| Injection molding is a typical batch process,consisting of injection,packing,cooling stages,which can be treated as a 2-dimentional(2D)system with a time dimension and a batch dimension.The nonlinearity characteristics appear on the time dimension while the repetitive nature appear on the batch dimension.Packing is one of the important stage in the process,during which the packing pressure is the main controlled variable.In order to improve the product quality,it is critical to enhance the tracking performance of packing pressure.In this paper,time-partitioned piece-wise affine(PWA)model for packing stage of injection molding process is proposed,the main research work is as follows:First of all,we utilize the repetitive nature of batch process to simplify the general PWA model,and propose time-partitioned PWA model.Note that the identification problem has both discrete and continuous variables,the separate least-squares method is adopted to solve the problem.More clearly,the MDS method is used to optimize the integral piecewise points,and the LM algorithm is used to estimate the sub-model parameters.Simulation and experiment show the effectiveness of the algorithm.Afterward,time-partitioned PWA fusion model is proposed to avoid the hard switching of PWA model.To make it easier to identify the model,we use the linear weights of the adjacent sub-models to describe the piecewise interval.As before,the separate least-squares method is adopted to estimate the PWA fusion model.Results show the PWA fusion model is a feasible way to improve the modeling accuracy.Finally,the multi-model PID controller and iterative learning controller is design based on time-partitioned PWA fusion model.In order to make full use of the 2D information,we propose a compound controller which is composed of PID controller and iterative learning controller.Simulation and experiment show the compound controller has good tracking performance and the ability to suppress non-repetitive disturbance. |
PDF Full Text Request