Modeling And Optimization Of Batch Chemical Process Scheduling

Facing the competition in worldwide market and the challenge of joining the WTO, the chemical complexes in our country must make the best use of information technology in order to reduce costs, improve quality of products by responding to market changes as fast as possible, such that it can improve economic performance, remain the superiority of competence. On the basis of analysing the characteristics of production economy and management, computer integrated manufacturing system (CIMS) had been achieving good acceptances in mechanical industries. However, the extension of CIMS to the field of process industries, i.e., the computer integrated process system (CIPS) is still far from mature. CIPS should be an integrated information system, by the aid of which the process complexes may practice the modern information technics. From the macroscopic view, one of the key task of CIPS is to optimize the production schedule. Particularly, many tough problems exist in the production of chemical batch processes in comparing with the continuous ones. Therefore, the studies in this domain were extremely needed and only small progresses were made until recently. Moreover, products made by batch processes take much more share of all chemical products due to the requirements of very diversified assortments of goods in modern market. So, the studies on batch chemical processes are meaningful both in academic and economic senses.The research work of this thesis consists of the following major contribution:1. The basic theory of Petri net was discussed in this paper. As a graphic and mathematic tool, Petri net provides an integrated environment with which the discrete event dynamic systems can be modeled and analyzed. Multi-product and multi-purpose batch process model with Petri net was addressed for a review of history. However, the Petri net describing real processes usually tends to be very complex and extremely large. To circumvent this situation, some extensions such as abstract and hierarchical methods were proposed. On the basis of object-oriented concept, an uniform batch chemical process model was plotted with colored and timed Petri net. In this way, system can be simulated compactly and performance can be controlled effectively.2. Integrated modeling is one of key technics of CIPS. Hierarchical and composition operations as the two main configuration methods of Petri net were introduced in this work. On the base of object-oriented message communicationmechanism, a hierarchy constitution method of Petri net was proposed. With this method, different levels of object could be described by different kind of Petri net, and hence the hiberarchy of chemical batch scheduling model become very clear. Through case study, a category of batch chemical scheduling problem can be simplified.3. The aim of scheduling is to access more profit by optimization. Some randomized optimization algorithms were applied to chemical batch process and some good results were achieved, however they should be further improved since the degree of optimization could not be judged easily. Ant colony algorithm is a novel evolutionary algorithm originally designed for discrete optimization problem, and has good performance. It was discussed in this work and some improvements were made to match the requirement in our studies. Based upon evolutionary concept, the colony algorithm was extended for optimization in continuous space. The reinforcement learning algorithm was also introduced, since it has some relations with the colony algorithm and can be need in the problem of scheduling.4. Some new concepts and scheduling algorithms for batch chemical process were proposed in our studies. For stationary scheduling problem, probability theory was used in production scheme to avoid from time conflicts. For dynamic scheduling problem, two novel algorithms were proposed. The first one is an online scheduling algorithm which combine predictive modification algorithm and extra equipment reschedu...