Development of model identification methodology based on form recognition for Computer-Aided Process Planning

Abstract

Computer-Aided Process Planning (CAPP) systems have become essential in manufacturing environments to integrate the information between CAD and CAM systems, and to automatically generate the NC code from the CAD model. Though the future of these systems seems to belong to the use of Artificial Intelligence to create knowledge-based algorithms which emulate human decisions, the CAPP systems based on feature recognition and model matching, which use databases of previously known mechanical components to generate new process plans, are also a very interesting option due to their accuracy and smaller development costs. Many researchers have proposed different kind of feature recognition algorithms before. However, these algorithms are usually application-dependent and require external codes to identify the features of wireframe models. This paper proposes a new methodology for shape recognition and model matching stages which improves the accuracy of the recognition tasks, uses solid models instead of wireframe models and can be successfully applied to any kind of part. The methodology is based on an original coding system that links the geometric information extracted from the CAD model with the features of the part by means of an identification sequence which is detailed in the text. Also, a score system has been created for the model matching stage. The obtained results show that the system presents high accuracy in shape recognition, feature identification and model matching tasks, even when the analyzed part is similar to the ones in the database. In addition, quantitative geometric data is also extracted from the CAD model on behalf of future steps of the CAPP system, such as the NC code generation stage. In contrast to other systems, this methodology can be easily applied to the industry since it makes use of the CAD model only.