Abstract:Aiming at the problems of easy to fall into local extremism in training and slow convergence speed appeared during computer numerical control (CNC) faults forecasting in BP neural network, an optimizing BP neural network method (IMBP) of CNC machine tool fault diagnosis based on artificial immune algorithm was proposed. The common types and classification of CNC machine tool faults were introduced, and the workflow was described of the artificial immune algorithm, the BP neural network, the optimized BP neural network based on artificial immune algorithm. By using the global search capability of immune algorithm to global optimize the weights of neural network and threshold values, the training process and the convergence speed of BP algorithm were quicken up, and the time in need of training process was reduced. Through the analysis of performance and simulation test, compared with three algorithms of BP, GABP and IMBP, the results show that the fault diagnosis forecasting of CNC machine tool is improved by 18.3% as compared to BP, and 12.05% as compared to GABP, which improve the accuracy of fault diagnosis of CNC machine tool.