Optimization-Based Production Scheduling For Large Enterprises

Xiaohong Guan

2008-01-01

Abstract:Production scheduling of many industrial systems with complicated operating dynamics and constraints in large enterprises such as electric power generation, batch chemical process, etc is very important with significant economic impact. In this speech, a new method is presented to solve the scheduling problem of a class of production systems with hybrid dynamics and constraints. Within Lagrangian relaxation framework, the exact optimal solutions to the subproblems are efficiently obtained without discretizing the continuous production levels or introducing intermediate levels of relaxation. A novel definition of the discrete state associated with a consecutive time period is introduced so that solving each subproblem is decomposed into solving a continuous and a discrete optimization problem separately. The optimality principle is applicable for both continuous and discrete problems. A double dynamic programming method is developed to solve the entire subproblem. To deal with the issues caused by homogenous subproblems, the successive subproblem solving (SSS) method is presented in this speech. With the introduction of the convex penalty terms associated with the system constraints, individual subproblems are solved successively to obtain a proper surrogate subgradient direction for the high level dual problem. In this way, the solutions to the homogenous subproblems can be differentiated in the dual solution. More dual solution patterns can be generated by the SSS method than by the standard LR method, and it is possible to modify the dual solution into better feasible schedules. The testing results for the practical problems of power generation scheduling and generation resource bidding demonstrate that the methods presented in the speech are efficient and effective.