Abstract:Aiming at the problem that model parameters have great influence on the prediction accuracy of residual life of mechanical equipment in traditional neural network, A a method for predicting residual life of mechanical equipment based on Bayesian optimization generalized regression neural networks (BO-GRNN) is proposed. The collected sensor data is smoothed and filtered, and principal component analysis (PCA) is used to extract the features of the smoothed data, and the first principal component is regarded as the degraded feature.Firstly, the collected sensor data is smoothed and filtered, and the principal component analysis (PCA) is used to extract the features of the smoothed data, and the first principal component is regarded as the degraded feature of the device. A Bayesian optimal generalized regression neural network residual life prediction model is constructed, and the smoothing filter window width and the smoothing factor in GRNN are optimized by Bayesian optimization algorithm, so that the average error of prediction results of BO-GRNN model is minimumSecondly, the Bayesian optimization generalized regression neural network residual life prediction model is constructed, and the smoothing filter window width and the smoothing factor in GRNN are optimized by Bayesian optimization algorithm, so that the average error of prediction results of BO-GRNN model is minimum. The test data set is input into the trained BO-GRNN model for residual life prediction. The proposed method is verified by using C-MAPSS data set, and the prediction error of the proposed method is the smallest compared with other methods. Experimental results show the effectiveness and accuracy of the proposed method.