基于神经网络和D-S证据理论的回热系统故障诊断

Regenerative System Fault Diagnosis Based on Neural Network and D-S Evidence Theory

火电机组中的回热系统的故障原因复杂,且具有相关联,在深入分析D-S证据理论和神经网络理论的基础之上,将这两种方法进行融合。通过对回热系统典型故障的数据流参数进行分析,先由神经网络进行初步诊断,并将诊断结果处理后作为证据理论的基本可信度分配值,得到最终的诊断结果。经过试验分析表明:该方法能够使得火电机组回热系统故障识别能力得到提高。

In thermal power unit,the factors of regenerative system fault is very complexity and also have relativity,so fusing the ANN and D-S evidence theory had been put forward.Through analysis regenerative fault parameters data flow,it take each low-dimension neural network output value as the basic belief assignment value,then through D-S evidence fusing to get the final result.The result shows that,this method could promote the fault diagnosis ability.