Fault diagnosis of analog circuit based on SFO optimized SELM

In order to improve the fault diagnosis accuracy of analog circuits and solve the problem of difficult selection of network hidden layer parameters, a fault diagnosis method for analog circuits based on stacked extreme learning machine (SELM) optimized by sailfish algorithm (SFO) was proposed. The SELM network was formed by training the extreme learning machine autoencoder (ELM-AE). ELM-AE has strong characterization capabilities, but the randomization of its hidden layer parameters will lead part of the effective feature information of the data itself to the loss, and produce some training errors. However, SFO has the characteristics of fast convergence speed and high optimization accuracy. Therefore, SFO is used to optimize the network parameters of SELM, to make that SELM have stronger generalization ability. Finally, the two-stage four-op-amp biquad low-pass filter circuit was used as a simulated experimental circuit, and further compared with the SELM optimized by genetic algorithm (GA) and particle swarm optimization (PSO), the experimental results show that SFO has strong optimization ability and can accurately diagnose the faults, it proved the feasibility of the algorithm.