An Improved Evolutionary Neural Network Algorithm and its Application in Fault Diagnosis for Hydropower Units
In order to overcome the conventional genetic algorithms shortcoming such as premature convergence and low global convergence speed, a help operator was added in genetic algorithm and the selection method and mating method were improved. Based on this improved genetic algorithm, an improved evolutionary neural network algorithm named IGA BP algorithm was presented in this study. In IGA-BP algorithm, the improved genetic algorithm was used firstly to evolve and design the structure, the initial weights and thresholds, the training ratio and momentum factor of neural network completely. Then, training samples were used to search for the optimal solution by the evolved neural network. The disadvantage of neural networks that their structure and parameters were decided stochastically or by ones experience was overcome in this way. IGA-BP algorithm was used to diagnose hydropower units fault. A fault diagnosis model for hydropower units was found based on neural network. The illustrational results show that IGA-BP algorithm is better than traditional neural network algorithm in both speed and precision of convergence. We can realize a fast and accurate diagnosis for hydropower units fault using this algorithm.
Genetic algorithm Neural network Complete evolution Hydropower units Fault diagnosis
YAN Tai-shan
School of Information and Communication Engineering, Hunan Institute of Science and Technology, 414000 Yueyang, Hunan, China
国际会议
长沙
英文
548-551
2010-05-11（万方平台首次上网日期，不代表论文的发表时间）