61

Computer Assistance Analysis of Power Grid Relay Protection

Based on Data Mining
Na Suo 1 and Zheng Zhou 2

1Department of Electrical Engineering, Zhengzhou Railway Vocational and Technical University,

Zhengzhou, Henan 450000, China, [email protected]
2Department of Electrical Engineering,Zhengzhou Railway Vocational and Technical University,

Zhengzhou, Henan 450000, China, [email protected]

Corresponding author: Na Suo, [email protected]

Abstract. Relay protection and fault diagnosis are one of the important guarantees
for the safe operation of power systems. With the widespread application of digital
technology in the power system, the power system has produced data that is
growing at an exponential rate. For this reason, this paper proposes a computer-
aided analysis system for relay protection based on data mining. Apply data mining
technology to the relay protection fault information processing system, and use
data mining technology to extract many potentially important factors, facts, and
correlations contained in a large amount of fault information. First, analyze the
characteristics of data mining technology and power grid protection, and design the
system structure of the power grid data intelligent analysis system. Secondly, the
relevance analysis of the data of grid relay protection adopts a class-based
frequent item set mining algorithm, and encapsulates the frequent item sets into
classes, which is convenient for the calculation of the minimum support and the
minimum confidence, weighted synthesis of exponential smoothing analysis model
and regression analysis model to obtain prediction algorithm model. Finally, the
historical data of a 110kv substation is analyzed and the proposed analysis
framework and algorithm are verified.

Keywords: computer assistance; grid relay protection; data mining

DOI: https://doi.org/10.14733/cadaps.2021.S4.61-71

1 INTRODUCTION
Through the relay protection intelligent analysis system based on the platform of big data, the
centralized remote intelligent diagnosis and fault analysis of the intelligent substation protection
system can be realized. Under the conditions of centralized protection of professional technical
resources and human resources, the protection system can be comprehensively improved.
Operation level, fault analysis level, intelligent diagnosis level, improves the reliability and safety
of smart grid operation [2]. How to make full and effective use of these valuable historical data,
adopt appropriate data analysis techniques, and conduct targeted analysis and mining on it,
obtaining valuable knowledge from it and providing decision support for the safe operation of the
power grid has become a problem to be solved. Data mining technology is the process of

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 62

extracting potential and valuable knowledge models or rules from massive data, that is, exploring
and analyzing large amounts of data according to predefined goals, revealing the hidden rules, and
further integrating Its modeled advanced and effective technical process [3]. With the continuous
development of mining technology, it has been widely used in various fields [4]. The application of
data mining technology to the intelligent analysis of the data of grid relay protection is one of its
important application fields.
In recent years, data mining specifically for the safe and stable operation of power systems
has developed rapidly. Among them, Lu et al. scholars have made outstanding contributions in this
regard. The ATDIDT (currently renamed PEPITO) power system safety and stability-mining
software developed by them has been put into commercial release [5]. Yue et al. proposed an
expert system for fault diagnosis and recovery processing in substations. It integrates two
knowledge expression modes: rule and process [6]. Gaber et al. proposed the fault diagnosis of
substation based on artificial neural network, which used the inherent fault tolerance of neural
network itself, but did not conduct special fault tolerance research, so the fault tolerance of the
network is very limited [7]. Sekar et al, in the power system substation, which requires accurate
and fast online fault identification, proposes a method combining cause-effect network and fuzzy
rule library, and uses the good parallel processing capability of cause-effect network to retrieve
faulty components [8]. Yang et al. use the basic regression analysis method to analyze the fault
and establish the prediction model. When other factors are relatively stable, the prediction
accuracy is relatively high [9]. The automation construction of the power system has been carried
out relatively early and the overall level is relatively high. For a typical large-scale power system,
with the increasing development of various computer monitoring equipment, geographic
information systems and management information systems, the data in the power system
database has exploded [10]. Regression analysis method is a mathematical method for studying
the dependent relationship between variables and variables. Given multiple sets of independent
variables and dependent variables, the regression equation is formed by studying the relationship
between the respective variables and dependent variables. At the same time, the application of
various automation systems and the continuous improvement of the level of information
technology make the data in the power system have the characteristics of a wide range of data
sources, a wide variety of data, and poor data quality.
This paper firstly summarizes the data mining technology from the concept of data mining,
mining mode, and related technologies used in mining, and propose the architecture of the
computer-aided analysis system for power grid data. Secondly, it analyzes the correlation analysis
and time series prediction of the data-mining model for intelligent analysis of power grid data. It
focuses on the concept of association analysis and typical frequent item set mining algorithms. On
this basis, a cluster-based frequent item set mining algorithm suitable for the computer-aided
analysis system of power grid data is proposed; the forecasting methods and technologies are
analyzed, and a comprehensive time series-forecasting model suitable for intelligent analysis
systems is proposed. Then, correlation and timing analysis using the collected test data 110kv
substation computer-aided analysis of the proposed system is the result of the prediction. Finally,
conclusions and prospects are given.

