Study On Reliability Evaluation Of Power System Based On Monte Carlo Simulation

With the development of power system towards high voltage, long distance, large capacity, area interconnection and deregulation, the blackout accident are always followed by enormous loss of economy and society. Hence, the accurate, rapid and comprehensive reliability evaluation of the power system is of great importance for the programming, operation and transaction of the power system.There are two basic methods for power system reliability evaluation, i.e. the certainty analysis and probabilistic analysis. Due to the stochastic characteristics of the components states and the load level, the probabilistic analysis is more reasonable and precise than the certainty analysis. The analytical method (AM) and Monte Carlo simulation (MCS), two basic algorithms of the probabilistic analysis, are different from the way how the system stochastic states and the probability distribution are obtained. For the AM, the system stochastic states arise from contigency enumeration and the probability distributions of the stochastic states are obtained by analytical calculation. The stochastic states, in MCS, come from the random sampling and the statistical frequency of the stochastic states are employed as the approximation of the probability of the stochastic states. Along with the superiority of the exactness of the mathematic model, three main disadvantages of AM limite its application to the power system reliability evaluation. First, the number of system states analyzed in AM increases exponentially to the number of system components, so it is unpractical to apply AM to the reliability evaluation of bulk power system. Second, it is difficult for the AM to get the frequency and duration (F&D) indices which are important informations of the power system. Last, AM can not simulate the stochastic incidents, such as the the load fluctuation and the reservior water level, and the control actions of the operators either. Compared with AM, MCS is more effective for the power sysetm reliability evaluation. The notable advantage of MCS is that the sampling number with a given accuracy is independent on the system size. In addition, the algorithm based on MCS can obtain the probabilistic indices as well as the F&D indices of power system reliability. Furthermore, by simulating the stochastic incidents and the control actions of the operators, MCS will supply dependable and credible reliability indices. Therefore, there has been a continuing and increasing interest in the application of MCS to the power system reliability evaluation. Unfortunately, the critical limitation of MCS is that the computational burdens increase dramatically with the advancement of required accuracy of the evaluation. The efforts devoted to improve the computational efficiency of MCS are of great significance.The principles and characteristics of the three algorithms of MCS, sequential simulation, non-sequential simulation and pseudo-sequential simulation, are expressed in the dissertation. The steps of power system reliability evaluation based on MCS as well as the indices are also listed. The reliability evauations of test system RBTS and IEEE-RTS are performed using uniform sampling to study the convergence of the conventional MCS. The results of the case study show that the convergence of conventional MCS is very time-consuming, and that the the poor efficiency of the sampling calculation leads to the conflict between the computational burden and computational accuracy. In the dissertation, the study of improved algorithms for variance reduction is performed to speed up the convergence of MCS in power system reliability evaluation. The new algorithms presented in the work are listed as follow.Combining scattered sampling technique with importance sampling algorithm, an improved importance sampling algorithm for power system reliability evaluation is proposed. The importance sampling algorithm is adopted to optimize the probability distribution of the system stochastic states, and the scattered sampling technique is employed to improve the sampling efficiency. Case studies show that the the improved importance sampling algorithm is a proper method for the reliability evaluation of the generation system, and that the computational efforts of the proposed algorithm are less than those of conventonal importance sampling and scattered sampling. The optimal sampling algorithm with good adaptivity and practicability is developed. The sampling calculation falls into two successive phases, the pre-sampling phase and normal sampling phase. In the pre-sampling phase, the optimal sampling density functions are obtained by iterative calculations. In the normal sampling phase, the sates of the components are sampled from the optimal sampling density functions, which lead to notable reduction of computational efforts and variance. It can be transplanted to the reliability evaluation of transmission system, composite system, distribution system etc., which indicates its practicability and robustness.The hybrid method of MCS is adopted in many papers to reduce the variance in the power system reliability evaluation. There is little theory for the selection of analytical variables, and the selective analytical algorithm is proposed to solve the question in the dissertation. The concept of variance projection is defined to indicate the contributions of the states variables to the variances of the test functions in the power system reliability evaluation. Then, the priority of the states variables determining the selection of analytical variables is obtained based on the values of variance projections. The selective analytical algorithm, helpful to select the reasonable analytical varibles, can improve the computational effiency of the hybrid method of MCS. The improved hybrid method is developed by combining the principles of selective analytical algorithm and optimal sampling algorithm. There are two successive phases in the improved hybrid method, i.e. the pre-sampling phase and normal sampling phase. In the pre-sampling phase, the optimal sampling density functions and the variance projections are calculated by iterative calculations. In the normal sampling phase, the values of stochatic variables are sampled from the optimal sampling density functions, and the values of selected analytical variables are enumberated. With notable reduction of the computational effects, the improved hybrid provides prospective on-line application of the power system reliability evaluation grogram.The optimization of the stochastic states of power system is the essential part of the reliability evaluation. Many efforts have been devoted to the modeling and solution algorithm of power system optimization. This dissertation focuses on the study of power system optimization referred in the reliability evaluation of composite power system which is the incorporation of generation and high-voltage transmission systems. The traditional optimizations, used in the composite power systme reliability evaluation, perform the active power rescheduling to to alleviate the operating constraint violations and minimize the load shedding. Without considering the influences of the reactive standby, reactive power flow and the bus voltages, the indices are not the reasonable indication of the power system reliability, but an optimistic assessment. Some scholars have noticed the necessity of including the constraints of reactive power and bus voltages in the reliability evaluation of power system, and established the corresponding algorithms. But in their algorithms the quantitative analyses of the influences of those constraints mentioned above on the reliability have not been made, and the correlated indices such as expected voltage (EV), which are important datum for assessing the system operation state and power quality, is not represented as a basic index for the reliability evaluation of power system. In this work, a new model for the reliability evaluation of power system has been presented. The constraint of reactive power adequacy, the integration of the constraints of the reactive power flow, reactive power capacity and bus voltages, is taken into account in the program of power system reliability evaluation. Case studies to RBTS and IEEE-RTS show the notable influences of the reactive power adequacy on the power system reliability. It is pointed out that the error arising from the neglect of the constraints of the reactive power adequacy is beyond the acceptability in the power system reliability evaluation and programming.In the dissertation, the dependences of the reliability indices on unit capacity and reactive power adequacy are analyzed in detail, and a new method for improving the system reliability is suggested. With the assessment of the bottleneck of the power system reliability, the effective strategies for the improvement of power system reliability will be presented. The new method, with reasonable balance between the reliabiltiy level and facility investment, are of significance for the programming and operation of the power system.