Fault Diagnosis for Electricity Station of SAM Test Vehicle Based on Improved BP Neural Network
Aimed at the scarcities of B P neural network that its learning rate was slow and it had the least value in some regions of the system, a new calculation method which could adapt the learning rate itself and transmit conversely the values of momentum gradient reduction get along with learning rate gradient reduction, was presented to improve the network and train it. By collecting lots of fault phenomenon 、studying the reasons of them and cleaning up these fault stylebooks for the electricity station system of test vehicle of SAM, a new fault diagnosis model based on B P neural network was established. The simulation results showed that the model based on improved B P neural network had such excellent characters as fast learning rate, good performance of error container and stabilization, strong capability of dealing with nonlinear problem and uncertainty factor. Conclusions proved the model designed in paper was quietly efficient in fault diagnosing for the electricity station system of test vehicle of SAM.
improved BP neural network electricity station system of SAM fault diagnosis gradien treduction
Chen Zhixiang Lei Humin Wang Fengchao Wang wei He Zewei
The Missile Institute, AFEUAir Force Engineering University
国际会议
北京
英文
2007-08-05（万方平台首次上网日期，不代表论文的发表时间）