Remaining Life Prediction of Complex Devices Based on Mamba

Complex industrial equipment is expensive, so accurate prediction of the remaining useful life(RUL) of complex industrial equipment has important industrial application value. In order to solve the problems of low accuracy of existing prediction algorithms for complex industrial equipment and insufficient fitting of equipment in the decline period, a segmented prediction method of the remaining useful life of complex equipment based on Mamba network is proposed. First, data preprocessing technology is applied to normalize the equipment data from Supervisory Control And Data Acquisition System (SCADA) to eliminate the effect of data magnitude differences; then the Mamba network frame is used to construct the model’s main and end models; the main model performs preliminary life prediction based on preprocessed SCADA data and makes realtime decision-making for model switching; and when the equipment is judged to be in the decay period, the end model is switched to focus on the end life prediction of the equipment to ensure the robust operation of the equipment through accurate prediction; finally, the CMAPSS dataset is used as the experimental data for the test, and the RMSE is reduced to 10.97, which strongly verifies the effectiveness of the proposed prediction algorithm.