Method for judging mistaken accelerator pressing based on competitive type neural network

The invention provides a method for judging mistaken accelerator pressing based on a competitive type neural network. According to the method, the amplitude of an accelerator sensor signal and the change rate of the amplitude within corresponding time T0 are taken as an input sample by acquiring a large number of amplitude change curve samples of the accelerator sensor signal under the normal accelerator pressing condition and the mistaken accelerator pressing condition, the competitive type neural network including two nerve cells is established for self classification, a well trained working network model is obtained, a mistaken accelerator pressing amplitude and changing rate threshold table is calculated on this basis, the current accelerator pressing amplitude and changing rate are acquired in real time during vehicle traveling, and a mistaken accelerator pressing judgment is made by comparing the current accelerator pressing amplitude and changing rate with the amplitude and changing rate threshold table. The method solves the technical problems that according to the prior art, judgment making is slow and misjudgment occurs easily, the mistaken accelerator pressing judgment can be made early and accurately, and accident probability can be reduced greatly.