Ứng dụng giải thuật thông minh trong sa thải phụ tải hệ thống điện

Abstract

This article presents the application of hybrid intelligent algorithms by combining
Back Propagation Neural Network algorithms (Back Propagation Neural Network: BPNN)
with Cuckoo Search algorithms (Cuckoo Search: CS) to improve neural network structure.
Subsequently, model is proposed in identifying power system faults, supporting load shedding
to restore the power system. Besides, the article also mentions the application of voltage
correlation sensitivity index to divide the amount of load shedding. This load shedding
method is tested on IEEE 37-Bus power system and simulated with POWERWORLD
software. The results of the method satisfy the requirement of restoring the frequency within
the allowable value. Furthermore, the amount of load shedding is reduced by nearly 2 times
compared to the traditional method. The data input into different neural network models to
compare performance and choice the structure that suitable with the data.
Keywords: Power System Stability, Load Shedding, Artificial Neural Network (ANN),
Cuckoo Search algorithms (CS)