外文原文.pdf

International Journal of CAD/CAM Vol. 2, No. 1, pp. 6975 (2002)An Intelligent Cavity Layout Design System for Injection MouldsWeigang Hu and Syed Masood*Industrial Research Institute Swinburne (IRIS), Swinburne University of Technology, Hawthorn, Melbourne, Australia 3122Abstract This paper presents the development of an Intelligent Cavity Layout Design System (ICLDS) for multiple cavityinjection moulds. The system is intended to assist mould designers in cavity layout design at concept design stage. Thecomplexities and principles of cavity layout design as well as various dependencies in injection mould design are introduced.The knowledge in cavity layout design is summarized and classified. The functionality, the overall structure and generalprocess of ICLDS are explained. The paper also discusses such issues as knowledge representation and case-based reasoningused in the development of the system. The functionality of the system is illustrated with an example of cavity layout designproblem.Keywords: Intelligent design, cavity layout design, injection mould design, case-based reasoning, design support system1. IntroductionIn manufacturing, the injection moulding is one ofthe most widely used production processes for producingplastic parts with high production rate and little or nofinishing required on plastic products. The processconsists of injecting molten plastic material from a hotchamber into a closed mould, allowing the plastic tocool and solidify and ejecting the finished product fromthe mould. For each new plastic product, the injectionmoulding machine requires a new injection mould. Designand manufacture of injection mould is a time consumingand expensive process and traditionally requires highlyskilled tool and mould makers. An injection mouldconsists of several components, which include mouldbase, cavities, guide pins, a sprue, runners, gates, coolingwater channels, support plates, slides and ejectormechanism 1. Design of mould is also affected byseveral other factors such as part geometry, mouldmaterial, parting line and number of cavities per mould.With the advances in computer technology and artificialintelligence, efforts have been directed to reduce thecost and lead time in the design and manufacture of aninjection mould. Injection mould design has been themain area of research since it is a complex processinvolving several sub-designs related to various com-ponents of the mould, each requiring expert knowledgeand experience. Mould design also affects the productivity,mould maintenance cost, manufacturability of mould,and the quality of the moulded part. Most of the workexpert systems, knowledge based systems and artificialintelligence to eliminate or supplement the vast amountof human expertise required in traditional design process.Kruth and Willems 2 developed an intelligent supportsystem for the design of injection moulds integratingcommercial CAD/CAM, a relational database and anexpert system. Lee et. al. 3 proposed a systematicmethodology and knowledge base for injection moulddesign in a concurrent engineering environment. Ra-viwongse and Allada 4 developed a neural network-based design support tool to compute the mouldcomplexity index to help mould designers to assesstheir proposed mould design on mould manufacturability.Kwong and Smith 5 developed a computational systemfor the process design of injection moulding based onthe blackboard-based expert system and the case-basedreasoning approach, which includes mould design,production scheduling, cost estimation and determinationof injection moulding parameters. Britton et. al. 6discussed the injection mould design from a functionalperspective using functional design knowledge and anumber of knowledge libraries. Mok et. al. 7 developedan interactive knowledge-based CAD system for injectionmould design incorporating computational, knowledgeand graphic modules.Several studies have also been made on improvingthe design of specific components of an injection mould.Ong et. al. 8 developed a knowledge-based and object-oriented approach for the design of the feed system forinjection moulds, which can efficiently design the type,in mould design has been directed to the application of location and size of a gating system in the mould. Iraniet. al. 9 also developed a software system for automaticdesign of gating and runner systems for injection mouldsand provide evaluation of gating design based onspecified performance parameters. Nee et. al. 10proposed a methodology for determination of optimal*Corresponding author: Tel : +61-3-9214-8260Fax : +61-3-9214-5050E-mail : S.auG01G02 International Journal of CAD/CAM Vol. 2, No. 1, pp. 6975parting directions in injection mould design based onautomatic recognition and extraction of undercut features.Chen and Chou 11 developed algorithms for selectinga parting line in mould design by computing the undercutvolumes and minimising the number of undercuts. Parkand Kwon 12 worked on the design of cooling systemsin injection moulds and proposed an optimal designbased on thermal analysis and design sensitivity analysisof the cooling stage of the injection moulding process.Lin 13 worked on the use of gate size and gate positionas the major parameters for simulated injection mouldperformance prediction.One area in injection mould design, which hasreceived little attention, is the design of cavity layout ina multiple cavity injection mould. Cavity layout designaffects the whole process of injection moulding directly,since it is one of the most important phases in moulddesign process. Consideration of cavity layout designin injection mould at concept design stage will improvethe quality of injection moulded products because it isassociated with the determination of many key factorsaffecting the design and quality of mould. Such factorsinclude number of cavities; parting