International Journal of Power Electronics and Drive System (IJPEDS)

Vol. 16, No. 2, June 2025, pp. 1314~1324

ISSN: 2088-8694, DOI: 10.11591/ijpeds.v16.i2.pp1314-1324  1314

Economic and ecological constraints of hybrid systems with

Bayesian networks

Amadou Fousseyni Toure1,2, Abdoulaye Sissoko1, Fadaba Danioko1, David Tchoffa2,

Abderrahman El Mhamedi2
1
Laboratoire, Centre de Calcul de Modélisation et de Simulation, Université des Sciences, Techniques et Technologies de Bamako,
Bamako, Mali
2
Equipe MGSI, Laboratoire Quartz, Université de Paris 8, Paris, France

Article Info ABSTRACT

Article history: To confront climate change and the shortage of fossil fuels, countries are
turning to renewable energies, particularly solar energy, which is an
Received Jan 7, 2025 abundant, inexhaustible energy with a low environmental impact. In Mali,
Revised Mar 30, 2025 hydraulic and thermal energy sources are insufficient, and the country has
Accepted May 6, 2025 been facing a huge energy crisis in recent years. For this reason, the
production of solar energy is a major solution. This work proposes a
photovoltaic system coupled to the national distribution network to reduce
Keywords: this energy crisis or even resolve it definitively. A preliminary analysis was
carried out to define the technical, economic, and ecological conditions for
Bayesian network the construction of a 30 MW solar power plant at three interconnection
Economic and ecologic points of the national distribution network of Mali. The objective of this
Grid management analysis is to determine the economic and ecological constraints of the
Hybrid system proposed hybrid system. For modeling, we used the Bayesian network, and
Modeling for simulation, the BayesienLab simulation tool was used. The simulation
results showed a considerable contribution of these three solar power plants
in terms of energy deficits, economic deficits, and climate change.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Amadou Fousseyni Toure
Laboratoire, Centre de Calcul de Modélisation et de Simulation
Université des Sciences, Techniques et Technologies de Bamako
Bamako, Mali
Email: [email protected]

1. INTRODUCTION
The demand for energy is evolving exponentially, whereas conventional fossils-based energy
sources are decreasing and climate change is becoming more and more evident. In this context, the use of
renewable energies, a solution to be considered, is particularly interesting for Africa in general and Mali in
particular. Since the 1980s, Mali, in cooperation with many development partners, has initiated various
development projects and programs to increase the use of renewable energy sources. This strategy combines
efforts to reduce poverty, validate national energy resources, and guarantee long-term security and
environmental sustainability of energy supply.
Given the rapid increase in fuel prices imported, renewable energy sources are of growing interest.
These renewable energy sources with less impact on the environment are abundant in Mali and contribute to
the country's development. Currently, electricity demand is increasing by approximately 10% per year, with
fuel transportation costs rising even further (African Bank for Development, 2010). This situation results in
enormous difficulties for the Malian government and national operators and seeks to reduce imports of fossil

