Fault Diagnosis of CNC Machine Tools based on Support Vector Machine Optimized by Genetic Algorithm

Abstract

To enhance the accuracy of CNC machine tool fault diagnosis, this study proposes an intelligent optimization method based on the combination of Particle Swarm Optimization (PSO) and Bacterial Foraging Algorithm (BFA), referred to as PSO-BFA. By simulating the local foraging behavior of bacteria, the PSO-BFA algorithm demonstrates characteristics of local convergence, replicability, and migratory properties during parameter selection, effectively improving the local optimization capability and fitness value of the model. This leads to faster convergence to the optimal solution in the fault data training process. The study utilizes a Deep Confidence Network (DCN) model, known for its strong adjustability of model structure, for training the fault feature set. The PSO algorithm is employed to search for the optimal value in the global range. Simulation data indicate that the PSO-BFA intelligent optimization method significantly outperforms traditional swarm intelligence methods in multi-fault diagnosis and classification, achieving the peak fitting value in fewer iterations.