Abstract

The simulation method is used to obtain approximate solutions to operations planning problems. Its power comes from the ability to model complex systems effectively. This property makes simulation preferable in practice over programming methods which provide exact solutions but are limited in the range of problems they can be applied. Flexibility of simulation modeling is desirable especially in resource constrained operations planning problems. When internal decision mechanisms like heuristic methods are added to the scope of simulation, performance can be enhanced significantly. As a result of application of neural networks and fuzzy associative memories to operations planning through the context of simulation, we expect to improve the plan quality in terms of accurate projection of schedule, cost and performance over the project duration. For this purpose, we partition the general system model into three parts. We implemented the functions of the activity network, the first part, using a neural network. Our analysis of the proposed network predicted hundred percent recall success, and we verified this through experiments. The second part which is to resolve resource constraints is implemented with a fuzzy associative memory. This approach allows us to use multiple heuristics to obtain the best results in plan quality. Our work is based on the success of earlier studies of multi-heuristic techniques in solving resource constrained operations planning problems. The rest of the process is taken care of by the third part of simulation model which includes everything that is not included in activity network part or in determining the priorities using heuristics. This part allocates the available resources to pending activities in the order of priorities assigned by the FAM, until the skyline of resources are reached. It puts the activities which did not receive service from a resource to hold status to evaluate its request again when resources become available.

Date of this Version

December 1992

Share

COinS