Automobile rapid acceleration deviation compensation method based on learning

The invention relates to the technical field of automobile design, in particular to an automobile rapid acceleration deviation compensation method based on learning. A learning module judges whether a deviation compensation learning condition is met or not, and if the learning condition is met, a learning working condition is recognized and a self-adaptive value is calculated; the verification module is used for judging whether the adaptive value obtained through learning is effective or not according to the average value of the TBT and verifying the learning working condition, and the TBT is the steering wheel torque; and the calculation module is used for calculating the motor torque required for generating torque steering according to the self-adaptive value and the vehicle acceleration, and correcting according to the vehicle speed and the steering angle. Compared with the prior art, the method comprises a learning module, a verification module and a calculation module, the learning module judges whether