Study on acquiring key performance parameters and comprehensive performance degradation model of relays

Study on the degradation performance mode has great significance for the life prediction of the relays. The contact resistance, pull-in time and other parameters (e.g., temperature rise) reflect the strengths and weaknesses of the relays from different extent. In order to study the performance degradation rules of the relays, we should process lots of sequence information that contains so many parameters and sampling points. We have to compress masses of data related. First of all, use a couple of ways to select the sequence information with the neural network such as selecting the sample points continuously, every two points, every three points and so on, so that we can determine which method to choose by comparing the errors we get. And then we can identify the rows of the sequence information matrix. To extract the key performance parameters, we should deal with the column vectors of the sequence information matrix with PCA (i.e., reducing the dimension of the matrix). And construct value function to acquire the characteristic quantity which can describe the comprehensive performance reliability of the relays. Finally we can get the comprehensive performance degradation model of the relays.