改进的BP网络在凝汽器故障诊断中应用

APPLICATION OF IMPROVED BP NEURON NETWORK IN CONDENSER FAULT DIAGNOSIS

针对BP网络学习收敛速度慢和易陷入局部最小点的不足 ,提出了一种恒误差修正率控制的网络学习率自适应调整方法。将这种改进的BP网络算法应用于凝汽设备故障诊断中 ,在 1台 30 0MW机组上取得了实效。

In view of low speed selflearning convergence concerning back propagation(BP) neuron network, and the said network being easily in insufficient of partial lowest points, a method of adaptive adjustment for the network selflearning rate controlled by correction rate of constant error has been put forword. This improved BP network has been used in condenser fault dragnosis, and obtaning practical effectiveness on a 300 MW unit.