Nonlinear Model Method of Microwave Power Device Based on Extreme Learning Machine

In large-signal simulations, building accurate models of power devices is essential for microwave circuit design. The traditional non-linear vector network measurement system is very complicated, and has the disadvantages of long test cycles and many measurement data. However, the powerful learning ability and accurate prediction ability of the extreme learning machine (ELM) neural network algorithm can be used to make up for the shortage of traditional measurement systems and improve the design efficiency. In this paper, an X-parameter model based on ELM neural network is established, the comparison of the model simulation with the actual circuit simulation shows that the fundamental error is within ±0.1 dBm and the second harmonic error is within ±0.3 dBm, which further verifies the correctness of the prediction model based on X-parameters of the ELM neural network.