Hybrid neuro-fuzzy approach to the generation of measuring points for knowledge-based inspection planning

Abstract

A knowledge-based inspection planning system is presented that can generate effective and consistent inspection plans automatically. The knowledge-based inspection planning system integrates part geometry information, tolerance information and heuristic knowledge of experienced inspection planners to determine the numbers and positions of measurement points. The system receives the tolerance information from users and stores it in the common database with 3D CAD geometry. A set of fuzzy rules and membership functions is automatically extracted from historic learning data using a hybrid neuro-fuzzy method. After the fuzzy rules are generated by the hybrid neuro-fuzzy model, a genetic algorithm is applied to optimize the weight parameters to find the best values for the constants. The proposed knowledge-based inspection planning system provides the stable and consistent inspection plan by removing the subjectivity of a human planner.