line; type of mould;type and position of gate; runner system; cooling systemand ejection system. Some of these factors are difficultto build as true mathematical models for analysis anddesign. This paper presents the development of a designsupport system, called Intelligent Cavity Layout DesignSystem (ICLDS), for multiple-cavity injection mouldsbased on knowledge based and object oriented approaches.It uses the case-based and ruled-based reasoning inarriving at the layout solution 14. It is based on thecommercial software system named “RETE+”, whichis an integrated development platform for customers todevelop their own knowledge-based systems 15. Theobjective is to make full use of available techniques inartificial intelligence in assisting mould designers atconcept design stage.2. Cavity Layout Design in Injection MouldsCurrent practice for injection mould design, especiallycavity layout design, depends largely on designers ex-periences and knowledge. It would therefore be desirableto use knowledge engineering, artificial intelligence andintelligent design techniques in generating an acceptablecavity layout design in injection mould accurately andefficiently. In mould design, most of patterns of cavitylayout and rules and principles of cavity layout designcan also be easily represented in the form of knowledge,which can be used in most of knowledge-based designsystems.For example, for the layout patterns shown in Fig. 1,the criteria to select the suitable layout pattern fordesign are mainly dependent on working environments,conditions and requirements of customer and are mainlybased on designers skill and experience. To make achoice of contradictory factors will rely obviously ondesigners knowledge and experiences. It is rather suitablefor intelligent design techniques to be used in systemsdesigned for such situations, especially for routine orinnovation design.Design of injection mould mainly involves considerationof design of the following elements or sub-systems:(1) mould type(2) number of cavitiesFig. 1. Some patterns of cavity layout with multiple cavities.Weigang Hu and Syed Masood An Intelligent Cavity Layout Design System for Injection Moulds G01G03(3) cavity layout(4) runner system(5) ejector system(6) cooling system(7) venting(8) mounting mechanismMost of the elements are inter-dependent such that itis virtually impossible to produce a meaningful flowchart covering the whole mould design process. Someof the design activities form a complicated designnetwork as shown in Fig. 2.Obviously, in injection mould design, it is difficultfor designer to monitor all design parameters. Cavitydesign and layout directly affects most of other activities.The application of advanced knowledge based techniquesto assist designer in cavity layout design at conceptdesign stage will greatly assist in the development of aIt is noted from Fig. 1 that a number of differentlayout patterns are possible with multiple cavities inside amould. Higher the number of cavities of mould, higherthe productivity of the injection mould. But this maylead to difficulties with issues such as balancing therunners or products with the complicated cavity shapes,which in turn may lead to problems of mouldmanufacturability. It is also possible that the number ofcavities and the pattern of cavity layout will influencethe determination of parting line, type of gate, positionof gate, runner system and cooling system. Most of themain activities of mould design are therefore linked tocavity layout design. Fig. 3 shows the relations betweencavity layout design and other design activities. Thecavity layout design problem therefore depends upon anumber of functionalities of the overall mould designsystem, which includes:Fig. 2. Design network of injection prehensive computer-aided injection mould designand manufacturing system.(1) definition of design specifications includinganalysis and description of characteristics ofG01G04 International Journal of CAD/CAM Vol. 2, No. 1, pp. 6975design problem(2) determination of mould type(3) determination of number of cavities(4) determination of orientation of product(5) determination of runner type and runnerconfiguration(6) determination of type and position of gate(7) cavity layout conceptual design(8) evaluation of ejection ability, manufacturingability and economic performances(9) determination of cooling system(10) graphic results display and output3. Structure of ICLDS and the Design ProcessThe structure of the Intelligent Cavity Layout DesignSystem (ICLDS) is based on case-based reasoning andruled-based reasoning designed around the RETE+Fig. 3. Relationship diagram between cavity layout design and other modules of mould.Fig. 4. Overall structure of ICLDS.Fig. 5. The general design process of ICLDS.Weigang Hu and Syed Masood An Intelligent Cavity Layout Design System for Injection Moulds G01G05software system. Fig. 4 shows the overall structure ofICLDS schematically. Fig. 5 shows the general designprocess of ICLDS. The design process starts with thedefinition of design specifications. The ICLDS systemretrieves similar cases from case base by computing thesimilarity between the cases and the new case. If thesolution is satisfactory, then results are displayedgraphically. If the solution is not satisfactory, then ICLDSwill use rule-based reasoning with forward or backwardchaining or a mixture of both to arrive at a solution. Ifthe solution is still unsatisfactory, then the user has tomodify some of the initial design specifications. The useof case-based technology in the design process inICLDS allows the user to obtain the solution(s) ofdesign problem more quickly and flexibly. The structure of knowledge base and database usedin the development of ICLDS is based on the underlyingknowledge base and database structure from the RETE+software system, which is a