2 SYSTEM STRUCTURE AND TECHNICAL BACKGROUND OF POWER GRID RELAY

PROTECTION DATA ANALYSIS SYSTEM

2.1 Data Mining Technology

Data mining is essentially a new information processing technology. Data mining technology
improves people's application of data from online query operations to more advanced applications
such as decision support, analysis and forecasting. This knowledgeable information can be used to
guide advanced decision-making activities. From a narrow perspective, data mining can be defined
as a process of extracting knowledge from a specific form of data set.

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 63

From a broad perspective, data mining is based on the fact that large data sets may be
incomplete, noisy, uncertain, and in various storage forms. People do not know in advance of
mining, hidden knowledge useful for decision-making process. Based on the broad view of data
mining functions, the structure of a typical data mining system and its implementation steps are
shown in Figure 1.
As shown in Figure 1, the source information repository is a database or a group of databases,
data warehouses, spreadsheets or other types of information repositories for storing business
information. Data cleaning, sorting and loading mainly preprocess information, the database or
data warehouse server is responsible for extracting relevant data; the data mining engine is the
core of the data mining part and consists of a set of functional modules for performing tasks such
as characterization, association and correlation analysis, classification, prediction, and cluster
analysis. The user interface communicates between the user and the data mining system to realize
the interaction between the user and the system. In fact, data mining is a basic step in the
knowledge discovery process. Knowledge discovery first extracts the data of interest from the data
source and organizes it into a data organization form suitable for mining. Finally, the knowledge
generated models to assess and integrate valuable knowledge to the intelligent system.

Data Data warehouse Data mining

Source information
preprocessing server engine
repository

User interface

Data loading

Characterization

Data collation
Related
Analysis
Grid information
Classification
Data cleaning

Prediction

Data warehouse

Figure 1: Data Mining System Structure.

2.2 System Structure of Data Analysis System for Power Grid Relay Protection
With the wide application of substation monitoring system, the function of dispatching automation
system is different. With the increase of power interruption, the automation degree of power grid
dispatching operation has reached a very high level. However, the automation of protection
monitoring of relay, system failures and the analysis of relay protection action behavior, and relay
protection management is relatively lagging behind, and its technical means can no meet the needs
of the development of power grid dispatching operation management automation. In order to meet
the requirements of grid dispatching, operation and management automation, the construction of
relay protection, safety automatic devices and fault recorders based on modern network
communication technology involves action devices when the power grid is abnormal. The fault
information processing system, namely the "relay protection fault processing system", can not only
realize the efficient use and full sharing of relay protection and related data and information, but
also realize the analysis of fault information and the operation of relay protection. Making the relay
protection professional Modernization of management is of great significance for improving the

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 64

normalization and intelligence of the dispatching system for safe operation of the power grid, and
for improving the overall management of the dispatching operation of the power grid.
The computer-aided analysis of the system is mainly to conduct multi-dimensional correlation
analysis and time sequence prediction on the historical data of grid relay protection. The reference
value of historical data is generally valid in the range of 5 to 10 years. No matter how long it
takes, it will lose its reference significance. Adopt multi-dimensional data warehouse storage based
on relational storage. Its architecture is shown in Figure 2. In this system, a SQL Server 2016
database is used to store historical data accumulated by the power grid. The data-preprocessing
module mainly completes the cleaning, sorting and loading of data in the business database.

