An Iterative Learning Predictive Control Method Based on Orthogonal Parameterized LTV Model

Abstract

The invention discloses an iterative learning predictive control method based on an orthogonal parameterized LTV model for the pressure holding section of the injection molding process. The implementation steps are as follows: (1) Establishing an orthogonal parameterized LTV model; (2) Estimating model parameters; (3) Order selection; (4) Deriving the ILC‑MPC control law. The invention makes full use of the technical characteristics of repeated operation and known track tracking in the injection molding process, and changes the characteristics of high complexity and poor extrapolation of the traditional modeling method by introducing an orthogonal parameterized LTV model. In the control process, the control strategy combining iterative learning control and model predictive control can not only make full use of historical batch information, but also perform online rolling optimization according to the predictive model, so that the control requirements can be completed more quickly and stably.

Description

technical field [0001] The invention relates to the field of injection molding process, in particular to an iterative learning predictive control method for an orthogonal parameterized LTV (linear time-varying, LTV) model of the pressure-holding section of the injection molding process. Background technique [0002] Plastic products are widely used in people's daily life. Injection molding is the main method of processing plastic products. At present, 80% of plastic products in the world are produced by this process. The main equipment in the injection molding process is the injection molding machine, referred to as the injection molding machine. Injection molding is developed according to the principle of metal die-casting. It is a typical batch production process by using injection molding machines and injection molds to convert plastic raw materials into shaped products. It uses the thermophysical properties of plastics to put the material from the hopper into the barrel...

AI Technical Summary

Problems solved by technology

[0003] The holding section is an important stage that determines the quality of the finished product. Its key variable is the holding pressure. However, since there is generally no steady-state operating point in the operating range of the holding process, the process variable will fluctuate violently in a wide range.

In addition, due to factors such as melt flow rate, uneven pressure distribution, materials, process conditions and other factors, the process will show strong nonlinear and time-varying characteristics. Therefore, the linear time-invariant model can no longer fully describe the injection molding process. , and the control strategy based on the linear model does not work well. These characteristics determine that the control of the pressure-holding process is more complicated than the control of the continuous process. Therefore, it is necessary to study a nonlinear model that can better describe the pressure-holding process, and based on This model controls the process variable

[0004] At present, the nonlinear models for the pressure-holding section of the injection molding process mainly include mechanism models, fuzzy models, neural network models, multi-models, etc., but for mechanism models, it is difficult to establish a mechanism model that can fully describe the process; the fuzzy model method The disadvantage is that the design of the fuzzy model lacks systematicness, and the simple fuzzy processing of information will lead to a decrease in the control accuracy and dynamic quality of the system; the main inconvenience of the neural network model lies in the difficulty of modeling and poor extrapolation; the disadvantage of the multi-model method is that the model Time consuming to select and design

[0005] However, the injection molding process has the process characteristics of repeated operation and known tracking trajectory. The above modeling methods do not make full use of this feature to reduce the difficulty of modeling

Images