外文翻译--解决金属切削问题的智能专家系统的开发 中文版

*,Departmeaccepted21In metal cutting, the problems need to be well analyzed in order to take precautions before any unexpected results are encoun-The objective of modern manufacturing is to have ious materials 2.andassuchnotanegativeprocessiftheanswersofwhen,how much and what type is known. There are certaintypesof wear mechanismsondierent typesofmetalcut-ting processes and similar/dierent types of problems areencountered as a result of these mechanisms. If these*Corresponding author. Tel.: +90 224 4428176； fax: +90 2244428021.E-mail addresses: .tr (M.C. Cakir), .tr (K. Cavdar).1Tel.: +90 224 4428176； fax: +90 224 4428021.Materials and Design 27 (2006)Materials0261-3069/$ - see front matter C211 2005 Elsevier Ltd. All rights reserved.ecient control over the organizational facilities in or-der to produce high quality products at lower priceswithin a shorter time period. To achieve better qualityat a lesser price, every attention has to be paid in themanufacturing division by employing better cuttingtools, inserts and high precision machines, etc. 1. How-ever, this is not sucient for productivity in most of thecases. Choosing the right tool for the right job is impor-tant, but tool wear should also be considered for moreMetal cutting is a chip forming process. Although theprocess is directed at cutting metal to shape and size, thishas to be done by creating defined chips. The chip-form-ing process means that a fresh metal interface is contin-ually produced and forced at very high pressure andtemperature along the tool material 3. The zones pro-duced make it an attractive environment for diusionand chemical reactions of metal. All cutting tools areworn during machining and continue to do so until theycome to the end of their tool-life. Tool wear is inevitabletered. This process plays a significant role in achieving consistent quality and in controlling the overall cost of manufacturing. How-ever, it is a dicult task that needs an expert who has a great deal of information and experience in metal cutting. In the presentpaper, a knowledge-based expert system (COROSolve) that investigates problems that are encountered in three main metal cuttingareas: turning, milling and drilling is developed. A great deal of metal cutting operations such as external/internal turning with neg-ative/positive inserts, aluminum turning, parting bars/tubes, grooving, profiling, recessing and threading operations in turning； facemilling, square shoulder milling, end milling, multi-purpose milling and side and face milling operations in milling； and drilling oper-ations that use solid drills or drills with indexable inserts in drilling are taken into consideration. COROSolve gives recommenda-tions for the cutting data (i.e., cutting speed, depth of cut, and feed) and updates the problem, cause and remedy database, thus thenumber of problems that the system can handle is increased.C211 2005 Elsevier Ltd. All rights reserved.Keywords: Metal cutting problems； Expert systems； KBS； Tool wear1. Introduction ecient manufacturing. In the literature, many publica-tions do exist on the tool wear in manufacturing of var-Development of a knowledge-basedsolving metal cuttingM. Cemal CakirUludag University, Mechanical EngineeringReceived 8 September 2004；Available onlineAbstractdoi:10.1016/j.matdes.2005.01.022expert system forproblemsKadir Cavdar1nt, Gorukle 16059, Bursa, Turkey31 January 2005March 2005/locate/matdes10271034& Designproblems are well analyzed, it could be possible to findthe right solutions to each of them.To analyze a problem in metal cutting, the problemshould be defined precisely and its possible causesshould be well determined. This is a dicult task thatneeds an expert who has a great deal of informationand experience in metal cutting. Today in metal cutting,the expertise facilities about the problems in chip form-ing are mainly provided by the cutting tool producingcompanies. Since the number of people working in thisarea is limited, it is not always possible to find an expertwhen he/she is really needed. Since expert systems havebeen particularly welcome in fields where existing ex-perts are expensive and in short supply, an expert systemthat solves metal cutting problems will be very useful.and an expert system.A more obvious problem is that of gathering therules. Human experts are expensive and are not extre-mely likely to want to sit down and write out a largenumber of rules as to how they come to their conclu-sions. More to the point, they may not be able to.Although they will usually follow a logical path to theirconclusions, putting these into a set of IF . THENrules may actually be very dicult and maybeimpossible.It is quite possible that many human experts, thoughstarting o in their professions with a set of rules, learnto do their job through experiential knowledge and C212justknowC213 what the correct solution is. Again they may havefollowed a logical path, but mentally they may haveC212skipped some stepsC213 along the way to get there. An Ex-pert System cannot do this and needs to know the rules1028 M.C. Cakir, K. Cavdar / Materials and Design 27 (2006) 102710342. Expert systems for aiding manufacturing processesKnowledge-based systems, or expert systems, arecomputer programs embodying knowledge about anarrow domain for solving problems related to thatdomain. An expert system usually comprises two mainelements, a knowledge base and an inferencemechanism (Fig. 1). The knowledge base containsdomain knowledge which may be expressed as anycombination of C212IF-THENC213 rules, factual statements,frames, objects, procedures and cases. The inferencemechanism is that part of an expert system thatmanipulates the stored knowledge to produce solu-tions to problems 4.A human expert uses knowledge and reasoning toarrive at conclusions, so does an expert system. Thereasoning carried out in an expert system attemptsto mimic human experts in combining pieces ofknowledge. Thus, the structure or architecture of anexpert system partially resembles how a human expertperforms. Thus, there is an analogy between an expertFig. 