Journal homepage: http://ijpeds.iaescore.com

Int J Pow Elec & Dri Syst ISSN: 2088-8694  1315

fuels, as well as for the public utility and private investors striving to provide enough electricity at a
reasonable price.
Much of the energy generation comes from large-scale hydro-electric power station produced on the
Senegal and Niger rivers, but small and medium diesel generators still provide about 20% of the total output.
Even if installations of electrical energy interconnections with Ghana and Ivory Coast are planned, political
and economic decisions are being put in place to exploit renewable energies such as wind, solar and
hydroelectric energy in the country. According to studies by The European Association of the Photovoltaic
Industry (EAPI, 2013) [1], the capacity of photovoltaic production was estimated at 100 GWp in 2012.
In this context, the implementation of hybrid photovoltaic power plants and the electricity network
is relevant. Besides, once these hybrid power plants are realized, the problem of managing solar energy
produced according to the availability of solar irradiation poses a significant problem. We will find in the
literature many studies on photovoltaic systems, among others, the analysis of the performance of a
171.36 kWp solar power plant connected to the network of the Island of Crete by Mungkin et al. [2]. Thirteen
photovoltaic systems coupled to electrical networks of different technologies were evaluated by Toure et al.
[3] in two locations in different countries (Cyprus and Germany).
Solar Energy Center in India conducted a performance evaluation study of different photovoltaic
technologies consisting of, thin films, monocrystalline silicon, and polycrystalline silicon [4]. Nouadje et al. [5]
evaluated the performance of a 300 KWc solar photovoltaic power plant connected to the network at DJIBOUTI
using the various methods. Sharma et al. [6] analyzed the reliability of a single-phase converter with reactive
power injection at night, considering mission profiles. A new approach to the study of reliability and reliability on
renewable energies was provided respectively by Ichim-Burlacu et al. [7] and Anurag et al. [8]. Ma et al. [9]
worked on the study of design and selection of reliable converters for hybrid photovoltaic power grid systems.
The flexibility of power control of photovoltaic systems has been discussed by Blaabjerg et al. [10]. In an urban
area of Mexico, Ferreira et al. [11] evaluated an autonomous photovoltaic system. A study on the configuration
of a multi-state converter was carried out by Blaabjerg et al. [12].
Most of this work has focused on evaluating the performance of solar systems, and the reliability of
converters and the maximum power point tracking for installed PVs. Our work will be focused on modeling
the availability of solar production, to help the dispatcher, predict a reliable estimate of purchases from
Manantali and the interconnection of Côte d'Ivoire suppliers using the Bayesian network to provide the
interconnected system to meet three essential requirements: stability, economy and above all continuity of
service. This is not always the case, because the network is often exposed to incidents that can interrupt this
service and cause significant financial losses for industry and inconvenience for ordinary consumers. Our
major contribution has been to combine analytical and probabilistic models with the Bayesian network in
order to have a real decision-making tool that will allow the dispatching service to make better purchasing
precision for energy and thermal productions. The objective is to reduce the overall costs of production and
the emission of greenhouse gases. For the simulation, we have considered three scenario cases. The first
scenario is to maintain thermal production as we have found to see the impact of solar production on the
economic and ecological level. The second scenario is to maintain the energies purchased by allowing solar
productions to reduce thermal productions with the same objectives that the first scenario. The third scenario
is to play simultaneously on thermal production and purchased energy to allow the dispatching service to
reduce economic and ecological impacts as much as possible. We have applied our approach on the
interconnection networks of Mali.
The rest of the paper is structured as follows: i) Section 2 introduces the Bayesian estimation
approach; ii) Section 3 presents a case study; iii) Section 4 discusses the simulation results for the best
prediction; and iv) Finally, Section 5 provides the discussion and conclusion.

2. PROPOSED METHOD
2.1. Introduction
In the literature, Bayesian networks have been applied in several studies, but most of these studies use
them for predictive maintenance. Perea-Moreno et al. [13] and Chen et al. [14] conducted a study on predictive
maintenance using dynamic probabilistic networks. Weber et al. [15] and Samé et al. [16] have studied
predictive maintenance of complex multi-state systems with a reliability structure. Kareem and Owolabi have
optimized maintenance planning in the industrial production sector with the Bayesian network [17].
Nevertheless, some studies have used Bayesian networks in the energy sector [18], [19]. We can
mention, among others, Sica et al. [20], who studied a cognitive system for the default prognosis in power
transformers, and Lee and Pan [21] have a cost-benefit analysis of offshore wind energy. However, the
Bayesian network can be used in several domains to make the right prediction and diagnosis [22]. In our
work, we used it in the photovoltaic system connected to the electricity grid, which constitutes our
contribution in this context, as shown in Figures 1(a) and 1(b). The PV generator node represents the
Economic and ecological constraints of hybrid systems with … (Amadou Fousseyni Toure)
1316  ISSN: 2088-8694

photovoltaic source, which has as child nodes control MPPT and converter DC/DC nodes. The MPPT control
node controls the DC/DC converter to maintain the PV generator at its maximum output. It is based on
information from the PV to drive the DC/DC converter.
The DC/AC inverter node, whose parent node is the DC/DC converter node, converts the direct
current produced by the PV generator into alternating current of the same nature as the grid. The node's
thermal production, hydroelectric production, and interconnection are different sources of production that
constitute the children of the grid node. The grid node is the sum of its child nodes that have a single point of
production. The station distribution node has for parents DC/AC and grid inverter nodes, which interconnect
the two sources and distribute them to consumers.
In our approach, we have combined analytical and probabilistic models with the Bayesian network
in order to have a real decision-making tool that will allow the dispatching service to make better precise
purchase of energy and thermal production. The objective is to reduce the overall costs of production and the
emission of greenhouse gases.

2.2. Global approach

Based on the identification of the powers of the PV modules, a photovoltaic system is designed with
a maximum power point tracking controller for PV modules. Then, we create a hybrid system with
production sources including: hydroelectric power stations, imported energy, and thermal power stations.
Once the hybrid system photovoltaic/power grid, we model and optimize this system in three steps [23].
The overall approach, which summarizes my approach, is as follows: this approach was developed
in three phases (Figure 2). Step 1 involves designing a controller for maximum power point tracking (MPPT)
of the PV system and integrating it with the electric grid. In step 2, the impact of productions photovoltaics
on different productions has been developed, and we have analyzed the cost of the investigation supply
service and CO2 emissions.

