摘要
关于汽轮发电机系统一定要保证可靠安全。研究汽轮机发电机组的故障快速诊断问题上,针对汽轮发电机故障具有多样性和不确定性,传统BP神经网络不能很好的识别这种特性,存在训练时间长、误差收敛速度慢的缺陷,故障诊断正确率低。为了提高汽轮发电机组的故障诊断准确率,提出一种附加动量法和自适应速率相结合的BP神经网络故障诊断模型。采用附加动量调整了BP神经网络的权值,加快了网络的收敛速度,用自适应速率动态地调整了学习速度,减少了迭代次数,最后利用得到的BP神经网络对故障进行了诊断。在matlab上采用实测汽轮发电机故障数据对故障诊断模型进行测试,相对于传统的BP算法,该算法不仅迭代次数少、学习速度加快,而且故障诊断准确率高。结果表明,有效地克服了传统的梯度下降的BP算法的缺陷,诊断结果可为保证汽轮发电机安全运行提供保障。
Study the fault diagnosis of turbine generator.Turbine generator fault has diversity and uncertainty.Traditional BP algorithm has the defects of long training time,slow convergence speed,and low accuracy of fault diagnosis.In order to improve the accuracy in the fault diagnosis of turbine generator unit,we put forward a kind of additional momentum method combined with adaptive rate of BP network model.The algorithm adopts additional momentum to adjust the weights of BP neural network and accelerate the convergence rate of the network,and uses adaptive rate dynamically to adjust the speed of learning and reduce the number of iterations.Finally,BP neural network is used to diagnose the fault in turbine generator unit.In matlab platform,the actual monitoring data are used to test the fault of turbine generator unit based on the improved BP algorithm.Compared with the traditional BP algorithm,this algorithm has fewer iteration times,faster learning speed,and higher fault diagnosis accuracy.The experimental results show that this method can effectively overcome the traditional gradient descent of BP algorithm,and the diagnosis results also conform to the real faults.
出处
《计算机仿真》
CSCD
北大核心
2011年第7期325-328,332,共5页
Computer Simulation
