RBF神经网络在水轮发电机故障诊断中的应用

Research on Hydraulic Generator Faults Diagnosis Based on PSO-RBFNN

研究发轮电机故障准确诊断对水电站正常运行有着重要意义。由于水轮发电机是将水动能转换为电能,结构复杂,传统故障诊断方法难以解决水轮发电机的高维、非线性和不确定输出等问题,故障诊断准确率低,不利于实时诊断。为了实时进行发电机故障诊断,保证系统安全性能,提出一种改进的神经网络故障模式分类算法。首先采用粒子群优化算法对基本RBF神经网络进行优化和改正,提高网络学习性能,然后用改进算法对水轮发电机故障进行故障诊断。对水轮电机振动数据进行测试实验,结果表明算法提高了水轮发电机故障诊断速度和准确率,结构简单,可以为水轮发电机故障实时识别提供科学依据。

Diagnosing the hydraulic generator faults is important for the power stations normal running.The hydraulic generator can transform water energy into electrical energy,but the system structure is very complex,and it is difficult for the traditional fault diagnosis method to solve the problems in hydraulic generator,such as high dimension,nonlinear and uncertain,which causes low fault diagnosis accurate rate is and poor real-time diagnosis.In order to improve the fault diagnosis accuracy of generators and ensure the system security performance,this paper presented an fault pattern classification algorithm based on the improved neural network.Firstly,RBF neural network was optimized by particle swarm optimization algorithm to improve the performance of neural network and then the hydraulic generator faults was diagnosis by building model.The model was tested by the hydraulic generator data,and the results show that the presented algorithm improves the hydraulic generator fault diagnosis speed and accuracy,simplifies the structure,and can provide a scientific basis for turbine generator fault diagnosis.