Research On The Key Issues Of Knowledge Based CAD System

Advances in artificial intelligence technologies have led to the emergence and development of Knowledge Based Engineering (KBE) techniques. KBE techniques are expected to promote the practicability and automation of computer aided design, and has great significance to improve product quality and design efficiency. Currently knowledge based product design technology has attracted much attention, and is considered as an important way to improve the application and development of computer aided design. Based on the overview of KBE, Several key issues to implement the knowledge based CAD system are studied in this dissertation, which is sponsored by the National Key Research Project of the 10th five-year-plan of China (Grant No. 2001BA201A02). Knowledge modeling of product design is one of the essential issues in knowledge based CAD system. To expedite knowledge acquisition process and to facilitate the maintenance of design knowledge, a knowledge fusion mechanism is introduced which is based on the extension of traditional feature model. Knowledge primitive that represents the basic knowledge form is employed to simplify the knowledge modeling process. Based on the extended feature, knowledge primitive feature and agglomerated feature are introduced. Knowledge primitive feature offers many merits, such as easy manipulation and suitable to construct design rationales. The design rationales are then applied to carry out complex computation, reasoning, validation, self-adjustment etc. Encapsulating design knowledge, geometry form and conjunction relationship, the agglomerated feature in system-level is generated to maintain the whole semantic elements of a function. Knowledge constraint modeling establishes a foundation to process the knowledge primitive model. To ensure the feasibility for engineering design and the consistency of constraint management, a directed bi-partite graph is proposed to model the knowledge constraints. According to characteristics of different knowledge constraints, some methods and algorithms are employed to direct, group and reduce the graph. A control strategy of reasoning and computation process can be formulated by using of solving sequence generated by knowledge constraint model. The control algorithm ensures the stability and effective control of hybrid reasoning and computation process. Knowledge repository is the indispensable component of the knowledge based CAD system. A DTS (Design task structure) oriented approach to construct a hierarchical and systematic knowledge repository is introduced. Thus the knowledge repository is more manageable and expandable. A set of knowledge primitives can be used to realize a specific problem solving method (PSM) of a design task in DTS. By using of knowledge primitive and design task, design knowledge can also be reused in multi-granule. The design task network (DTN) is employed to guide the automation of design process. On the basis of the above theoretic achievement, a knowledge based CAD prototype system has been developed which passed the project assessment. Finally, the feasibility and validity of this research is testified by a gear reducer design.