Reliability Assessment of Power Generation System Using an Optimized State Enumeration Method

Background: The single state enumeration method cannot meet the requirement of accuracy and high efficiency in the reliability assessment of complex power systems because of many uncertain factors and the large scale of the power grid. Methods: A new method of generating system reliability assessment based on Self-Organizing Map (SOM) neural network and state enumeration is presented. First, the input parameters of the state enumeration method are optimized by using the feature of the SOM neural network algorithm that can automatically, quickly, and accurately classify the sample parameters in this method. Second, combining with Markov Model, the optimized system state samples are divided into fault state and normal state, and then the reliability indexes are enumerated. Finally, this method is used to calculate the reliability indexes of IEEE-RTS single-stage power units under different operation conditions. Results: The results show that this method is superior to the single state enumeration method in calculating time; it can be used to evaluate the reliability of modern complex power systems. Conclusion: The optimized state enumeration method is more suitable to evaluate the reliability of the system with a large network scale, and its reliability index is more accurate; while retaining the higher calculation accuracy of the state enumeration method, it can promote the safe, reliable, and economical operation of the power system.