1. Knowledge basedvery clearly.A very strong benefit of expert systems is being ableto widely distribute the knowledge of a single expert,or being able to accumulate the knowledge of severalwidely separated experts in one place. Expert systemsare especially helpful when a task is performed onlyoccasionally and the expert has to relearn the procedureeach time it is performed.Expert systems are used to standardize operations.If you have three machine operators (or engineers) thatperform the same task but each does it dierently, anexpert system will do it the same way everytime. Thesesystems can be used to train employees, advise them, orcan actually perform a task such as a calculation. An-other use for expert systems is as job aids for your ex-perts. They will allow them to perform moreaccurately, more consistently, and faster, freeing uptime for the expert to be more creative performingthe task. This is especially helpful when dealing with te-dious, repetitious tasks.Consequently, once the domain knowledge to beincorporated in an expert system has been extracted,the process of building the system is relatively simple.expert system.The ease with which expert systems can be developed hasled to a large number of applications of the tool. In engi-neering, applications can be found for a variety of tasksincluding selection of materials, machine elements, tools,equipment and processes, signal interpreting, conditionmonitoring, fault diagnosis, machine and process con-trol, machine design, process planning, productionscheduling and system configuring. Some recent exam-ples of specific tasks undertaken by expert systems are: identifying and planning inspection schedules for crit-ical components of an oshore structure 5； training technical personnel in the design and evalua-tion of energy cogeneration plants 6； configuring paper feeding mechanisms 7； carrying out automatic re-meshing during a finite-elements analysis of forging deformation 8； storing, retrieving and adapting planar linkagedesigns 9； designing additive formulae for engine oil products10； selecting cutting tools or cutting data 1113.Several potential research areas were identified withrespect to expert systems in manufacturing 14,15.Kojiyama et al. 16 have discussed in their articles thata framework for machining operation planning systems,in which machining know-how extracted and organizedfrom electronic tool catalogs and machining instancedatabases available in the Internet environment plays aprincipal role. On the use of reference machining datacan be constitute, which is derived from the investiga-tion of tool catalogs, related international standards,reference textbooks, and handbooks.M.C. Cakir, K. Cavdar / Materials and Design 27 (2006) 10271034 1029Fig. 2. Inputs of the software.The application of Mookherjee and Bhattacharyya11 on expert system in the general turning and mill-ing operation is very useful for solving some of thechallenging problems presently faced by manufactur-ing engineers during the integration of CAD and var-ious CNC machining centers. Jiang et al. 17 havedeveloped an expert system to optimize the millingoperations for prismatic components. They havedescribed a new GT based coding scheme that repre-sents the surfaces to be machined of prismaticcomponents.Limsombutanan 18 has presented an algorithm toselect the cutter size and tool orientation in 5-axis sur-face machining that can act as a holon. Milling the sur-face is divided in to three phases, namely, roughing,semi-roughing and finishing. This algorithm selects thebest tool and plans the tool path autonomously for abicubic (convexconcave) surface based on analysis ofthe curvature.The main objective of the present paper is to buildsuch system that covers most of the major metal cuttingproblems and helps the people who are involved in metalcutting to improve the quality.3. A knowledge-based expert system for metal cuttingproblems3.1. Factors considered in solving metal cutting problemsMost of the problems in metal cutting are the resultof wear mechanisms. All cutting tools wear duringmachining and continue to do so until they come tothe end of their tool-life. Tool wear is inevitable and1030 M.C. Cakir, K. Cavdar / Materials and Design 27 (2006) 10271034Fig. 3. Flowchart of COROSolve.as such not a negative process if the answers of when,how much and what type is known. Thus, if the prob-lems resulted from tool wear as well as the other harmfulfactors eecting the machining operations are well ana-lyzed, it could be possible to find the right solutions toeach of them. This would reduce