Intelligent Fault Diagnosis of Inverter Based on Improved PSO Algorithm

In order to overcome drawbacks of the BP network, an improved non-linear dynamic adaptive particle swarm algorithm is used in this paper. The value of inertia weight can automatically change with the change of fitness value for avoiding particles premature. The particle velocity avoids getting into local minima by adding a negative disturbance. This method is applied to the inverter circuit. Simulation results indicated that: improved particle swarm algorithm was superior to BP algorithm in accuracy, convergence speed, ability to search the optimal solution.