Learning The Best Scheduling Policies In Manufacturing

This research aims to propose a framework for the integration of dynamic programming and machine learning techniques (e.g., neural networks) to take advantage of learning procedures that self-adjust to meet goals for scheduling in a manufacturing environment. Our proposal contributes by understanding the decision process for scheduling optimization and allows the learning of good scheduling policies. Furthermore, the proposed system is designed to achieve learning using this hybrid modeling approach and with the use of signals of the environment measure the achieved state or goal and calculate the performance criteria. An example illustrates how the learning mechanisms allow the system to adjust itself to new situations. In addition, different paradigms in reinforcement learning such as Q-Learning is discussed