In the field of manufacturing feature recognition, the early research usually only considers that all the shapes and functional structures of casting parts are completely formed from raw materials through multiple machining. In reality, many mechanical casting parts are processed from castings, weldments or forging blanks, and the geometry and generation mode of blanks have a significant impact on the manufacturing process planning. For example, for box casting parts, the geometric model of casting blank is closely related to the geometric model of casting parts. After the design of a casting part model is completed, according to the non geometric information of the model, which casting parts of the casting part are completed by casting, machining or other processes will be determined, Both shape features and machining features depend on the original blank model. In fact, in the whole process from product design to manufacturing, when a product design model is given, the subsequent work of product production should be understood as the inverse transformation from the design model to the blank model. Therefore, how to reasonably convert the design model of casting parts into blank model must be considered.
Feature recognition and generation technology provides a technical basis for obtaining blank model from casting part design model. It expresses complete product information and contains rich semantic information. It is also the key technology of CAD / CAM integration. The problem of feature recognition from CAD model can be traced back to an early work report of kypranou at Cambridge University in 1980. In the report, it is proposed to use GT code description of casting parts for the development of automatic process planning system, In order to give the description of GT code of casting parts, it is proposed to identify some concave and tensile features from B-rep model. Since then, a lot of research work on solving the problem of feature recognition has been carried out widely. At present, many research units and scholars at home and abroad are doing research in this field.
In 1982, arbab first proposed feature decomposition modeling. Its main idea is to apply a series of Boolean subtraction operations on the blank to generate casting parts. Pratt and Wilson of Cranfield Institute of technology proposed a model for CAM-I to classify shape features according to shape and structural features; Falcidieno of the Institute of Applied Mathematics in Genoa, Italy, put forward the description method and feature recognition method of boundary model to represent feature objects, and developed the corresponding system; Beitz of Berlin Technical University in Germany developed the feature-based modeling system GEKO; Douglas et al. Studied the geometric reasoning technology of machining features by convex polyhedron decomposition method; Turner et al. Studied the establishment of tolerance feature model; Roy et al. Studied the representation and processing of dimensions and tolerances; Jaroslaw et al. Studied feature editing and query technology.
For the machining feature recognition of processing multiple processes, Boothroyd et al. Described a method to select materials and machining processes at the high level of process planning, but the scope is limited to simple casting parts. Herrmann et al. Studied the assumption of high-level process planning. From the perspective of virtual enterprise, they once again use simple geometry casting parts for ordinary casting parts and EDM casting parts. They believe that simple geometry casting parts are enough to reflect the ability to deal with general casting parts. Wang and Kim described a feature recognition method using convex body decomposition. This method recognizes shape features from the inherent geometric information of casting parts. It provides local and global geometric relationships that support multi process processing.
Dong and wozny generate volume features from face features by extending adjacent faces of casting parts. Their method only considers the feature recognition of casting parts with simple geometry, and can not be used to deal with casting parts that need to be processed through multiple processes. Xu and hinduja combine feature recognition and incremental workpiece repair technology to identify rough machining features from the intermediate state of the workpiece, but they assume that the blank or initial workpiece is provided as an additional input to the casting part model, and their method can not guarantee the correct geometric relationship between the identified features, This relationship is important to support process planning. In the field of feature recognition, the current focus has turned to the application of feature recognition technology in manufacturing.