(a) (b)

Figure 1. Modeling a PV system: (a) connected to the electricity grid and (b) global interconnection grid

Figure 2. Global approach for our approach

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1314-1324
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1317

2.3. Modelling of the hybrid system

Moreover, once these hybrid plants are installed, the problem of managing the variability of solar
energy produced according to the availability of the sun poses a major problem. The Bayesian network can
combine analytical and probabilistic models in order to have a real decision-support tool that allows the
dispatching service to make better purchasing accuracy for energy and thermal production [24], [25]. The
different production sources, distribution stations and loads will be modeled by Bayesian networks and then
we will develop a timed influence diagram where the distribution will be injected into each node of interest at
each moment.
From this Bayesian network, with the help of expert knowledge and the availability of a database,
we can act on thermal production, which is a source of pollution, depending on the availability of alternative
photovoltaic sources and the consumer demand. In addition, the dispatcher can be based on the hours of
production of PV sources to reduce daily, monthly, or even annual purchases near interconnected suppliers.
For the mathematical modeling, we will use the principle of energy conservation, which states that the sum of
the energies produced must be equal to the sum of the energies consumed plus the sum of the losses. For 𝑛
sources and consumption point we have (1).

∀ 𝑛𝜖ℕ ∑𝑛𝑖=1 𝑃𝑖 (𝑃𝑟𝑜𝑑𝑢𝑖𝑡𝑒𝑠) = ∑𝑛𝑖=1 𝑃𝑖 (𝐶𝑜𝑛𝑠𝑜𝑚𝑚é𝑒𝑠) + ∑𝑛𝑖=1 𝑃𝑖 (𝑃𝑒𝑟𝑡𝑒𝑠) (1)

Figure 1(b) consists of the model of our approach, which includes the essential elements of an
electrical interconnection network, namely: the sources of production, a post of distribution, and a point of
consumption (load). The operating principle is such that at post of distribution 1 (PD1), if thermal production
(ThP) and hydroelectric production (HP) are not sufficient to assure the electricity demands of the load 1
(L1), post of distribution 2 (PD2) will support it to compensate for the overload. At post of distribution 2, if
the electricity demands of the load 2 (L2) and the demand of the post of distribution 1 (PD1) are higher than
solar production (SP), then the interconnection network will come into support (IN). We have modeled this
operating principle using (2) and (3):
- At the post of distribution 1.

𝑖𝑓 𝑃𝐷1 − 𝑇ℎ𝑃 − 𝐻𝑃 ≥ 0 ⇒ 𝑃𝐷2 = 𝑃𝐷1 − 𝑇ℎ𝑃 − 𝐻𝑃

{ (2)
𝑖𝑓 𝑃𝐷1 − 𝑇ℎ𝑃 − 𝐻𝑃 < 0 ⇒ 𝑃𝐷2 = 0

- At the post of distribution 2.

𝑖𝑓 𝑃𝐷2 + 𝑃𝐷1 − 𝑆𝑃 ≥ 0 ⇒ 𝐼𝑁 = 𝑃𝐷2 + 𝑃𝐷1 − 𝑆𝑃

{ (3)
𝑖𝑓 𝑃𝐷2 + 𝑃𝐷1 − 𝑆𝑃 < 0 ⇒ 𝐼𝑁 = 0

For the construction of the Bayesian model, we use these mathematical models, (9) and (10) to define a
binary state variable of the post of distribution PD2(t) and the interconnected network IN(t) as (4).

∀ 𝑡: 𝑃(𝑃𝐷2(𝑡)) = 𝑃(𝑃𝐷1|𝑇ℎ𝑃, 𝐻𝑃) 𝑒𝑡 𝑃(𝐼𝑁(𝑡)) = 𝑃((𝑃𝐷2 + 𝑃𝐷1)|𝑆𝑃) (4)

In other words:

1 𝑖𝑓 𝑃𝐷1 − 𝑇ℎ𝑃 − 𝐻𝑃 ≥ 0
𝑃𝐷2(𝑡) = { (5)
0 𝑒𝑙𝑠𝑒
1 𝑖𝑓 𝑃𝐷2 + 𝑃𝐷1 − 𝑆𝑃 ≥ 0
𝐼𝑁(𝑡) = { (6)
0 𝑒𝑙𝑠𝑒
The probabilities for all intervals must be equal to 1. In the case of variables that have parent nodes,
such as Post Bamako, it uses the intervals defined for the thermal power station node and those described for
the load Bamako node. The user must be expert enough to determine appropriate discrete intervals for each
context variable so that all scenarios (a combination of parent intervals and variable) include representative
data for probability function calculations. The process of calculating the probability functions of the model is
called formation when BN defines the probability functions according to the values observed from the data.
By using probability values with possible reasoning, the BNs can deduce based on the evidence (observed
data). Indeed, once formed, the BNs can answer the following question: "What is the probability of the
presence of a target node (eg, purchase RCI), given the values observed for the variable context (for example,
thermal power station and solar).

