基于ReLU深度信念网络的35kV开关柜故障诊断

Fault diagnosis of 35kV switch cabinet based on ReLU-depth belief network

开关柜内环境复杂,状态类型多样,为了准确诊断开关柜故障,提出基于ReLU深度信念网络的35kV开关柜诊断模型。首先布置不同传感器测点,获取各类信号源后从中提取敏感状态特征参量,输入到深度信念网络模型中,最终根据网络输出结果进行开关柜状态判定。实测数据验证了所述方法对35kV开关柜故障诊断的可行性和优越性,对于实际工程应用具有一定参考借鉴价值。

The inner environment of switch cabinet is com plex,and the status types are various.In order to accurately diagnose the switch cabinet fault,a 35kV switch cabinet diagnosis model based on rectified linear units deep belief network(R eL U-D B N)is proposed.Firstly,the different sensor monitoring points are arranged,after the signal sources are obtained,the sensitive feature parameters are extracted and inputted into the deep belief network.Finally,the condition of switch cabinet is judged based on the network output results.The measured data verifies that the feasibility and the superiority of the proposed method in diagnosing the 35kV switch cabinet,and it has a reference value for practical actual engineering application.