the unproductive faultsearching time, therefore reduce machining time, unpro-ductive stoppage times of machine tools, machining costand therefore increase the eciency of the process.Architecture of the expert system used in solving metalcutting problems is given in Fig. 1. Inputs that are takeninto consideration by the software are given in Fig. 2.3.2. Features of COROSolveThe software has been developed using DELPHI Vi-sual programming language. A major diculty in build-ing a consultation system was in capturing andexamining the knowledge elements used in the solutionof the problem. Based upon the design for solving prob-But although the tooling is right, if the machining con-ditions are not up to standard, especially as regards cut-ting data and general stability, problems arise andoptimum tool-life will not be reached. Incorrect cuttingdata, vibrations and lack of rigidity in tool holders andclamping are the main reasons of the metal cuttingproblems.Apart from being a problem solver, the softwareprovides cutting data recommendations for each typeof operation (types of operations were mentionedpreviously), for each type of material group (ISO P,M or K), for each type of application (rough, mediumor finish), for each type of machining condition (good,average, dicult) and displays the list of insertsavailable (Fig. 4). Here, it is not intended to findthe most suitable insert for the application, but tomonitor the starting values and the working rangesof cutting data (i.e., the cutting speed, feed rate andthe depth of cut) for the insert used in metal cutting.At this stage, it is possible to see the suitable gradeM.C. Cakir, K. Cavdar / Materials and Design 27 (2006) 10271034 1031lems in metal cutting study, once the operation type isselected, four major stages have been identified for eachtype of operation: Cutting data recommendations. Problem definition. Cutting data evaluation. Problem editor.The flowchart of the software is given in Fig. . Cutting data recommendationsThe selection of the right cutting tool is critical forachieving maximum productivity during machining.Fig. 4. Cutting data recommenthat was determined due to the selected operation,material group, application, etc. Besides, the list ofmaterials that each material group consist of can alsobe observed.Recommended cutting data values were acquiredfrom Sandvik Coromant 20, the cutting data fromthe other cutting tool vendors can be used as the rec-ommended values as well. Cutting data values dis-played are for a certain material hardness and for acertain tool-life. For any workpiece material havingdierent hardness values than the specified values,and the tool-life dierent than 15 min, the cutting dataprovided should be multiplied by the correction factors(see Table 1).dations.causesIncreased hardness0 +20 +40 +60 +80 +1001 0.91 0.84 0.77 0.72 0.671 0.91 0.84 0.78 0.73 0.681 0.95 0.9 0.86 0.82 0.7925 30 45 600.88 0.84 0.75 0.701032 M.C. Cakir, K. Cavdar / Materials and Design 27 (2006) 102710343.2.2. Analysing the metal cutting problemsFig. 5. Problems,Table 1Material hardness and tool life considerations 19ISO/ANSI HB Reduced hardnessC060 C040 C020P 180 1.44 1.25 1.11M 180 1.42 1.24 1.11K 260 1.21 1.13 1.06Tool life (min) 10 15 20Correction factor 1.11 1.0 0.93There are various dierent/similar types of problemsin metal cutting processes. In appearance some of theseproblems are identical and very hard to be distinguishedfrom each other. The classification of problem types hasbeen developed to form an important basis for assessingthe machining operation and to optimize productivityby getting the tool grade and machining conditions rightfor the type of cut and material. The right tool, goodstarting values for cutting data, expert support, ownexperience, good quality of workpiece materials and ma-chine conditions are important ingredients for success inmachining. Aggregation of problem lists from varioustool manufacturers is classified to form KBS used bythe software.The user analyses the metal cutting problems either inone or two stages: by straightforwardly picking outProblem Definition option from the menu or by checkingout the cutting data used in the metal cutting processfirst and accessing to the problem definition sectionafterwards. Problem Definition section displays the listof problems, their possible causes and the remedies withrespect to the causes. Once a problem is selected (i.e.,flank wear in turning or chip jamming in drill flutes indrilling), a picture that provides a clear definition tothe problem and the list of possible causes are displayed.Consequently once a possible cause is selected, the rem-edies are determined. More information about the prob-lem is obtained by clicking on ? (Fig. 5).3.2.3. Cutting data evaluationand remedies.In metal cutting, most of the problems are the resultof unsuitable cutting data for the application. Thus,before straightforward listing of the problems, the useris advised to check the cutting data for the operation,for the application and for the machining conditionsto find out whether the cutting speed, feed rate andthe cutting depth are correct for the length, thicknessand nose radius of the insert used. The Cutting Dataoption checks if the cutting data in the operation areright for the insert in use and this task is basically theFig. 