Economic and ecological constraints of hybrid systems with … (Amadou Fousseyni Toure)
1318  ISSN: 2088-8694

3. MODELLING AND SIMULATION

We applied our approach to the Mali energy interconnection grid (EDM-sa), as shown in Figure 3,
which ensures the supply of electricity to 4 regions and their major cities, as well as the capital of Mali.
Hydroelectric power plants specific to the EDM, thermal power plants, and a large part of energy purchased
either from a neighboring country (Ivory Coast) or from the production of Manantali (a private sector) ensure
Mali's electricity demand. Three 30 MW solar power plants were connected to 3 electricity distribution points
(Kayes, Kita, and Fana). Thermal production is colored red, EDM-sa's own hydroelectric production sources
are colored blue, photovoltaic production is green, and purchase points are purple. Distribution stations and
consumption points are coded in yellow and black, respectively. From the analysis of these data, we noted that
the electricity supply service of Mali bought 54% against 46% from its own sources of production. Among
these 46%, thermal production is 34%, and 12% hydroelectric production, which is insignificant.

3.1. Environmental assessment

For our case study, the thermal power plants use fuel oil as a fuel source, so their emission rates are
730 g/kWh and on this basis, we can calculate the amount of CO2 emitted by the sources of thermal energy
production in Mali. This rate will be multiplied by the annual energy produced by thermal production, which is
589,838.94 MWh according to our database. According to calculations, the thermal power plants emitted
430,582,428 kg of CO2 in a year. The emission rate for hydroelectric is 6 g/kWh and summing all the
hydroelectric generation in our system, we will have 1,136,325 GWh/year [23]. According to calculations, the
emissions from hydroelectric production are 6,817,950 kg/year. The total CO2 emission is 437,400,378 kg/year.

3.2. Construction of the Bayesian model

The Bayesian network in Figure 4(a) is the typical model of our interconnection network, and
Figure 4(b) is the dynamic model of Mali’s grid. The transition to the Bayesian network is done so that at the
Bamako distribution post, if thermal production and its own hydroelectric power stations are not sufficient to
cover the loads of the city of Bamako, the post of the city of Fana will come to support to compensate for the
overload. At the Fana post, if the load of the city of Fana and the demand from the Bamako post are higher
than the available solar production of Fana, then the Segou post will support. Sikasso's post works the same
way as Fana's. If no PV and thermal production is available, all the charges will be supplied by the own
hydroelectric production and the Ivory Coast and Manantali interconnection networks. The assumption is that
the availability of thermal production and solar energy should reduce purchases at the levels of Manantali and
the Ivory Coast. Our model was modeled and simulated using BayesianLab networks software (Figure 4).
In our case study, the nodes of the solar productions, each having a production capacity of 30 MW,
which don’t have parental nodes, were discretized using ten continuous intervals equidistant according to
probabilistic relationships. Their probability distributions are made on the basis of solar production data from
the different sites. The nodes of the different loads are discretized on continuous intervals with a probabilistic
distribution. The nodes of the posts followed deterministic relations with conditional equations and were
discretized on equidistant intervals. The Manantali and Ivory Coast purchasing nodes, which can take on all
the charges in the case of unavailability of other sources, are discretized over equidistant intervals and follow
probabilistic relationships. The probabilities for all intervals must be equal to 1. For the work, we relied on
the one-year database collected at the level of the electricity distribution structure in Mali.

Figure 3. Mali's electricity supply interconnection network

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1314-1324
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1319

(a)

(b)

Figure 4. Bayesian model of the Mali interconnection network: (a) dynamic Bayesian model of the Mali
interconnection network and (b) under BayesianLab software

4. RESULTS AND DISCUSSION

For the simulation, we injected the 8760 data from each production system (equivalent to one year
of measurement) into our model. The production costs of the SELINGUE and SOTUBA production systems
have not been examined, since they are EDM's own hydroelectric productions and which have negligible
ecological impacts compared to thermal productions. In Figure 5(a), we can see a reduction in greenhouse
gas emissions, which are still significant, but these values are only the emissions from the sources of
production of the energy purchased. The total CO 2 emitted has been reduced from 458,503,525 kg/year to
455,358,714 kg/year. Here, we have made a simulation using PV systems without playing on any of the
productions (thermal and purchased energy). The objective is to see if it would be beneficial to distribute the
energies produced by PV systems between the energies purchased and thermal production.
Another important objective of this part is to get an idea of how many MW per day the supply
service must buy to stay in the previous scenarios. For the simulation, we fixed the value of energy purchases
and increased it gradually so as to balance the distribution of solar production. For the first scenario, the
service must buy around 136.135 MW per day. For the second scenario, the service must plan to buy 164.157
MW per day to remain in this scenario.
After having found the conditions of purchase for the two scenarios, we proceeded to a series of
simulations by increasing the powers to buy. These simulations have the objective of playing with the
operating cost. We have analyzed the forecast cases of 175 MW, 200 MW, 250 MW, and 300 MW in
Figures 4-6. Figure 4 illustrates the trend of different productions depending on the power demand. We note

