ساخت بیزی شبکه برای رتبه بندی اهمیت قابلیت اطمینان اجزا در سیستم های قدرت کامپوزیت

In this paper, Bayesian Network (BN) is used for reliability assessment of composite power systems with emphasis on the importance of system components. A simple approach is presented to construct the BN associated with a given power system. The approach is based on the capability of the BN to learn from data which makes it possible to be applied to large power systems. The required training data is provided by state sampling using the Monte Carlo simulation. The constructed BN is then used to perform different probabilistic assessments such as ranking the criticality and importance of system components from reliability perspective. The BN is also used to compute the frequency and duration-based indices without time sequential simulation based inferences. The proposed approach provides the possibility of assessing the components importance in view of different load points. The validity and efficiency of the proposed approach is verified by application to the IEEE-Reliability Test System (RTS). Highlights ► A simple approach is presented to construct the Bayesian Network (BN) of a power system. ► The approach is based on the capability of the BN to learn from data. ► The approach can be applied to large power systems. ► Different probabilistic assessments are easily provided by the constructed BN. ► The importance of each component on system reliability can be easily evaluated.

The common purpose of power system reliability studies is to provide probabilistic analysis to determine various reliability indices to evaluate the adequacy of power system in supplying the total load. However, analyzing the effect of individual components in system reliability and ranking is also of high importance for a variety of purposes such as determining the background of outages, system reinforcement, maintenance scheduling, and expansion planning and so on, all performed to improve system reliability. Bayesian Network is one of the most efficient probabilistic graphical models to represent uncertain information and inferences thereof. BNs have found wide applications in many fields and they especially have been used in power system studies [1], [2] and [3]. BNs have recently been used as an appropriate probabilistic framework in reliability studies. Various applications of the BN are also reported in power system reliability assessments. The first step in applying the BN in system study involves determining its structure and parameters. BN is mainly utilized for reliability assessment of generating systems [4] as well as power distribution systems [5], [6], [7], [8], [9] and [10]. In [11], the BN is used in reliability assessment of a small composite power system constructed on the basis of the system’s physical topology interpreted by its fault tree and minimal cut-sets or tie-sets. Refs. [6], [7], [8] and [9] present methods based on time sequential simulation technique to perform inferences by the BN to compute frequency and duration based indices. A D–S evidence inference method with Bayesian Network is employed in [10] for reliability evaluation of distribution system in the case of lack of original data. This approach considers the impacts of uncertain information on the system reliability. In these studies, the BN is constructed on the basis of expert beliefs, cause-and-effect relationships, and physical topology of systems. Particularly, constructing the BN associated with the composite power systems necessitates access to the fault tree, minimal cut-sets or tie-sets of the given system. But for a composite power system with a generally non-radial topology, identification of minimal cut-sets or tie-sets or constructing the fault tree of system, especially with regard to the different operational conditions in the system is not practical. In [12], the common structure learning algorithms were used to construct the BN for a power system with a relatively large burden of computation. In this paper, a simple approach based on the learning capability of the BN from data is proposed to construct the BN associated with a composite electric power system. The required training data is generated by state sampling using the Monte Carlo simulation. The obtained BN is then used for importance ranking of individual system components, computation of frequency and duration-based reliability indices and to perform other probabilistic assessments that may not be easily handled by the conventional methods. The rest of this paper is organized as follows. In Section 2, the BN is briefly introduced. The approach used for data generation is presented in Section 3. A method to construct the BN is proposed in Section 4 and then it is applied to the IEEE-RTS in Section 5. In this section, the obtained BN is used for different inferences and components ranking from different aspects. The paper is concluded in Section 6.

In this paper a simple approach is presented to construct the BN of composite power systems. The approach is based on the learning feature of the BN from training data and can be easily applied to large power systems. The required training data is provided by state sampling using MC simulation. Once the BN associated with the power system is constructed, it can be used for different reliability analyses. Using the BN provides the possibility of assessing components importance from different reliability aspects easily. In this paper, it is shown that by the BN it is also possible to compute the frequency based indices of system without using time sequential simulation. The proposed approach in constructing the BN is applicable to large and complex power systems and allows for evaluating the system from the viewpoint of different load buses. The approach was applied to the IEEE-RTS. The results showed the efficiency and superiority of the proposed approach in performing different analyses not easily possible by other reliability assessment methods.