Electric servo position feedback dynamic tuning method based on deep reinforcement learning in semi-closed loop scene

The invention discloses an electric servo position feedback dynamic tuning method based on deep reinforcement learning in a semi-closed loop scene, and the method aims at a semi-closed loop control scene in which only motor feedback signals such as the rotor angle of a permanent magnet synchronous motor can be measured, but the actual position of a load mechanism cannot be measured. In consideration of high-order nonlinear characteristics of a load model, a traditional PID three-loop controller is used as a basic control method under a permanent magnet synchronous motor FOC control framework, and a double-delay depth determination strategy gradient algorithm is used to train an adjustment and optimization strategy network, so that the feedback quantity of the permanent magnet synchronous motor is observed, and an adjustment and optimization value of a feedback position of a position loop is output. The control precision and the response speed of traditional PID three-loop control in the face of a high-order n