Economic and ecological constraints of hybrid systems with … (Amadou Fousseyni Toure)
1320  ISSN: 2088-8694

that the more the forecast of energy purchases increases, the more the production of thermal energy
decreases. Thermal production becomes lower than solar production when we plan to buy 250 MW/day. For
a purchase forecast of 300 MW/day, thermal production is almost insignificant Figure 5(b).
Figure 6 summarizes the emission balance for carbon dioxide (CO 2), which is the most polluting
element in greenhouse gases. We observe that the CO2 emissions are proportional to the thermal production
and inversely proportional to the purchase of energy. For a purchase forecast of 136,135 MW/d, there was an
annual energy purchase of 803.36 GWh/year and thermal production of 589.84 GWh/year, resulting in an
emission of 458,500,000 kg/year (Figure 5). This emission decreased to 87,840,000 kg/year and 15,660,000
kg/year for a purchase forecast of 250 MW and 300 MW, corresponding respectively to an annual thermal
production of 103.86 GWh/year and 9.54 GWh/year (Figure 6).

(a)

(b)

Figure 5. Ecological and economic impact of the use of PV systems: (a) greenhouse gas emissions before and
after the use of PV systems; and (b) the trend of different productions according to forecasts

Figure 7 shows the costs of thermal production, the costs of energy purchased, and the total cost of
using these energy sources by the electricity supply service in Mali. It shows that the use of PV systems has
helped to reduce production costs. It can be observed that by increasing the forecast to buy more, the total
cost of use is considerably reduced. The forecast to buy 136.135 MW/day is the most expensive at around
136.55 billion FCFA for a cost of thermal production of 90.36 billion and a cost of energy to buy of 46.19
billion (Figure 7, curve cost_total). By increasing the forecast to buy, the cost of thermal production
decreases by reducing the total cost of use. This total cost will be reduced to 90.2 billion and 80.035 billion
FCFA for a forecast of 250 MW and 300 MW per day, respectively (Figure 6). Through the analysis of
Figures 5 and 6, we note that for the forecast to buy 300 MW/day, thermal productions to decrease from
103.86 GWh/year to 9.54 GWh/year (Figure 6) but the cost of thermal production has only decreased from
16 billion to 14.75 billion FCFA Figure 7, the ratio is really low. So, it would not be desirable to exceed a

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1314-1324
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1321

250 MW/day purchase forecast. We also notice that thermal production is expensive and not recommended
for supplier service.

Figure 6. The amount of carbon dioxide emitted for the different forecasts

Figure 7. The annual production costs of the different productions

5. CONCLUSION
The Mali electricity supply service called Mali energy (EDM-sa) includes its production systems,
constituting hydroelectric power stations and thermal power stations, production systems Manantali and the
interconnection network Ivory Coast, whose service buys energy. The total energy produced during 2018 was
1,726,164 GWh/year, of which 237,046 GWh/year was purchased from the neighboring country, Ivory
Coast, i.e. 14% of total production, and 697,818 GWh/year was purchased on the interconnected grid of
Manantali, 40% of the total production. The total energy purchased by Mali's electricity supply service was
934,864 GWh/year (75% for Manantali and 25% for Ivory Coast). Thermal production was 589,839
GWh/year or 34% of total production, and the hydroelectric production specific to EDM Selingue and
Sotuba was, respectively, 169.73 GWh/year and 31,731 GWh/year, i.e. 10% and 2% of total production. We
have used Bayesian Networks to model our case study. The goal was to allow the dispatching service to have
a decision support tool that would allow the dispatching service to make better purchasing precision for
Economic and ecological constraints of hybrid systems with … (Amadou Fousseyni Toure)
1322  ISSN: 2088-8694

energy and thermal production. For that, we have combined analytical and probabilistic models with the
Bayesian network.
A first simulation was made without the photovoltaic systems to make the economic and ecological
assessment of the year. The simulation of this scenario does not bring enough effect; the total cost of
production only reduced from 145.76 billion FCFA to 136.55 billion FCFA per year, a gain of 9.21 billion
FCFA, and CO2 emissions have been reduced from 458,503,525 kg/year to 455,358,714 kg/year. The second
scenario was to play simultaneously on thermal productions and purchased energies in order to reduce the
investigation cost of the electricity supply service in Mali using PV. In this part, we have increased the
forecast of the energies to buy and observed the economic and ecological effects. The results of the
simulation showed us that by increasing the purchase forecast, the total investigation cost decreases, as well
as the CO2 emissions. In conclusion, thermal production is the costliest economically and ecologically.

