Deep Learning Model And Experiment Of Electrical Lifetime Prediction Of AC Contactor

Alternating current(AC)contactors are a type of switchgears that are switched on and off frequently.They are widely used in various power systems.The key factor in reducing the electrical lifetime is the erosion of the contacts by the breaking arc,resulting in an increase in the loss of contact mass.Therefore,the electrical lifetime of an AC contactor is much shorter than its mechanical lifetime.Electrical lifetime prediction which is related to power safety,is a very important research topic.Firstly,the Schneider LC1D80M7 C AC contactor was used as an experimental sample.The initial broken arc waveform of the contact is collected by the electrical lifetime experimental system and saved by the data acquisition system.The data waveform software is used to obtain the arc sampling point data as the input feature.The broken arc waveform characteristic data is processed by linear interpolation and normalization.Secondly,all contact mass data for each measurement was acquired by the electronic scale during the electrical lifetime test.The sum of the mass loss of all contacts for each measurement is then calculated as thelabel for the model in this paper.The linear interpolation is used to increase the amount of contact mass loss label data.Thirdly,deep learning software TensorFlow and convolution neural networks(CNN)are used to obtain electrical lifetime prediction model of the AC contactor by regression method in this paper.The model adopts two convolution layers,two pooling layers and two fully connected layers.The model is evaluated by finding the total mass loss of the contact and the mean squared error of the model output.Lastly,the feature and label data are randomly divided into training and testing sets,which are read into the prediction model respectively.The way to predict electrical lifetime is to determine the minimum test mean square error value and calculate the total mass loss error percentage of the contact.Since the broken arc generated when the contacts are separated will ablate the contacts,the mass loss of the contacts is increased,and the electrical lifetime is reduced.The error percentage of the total contact mass loss is about 12.03% after calculation,which indicates that the model can predict the electrical lifetime of the AC contactor.