Analysis results

Visualization Main interface

Multi-dimensional correlation analysis Predictive model analysis

Analysis item Electricity

Association Failure Load
association consumption
rules prediction Forecasting
rules forecast

Association Frequent Predictive

rule algorithm itemset table model

Multidimensional
data warehouse
Current theme Voltage theme Load theme

Generalized
processing
Correlation Data Data Predictive
analysis cleaning collation analysis
Data Data
cleaning collation

Mainnet Electric Power

Data source Other sources
SCADA MIS System

Figure 2: Teaching parallel corpus construction design.

Since association, rule mining can discover the relationship rules between items or attributes and
attributes that cannot be discovered by traditional methods, it has important research value. Multi-
dimensional association rules are mainly embodied in discovering frequent patterns, correlations or
causal relationships between classification attributes and decision-making attributes from massive
data, to grasp the correlation characteristics between grid components from a macro perspective.
Use the K-Means algorithm to discredited the data, divide the continuous attribute value into

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 65

discrete intervals, and divide the discrete attribute value into several different value ranges,
thereby reducing the number of attribute values and improving the connotation of attribute values,
it is convenient for the process of data mining and the visual display of mining results.
The fault diagnosis of relay protection is essentially a problem of pattern classification. It is
necessary to establish fault phenomena and faults. Accurate mathematical models between the
causes of obstacles are very difficult. This paper designs an intelligent analysis system to predict
grid failures. First, extract the required data from the grid business data, and then use the K-
Means clustering algorithm to discredited the data, establish a multi-dimensional data model data
warehouse, and finally perform correlation analysis and Failure prediction

3 ALGORITHM MODEL OF DATA ANALYSIS SYSTEM FOR POWER GRID RELAY

PROTECTION

3.1 Data Protection Association Rule Mining Algorithm

Association rule mining can discover the relationship rules between items and items or
attributes and attributes that cannot be discovered by traditional methods. The problem of
association rule mining is to specify the minimum support and minimum confidence in a given
transaction number by the user. The process of finding suitable association rules in the database.
Generally, the problem of association rule mining can be divided into two problems: one is to find
frequent item set, and find all frequent item set through the minimum support min_S given by the
user, that is, item set with support not less than min_S. The other is to generate association rules,
and find association rules with no less than confidence in each maximum frequent item set given
by the user.
Tian et al. improved an association rule-mining algorithm called Apriori based on previous
work, which has been cited as a classic association rule-mining algorithm [11]. But it has two fatal
performance bottlenecks: scanning the transaction database multiple times requires a large I/O
load; it may produce a huge candidate set. In this paper, combining the characteristics of grid
protection data, the classic Apriori algorithm is improved.

Horizontal scan

Vertical scan
Databases
Two-dimensional array

Frequent k-itemsets

FreqTmpk={Ik1,Ik2,Ik3 IkN},

Frequent k+1-itemsets Recursion Items converted to

classes

Vector<FreqTmpT>ClassItemSet=new Vector<FreqTmpT>()

Vector set

Association rules

Figure 3: The processing process of frequent item set mining algorithm based on clusters.

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 66

Aiming at the shortcomings of the algorithm, the algorithm has been improved from the following
aspects in order to realize the efficient mining of frequent item set of grid data. 1) Use classes to
encapsulate frequent items, and the generated frequent item sets are stored in the form of
classes. This class includes item names, support and confidence. Therefore, when calculating the
support of candidate item sets, you can directly use the class items Obtained. 2) In order to
overcome multiple scans of the transaction database, the performance bottleneck of a large I/O
load is required, the data in the database is first stored in a two-dimensional array, and the
frequent set mining operation is performed on the two-dimensional array, which avoids frequent
transaction database scanning and improves mining efficiency.
Frequent set mining algorithm based on clusters. First, scan the transaction database and
store the mining data in a two-dimensional array. Then conduct association mining on the data in
the two-dimensional array, and the obtained frequent item set are stored in the class set in object
format, and the calculation of support and confidence is realized by the class set. The processing
process is shown in Figure 3.