FUNDING INFORMATION
The authors would like to thank for the support from Higher Institute of Technologies (TechnoLAB-
ISTA) of Bamako, Mali.

AUTHOR CONTRIBUTIONS STATEMENT

This journal uses the Contributor Roles Taxonomy (CRediT) to recognize individual author
contributions, reduce authorship disputes, and facilitate collaboration.

Name of Author C M So Va Fo I R D O E Vi Su P Fu
Amadou Fousseyni Toure ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Abdoulaye Sissoko ✓ ✓ ✓ ✓ ✓ ✓ ✓
Fadaba Danioko ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
David Tchoffa ✓ ✓ ✓ ✓
Abderrahman El Mhamedi ✓ ✓ ✓ ✓ ✓ ✓

C : Conceptualization I : Investigation Vi : Visualization

M : Methodology R : Resources Su : Supervision
So : Software D : Data Curation P : Project administration
Va : Validation O : Writing - Original Draft Fu : Funding acquisition
Fo : Formal analysis E : Writing - Review & Editing

CONFLICT OF INTEREST STATEMENT

Authors state no conflict of interest.

DATA AVAILABILITY
Derived data supporting the findings of this study are available from the corresponding author,
[AFT], on request.

REFERENCES
[1] A. F. Touré, S. Addouche, F. Danioko, B. Diourté, and A. Mhamedi, “Hybrid systems optimization: Application to hybrid
systems photovoltaic connected to grid. A Mali case study,” Sustainability, vol. 11, no. 8, p. 2356, Apr. 2019, doi:
10.3390/su11082356.
[2] M. Mungkin, H. Satria, D. Maizana, M. Isa, S. Syafii, and M. Y. Puriza, “Analysis of the feasibility of adding a grid-connected
hybrid photovoltaic system to reduce electrical load,” International Journal of Power Electronics and Drive Systems (IJPEDS),
vol. 14, no. 2, p. 1160, Jun. 2023, doi: 10.11591/ijpeds.v14.i2.pp1160-1171.
[3] A. F. Toure, F. Danioko, B. Diourte, and N. Keita, “Study of decision support for the installation of photovoltaic systems coupled
to the electricity grid: Mali’s case study,” Journal of Emerging Science and Engineering, vol. 3, no. 1, p. e29, Dec. 2024, doi:
10.61435/jese.2025.e29.
[4] E. Kymakis, S. Kalykakis, and T. M. Papazoglou, “Performance analysis of a grid connected photovoltaic park on the island of
Crete,” Energy Conversion and Management, vol. 50, no. 3, pp. 433–438, Mar. 2009, doi: 10.1016/j.enconman.2008.12.009.
[5] B. A. M. Nouadje, P. Tiam Kapen, V. Chegnimonhan, and R. Tchinda, “Techno-economic analysis of a grid/fuel
cell/PV/electrolyzer system for hydrogen and electricity production in the countries of African and Malagasy council for higher
education (CAMES),” Energy Strategy Reviews, vol. 53, p. 101392, May 2024, doi: 10.1016/j.esr.2024.101392.
[6] V. Sharma, A. Kumar, O. S. Sastry, and S. S. Chandel, “Performance assessment of different solar photovoltaic technologies
under similar outdoor conditions,” Energy, vol. 58, pp. 511–518, Sep. 2013, doi: 10.1016/j.energy.2013.05.068.

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1314-1324
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1323

