Parameter Identification and Fault Diagnosis of Servo System Based on Improved PSO Algorithm

The capture of nonlinearity in servo systems during operation is significant for fault diagnosis, which can greatly reduce the probability of system failure throughout the life cycle. To address the problem, a step-by-step identification method based on iterative optimization mutation particle swarm optimization (S-IPV-PSO) is proposed for fault diagnosis. Firstly, an iterative optimization mutation particle swarm optimization (IPV-PSO) is presented to improve convergence speed and algorithm accuracy by optimizing the learning factor and inertia weight coefficient. Secondly, a Hammerstein model is established to identify the parameters for predicting different states of the system by using the IPV-PSO algorithm, where inverse M signal is used as an output to separate between the nonlinear and linear parts of the system. Finally, a servo system simulation with dead time characteristics is built to verify the effectiveness of the proposed algorithm. Extensive experiments demonstrate that the proposed method can effectively identify fault states and has better convergence speed and recognition error.