论文部分内容阅读
为求解车间作业调度问题(JSSP),本文提出了一种新颖的多小组协同学习的教学算法,实现小组间学习的协同及基于学习能力的深度和广度搜索策略。针对JSSP问题因其复杂度较高容易导致算法陷入局部最优的不足,引入学习小组协同学习,通过组内学习和组内交流,使学习过程跳出当前的局限。为了兼顾局部和全局搜索能力,引入基于学习能力的深度和广度搜索策略,小组内学生按照学习能力强弱进行学习,较优的学生进行深度的学习,较差的学生进行广度的学习。最后,通过对OR-Library中的标准仿真实例进行实验,结果表明本文所提的教学算法在JSSP问题上的收敛精度和搜索能力均得到了有效的提高。
In order to solve the job shop scheduling problem (JSSP), this paper proposes a novel multi-group collaborative learning teaching algorithm to achieve inter-group learning and depth-based and breadth-based search strategies based on learning ability. Due to the complexity of JSSP, the problem of JSSP is easy to cause the algorithm to fall into the local optimum. The study group is introduced into collaborative learning to make the learning process out of the current limitations through intra-group learning and intra-group exchange. In order to take both local and global search capabilities into consideration, a deep and breadth-based search strategy based on learning ability was introduced. The in-group students studied according to their ability to learn, the better students conducted in-depth learning, and the poorer students conducted in-depth learning. Finally, through the experiment on the standard simulation examples in OR-Library, the result shows that the teaching algorithm proposed in this paper can effectively improve the convergence precision and search ability of JSSP.