大型水轮机组故障诊断的神经网络方法研究

The Research of Neural Network Method in the Faults Diagnosis of Lrge-scale Turbo Units

本文针对大型水轮机组状态监测与故障诊断系统 ,对采用神经网络进行故障模式识别的方法进行研究 ,为克服单一神经网络模型诊断方法的局限性 ,对现有的网络模型进行了分析和比较 ,提出并探讨了多种神经网络技术应用于故障诊断的诊断方法。该方法综合了多层感知模型和自适应谐振网络各自的优势 ,建立了适用于水轮机组故障诊断的混合网络模型 。

With the application demand of neural network in faults diagnosis of large scale turbo units,facing with the limits of single neural networkd model,the article analyses current neural network models,and designs and discusses the hybrid networkd for diagnosis.The model integrates both the advantage of MLP model and ART model,then builds the HANN models for turbo unit's faults diagnosis,and is applied in the system of monitor and faults diagnosis for largescale turbo unit.