AI in CNC machining

8/7/2019 AI in CNC machining

1/8

ELSEVIER

Artificial Intelligence in E ngineering 12 (1998) 121-1340 1997 Elsevier Science LimitedPrinted in Great Britain. All rights reserved

PII: SO954-1810(97)00011-3 0954-1810/98/$19.00

Artificial intelligence approaches to determinationof CNC machining parameters in manufacturing:a reviewKyung Sam Park & Soung Hie Kim*

Graduate School of Management , Korea Advanced Inst it ut e of Science and Technology, 207-43 Choengryan gri, Dongdaemun,Seoul, Korea(Received 1 March 1995; in revised version 13 May 1996; accepted 5 February 1997)

In Computer Numerical Control (CNC) machining, determining optimum orappropriate cutting parameters can minimize machining errors such as toolbreakage, tool deflection and tool wear, thus yielding a high productivity orminimum cost. There have been a number of attempts to determine themachining parameters through off-line adjustment or on-line adaptive control.These attempts use many different kinds of techniques: CAD-based approaches,Operations Research approaches, and Artificial Intelligence (AI) approaches.After describing an overview of these approaches, we will focus on reviewingAI-based techniques for providing a better understanding of these techniques inmachining control. AI-based methods fall into three categories: knowledge-basedexpert systems approach, neural networks approach and probabilistic inferenceapproach. In particular, recent research interests mainly tend to develop on-lineor real-time expert systems for adapting machining parameters. The use of AItechniques would be valuable for the purpose. 0 1997 Elsevier Science Limited.Key words: CNC machining, machining parameter, knowledge-based expertsystem, neural network, influence diagram.

1 INTRODUCTIONComputer Numerical Control (CNC) machining iswidely used in mold/die industries and airframecomponent manufacturing, because of its suitabilityfor high accuracy in machining complicated parts. Inthe CNC machining, determining optimal cuttingconditions or parameters under the given machiningsituation is difficult in practice. Conventional way forselecting these conditions such as cutting speed andfeedrate, has been based upon data from machininghandbooks and/or on the experience and knowledgeon the part of programmer. The selected parameters,in most cases, are extremely conservative to protectexcessive matching errors from tool failures such as tooldeflection, wear, breakage, etc. As a result, the metalremoval rate is low because of the use of suchconservative machining parameters.As frequently encountered in complex or free-formedsurface machining, the geometry of the part or work-piece prevents a constant depth and width of cut.*Author to whom correspondence should be addressed.

127

Consequently, the conservative cutting conditionsassuming a constant depth and width of cut do notperform high productivity. To overcome such a prob-lem, the machining parameters should be adjustedaccording to the current in-process part geometry.The objective of this paper is to review prior work ondetermining machining parameters in order to give abetter understanding to researchers and practitionersin machining domain, since we have not found anypublication on the survey. In particular, we will focus onexamining Artificial Intelligence (AI)-based methods. Inthe next section, we describe an overview of relatedwork that attempts to select an optimal machiningparameters, and present AI-based approaches in thesubsequent sections.

2 AN OVERVIEWRecently, there have been several attempts to deter-mine the optimal machining parameters from off-lineadjustment or on-line adaptive control, thus the partprogrammer does not have to spend time and effort to


2/8

128 Kyung Sam Park, Soung Hie Kimcalculate their optimal values. These attempts arecategorized as CAD-based approaches, OperationsResearch (OR) approaches, and AI approaches.32.1 CAD-based approachesThe off-line approach uses machining process models,cutting force and tool wear models, based on a priorknowledge gathered from off-line experiments. Based onthe process models, cutting force and tool wear arecalculated through computer machining simulation(CMS) using information on NC-code with initialmachining parameters, tool shape and workpiecegeometry. Using the results, an optimum machiningparameter for each tool motion is achieved by maxi-mizing the metal removal rate (MRR) without violatingmachining constraints.

The basic concept of optimizing machining param-eters is that when the cutting force is too large at thelarge depth and width of cut, either low feedrate or highcutting speed, or both can be added to the NC-code.However, note that too high cutting speed can not beselected since the tool life is largely due to the cuttingspeed.5 Most CAD-based approaches belong to the off-line adjustment. Advantages of these methods are thatthey are easy and effective in practical applications.Figure 1 shows a conceptual framework for simulationand optimization of machining.Machining process modelsThe machining process models represent the rela-tionships between the machining responses (i.e., cuttingforce and tool wear) and the machining conditions in aspecific tool and workpiece. These models can be builtby prior knowledge obtained from field and laboratoryexperiments. An example of machining process modelbased on a multiplicative model is given by

Cutting Force (N), FC = ald2f a3dn4w aS,Tool Life (min.), TL = blvb2f b3db4wb5,

where V, f, d and w are, respectively, cutting speed(mm/min), feedrate (mm/tooth), depth of cut (mm),and width of cut (mm); and oi and J3i are the modelparameters.

