For improving the accuracy of IGBT fault diagnosis in H-bridge power modules

a particle swarm optimization least squares support vector machine is proposed for IGBT fault diagnosis in H-bridge power modules. The signal to be collected in the power module is analyzed

and the capacitive voltage at the DC side of H-bridge is used as original signal of the fault. The fault features are extracted by wavelet multi-resolution. The PSO algorithm is used to optimize the kernel function and the regularization parameter in the least squares support vector machine. It is shown by the simulation experiment that

compared with the default parameters least squares support vector machine

particle swarm optimization support vector machine and genetic algorithm least squares support vector machine

have such advantages as high diagnostic accuracy and short diagnosis time as well as good practicability.