Performance degradation prediction of aircraft cables for complex mechatronic products based on TCN-LSTM

To address the phenomenon of performance degradation and safety problems of aircraft cables under the action of various causes, we proposed a prediction method based on time-series convolutional networks combined with long and short-term memory network models to predict the cable performance status. First, we analysed the overall causes of cable performance degradation, and after concluding that the cable insulation state is the main cause, we selected four parameters that can reflect the cable insulation state information and determined the data set based on the analysis results. After that, we constructed a network model based on the data features and extracted the four parameters reflecting the cable insulation status under the convolutional network. Finally, the prediction results were obtained by bringing the feature data identified by Pearson correlation coefficient analysis and the influencing factors with high enough correlation into the TCN-LSTM based prediction model. The case study and the comparison of the prediction results of different networks verified that the prediction results of the constructed method are more relevant to the actual situation.