Multi-Agent Based Dynamic Scheduling For The Process Of Oil Refining

Nowadays, modern production corporations are facing with many severe problems which include austerity circumstances, various production systems and complex produce process. So, traditional production approaches cannot satisfy the requirements of the changing markets. In addition, the high production requirements of the consumers make the production more diverse and individual. However, at the same time the lifecycle of the production becomes shorter. These factors requiring the production systems cannot only finish the tasks within short time with high quality, but also have the ability to solve the emergency quickly, so as to correspond to the changing markets and to prepare to get the opportunities and face to the challenges effectively. Therefore, the research on the production scheduling, especially in the dynamic scheduling area is important and urgent.Oil refining industry is the main body of the nation-owned enterprises. It is located on the top of the production industry. Its production value is account for 1/3 of the country economic value. Oil refining industry is the mainstay industry of the country economics and it impels the development of the other industries. The whole optimization and the comprehensive utilization of the resource and energy, not only can impel the production to be safe, environmental-protected, energy saving and consumption deducted, enhance the economics benefit and social benefits, but can ensure the realization of the economy sustainable development. Therefore, this thesis mainly does research on the dynamic scheduling for the process of oil refining.The contributions are summarized as follows:Firstly, based on the analysis of the background and meaning of the oil refining process, some problems in the dynamic production scheduling for the oil refining process are pointed out. By analyzing the current research status of the dynamic scheduling, the different research approaches of dynamic scheduling are listed and compared. The research approach of this thesis is based on Multi-Agent System. Secondly, by analyzing the structure of the oil refining process, production requests, production objects and specific production process of ShengLi oil refining factory, China Petroleum and Chemical Corporation, some principles for simplify the equipments are given, and according to which the oil refining system is simplified effectively. This makes the system can almost characterize the original manufacture process. Based on the simplified system, a Multi-Agent model is built, and then each kind of Agent is described.Thirdly, the approach for defining the products flow proportion distribution from different direction is proposed, combined it with all of the restrictions in the production process, the goal of reasonable distribution and fully use of the equipments is gained. After the dynamic events, the Agent negotiation is used to solve the new products flow proportion rapidly, to guarantee the production process achieves a relatively stable state in a short time.Finally, the results of the algorithm are compared with the results from grouping heredity algorithm and particle swarm algorithm.