基于软计算技术的实时调度控制系统理论及工具软件的开发

Modern manufacturing systems became more and more complicated and the automation was applied popularly in this field. It is urgent to solve the dynamic scheduling problems for the complicated production line.Based on Windows mechanism, a new dynamic scheduling structure is proposed here. In accords with this structure, performance and evaluation criterion of the dynamic scheduling have been dealt with.For real-time dynamic scheduling, several methods such as knowledge discovery based scheduling, predictive control based scheduling, neuro-dynamic programming based scheduling, and re-entrant line scheduling heuristics were studied in details. Those scientific methods introduced to traditional experience based research field promoted dynamic scheduling into a new level..Planning and collaborative problems among different production procedures were another important issue discussed. The model of the supply chain between the cold rolling and hot rolling was established, and dynamic Game Theory was applied for planning negotiation. A Lagrangian Relaxation algorithm was suggested for integrated production scheduling of steelmaking, continuous-casting and continuous-rolling. Finally, a simulation platform for scheduling knowledge acquisition is designed and developed..Scheduling is also very important in real-time system. The method to construct the FF schedule time are studied. The extended compact mode algorithm based on job rate-monotonic is presented to meet time constraints and precedence constraints of remote periodic message and function blocks.The former dynamic scheduling mostly depends on the experience or heuristics. The mathematic optimization can solve the static scheduling problem perfectly, however, it doesn't adapt to the dynamic scheduling problems. The proposed methods here were closely dependent on computing science greatly promote scientific progress of dynamic scheduling. In general a new theoretic foundation of scheduling for industrial production has been established, which expands contents of traditional scheduling research.

运用人工神经网络、遗传算法、进化计算等软计算技术，以及机器学习和仿真技术，研究实时调度控制系统理论和方法，并开发相应的软件工具包，为现代制造企业生产过程开发高效率的调度指挥系统，提供完整、实用的理论、方法和计算机辅助工具。.该项研究包括1)根据调度案列，应用人工神经网络、ID3等机器学习算法为代表的数据挖掘技术，获取调度策缘乃惴ㄓ肜砺垩芯浚?) 基于遗传算法的调度控制器设计（实验）方法学研究。.