【正文】 transfer to other moulding personnel. Since the nature of their heuristic knowledge and experience is fragile and not well structured, it is not acquired easily or represented well in an expert system. Simulation is another approach for determining moulding parameters. Various mercial simulation packages such as CFlow from Advanced CAE Technology Inc. and MoldFlow from MoldFlow Australia are available. However, owing to the plexity of the injection moulding process and difficulties in obtaining an accurate rheological description of the actual material being processed, a general prehensive mathematical model giving a satisfactory picture of the process is not yet available.2. Combining a BlackboardBased Expert System and a CaseBased Reasoning ApproachThe literature review shows that an effective architecture of expert systems is required to manage the heterogeneous knowledge base. Various strategies and architectures have been proposed to implement such a design procedure, where more than one domain expert is involved. One such popular architecture is the blackboard, which simulates contradictions arising and the arriving at promises to resolve the contradictions, when experts representing different domains interact with each other to arrive at an acceptable solution. A detailed description of the blackboard architecture can be found in the literature [8,9].Blackboard architecture has been adopted in organising expert systems for concurrent engineering [10,11], part design [12,13],and handling engineering design data [14]. The literature review has also revealed that a search for other techniques which could capture the experience in setting injection moulding parameters is necessary. In this paper, a novel approach, casebased reasoning (CBR), is described to derive proper moulding parameters based on old reasoning can mean adapting old solutions to meet the new demands,or using old cases to explain new situations,or to assess new solutions. There are several benefits of applying CBR technology in the determination of injection moulding parameters. It allows the reasoner to propose solutions quickly, hence reducing the time needed to develop them. In addition, remembering previous experiences is particularly useful for injection moulding, previous experience helps to avoid the repetition of past mistakes. The learning process available with a adapting them. Schank [15], and Riesbeck and Schank [16] pioneered the casebased reasoning technique as an alternative to the more traditional rulebased and modelbased reasoning reasoning has been employed for a multitude of problems [17]。 caterers, recipe makers, dispute mediation, criminal sentencing, process planning [18], are some systems that have been documented as design systems in the CBR literature [19]. A putational system for the process design for injection moulding, named CSPD, was developed based on the bination of the blackboardbased expert system and casebased reasoning approach. In Section 3, architecture of the CSPD is presented. This is followed by a description of its two major ponents: the blackboardbased expert system for process design (KPC)。 and the casebased reasoning system for process design (CBRS), An illustrative example of the CSPD is given in Section 6. Finally, a conclusion is given at the end of this paper. Architecture of the Computational System for Process Design of Injection Moulding, CSPDFigure 1 shows the architecture of the CSPD, which mainly consists of four ponents: a user interface。 a blackboard based expert system for process design (KPC)。 a casebased reasoning system(CBRS)。 and a database. KPC is a plex expert system that involves multidisciplinary expert knowledge,which is used to generate a process solution including the selection of injection moulding machine, and mould base,tooling and processingcost estimation, and production is used to retrieve the previous cases and experience,and works out the setting of injection moulding parameters in a very short time. The database is used to store the information and data of the shopfloor injectionmoulding machines, standard mould bases, and mon thermoplastic materials. CSPD first derives the partial process solution based on a blackboardbased expertsystem approach. The partial solution together with the part information are then read by the CBRS for case retrieval and revision from which proper injectionmoulding parameters would be remended to the user, based on the casebased reasoning approach. Detailed descriptions of the KPC and CBRS will be provided in Sections 4 and 5, respectively. Expert System for Process Design, KPCAs mentioned in Section 2, blackboard architecture is adopted to develop the KPC in this research project, which mainly consists of three ponents: the blackboard, knowledge sources, and a control module. The architecture of the KPC is shown in Fig. 2. BlackboardThe blackboard is a global database which contains the data input to the problem, and intermediate and partial solutions of the problem and the process solution. The data and partial solutions are needed or are produced by the knowledge solution space is organised into an applicatondependent hierarchy. Information at each level in the hierarchy represents a partial solution. The domain knowledge is partitioned into independent knowledge sources that transform information on one level of the hierarchy into information on the same or other levels, using algorithmic procedures or heuristic choice of a knowledge source is based on the solution state posted on the blackboard and on the existence of knowledge sources capable of improving the current state of the solution. The blackboard is the only channel through which knowledge sources can interact with each other. In the KPC, the solution is organised in a frame hierarchy in the blackboard. Fig. 3 shows the structure of the blackboard and its corresponding solution partiti