3.2 Time Series Prediction Algorithm Mode

Forecasting is to make time-series forecasts on accumulated historical data, dig out future
development trends, and provide a basis for macro decision-making. It is difficult for a single
prediction model to guarantee satisfactory prediction results under any circumstances. Therefore,
the power grid data intelligent analysis system adopts a comprehensive forecasting model, which
combines the exponential smoothing analysis model and the linear regression analysis model to
predict the data of power time series. By integrating different forecasting information and
forecasting accuracy provided by different forecasting models, the forecasting results are
optimized and the forecasting performance is improved. The comprehensive forecasting model is
the appropriate weighted average of the different forecast values obtained by two or more
different forecasting models. Finally, take its weighted average as a prediction model for the final
prediction result. The predictive model used in the computer-aided analysis of the data of grid
relay protection is shown in Figure 4.
In order to illustrate the effectiveness of the prediction algorithm model, a simple distribution
network example is used to illustrate. As shown in the figure, there are four bus bars, two
transformers, and three lines in the distribution network. Equipped with bus differential protection
BR1 to BR4; current protection OR1 to OR8,: transformer differential protection TR1, TR2.
According to the operation principle of power grid protection, according to the various faults that
occur during the operation of the power grid the protection and circuit breaker actions that may
occur under the following conditions. Extract relevant data and perform correlation analysis, and
input the obtained data into the prediction algorithm model to predict and analyze the fault
information of grid relay protection.
The exponential smoothing analysis model and the regression analysis model simultaneously
accept the original data, that is, the historical data sequence of the amount to be predicted to
generate their own predicted values, and then these predicted values are weighted and integrated
to obtain the final predicted data. The exponential smoothing analysis model is also called the
exponential weighted average method. It is actually a weighted moving average method. A
method selects the weight value of each option as the mean value of a decreasing exponential
series. The exponential smoothing method solves the shortcomings of the moving average method
that requires several observations and does not consider the data of the period before t-n.
Through a certain averaging method, the random fluctuation in the historical statistical sequence is
eliminated, and the main development trend is found. For time series x1 , x2 , x3 , …xn , the one-time
smoothing index formula is:
Ft =  xt + (1 −  ) Ft −1 (1)

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 67

Where  is the smoothing coefficient, 0    1 , xt is the observed values of historical data

sequence x at t, and Ft and Ft −1 are the smoothed values at t and t-1.

BR1 BR2 BR3

I-I I-I I-I

OR1
I data input
t 2QF 3QF 4QF 5QF 6QF
OR2 OR3 OR4
I I I
1QF t t t
Exponential smoothing
analysis model
I-I
m
1
ki = ( i2  ) −1
TR1
j =1  2j

OR6 OR7
I I Weighted synthesis
t t
7QF
data input
OR5
I 8QF 9QF 10QF
t
OR8
I Linear regression analysis
I-I
t model
TR2

110kv Substation

Figure 4: Predictive model of analysis system.

The university regression model is suitable for prediction problems where the prediction object is
mainly affected by a related variable and the relationship between the two is linear. The working
procedure is as follows: The important step is to establish a university regression model. The
university linear regression model is a mathematical equation Ŷ = a + bX used to analyze the linear
relationship between an independent variable and a dependent variable.
A very important problem that needs to be solved in comprehensive forecasting is how to
appropriately determine the weighted weights of each single forecasting method. Weights are also
called coefficients, which determine the status or proportion of a single model in combined
forecasting. This system uses linear combination forecasting to determine the weight of each
individual forecast model.
With m single prediction methods, the error of the i-th prediction method is

i = F − F0t
2
it (2)
t

The idea of linear combination forecasting is to select the appropriate weight k i to minimize the
m m
total error  = k 
i =1
i i .Among them, k i =1
i = 1, ki  0 and i = 1, 2, …m , combined with the least square

method, can get the weight:

m
1
ki = ( i2  )−1 (3)
j =1  j
2

After calculating the predicted value obtained by the exponential smoothing analysis model and
the regression analysis model, the predicted value is obtained after weighting processing and
comprehensive analysis.

