncritical load in department and hostel
9 CA Control agent
10 SDA and SHA Department and hostel solar agents
11 LDA and LHA Department and hostel load agents
12 BDA and BHA Department and hostel battery agent
13 GA and DA Grid agent and diesel agent
14 (DSL-LD-CA) LD and DSL switch controlled by CA
15 (DSL-LH-CA) LH and DSL switch controlled by CA
16 (GRD-LD-CA) LD and GRD switch controlled by CA
17 (GRD-LH-CA) LH and GRD Switch controlled by CA
18 SD-BD Switch connecting solar and battery in department
19 SD-BH Switch connecting SDP and battery in hostel
20 SD-GRD Switch connecting solar department and grid
21 BD-LD Switch connecting battery and load in department
22 BD-LH Switch connecting BD and load in hostel
23 SH-BH Switch connecting solar and battery in hostel
24 SH-BD Switch connecting solar hostel and BD
25 SH-GRD Switch connecting solar hostel and Grid
26 SH-LD Switch connecting solar hostel and LD
27 SD-LH Switch connecting SD and load in hostel
28 BH-LH Switch connecting BH and load in hostel
29 BH-LD Switch connecting BH and load in department
30 DSL-LD Switch connecting diesel unit and LD
31 DSL-LH Switch connecting diesel unit and LH
32 GRD-LD Switch connecting grid and LD
33 GRD-LH Switch connecting grid and LH

Table 2: Case study values (kW). management and fair allocation of resources for demand
side management of solar microgrid. In the microgrid,
Time SDP LDP SHP LHP DPR GPR the update rate of the unit dispatch command should be
0 400 200 300 100 10 8 sufficiently fast to follow the unexpected changes of load
1 200 400 300 100 10 8 and nondispatchable generators. So the inherent features
2 400 200 100 200 8 10 of MAS increase switching speed and reduce the network
3 200 400 100 300 8 10 load leading to achievement of lowest possible cost of power
4 200 400 100 300 10 8 generation under intermittent nature of solar power and
5 200 300 300 100 10 8 randomness of load. JADE augments the MAS with its
6 400 200 100 200 8 10 features and further improves the operational efficiency.
7 200 300 100 200 8 10
The proposed framework gives the intelligent consumer the
ability to explore all possible logical sequences of options,
8 200 300 100 200 8 10
understand the stochastic environment, and implement the
9 200 400 100 300 10 8 optimal energy management actions to increase operational
efficiency autonomously in a distributed environment, bring-
ing MAS closure to real-time implementation. Future work
MACSimJX for realizing the practical implementation of will focus on extension to multiple agents integrating diverse
MAS. A novel control strategy is used for dynamic energy renewable generators (solar and wind) with several intelligent
The Scientific World Journal 13

consumers with conflicting requirements. Furthermore, the

SH-LH

1 proposed approach can be implemented in actual microgrids

0 by sending and receiving the environment variables values
0 1 2 3 4 5 6 7 8 9 10 11 using Internet Of Things (IOT) technology.
Time (hr)

Competing Interests
SH-LD

1
0 The authors declare that they have no competing interests.
0 1 2 3 4 5 6 7 8 9 10 11
Time (hr)
References
SH-BH

1
[1] A.-R. Sadeghi, C. Wachsmann, and M. Waidner, “Security
0
0 1 2 3 4 5 6 7 8 9 10 11 and privacy challenges in industrial internet of things,” in
Time (hr) Proceedings of the 52nd ACM/EDAC/IEEE Design Automation
Conference (DAC ’15), pp. 1–6, IEEE, San Francisco, Calif, USA,
June 2015.
SH-BD

1
[2] M. E. El-Hawary, “The smart grid—state-of-the-art and future
0 trends,” Electric Power Components and Systems, vol. 42, no. 3-4,
0 1 2 3 4 5 6 7 8 9 10 11
pp. 239–250, 2014.
Time (hr)
[3] N. D. Hatziargyriou, “Special issue on microgrids and energy
management,” European Transactions on Electrical Power, vol.
SH-GRD

1 21, no. 2, pp. 1139–1141, 2011.

