Radio signal counterfeiting method for counterfeiting deep learning

The invention provides a radio signal counterfeiting method for confronting deep learning, and solves the technical problem that a signal classifier based on deep learning is difficult to confront in the technical field of artificial intelligence and electronic confrontation. The method comprises the following steps: randomly generating candidate adversarial signals for modulated radio signals, taking the candidate adversarial signals as an initial parent population, and generating a signal interference scheme set through a differential evolution method based on visual limitation; and performing evaluation by adopting a radio signal classifier based on deep learning to obtain an interference signal and an adversarial signal, and completing adversarial signal counterfeiting. The forged signal and the original radio signal have extremely high similarity, a signal classifier based on deep learning is effectively resisted, and the radio signal modulation type classification precision is reduced. The counterfeited