基于经验模态分解的水力发电机组故障自动化诊断技术

Automatic Fault Diagnosis Technology of Hydropower Unit Based on Empirical Mode Decomposition

我国水电机组大多采用事后检修和计划检修两种,随着检修管理水平的提高,这些方式已经不能满足自动化要求。为此,本文提出基于经验模态分解的水力发电机组故障自动化诊断技术。通过发电机组结构,建立引水系统数学模型;通过舍去前几阶分量的方式完成信号去噪处理;将变异系数、偏态系数与峰均比作为无量纲特征参数,确定神经网络训练和测试样本,选择输入向量与神经元数量,调整阈值,反复训练信息正向传播与误差反向传播过程,输出最终诊断结果。仿真实验证明方法能够准确诊断出故障类型,实现自动化诊断。

With the improvement of maintenance management level,these methods can no longer meet the automation requirements.Therefore,an automatic fault diagnosis technique based on empirical mode decomposition is proposed in this paper.Through the structure of generator set,the mathematical model of water diversion system is established.The signal denoising process is completed by discarding the first several order components.Coefficient of variation,coefficient of skewness and peak-to-average ratio were used as dimensionless characteristic parameters to determine the training and testing samples of the neural network,select the input vector and the number of neurons,adjust the threshold,repeatedly train the process of information forward propagation and error back propagation,and output the final diagnosis results.The simulation results show that this method can diagnose the fault type accurately and realize automatic diagnosis.