【導讀】一個設計巧妙的CAPP系統(tǒng)能在CAD和CAM之間架起相互聯系的橋梁。同時，現代的計算機集成制造系統(tǒng)要求CAPP系統(tǒng)。在實際運用中具有良好的可擴展性和柔性。建工業(yè)CAPP系統(tǒng)是非常困難的。為了克服上述的缺點，本文提供了一種新的協同代。它具有以下優(yōu)點：良好的獨立性、柔性、通用性、模塊性和可分。依據這框架我們特別設計了這種協同過程計劃系統(tǒng)即機械協同CAPP來驗證上。它的系統(tǒng)模擬、代理結構設計、協作、協同和協同CAPP系統(tǒng)的結構研究將。近幾十年來，計算機技術的快速發(fā)展極大的促進了CAPP. 一般來說，CAPP的實現方法有兩種：差異法和生成法。一些標準或成熟的CAPP系統(tǒng)都被使用了成組集成技術并且儲存在一個數。多早期的CAPP工具都是根據不同的過程計劃方法被分類處理的。·獨立性意味著CAPP系統(tǒng)是作為一個獨立的系統(tǒng)發(fā)展。圖一給出了機械協同CAPP系統(tǒng)在CAD/CAP/CAM環(huán)境下的總體結構。息反饋給設計部門或其他相關部門。如此反復，最后，切實可行的替代方案就會傳遞

　　

【正文】 ironment of heterogeneous and scalable architecture suitable to solving different problems. 2. Process planning problem A machining process generally involves many machine tools, operations, fixtures, and cutting tools. Its planning requires knowledge from diversified fields. Generally, a machining process planning includes the following parts: 175。 feature recognition。 175。 machining operation selection。 175。 machine selection。 175。 cutting tool selection。 175。 fixture selection and design。 175。 sequencing operation and cost estimate. The feature recognition part identities manufacturing features from the product design data. The machining operation selection part selects the relevant machining operation according to the feature characteristics and the manufacturing environment. The required machine equipment is selected for implementing the selected operations after considering the nature of the parts and the machine processing capabilities such as the working volume, accuracy, power, fixturing, and other functions. The fixture selection part chooses the fixtures according to the part geometric shapes and dimensions as well as manufacturing features. The main concerns of the cutting tool selection include the tool types, materials, shapes, and tool dimensions. 3. The proposed method From the technological viewpoint, CAPP is still a very plex and difficult problem. Many research efforts have focused on CAPP system development, using different methodologies and strategies. However, most systems are developed by using standalone expert systems. Due to the plexity of CAPP, such a system structure is hardly able to solve the problems normally found in the manufacturing industry. A cooperative CAPP framework is proposed to reduce the limitation of currently available CAPP systems. In particular, it highlights the requirements that a modern CAPP system should meet in order to facilitate practical development: flexibility, modularity, interoperability, autonomy, and scalability. 175。 Autonomy means that the CAPP system is developed as independent system. Once developed, it can readily be integrated into the CAD/CAM system. Each agent is also treated as an independent and autonomous system. 175。 Flexibility permits new technologies and new methods to be easily added into the CAPP system. 175。 Interoperability permits multiple heterogeneous machines or approaches to work smoothly together in solving problems. 175。 Modularity enables the CAPP to function as an integration of multiple subsystems. 175。 Scalability offers the ability to scale the CAPP system architecture according to the user’s transaction requirements. The design of the CoCAPP framework is discussed with relation to a machining process planning(Machining CoCAPP). Its structure, implementation, and its application case study are detailed in this paper. . Overview of Machining CoCAPP The overall structure of the Machining CoCAPP system in the integration environment of CAD/CAPP/CAM is shown in Fig. 1. The following six process planning agents (Pagents)are in general included in the implementation. 175。 feature recognition agent。 175。 machining operation selection agent。 175。 machine selection agent。 175。 cutting tool selection agent。 175。 fixture selection and design agent。 175。 sequencing operation and cost estimate agent. Each agent may run in a different puter connected to the inter. The Bagent supplies the global state information of the problem and monitors the operational depend encies among all the individual Pagents. Once there is a problem, the Bagent will notify all registered Pagents for actions. The results from the Pagents will be posted on the B agent. The Bagent maintains the consistency and integrity of the decision space within the given product design data are sent to the CoCAPP system through the D agent from the puteraided design system. The CoCAPP system gets the production constraints from the scheduling/ shopfloor system. The CoCAPP system generates the process plans according to the product design data and production constraints. If no accept able results can be obtained, the CoCAPP system will feed the conflict information back to the design department or other relevant departments. The feasible process plan alter natives will eventually be transmitted to the scheduling/shopfloor system for scheduling. . Agent infrastructure The general structure of process planning agents(Pagent) is shown in Fig. 2. The agent is posed of four parts: agent controller, inference engine, functional adapters, and application libraries. The configuration file is used to construct the agent. The rules and facts form the application libraries. The agent controller, rulebased engine, inter adapter, file adapter, keyboard adapter, information view adapter, and schedule adapter are the monly used ponents for all the Pagents. Other adapters are also shown in Fig2. The solver adapter is a very important adapter in the Pagents because it is used to acplish proposal generation, conflict resolution and proposal evaluation of the process planning. In order to utilize the knowledge of each domain in the bination of produc tion rules and objects, the adapter is differently implemented for different agents. It must be specifically designed to deal with the knowledge of objectoriented description. The database adapter is used to store data useful to the Bagent and Pagents, such as problem definitions proposals, conflicts, evaluations, solutions, etc. Because KQML [17] is the most monly used language for munication among agentbased programs, particularly when they are autonomous and asynchronous, the Co CAPP system has chosen a KQMLbased munication protocol as a munication language used by each agent. The NetKQML adapter is used to municate with the Bagent According to