Reconfigurable Control in Discrete Event Dynamic Systems Applied to Manufacturing Systems

Abstract

Production management in an automated manufacturing system entails the implementation of the following two decision functions: Operational Planning and Resource Allocation, and Production Control. In this work we present scheduling and production control algorithms for manufacturing systems. We present a general manufacturing system model and formalize the concept of a schedule as a single sequence of operations. Using a general performance measure we formulate the operations scheduling problem as a combinatorial optimization problem. A simulated annealing based optimization algorithm is developed for this job shop scheduling problem. We present a three level hierarchical on-line reconfigurable control scheme. For the "process" level control, we present a reactive operations scheduling scheme. Based on the control specification at the process level, the lower "operation" level parameters are defined. We present operation control algorithms for both continuous and batch mode processing. Using dynamic programming principles, we present a Quasi-Variational Inequality based impulse control algorithm for online control of processing rates for a single batch. Perturbation Analysis, in conjunction with stochastic approximation techniques, is used for continuous mode, online processing rate control. Algorithm extensions to other discrete event systems are also discussed.