The high similarity between active deceptive jamming and echo of air surveillance radar results in the difficulty in identificating the active deception jamming. Based on jammer′s power amplifier nonlinear characters, a new active deceptive jamming identification method is proposed by extracting piecewise autocorrelation maximum variance and singular value spectral entropy of radar receiving signal. Simulation shows that the recognition probability for deceptive jamming is 100% with signal-to-noise ratio of 8 dB, meeting the actual engineering needs.