Depth deterministic strategy gradient control method of AC-DC converter

The invention relates to the field of AC-DC converters, in particular to a depth deterministic strategy gradient control method for an AC-DC converter, and solves the problem that the AC-DC converter does not have self-learning and self-perfection for different environments. According to the method, a depth deterministic strategy gradient method is adopted, a depth deterministic strategy gradient algorithm based on an action evaluation neural network architecture is designed according to the input and output states of the AC-DC converter, continuous self-correction and evolution are learned online according to system data generated in the decision making process, and decision making behaviors are evaluated. The decision precision of the output action is improved by maximizing rewards, the adaptive capacity of the AC-DC converter to deal with new problems is enhanced, the stability of the output voltage is ensured, and the self-adaption and self-stability of the AC-DC converter are realized. The method is suit