Probability Based Stochastic Iterative Learning Control for Batch Processes With Actuator Faults

Limin Wang,Bingyun Li

doi:10.1109/access.2019.2943699

Limin Wang, Bingyun Li

Open Access

https://doi.org/10.1109/access.2019.2943699

Copy DOI

Abstract

This paper proposes a new stochastic composite iterative learning control for batch processes with actuator faults that happen with a certain kind of probability. The main contribution lies in the fact that stochastic control theory is used to analyze the stability issues and consider controller designs of batch processes with actuator faults that happen under various kinds of probability, which paves the way for reliable control theory. Firstly, the batch process is represented as a system with stochastic uncertainty for iterative learning control design; and with the state and output error, it is further treated into a two dimensions (2D) Rosser system with stochastic errors. Using such transformation, the iterative learning control design is transformed into the updating control law design. Secondly, 2D Lyapunov stochastic theory is used to design the controller with actuator faults under various kinds of probability, together with the stability results of the control system in terms of LMI conditions. Finally, the proposed method is tested on the injection molding process to demonstrate the effectiveness.

Highlights

The batch process is a mass production of operation through sequential operations, and is widely used in fine chemical engineering, pharmaceutical production, biological products, modern agriculture and other fields [1], [2]
In order to adapt to the variety and quality requirements of the market together with the trend of industrial production flexibility, batch production processes have attracted more and more attention, many new upsurges have appeared in the study of batch process optimization and advanced control
Since the probability of failure in time direction and batch direction is different, the probability of failure on both axes should be calculated based on the difference between time and batch direction. Under this probability, according to the characteristics of batch process, the stochastic iterative learning control law has been designed to achieve the stochastic stability of the batch process that has a random failure with a certain probability

Summary

INTRODUCTION

The batch process is a mass production of operation through sequential operations, and is widely used in fine chemical engineering, pharmaceutical production, biological products, modern agriculture and other fields [1], [2]. Of batch processes has mature, and it has been shown that pure ILC has some inherent disadvantages It cannot be used in systems which initial values are uncertain, non-minimum phase systems and uncertain timedelay systems [14]. This paper designs a new control algorithm, which can determine the type of controllers according to the probability of failure in the system. Since the probability of failure in time direction and batch direction is different, the probability of failure on both axes should be calculated based on the difference between time and batch direction Under this probability, according to the characteristics of batch process, the stochastic iterative learning control law has been designed to achieve the stochastic stability of the batch process that has a random failure with a certain probability. The packing-holding pressure in the injection molding is taken as an example to verify the effectiveness and practical value of the above-mentioned ideas

PROBLEM FORMULATION

CASE STUDY

CONCLUSION