High Power Supply Signal Filtering Based on Adaptive Algorithm

In the polar-field high-power power supply system of a tokamak fusion device, key electrical signals such as voltage and current are inevitably affected by electromagnetic noise. To improve the control performance of the polar-field power supply system and reduce the number of triggering times of the protection mechanism, the electrical signals must be filtered after sampling to improve the signal-to-noise ratio of the signals. In this study, a controlled step size LMS (Least Mean Square) algorithm based on the hyperbolic tangent function (tanh) was proposed, which adjusts the weight vector by constructing the affiliation function of the step size factor to balance the relationship between the convergence speed and the steady state error. In this study, taking control of the output signal of high power supply as the target, the simulation model was constructed to verify the algorithm's convergence speed and steady-state performance, and then experimental tests were carried out to verify the algorithm's feasibility. Moreover, the algorithm's delay is controlled at the nanosecond level, which gives the algorithm good real-time performance.