Wireless signal perception model construction and perception method and system based on adversarial training

The invention discloses a wireless signal perception model construction and perception method and system based on adversarial training, and the method comprises the steps: collecting non-interference data and interference data, marking the non-interference data with an action classification label, and obtaining a data set; performing denoising and data conversion processing on the data set to obtain preprocessed interference data and interference-free data with an action classification label; training an adversarial network model by taking the interference data and the non-interference data with the action classification label as input and taking a judgment result of judging whether the interference data is interference and an action classification result as output, and obtaining a wireless signal sensing model after training is completed; and carrying out perception classification on wireless signal data to be processed by adopting the wireless signal perception model. Through adversarial training of the fea