[7] C. Ichim-Burlacu, C.-L. Rat, C. Panoiu, and C. Cuntan, “PV panel monitoring using Arduino and LabVIEW,” in 2023 5th Global
Power, Energy and Communication Conference (GPECOM), IEEE, Jun. 2023, pp. 270–275. doi:
10.1109/GPECOM58364.2023.10175667.
[8] A. Anurag, Y. Yang, and F. Blaabjerg, “Reliability analysis of single-phase PV inverters with reactive power injection at night
considering mission profiles,” in 2015 IEEE Energy Conversion Congress and Exposition (ECCE), IEEE, Sep. 2015, pp. 2132–
2139. doi: 10.1109/ECCE.2015.7309961.
[9] K. Ma, H. Wang, and F. Blaabjerg, “New approaches to reliability assessment: Using physics-of-failure for prediction and design
in power electronics systems,” IEEE Power Electronics Magazine, vol. 3, no. 4, pp. 28–41, Dec. 2016, doi:
10.1109/MPEL.2016.2615277.
[10] F. Blaabjerg, D. Zhou, A. Sangwongwanich, and H. Wang, “Design for reliability in renewable energy systems,” in 2017
International Symposium on Power Electronics (Ee), IEEE, Oct. 2017, pp. 1–6. doi: 10.1109/PEE.2017.8171658.
[11] V. de N. Ferreira, A. F. Cupertino, H. A. Pereira, A. V. Rocha, S. I. Seleme, and B. de J. C. Filho, “Design and selection of high
reliability converters for mission critical industrial applications: A rolling Mill case study,” IEEE Transactions on Industry
Applications, vol. 54, no. 5, pp. 4938–4947, Sep. 2018, doi: 10.1109/TIA.2018.2829104.
[12] F. Blaabjerg, A. Sangwongwanich, and Y. Yang, “Flexible power control of photovoltaic systems,” in Advances in Renewable
Energies and Power Technologies, Elsevier, 2018, pp. 207–229. doi: 10.1016/B978-0-12-812959-3.00006-X.
[13] A.-J. Perea-Moreno, Q. Hernandez-Escobedo, J. Garrido, and J. Verdugo-Diaz, “Stand-alone photovoltaic system assessment in
warmer urban areas in Mexico,” Energies, vol. 11, no. 2, p. 284, Jan. 2018, doi: 10.3390/en11020284.
[14] W. Chen, E. Hotchkiss, and A. Bazzi, “Reconfiguration of NPC multilevel inverters to mitigate short circuit faults using back-to-
back switches,” CPSS Transactions on Power Electronics and Applications, vol. 3, no. 1, pp. 46–55, Mar. 2018, doi:
10.24295/CPSSTPEA.2018.00005.
[15] P. Weber, P. Munteanu, and L. Jouffe, “Dynamic Bayesian networks modelling the dependability of systems with degradations
and exogenous constraints,” IFAC Proceedings Volumes, vol. 37, no. 4, pp. 207–212, Apr. 2004, doi: 10.1016/S1474-
6670(17)36120-7.
[16] A. Samé, L. Bouillaut, P. Aknin, and A. Ben Salem, “Dynamic Bayesian networks with continuous exogenous variable for the
classification of singular points on a railroad track (in French: Réseaux bayésiens dynamiques à variable exogène continue pour la
classification des points singuliers d’une voie ferrée),” Revue d’intelligence artificielle, vol. 21, no. 3, pp. 353–370, Jun. 2007,
doi: 10.3166/ria.21.353-370.
[17] A. F. Toure, D. Tchoffa, A. El Mhamedi, B. Diourte, and M. Lamolle, “Modeling and control maximum power point tracking of
an autonomous photovoltaic system using artificial intelligence,” Energy and Power Engineering, vol. 13, no. 12, pp. 428–447,
2021, doi: 10.4236/epe.2021.1312030.
[18] V. Olabi and H. Jouhara, “An assessment of current hydrogen supply chains in the Gulf Cooperation Council (GCC),” Energy,
vol. 299, p. 131576, Jul. 2024, doi: 10.1016/j.energy.2024.131576.
[19] G. Yingkui and L. Jing, “Multi-State System Reliability: A New and Systematic Review,” Procedia Engineering, vol. 29, pp.
531–536, 2012, doi: 10.1016/j.proeng.2011.12.756.
[20] F. C. Sica, F. G. Guimarães, R. de O. Duarte, and A. J. R. Reis, “A cognitive system for fault prognosis in power transformers,”
Electric Power Systems Research, vol. 127, pp. 109–117, Oct. 2015, doi: 10.1016/j.epsr.2015.05.014.
[21] D. Lee and R. Pan, “Predictive maintenance of complex system with multi-level reliability structure,” International Journal of
Production Research, vol. 55, no. 16, pp. 4785–4801, Aug. 2017, doi: 10.1080/00207543.2017.1299947.
[22] L. Hichem, O. Amar, and M. Leila, “Optimized ANN-fuzzy MPPT controller for a stand-alone PV system under fast-changing
atmospheric conditions,” Bulletin of Electrical Engineering and Informatics, vol. 12, no. 4, pp. 1960–1981, Aug. 2023, doi:
10.11591/eei.v12i4.5099.
[23] A. F. Touré, “Decision support for the siting of grid-connected photovoltaic stations: a case study in Mali (in French: Aide à la
décision pour l’Implantation des Stations Photovoltaïques couplées au Réseau électrique : Cas d’application au Mali),” Université
de Paris8, 2020.
[24] F. Danioko, S.-A. Addouche, and A. El Mhamedi, “Formalization of an approach for improvement of maintenance policy on
multi-state systems,” International Journal of Engineering Research and Development, vol. 10, no. 8, pp. 36–44, 2014.
[25] N. Ghorbel, “Management of periodic replenishments of spare parts under uncertainty for the MCO of equipment fleets:
controlling the risk of stock shortages (in French: Gestion des réapprovisionnements périodiques de pièces de rechange sous
incertitudes pour le MCO de,” 2013.

