The Research Of Electro-hydraulic Hybrid Injection Molding Machine's Intelligent Control Algorithm

Along with the increasing demand towards plastic products as well as products' quality,the requirement towards control precision and speed of moving mold of injection molding machine is improved as well.The electro-hydraulic servo control system for electric-hydraulic injection molding machine has intrinsic properties as nonlinearity,time-variability and uncertainty etc.,which may also be influenced by interference signal as noise and bursts.Therefore,the study of appropriate software control method is one of the significant ways to improve system's efficiency,control accuracy and stability.In recent years,the intelligence algorithm has been developed quickly,which now has been penetrated into all walks of life.Therefore,the application of its idea into process control of injection molding machine's repetitive exercise has very important practical significance.This paper carries out research with design projects of D200 type electric-hydraulic injection molding machine control system in an embedded intelligent control laboratory of Hangzhou Dianzi University.The concrete research context is as follows:First,the feature and existing problems in electric-hydraulic injection molding machine which integrated advantages of all-electric control with all hydraulic control was analyzed.Then,this paper advocated the application of fine nonlinear approximation capability,simple and rapid RBF neural network as well as classical portfolio model of digital incremental PID algorithm into motion control of electric-hydraulic injection molding machine and thus realize on-line identification through this model as well as adaptively adjust three parameters of PID algorithm.Meanwhile,adaptive learning rate has been putting forward in this paper in order to improve convergence performance of gradient descent and the superiority has been verified through experiment;besides,this paper introduces into sensitivity critical value so as to avoid PID parameter adjustment caused by external disturbances;what's more,in order to increase the algorithm speed,one judging condition has been added in the algorithm flow and when identification error does not exceed critical limits,people don't have to update RBF neural network and PID algorithm respectively.At last,combined with genetic algorithm,this paper presents a G-PSO algorithm with both fine global optimization capability and high convergence rate,of which the optimization performance has been analyzed.The purpose of it is to looking for optimal initial parameters of improved RBF-PID model and then realizes on-line self-adaptation control with initialized RBF-PID model.This control model of two phases can realize self-adaptation control with no model and entirely online;besides,it can enhance the control precision,quick response,and track expectation curve commendably;apart from this,it can suppress the influences of varying factors including oil leaking and external disturbance etc.on the controlling results of velocity and pressure during injection molding process,so that the system is kept in a stable state and the production is ensured smooth and safe.In the end,the feasibility and advantage of this composite pattern is verified through experiment on simulation model so as to lay a more solid theoretical basis for the application on actual machine in the next step.