[4] C. Chen, S. Duan, T. Cai, B. Liu, and G. Hu, “Smart energy man-
0
0 1 2 3 4 5 6 7 8 9 10 11 agement system for optimal microgrid economic operation,”
Time (hr) IET Renewable Power Generation, vol. 5, no. 3, pp. 258–267, 2011.
[5] G. Kyriakarakos, A. I. Dounis, K. G. Arvanitis, and G.
Figure 11: Switching operations of solar hostel agent (SHA). Papadakis, “A fuzzy logic energy management system for
polygeneration microgrids,” Renewable Energy, vol. 41, pp. 315–
327, 2012.
[6] F. A. Mohamed and H. N. Koivo, “System modelling and online
optimal management of microgrid with battery storage,” in Pro-
ceedings of the International Conference on Renewable Energies
and Power Quality (ICREPQ ’07), EA4EPQ, Universidad de
SD-LD

1 Sevilla, Sevilla, Spain, March 2007.

[7] C. M. Colson, M. H. Nehrir, and S. A. Pourmousavi, “Towards
0 real-time microgrid power management using computational
0 1 2 3 4 5 6 7 8 9 10 11
intelligence methods,” in Proceedings of the IEEE Power and
Time (hr)
Energy Society General Meeting (PES ’10), pp. 1–8, IEEE, Min-
neapolis, Minn, USA, July 2010.
SD-LH

1 [8] Z. Jun, L. Junfeng, W. Jie, and H. W. Ngan, “A multi-agent

solution to energy management in hybrid renewable energy
0
0 1 2 3 4 5 6 7 8 9 10 11 generation system,” Renewable Energy, vol. 36, no. 5, pp. 1352–
Time (hr) 1363, 2011.
[9] Z. Jiang, “Agent-based control framework for distributed energy
resources microgrids,” in Proceedings of the IEEE/WIC/ACM
SD-BD

1 International Conference on Intelligent Agent Technology (IAT

’06), pp. 646–652, Hong Kong, December 2006.
0
0 1 2 3 4 5 6 7 8 9 10 11 [10] A. Dimeas and N. Hatziargyriou, “A multi-agent system for
Time (hr) microgrids,” in Proceedings of the IEEE Power Engineering
1 Society General Meeting, vol. 1, pp. 55–58, June 2004.
SD-BH

0 [11] F. Y. S. Eddy and H. B. Gooi, “Multi-agent system for opti-

−1 mization of microgrids,” in Proceedings of the 8th International
0 1 2 3 4 5 6 7 8 9 10 11 Conference on Power Electronics-ECCE Asia, pp. 2374–2381,
Time (hr) Institute of Electrical & Electronics Engineers (IEEE), Jeju,
Republic of Korea, May 2011.
1
SD-GRD

[12] A. L. Dimeas and N. D. Hatziargyriou, “Operation of a multia-

0 gent system for microgrid control,” IEEE Transactions on Power
0 1 2 3 4 5 6 7 8 9 10 11 Systems, vol. 20, no. 3, pp. 1447–1455, 2005.
Time (hr) [13] M. Pipattanasomporn, H. Feroze, and S. Rahman, “Multi-agent
systems in a distributed smart grid: design and implementa-
Figure 12: Switching operations of solar department agent (SDA). tion,” in Proceedings of the IEEE/PES Power Systems Conference
14 The Scientific World Journal

