带有偏差单元的递归神经网络在故障诊断方面的应用优化

Application Optimization Fault Diagnosis in for the Unit with Deviation Recursive of Neural Network

针对多层BP网络的不足,在BP网的基础上,给出一种带有偏差单元的递归神经网络模型及其算法,在学习过程中引入经验知识.仿真表明该算法能大大提高学习速度和训练精度.

Aimed at many shortage of the layer BP networks and based on BP net,a kind the nerve network model to have the deviation unit to pass to return and its calculate way were given,lead into experience knowledge in the learning process.Imitated calculation showed that the methods can raise study speed and train accuracy consumedly.