کنترل یادگیری تکراری با اسمیت زمان جبرانکننده تاخیر برای فرآیندهای دسته ای

How to improve the control of batch processes is not an easy task because of modeling errors and time delays. In this work, novel iterative learning control (ILC) strategies, which can fully use previous batch control information and are attached to the existing control systems to improve tracking performance through repetition, are proposed for SISO processes which have uncertainties in modeling and time delays. The dynamics of the process are represented by transfer function plus pure time delay. The stability properties of the proposed strategies for batch processes in the presence of uncertainties in modeling and/or time delays are analyzed in the frequency domain. Sufficient conditions guaranteeing convergence of tracking error are stated and proven. Simulation and experimental examples demonstrating these methods are presented.

While continuous processing has always been considered the ideal method of operation of a chemical plant, there are a lot of low-volume and high-cost materials are obtained in batchwise form in many chemical and petroleum plants. Improved performance of batch processes is becoming necessary because of competitive markets. Even though there exists an important amount of literature referring to batch unit optimal control methods [1], [2] and [3], these methodologies are rarely part of everyday industrial practice because of imperfect modeling, unmeasured variables and time delays. As a result of these issues, the degree of automation of many batch units is still very low. The concept of iterative learning control has been presented by many researchers. Iterative learning control (ILC) provides a method for increase of control efficacy by taking advantage of the accumulation of batch-to-batch data. Many ILC algorithms have been proposed [4], [5], [6], [7], [8], [9], [10], [11] and [12]. The usual approach of the existing works is to presume a specific learning control scheme in the time domain and then to find the required convergence conditions, and most of the works focuses on only finding open-loop control signals. In practice where unexpected disturbances are unavoidable, these algorithms may fail to work. Very few results, up to now, on the ILC are for dynamics systems with time-delay. Lee et al. [13] proposed a feedback-assisted ILC for chemical batch processes, but this method is sensitive to process order and time delays, and the general stability analysis is not available. Hideg [14] investigated the possibility of divergence of an ILC for a plant with time-delay. Park et al. [15] designed an ILC algorithm for a class of linear dynamic systems with time-delay. However, only uncertainty in time-delay is considered in the existing literature. In the present study, learning control algorithms together with a Smith predictor for batch processes with time delays are proposed and analyzed in the frequency domain. The dynamics of the process are represented by transfer function plus dead time. Perfect tracking can be obtained under certain conditions. Convergence conditions of the proposed methods are stated and proven. By using the past batch control information, the proposed ILC strategies are able to gradually improve performance of the control system. These results are evaluated through simulation as well as experiments

The problem of iterative learning control of batch processes with time delays has been addressed. The processes were represented by a transfer function plus dead time. Convergence conditions were derived in the frequency domain. Perfect tracking can be obtained under certain conditions. Since the proposed ILC schemes are directly attached to the existing control systems, they can easily be applied to batch processes to improve control performance. Simulations and experimental results on temperature control of a reactor show the effectiveness of the proposed methods.