Research On Predictive Self-tuning Of Speed Control Parameters For Servo System Under Polynomial Command

During operation,the servo system is easily affected by many factors such as external load variation and friction force variation,so that the original control parameters cannot maintain the best control performance.In order to improve the dynamic performance of servo system,this paper takes speed loop as the research object and puts forward two kinds of control parameter prediction and self-tuning methods.The research contents are as follows:The servo system’s dynamic models is established.A 2-degree-of-freedom N-type PI(2DOF NPI)controller is introduced,which has strong advantages in response speed and steady state error.Aiming at the generalized predictive control is restricted by reference command and the model mismatch error exists in the identification convergence process,a control parameters predictive self-tuning method under polynomial command is proposed based on extended sliding mode compensator.First,the recursive least squares identification method is used to estimate the model parameters online,which provides the basis for the design of parameter self-tuning method.Secondly,an extended prediction model containing only the velocity errors is established by means of the n-order difference operator,thus freeing GPC from the speed command form.Then,the quadratic index function is reconstructed and the rolling optimization is carried out to obtain the optimal predictive control law.The control law consists of a model term and a model-free term,wherein the model term provides control parameters for the 2DOF NPI controller,and the model-free term is used to construct an extended sliding mode compensator to weaken the velocity error spikes caused by the model mismatch.In view of the unknown modeling dynamics of the servo system,the predicted output of the model is different from the actual output,will affect the control performance of the GPC,a self-tuning method of control parameter prediction based on no model is proposed under polynomial command.First,the pseudo partial derivative algorithm is used to estimate the dynamic mapping relationship between output and input increments online,thereby transforming the servo model into a compact format dynamic linearized data model,thus avoiding the interference of unmodeled dynamics on the predicted output.Secondly,a new extended prediction model is reconstructed based on the data model,and the optimal control law obtained by rolling optimization is matched with the 2DOF NPI controller to complete the model-free parameter self-tuning process.Then,in order to ensure the stability of the system,the convergence of the system error is discussed and analyzed.Finally,through simulation comparison,the superiority and effectiveness of the above two control parameter prediction self-tuning methods are verified.