Research on fault diagnosis technology of inverter based on associative memory neural network

A modified associative memory model of artificial neural network (ANN) is introduced into the fault diagnosis of inverter. Firstly, the diagnosis algorithm is derived and the fault table is concluded. Secondly, the fault sample vectors are preprocessed by HADAMARD transform, and learned by the model. Finally, the samples and non-samples are parallel associatively recalled based on this network. Furthermore, the digital simulation results demonstrate that the novel strategy is reasonable. Additionally, a hardware implementation thought is presented, which uses large-scaled field programmable gate array (FPGA) circuits.