BIOGRAPHIES OF AUTHORS

Amadou Fousseyni Toure is an Assistant Professor in the Electrical Engineering

Department and Industrial Informatics at the University of Sciences, Techniques and
Technologies of Bamako, Mali. He obtained his general university diploma in 2005 at the
Faculty of Sciences and Techniques of Bamako, his engineering diploma in design Industrial
Engineering option: Electricity at the Bamako Engineering School in 2008 and his Ph.D. in
Productivity-Industrial Engineering at the University of Paris8, France in joint supervision with
the University of Sciences, Techniques and Technologies of Bamako, Mali in 2020. He is
currently a teacher-researcher at the University of Sciences, Techniques and Technologies of
Bamako, Mali. His research interests include the field of digital design, industrial informatics,
power electronics, renewable energy, Arduino applications, embedded systems, and artificial
intelligence. He can be contacted at email: [email protected].

Economic and ecological constraints of hybrid systems with … (Amadou Fousseyni Toure)
1324  ISSN: 2088-8694

Abdoulaye Sissoko is an Assistant Professor in the Electrical Engineering

Department and Industrial Informatics at the University of Sciences, Techniques and
Technologies of Bamako, Mali. He obtained his Master's in Electronic Telecommunications
System (EST) at the University Cheick Anta DIOP of Dakar (UCAD), Dakar, Senegal, in
2013, and his Ph.D. in Electronics at the University of Nantes, Nantes, France, in 2022. He is
currently a teacher-researcher at the University of Sciences, Techniques and Technologies of
Bamako, Mali. His research interests include antennas, electronics, high and microwave
frequencies, and transmission systems. He is responsible for the electronic research team in the
Laboratory, Center for Modeling and Simulation Computing, at the University of Sciences,
Techniques and Technologies of Bamako. He can be contacted by email at
[email protected].

David Tchoffa has been qualified as a university full professor since March 2021.
He is a Teacher/Researcher in computer science and industrial engineering. He graduated Ph.D.
and a Habilitation Diploma to Supervise Research (HDR) in computer science at the University
Paris8 Vincennes Saint-Denis (France), while working as a production and quality engineer at
Renault industry and also consultant/researcher partner with EADS Innovation works within
the framework of several projects. He is currently responsible for the relationships between
QUARTZ laboratory of the Polytechnics Institute of Grand Paris (IPGP), IUT of Montreuil/
MGSI, and Industries. He was recently elected a member of the board of the Quartz laboratory
(EA 7393). He supervised several doctoral theses. His research works are related to systems
lifecycle management, interoperability of enterprise systems, dynamic manufacturing network
(IMAGINE), Product data exchange and modeling standards, risk management, blockchain,
smart city, AI, and supply chain. He can be contacted at email: [email protected].

Fadaba Danioko has been a Professor since August 2022 at the Department of
Physics of the Faculty of Science and Technology in Bamako, Mali. He has just joined the
artificial intelligence and robotics center of Mali in July 2024. He obtained his engineer
diploma in design Industrial Engineering option: mechanics at the National School of
Engineers of BAMAKO (E.N.I) - Republic of Mali in 2001, his Master in Mechatronics and
advanced design in mechanics, University of Rennes 1, Beaulieu-Rennes and ENS de Cachan,
Antenne de Bretagne, France in 2009 and his Ph.D. in Productivity-Industrial Engineering,
University of Paris8, France. He is currently a teacher-researcher at the University of Sciences,
Techniques and Technologies of Bamako, Mali. His research interests include the field of
mechanics, electronics, informatics, science, robotics, and artificial intelligence. He can be
contacted at email: [email protected].

Abderrahman El Mhamedi is a Professor at the University of Paris 8, France. He

obtained his Bachelor's and Master's degrees in computer science at the Joseph-Fourier
University of Grenoble, (June 1985, his DEA in Computer Science in Social Sciences at the
University of Social Sciences of Grenoble (July 1986) and Doctoral thesis in Automation -
Production from the National Polytechnic Institute of Grenoble (INPG), supported on April 25,
1990. Its research activities are focused on modeling methods and tools, and improving the
performance of systems for producing. His research interests on logistics modeling and
optimization, and enterprise systems engineering. He can be contacted at email:
[email protected].

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1314-1324