commercially availablesoftware development platform.4. Development of ICLDS4.1. Classifications of KnowledgeFor various logic and steps involved in layout design,there are different kinds of knowledge that needs to bedescribed and represented in cavity layout design. Thetypes of knowledge can be classified into five kinds basedon object oriented (OO) concept as described below:(1) Design instance/case: previous design cases andcurrent design instances(2) Relation: superclass-class-subclass relation, class-instance relation(3) Attribute: design variables, features, attributes ofdesign problem(4) Rule: general design rules, design experiences(5) Procedure and/or model: numeric calculation,mathematical modeling, analysis, evaluation andprocedures.4.2. Knowledge RepresentationsTo describe each of these types of knowledge, theinternal data structures of the ECLIPSE language,included in RETE+ inherently, can be used to makethe object orientated representation of the design processas explained earlier. Some other considerations inknowledge representation are as follows:(1) For “design instance/case”, we combine “factdefinition” and “relation definition” plus databaseand case base to represent it(2) The “attribute” are represented as instances of“template definition” and/or “relation definition”(3) For “relation”, we use “relation definition” todescribe it(5) The “procedure/model” are defined by externalroutines using C+ languageFurthermore, “goal definition” and “goal generation”techniques are used to fulfil backward chaining reasoning,and “case-based reasoning” is used to carry out case-based design.4.3. Case-based ReasoningCase-Based Reasoning (CBR) is dependent firstly oncase retrieved. Case-based retrieval is based on “SimilarityMetric”. Therefore, how to calculate the similarity isobviously the key technique in CBR, and it is describedin detail as below.Similarity metric is a weighted distance function ina multi-dimensional space where each dimensioncorresponds to a field whose value is specified in thequery (new case) and which has a non-missing value inthe case being ranked. The distance between the caseand the query (which corresponds to a point in thismulti-dimensional space) is computed differently forordinal and nominal fields. An ordinal field is a fieldwhose values are ordered or sorted. A nominal field isone whose values represent qualitative information forwhich sorting makes no sense. In general, ordinal fieldsinclude dates, integers, and real numbers while nominalfields include Boolean, Symbols, and Text.4.3.1. Ordinal DistanceFor ordinal fields the distance, dij, between the i-thcases value, vij, and the querys value, Vj, for the j-thfield, is computed as:(1)where the maximum and minimum values for eachfield are determined during index construction.Since dijrepresents the distance along the j-th axis ofan n-dimensional similarity space, the similarity spacedistance Diis given by:(2)which, since dijmust range between 0 and 1, must alsorange between 0 and 1. When weights are used, theabove equation becomes:(3)which, since Wjand dijmust range between 0 and 1,must also range between 0 and 1.dijvijVjVjmaxVjmin-=Dij 1=Ndij2N-=Dij 1=NWjdij()2j 1=NWj-=(4) For “rule”, we combine “rule definition” and“rule set definition” to represent it4.3.2. Nominal Distance for TextTo determine the distance between the value of aG01G06 International Journal of CAD/CAM Vol. 2, No. 1, pp. 6975TEXT field in a case and that specified in a query, wedetermine a weight for each term by which a text fieldis indexed and a weight for each term in the query.These weights are computed according to the followingformula, where:(4)N is the number of cases.nkis the number of different cases in which term koccurs.Fikis the number of occurrences of term k in case idivided by the total number of terms in case i.Wikis the weight of the k-th term in case i.Let Wkbe the weight of the k-th term in the query,computed as in the above formula.Let T be the number of terms in the query.Given these weights, the similarity (expressed as anormalized distance) between two text fields is computedas :(5)4.3.3. Nominal Distance for SymbolsSymbols are merely a special case of text with onlyone term. The weights for symbol fields are computedas in the above equations for text fields with Fikalwaysbeing one. Given these weights, the similarity is computedexactly as for text fields.4.4. Validation of CaseValidation of case is to check up whether eachacceptable case is suitable for current problem and tofind out the most suitable one, so each case should beassociated with testing methods and tested results on it.Only the case, under the given conditions, for which alltested results on it match those of the current designproblem, can be considered as the solution prototypefor further refining.4.5. Criteria for Validity of Cost ReductionWith the application of ICLDS for cavity layout, twokinds of cost reduction can be expected. One is theoverall theoretical cost reduction achieved in using thesystem to carry out the conceptual design of injectionmoulds. The other is the practical cost reduction valuerecorded in the case base which may be used to do thecase-base reasoning if the case has the “cost reduction”attribute. For the theoretical one, there is no need ofany criteria for validity of cost reduction because thecost savings will obviously come out through lead timethe criterion of comparison can be used. For example,we can compare the “cost reduction” attributes for twocases and determine which one provides better fit forthe customers requirements. One can use the percentagecost reduction formula to do the comparison. Thepercentage cost reduction can be calculated by theformula:Cost Reduction=(previous cost-current cost)/previouscostIt is better first to work out the percentage for thecost reduction attributes and then d