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 68

4 ANALYSIS OF RESULTS

4.1 Correlation Analysis Test Results

The algorithm model is the core of data mining, and the corresponding algorithm model is selected
for different mining tasks for knowledge mining. According to the characteristics of power grid data
and the needs of knowledge mining, this computer-aided analysis system mainly considers the
association and prediction of two algorithm models. Correlation analysis is mainly to find frequent
patterns, correlations or causal relationships between classification attributes and decision
attributes from the power grid relay protection data, to grasp the correlation characteristics
between data elements from a macro perspective.
In order to verify the effectiveness of the proposed association algorithm for frequent item sets
data mining based on clusters, the association analysis of a 110kv substation data is carried out.
The six failure causes are numbered according to T1~T6, which respectively indicate:
manufacturing quality, design defects, accidental contact by personnel, bad weather, improper
maintenance, and others. Three types of faults O1~O3 are selected for correlation analysis,
namely: transformer fault, protection device fault and protection secondary circuit fault. Use the
association rules to analyze the provided fault data. When the minimum support min_S is set to
0.2, the correlation analysis result of the fault cause and the fault type is shown in Figure 5.
It can be seen from Figure 5 that "manufacturing quality" and "protection device failure" are
closely related, while "design defects" and "protection secondary circuit failure" are closely related.
Through the analysis of the association rules, we can get the knowledge and information we need
for the timely detection of faults and problem analysis.

Figure 5: The correlation analysis result of the fault cause and the fault type.

In order to verify the superiority of the cluster-based frequent itemset data mining association
algorithm (CB-DMA) proposed in this paper compared with the traditional non-cluster-based
frequent itemset data mining association algorithm (DMA), the "manufacturing quality" and
"protection device failure" are adopted. The correlation error is used to verify the algorithm. Using
nearly half a year's failure data, the mean square error of the correlation matching between the
failure cause and the failure type is analyzed. The analysis result is shown in Figure 6.
It can be seen from the figure that the classification of clusters in advance improves the
matching accuracy of fault types and fault causes. The algorithm proposed in this paper has lower
correlation mean square error.

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 69

Figure 6: Mean square error results of different correlation algorithms.

4.2 Time Series Prediction Test Results

Here, only the daily data of transformer failure in an 110kv substation data is selected, the data of
nearly one and a half months is selected to form the time series, and the data of the previous
month is used as the model-training sample. Use nearly half a month's data to test the model's
prediction effect and test the system. The main changed parameter is MINIMUM-SUPPORT, and the
average error (AE), root mean square error (RMSE), average relative error (ARE), and maximum
prediction error (MPE) are used as evaluation indicators. The prediction result is shown in Figure 7.

Figure 7: The prediction result of different error indicators.

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 70

According to the average relative error index, when the ARE range is between 10% and 20%, the
system is good. The results show that when MINIMUM-SUPPORT is equal to 30, its prediction
accuracy is a good fit.
In order to verify the superiority of the prediction model proposed in this paper, choose
different main changes to set the parameter MINIMUM-SUPPORT to analyze its prediction average
relative error (ARE) results. The average relative error results of the exponential smoothing
prediction model, linear regression analysis model, and weighted comprehensive model are shown
in Figure 8. It can be seen from the figure that the proposed prediction algorithm model is
smoother than a single prediction algorithm, and has a smaller average relative error on the
optimal MINIMUM-SUPPORT, which is more effective for the fault prediction of power grid relay
protection.

Figure 8: ARE results of different prediction algorithm models.

