The K value determination research of advanced breaking current weighted cumulative method for VCB electrical endurance detection

The integral action coefficient value K about the advanced breaking current weighted cumulative method (BCWC) used for the VCBs´ contact system electrical endurance monitoring is studied. At first, the general selecting method of integral current coefficient K value is discussed, and the existent problem in the K value selecting, calculating process and potential improving room are analyzed. Considering the distribution, characteristics and factors which impact K value, and according to the analysis results the LM-BP neural network algorithm is adopted to research how dynamically select the K value based on the engineering application. The construct process, establishing skill, selecting method and determination criteria of the hidden layer node of researched K LM-BP neural network are significantly emphasized. In order to verify the feasibility and accuracy of the proposed LM-BP neural network the K value predict process was carried out by the researched neural network and the result indicate that the net has a good network optimization ability and forecast precision. Thus the feasibility of provided method can be achieved of this work The LM-BP neural network algorithm introduced can promise the K value selected more approximate to practical engineering, which can ensure and effectively improve the predicting accuracy.