and Exposition (PSCE ’09), pp. 1–8, IEEE, Seattle, Wash, USA,
March 2009.
[14] T. Logenthiran, D. Srinivasan, A. M. Khambadkone, and H.
N. Aung, “Multiagent system for real-time operation of a
microgrid in real-time digital simulator,” IEEE Transactions on
Smart Grid, vol. 3, no. 2, pp. 925–933, 2012.
[15] H. S. V. S. Kumar Nunna and S. Doolla, “Multiagent-based
distributed-energy-resource management for intelligent micro-
grids,” IEEE Transactions on Industrial Electronics, vol. 60, no. 4,
pp. 1678–1687, 2013.
[16] C. M. Colson and M. H. Nehrir, “Comprehensive real-time
microgrid power management and control with distributed
agents,” IEEE Transactions on Smart Grid, vol. 4, no. 1, pp. 617–
627, 2013.
[17] D. E. Olivares, A. Mehrizi-Sani, A. H. Etemadi et al., “Trends in
microgrid control,” IEEE Transactions on Smart Grid, vol. 5, no.
4, pp. 1905–1919, 2014.
[18] J. J. Gomez-Sanz, S. Garcia-Rodriguez, N. Cuartero-Soler, and
L. Hernandez-Callejo, “Reviewing microgrids from a multi-
agent systems perspective,” Energies, vol. 7, no. 5, pp. 3355–3382,
2014.
[19] W. Zhang, Y. Xu, W. Liu, C. Zang, and H. Yu, “Distributed
online optimal energy management for smart grids,” IEEE
Transactions on Industrial Informatics, vol. 11, no. 3, pp. 717–727,
2015.
[20] W. Li, T. Logenthiran, and W. Woo, “Intelligent multiagentsys-
tem for smart home energy management,” in Proceedings of
the IEEE Innovative Smart Grid Technologies—Asia (ISGT ASIA
’15), pp. 1–6, November 2015.
[21] P. Vrba, V. Marik, P. Siano et al., “A review of agent and service-
oriented concepts applied to intelligent energy systems,” IEEE
Transactions on Industrial Informatics, vol. 10, no. 3, pp. 1890–
1903, 2014.
[22] T. Logenthiran, R. Naayagi, W. L. Woo, V.-T. Phan, and K. Abidi,
“Intelligent control system for microgrids using multi-agent
system,” IEEE Journal of Emerging and Selected Topics in Power
Electronics, vol. 3, no. 4, pp. 1036–1045, 2015.
[23] C. R. Robinson, P. Mendham, and T. Clarke, “Macsimjx: a tool
for enabling agent modelling with simulink using jade,” Journal
of Physical Agents, vol. 4, no. 3, pp. 1–7, 2010.
[24] A. Sharma, D. Srinivasan, and A. Trivedi, “A decentralized
multiagent system approach for service restoration using dg
islanding,” IEEE Transactions on Smart Grid, vol. 6, no. 6, pp.
2784–2793, 2015.
[25] A. Kantamneni, L. E. Brown, G. Parker, and W. W. Weaver, “Sur-
vey of multi-agent systems for microgrid control,” Engineering
Applications of Artificial Intelligence, vol. 45, pp. 192–203, 2015.
[26] Y. S. F. Eddy, H. B. Gooi, and S. X. Chen, “Multi-agent system
for distributed management of microgrids,” IEEE Transactions
on Power Systems, vol. 30, no. 1, pp. 24–34, 2015.
[27] E. Karatepe, M. Boztepe, and M. Colak, “Neural network based
solar cell model,” Energy Conversion and Management, vol. 47,
no. 9-10, pp. 1159–1178, 2006.
[28] F. Bellifemine, A. Poggi, and G. Rimassa, “Developing multi-
agent systems with JADE,” in Intelligent Agents VII Agent
Theories Architectures and Languages, pp. 89–103, Springer,
Berlin, Germany, 2001.
[29] P. Mendham, “Macsim: a simulink enabled environment for
multi-agent system simulation,” in Proceedings of the 16th IFAC
World Congress, P. Zı́tek, Ed., p. 1131, Elsevier, Prague, Czech
Republic, July 2005.
 Advances in Journal of
Industrial Engineering
Multimedia
Applied
Computational
Intelligence and Soft
Computing
The Scientific International Journal of
Distributed
Hindawi Publishing Corporation
World Journal
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Advances in

Fuzzy
Systems
Modelling &
Simulation
in Engineering
Hindawi Publishing Corporation
Hindawi Publishing Corporation Volume 2014 http://www.hindawi.com Volume 2014
http://www.hindawi.com

Submit your manuscripts at

Journal of
http://www.hindawi.com
Computer Networks
and Communications  Advances in 
Artificial
Intelligence
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014

International Journal of Advances in

Biomedical Imaging Artificial
Neural Systems

International Journal of
Advances in Computer Games Advances in
Computer Engineering Technology Software Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

International Journal of
Reconfigurable
Computing

Advances in Computational Journal of

Journal of Human-Computer Intelligence and Electrical and Computer
Robotics
Hindawi Publishing Corporation
Interaction
Hindawi Publishing Corporation
Neuroscience
Hindawi Publishing Corporation Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014