Integrating an Expert System And a Neural Network for Process Planning

This paper presents a computer aided process planner for metal furniture assembly, welding and painting using a rule based expert system integrated with an artificial neural network. The if/then rules create parts lists and process plans, while the neural network estimates standard processing times for individual product variations. Although essentially a variant process planner, the rules and neural network allows some generalization capability to new products. This development effort demonstrated that a composite intelligent approach can be useful for process planning in a real manufacturing situation. Accepted to Engineering Design and Automation October 1995 1 Part of this work was supported by the National Science Foundation under grant DDM-9209424. 2 Corresponding author. 2 Integrating an Expert System And a Neural Network for Process Planning Mark Wilhelm, Alice E. Smith and Bopaya Bidanda Department of Industrial Engineering University of Pittsburgh 1048 Benedum Hall Pittsburgh, Pennsylvania 15261 USA Abstract. This paper presents a computer aided process planner for metal furniture assembly, welding and painting using a rule based expert system integrated with an artificial neural network. The if/then rules create parts lists and process plans, while the neural network estimates standard processing times for individual product variations. Although essentially a variant process planner, the rules and neural network allows some generalization capability to new products. This development effort demonstrated that a composite intelligent approach can be useful for process planning in a real manufacturing situation. This paper presents a computer aided process planner for metal furniture assembly, welding and painting using a rule based expert system integrated with an artificial neural network. The if/then rules create parts lists and process plans, while the neural network estimates standard processing times for individual product variations. Although essentially a variant process planner, the rules and neural network allows some generalization capability to new products. This development effort demonstrated that a composite intelligent approach can be useful for process planning in a real manufacturing situation.