Research On STEP-based 3D Solid Manufacturing Feature Recognition

1. PrefaceThe integration of Computer Aided Design (CAD), Computer Aided Manufacturing (CAM) and Computer Aided Process Planning has drawn significant attention in the recent years with the development of faster computing power tools. However, the actual integration between CAD and CAM for the downstream applications such as process planning, can be achieved only when the manufacturing information is obtained directly from 3D solid model and automate the process planning functions. Currently, most CAD systems store product information in the form of geometry and topology, so the CAPP system and follow-up system couldn't understand the information of product. This automatic extraction and recognition of features from CAD systems can best be facilitated by CAD systems capable of generating the product geometry based on features and play an important role in the integration of CAD, CAM and CAPP.Standard for the Exchange of Product Model Data (STEP) is an international standard designed to provide a complete, unambiguous, computer-readable definition of the physical and functional characteristics of a product throughout its lifecycle. It can make the data of product communicate and exchange in different computer system and CAD systems. The EXPRESS is a powerful formal data describing language in STEP which almost can describe all objects and actions,and these objects and actions can be mapped into some programming language such as C/C++. STEP/AP203 is mainly on part data during part design, which provides the data structure of part-exchange and structure-identification. STEP neutral file is constructed based on STEP/AP203. STEP neutral file describes entity according to specific expression format which is carried out by the mutual reference among entities and operator related. STEP provides the interface for product related data exchange and share, which also guarantees the further development to CIM. With the development of CIM, the platform based on STEP can provide some functions such as 3D solid model browsing, finite element analysis for 3D solid and so on. In this paper, an efficient approach for the integration of CAD/CAPP/CAM -- Technology of 3D solid manufaturing feature recognition based on STEP is presented.2. Extraction information from STEP neutral fileThe traditional application of STEP is ST-Developer and SDAI provided by STEP Tools, which is available for large manufacturing system that needs a great deal of planning information such as CAPP. SDAI is an interface, which provides an efficient way for STEP software but cost highly. In this paper, an alternative way for STEP application is presented, which starts with the STEP neutral file directly. This paper presents a method for 3D solid information extraction from STEP neutral file. We also deeply research the EXPRESS, C++ and mapping between them, construct partly C++-data-structure of product, design data dictionary and Lexical analyzer, and construct the storage-working-form. The process of extraction feature from 3D model includes three parts, reading the STEP neutral file in internal storage in form of string firstly,then extracting basic geometric element (point, edge, loop, face) through data dictionary and lexical analyzer, finally constructing the storage-working-form of STEP neutral file.3. The recognition of manufacturing featureThe work of recognizing manufacturing feature started in CAD center of Cambridge University in 1970s. In 1975, Dr Grayer of this CAD center first tried to extract the meaningful geometrical shape (such as cavity of a mould) for calculation the cutter path from the solid model automatically, and calculate the tool path of the part based on the extracted features. In 1980, Kyprianou of this center proposed the idea of recognizing feature, and prepared the foundation for recognizing feature from B-rep part model. From then on, academia and industrial community gave prominence to the concept of recognizing feature and the technology of recognizing feature, the research work on recognizing feature started widely.This paper employs the approach of boundary matching which is based on attributed adjacency graph. First of all, in order to simplify part model, we recognize and suppress the round feature; secondly, we classify the edges as concave or convex edge, which is relative with the normal vectors of two faces and the edge direction; finally, we construct attributed adjacency graph and recognize manufacturing feature. The round features are classified in two categories: ring-type round and disc-type round, the different round features are suppressed with different approaches. The method of boundary matching could recognize simple features, but could not recognize intersecting features and complex features.In this paper, neural network is implemented for feature recognition. The intelligence and parallelism are analyzed. The attributed adjacency graph is decomposed into sub-graph using a set of heuristics which is converted into the representation vector of neural network by using approach of questions. The representation vector is regarded as input for the neural network to train the network which can recognize the features. This approach can recognize interacting feature and complex feature. If new manufacture feature appears, it only needs to increase the number of input neurons which match to the new manufacture feature and train the neural network, so the neural network is much flexible. Neural network for feature recognition is a new and effective approach.