5 CONCLUSION
After years of operation of the power grid, a large amount of sufficient fault data has been
accumulated. How to use these data to find out the objective laws hidden after accidental failures
is the focus of power workers' research. This paper discusses the framework model of computer-
aided analysis of grid relay protection based on data mining, System function realization, key
technology and testing in the research process. The proposed class-based frequent itemset mining
algorithm has achieved good results in the analysis of the correlation of fault data. At the same
time, the original regression linear prediction model is improved, and the exponential smooth
prediction model and the regression linear prediction model are analyzed. It adopts the method of
weighted synthesis to predict the possible causes of the failure. From the analysis results, the
prediction accuracy of the proposed method is significantly improved. From the overall concept
research and analysis of the power grid relay protection computer-aided analysis and development
program and practice it with the data of a 110kv substation, it proves that this program is effective
and provides a reference for the development of other analysis systems.

6 ACKNOWLEDGMENT
The 13th Five-Year Plan of Education Science of Henan Education Department in 2020 project
name: Research on evaluation index of online teaching under the background of epidemic disease
project number: 2020YB0547

Na Suo, https://orcid.org/0000-0002-3561-6468
Zheng Zhou, https://orcid.org/0000-0003-2519-8386

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net
 71

REFERENCES
[1] Amin, M.-S.; Chiam, Y.-K.; Varathan, K.-D.: Identification of significant features and data
mining techniques in predicting heart disease, Telematics and Informatics, 2019, 36: 82-93.
https://doi.org/10.1016/j.tele.2018.11.007
[2] Chen, W.; Tsangaratos, P.; Ilia, I.: Groundwater spring potential mapping using population-
based evolutionary algorithms and data mining methods, Science of The Total Environment,
2019, 684: 31-49. https://doi.org/10.1016/j.scitotenv.2019.05.312
[3] Wang, L.; Li, Y.; Zhu, H.: Research on Transformer Relay Protection System Based on
Ubiquitous Power Grid Technology[C]//IOP Conference Series: Earth and Environmental
Science, IOP Publishing, 2020, 558(5): 052011. https://doi.org/10.1088/1755-
1315/558/5/052011
[4] Wu, D.; Li, P.; Zhou, H.: Research and Application of Intelligent Maintenance of Relay
Protection Equipment Based on Internet of Things Technology, E&ES, 2020, 440(3): 032043.
https://doi.org/10.1088/1755-1315/440/3/032043
[5] Lu, J.; Zhang, L.: Data mining technology of computer testing system for intelligent
machining, Neural Computing and Applications, 2020: 1-11. https://doi.org/10.1007/s00521-
020-05369-6
[6] Yue J.; Li C.: Type Identification and Classification of Cascading Failures of Time Sections of
Dispatching and Operation Based on Data Mining, MS&E, 2020, 740(1): 012137.
https://doi.org/10.1088/1757-899X/740/1/012137
[7] Gaber, M.-M.; Aneiba, A.; Basurra, S.: Internet of Things and data mining: From applications
to techniques and systems, Wiley Interdisciplinary Reviews: Data Mining and Knowledge
Discovery, 2019, 9(3): e1292. https://doi.org/10.1002/widm.1292
[8] Sekar, K.; Mohanty, N.-K.: Data mining-based high impedance fault detection using
mathematical morphology, Computers & Electrical Engineering, 2018, 69: 129-141.
https://doi.org/10.1016/j.compeleceng.2018.05.010
[9] Yang, X.; Wang, W.; Li, Y.: Analysis of Power Grid Fault Diagnosis Based on Association Rules
and its Evaluation Index Optimization Method[C]//2018 Chinese Automation Congress (CAC),
IEEE, 3092-3097. https://doi.org/10.1109/CAC.2018.8623284
[10] Zhu, L.; Li, M.; Zhang, Z.: Privacy-preserving authentication and data aggregation for fog-
based smart grid, IEEE Communications Magazine, 2019, 57(6): 80-85.
https://doi.org/10.1109/MCOM.2019.1700859
[11] Tian, M.; Zhang, L.; Guo, P.: Data Dependence Analysis for Defects Data of Relay Protection
Devices Based on Apriori Algorithm, IEEE Access, 2020, 8: 120647-120653.
https://doi.org/10.1109/ACCESS.2020.3006345

Computer-Aided Design & Applications, 18(S4), 2021, 61-71

© 2021 CAD Solutions, LLC, http://www.cad-journal.net