6. Checking the cutting data.a newM.C. Cakir, K. Cavdar / Materials and Design 27 (2006) 10271034 1033comparison of the cutting data with the ideal values cal-culated from the multiplication of the catalogue valuesand the correction factors. If there is no match (thismeans the values of cutting data are not in the range)the user is advised to correct the cutting data. In millingand drilling only cutting speed and feed rate values areconsidered, in drilling this consideration takes the drilldiameter and the grade of the centre/periphery insertinto account (Fig. 6). In the figure, cutting data for facemilling of steel for medium feed rates (L for light, M formedium, H for heavy milling operations) are evaluatedand since the feed value is lower than the ideal valuescalculated for given hardness and tool life, a warningis displayed.Once the cutting data are reviewed (and corrected),Fig. 7. Addingaccessing to the problem definition section is the nextstep.Sincethecuttingdataarecorrected,thelistofcausesandtheremedieswillnowbedierentandnosuch causesasCuttingspeedistoohighorFeedrateistoolowwillnot be amongst the causes of the related problem.The software is capable of analysing more than 100types of problems in various types of operations andproviding remedies to the nearly 200 causes (340 reme-dies). The knowledge contained in the system has beencompiled from two main sources: from human expertsworking in the field of metal cutting and from thetechnical documents, catalogues or handbooks of vari-ous cutting tool producing companies 2. Problem editorThe knowledge contained in the system has beencompiled from two main sources: from human expertsworking in the field of manufacturing and from the tech-nical documents, catalogues or handbooks of variouscutting tool producing companies. The success of an ex-pert system is hidden in its expandable structure like ahuman expert who adds every new solution he comesacross to his knowledge and uses this knowledge in hisfuture analyses. Thus, COROSolve has an expandabledatabase structure that grows to handle more and moreproblems everyday.Since the system has a separate and modular knowl-edge base, it is very easy indeed to update the system bysimply getting into the database and editing the knowl-edge files. The more information the system contains,the more problems it can handle in metal cutting.Knowledge base is the heart of the system, therefore itis the task of a few people who are responsible fromthe production to add, delete or modify it. Thus, theuser needs to know the password to access to the knowl-edge base.remedy.Problem edit allows the new problems, picture andinformation files related to the problems, causes andthe remedies of the problems to be added to the knowl-edge base. Besides, it is possible to add new causes to theproblems, or new remedies to the causes that are alreadyexist in the knowledge base (Fig. 7).The system is multi-linguistic, therefore it is capableof handling the metal cutting problems both in Turkishand English. Once the language is determined, all theprogram menus and problem, cause and remedy listsare displayed in the language chosen.4. ConclusionsThis paper introduces an expert system approach forthe task of solving metal cutting problems in variousmachining operations. Since there are not many workexist in the literature about this subject and since thistype of system can mostly fulfil the requirements in me-tal cutting industry, the work described here can beconsidered as a useful piece of work. In manufacturingindustry, especially in small or medium-volume machineshops, incorrect cutting data are the main reason of theproblems. Tools were generally run at lower cutting datato make them last longer for less frequent changes. Thisthen obviously meant poor utilization of the metal cut-ting time. Since one of facilities that the system provideis the evaluation of the cutting data, it will help the userto choose the right speed, feed or the depth of cut for theapplication. If the cutting data are correct and the metalcutting problems are solved, the stoppages due to thebreakdowns will be shorten, good utilization of powercapacity will be achieved and the metal cutting time willtherefore be reduced. This means a great deal of reduc-tion in production cost.The system developed intends to educate the people2005；21(2):17583.5 Peers SMC, Tang MX, Dharmavasan S. A knowledge-basedplants. In: Proceedings of the third world congress on expertsystems, Seoul, Korea. February 1996. p. 3005.7 Koo DY, Han SH. Application of the configuration designmethods to a design expert system for paper feeding mechanism.In: Proceedings of the third world congress on expert systems,Seoul, Korea. February 1996. p. 4956.8 Yano H, Akashi T, Matsuoka T, Nakanishi K, Takata O,Horinouchi N. An expert systems to assist automatic remeshing inrigid plastic analysis. Toyota Tech Rev 1997；46:8792.9 Bose A, Gini M, Riley D. A case-based approach to planarlinkage design. Artif Intell Eng 1997；11:10719.10 Shi ZZ, Zhou H, Wang J. Applying case-based reasoning toengine oil design. Artif Intell Eng 1997；11:16772.11 Mookherjee R, Bhattacharyya B. Development of an expertsystem for turning and rotating tool selection in a dynamicenvironment. J Mater Process Technol 2001；13:30611.12 Wong SV, Hamouda AMS. Development of generic algorithm-based fuzzy rules design for metal cuttin