基于深度确信策略梯度的电动汽车异步电机参数标定方法

An Electric Vehicle Induction Motor Parameters Calibration Method Based on Deep Deterministic Policy Gradient

针对传统电动汽车异步电机参数标定任务中存在参数标定精度低、人工工作量大的问题,研究一种深度确信策略梯度的电动汽车异步电机参数标定方法,给出基于深度确信策略梯度的电动汽车异步电机标定算法框架,并通过实验对该方法的有效性和可行性进行了验证。可以发现,基于深度确信策略梯度的电动汽车异步电机参数标定方法具有如下特点:①标定的参数并非电机实际物理模型参数,而是使电机在任意给定转速及电流下均能够输出最优转矩的参数,因此更加适用于电动汽车技术领域;②它是一种端到端的方法,即所有工作全部由算法自动完成,无需人工辅助操作,大大减少了标定工程师的工作量。

Most existing calibration methods of electric vehicle induction motor parameters have the problems of low calibrated accuracy and large workload.Therefore,a method for calibrating electric vehicle induction motor parameters based on deep deterministic policy gradient is proposed,and the framework of algorithm for induction motor parameter calibration task is illustrated.The experimental results verify the effectiveness and feasibility of the proposed method.The proposed method has the following advantages:①the parameters calibrated by the proposed method are not the real modeling parameters,but the ones that enable the motor to run at the optimal torque in any given speed and current,which is more suitable for electric vehicle application;②the proposed method is an end-to-end method,that is,all work is automatically completed by computers without manual assistance,thereby greatly reducing the workload of the calibration engineers.