Screw Remaining Life Prediction Based on Quantum Genetic Algorithm and Support Vector Machine

Abstract

To predict the remaining life of ball screw, a screw remaining life prediction method based on quantum genetic algorithm (QGA) and support vector machine (SVM) is proposed. A screw accelerated test bench is introduced. Accelerometers are installed to monitor the performance degradation of ball screw. Combined with wavelet packet decomposition and isometric mapping (Isomap), the sensitive feature vectors are obtained and stored in database. Meanwhile, the sensitive feature vectors are randomly chosen from the database and constitute training samples and testing samples. Then the optimal kernel function parameter and penalty factor of SVM are searched with the method of QGA. Finally, the training samples are used to train optimized SVM while testing samples are adopted to test the prediction accuracy of the trained SVM so the screw remaining life prediction model can be got. The experiment results show that the screw remaining life prediction model could effectively predict screw remaining life.

Remaining capacity estimation of lithium-ion batteries based on the constant voltage charging profile

“…Accordingly, SVR is often adopted as the learning model in studies of lithium batteries [ 30 – 32 ]. Since the performance of SVR highly relies on the selection of model parameters especially the kernel parameters, many intelligent algorithms like genetic algorithm (GA) [ 33 , 34 ] and PSO [ 9 , 35 ] are used to optimize the SVR model. Compared with GA, the PSO has faster convergence speed [ 36 ].…”