The multiplicative model can be generated fromstatistically planned machining tests (see Ref. 11). Anadvantage of this model is easily obtained even thoughgeometry of cutting tool is complicated such as ball-endmill. Furthermore, it is reported the accuracy of themodels is quite good. J~J The above models will play arole of constraints in optimizing machining parameters.In addition, analytical process models for the predictionof cutting force have been studied,7~8110V17nd theirapplication to on-line feedrate adjustment in end millinghas been found in Ref. 6. However, a difficulty may existin using such analytic models in practice because of theirhigh computational complexity.Computer machining simulationThe main objective of CMS for determining machiningparameters is to compute the maximum depth of cut (d)and width of cut (w) for each tool motion from givenpart geometry, NC-code and tool configuration. Whycompute the maximum point? The reason is machiningerror from the tool failures is mostly occurred at themaximum point. CMS of in-process workpiece can berealized as a Boolean subtraction of the space occupiedby the tool movement along the tool path from initialpart geometry. Hence, it is first needed to represent thepart geometry for CMS.Solid modeling8>912 r Z-buffer techniques22>23havebeen used to model workpiece geometry for CMS. Apape? has proposed a method of feedrate adjustmentusing a swept volume generation technique based onsolid modeling. However, this method can calculateonly average cutting force, thus it does not provide theinstantaneous cutting force that is necessary forestimating the tool failures. Z-buffer model is aform of discrete nonparametric representation inwhich the Z-values of the surface are given at gridpoints on the XY-plane. More detailed descriptionon Z-buffer model and its application to controland monitoring of machining can be found in theliterature.2223Machining parameter optimizationBased on the CMS and the machining process models,feedrate (f ) and cutting speed (v) are determined in the

( /,+omputer Machining Simulation chining Process Models I

Fig. 1. A framework for simulation and optimizaiton of CNC machining.


3/8

Determination of CNC machining parameters 129optimization module. For increasing the productivity,MRR has to be maximized while maintaining anallowable load fluctuation on the cutting tool in spiteof variations in depth of cut and width of cut. The MRRis expressed as MRR = kvfdw , where k = n/(rD), n isthe number of tooth, and D the diameter of thetool. A mathematical model for such problem can beformulated as follows:

Model 1: Maximize MRR = kvfdwsubject to v,in 5 v 5 vmax

FC I JG,,,TLmin I TL 5 TL,,,HP 5 HPr,,,,,

where HP represents the spindle horsepower(Nmm/min) as a constraint for the machine capacity,and Imin and Vmax, respectively, are minimum andmaximum allowable values of V. HP is expressed basedon the FC as HP = c - FC - v, where c is 0.041 as aconstant.5 Once taking natural logarithms in Model 1,it is converted into the standard linear programming(LP) form. Thus, the LP problem can be solved byusing a general algorithm referred to as the Simplexmethod.24

2.2 Operations research approaches

Of course, the use of the above LP technique can beviewed as an OR approach. However, main researchinterest of OR approaches is to minimize globalmachining cost by considering multiple criteria relatedto machining, thus which problem is to solve a multiplecriteria optimization problem (for an overview of themultiple criteria optimization problem, see Refs 25,26). These methods should be used for off-lineadjustment because of the restriction of computationaltime. An advantage of these methods can provide areference model, i.e., a general model because anexhaustive consideration on selecting machiningparameters is involved. A typical research is found inRef. 27. According to the research, the model withoutdescribing full mathematical form can be expressed asfollows:

Model 2: Maximize {MMR}Minimize {surface roughness}Minimize {machining cost}subject to the constraints of Model 1.

On surface roughness, there are two methods of itsmeasuring:average.2y3 root-to-crest roughness and roughnessThe factors of measuring machining costper workpiece are cost of tool, cost of cutting, and costsassociated with machine idle time, due to setup, loading

and unloading and tool changes to replace worn-out ordamaged tools. Solving Model 2 is more complexbecause the model have multiple objectives and con-flicting between the objectives (a mathematical repre-sentation and the solution method for Model 2 appearin Ref. 27).2.3 Artificial intelligence approachesThe on-line approach is an attempt to automaticallyadapt and optimize the machining parameters based onsensor information on machining responses in real time,without CMS. Well-known sensor information is listedas cutting force, tool wear, tool temperature andacoustic emission. Note that the information of tem-perature and acoustic emission can not be used in off-line methods using CMS. These information, however,can play very important role in machining controlor adapting machining parameters, praticularly whenoccurring an abnormal machining due to unpredictablevariables such as unknown material properties, toolconditions, etc.For on-line control, the following components ortechniques are required: (1) sensing devises, (2) repre-senting the information from the sensor, and (3)optimizing machining parameters. A description ofsensing devises is not presented in this paper. For thedescription, refer to Refs 4, 16.AI approaches offer a possible technique in order tohandle the problems (2) and (3). One of the mostimportant factors for successive on-line control is theexecution time with respect to machining control ordetermining optimal machining parameters. Reaction tomachining conditions by tool wear, machine break-downs and other failures must be carried out withinseconds or milliseconds to guarantee the safety andreliability of the machining process. However, asimplistic adaptation of AI techniques to machiningcontrol would be inadequate, because execution time ofthese systems are generally too long as compared with thereaction time required for the machining control,particularly if the knowledge base becomes very complex.There have been a number of studies on theapplication of AI techniques to on-line control, whichwe categorize into knowledge-based expert systemsapproach, neural networks approach and probabilisticinference approach. Each approach is described in thenext subsequent sections.

3 KNOW LEDGE-BASED EXPERT SYSTEM SAPPROACH3.1 An overviewKnowledge-Based Expert Systems (KBES) are intel-ligent computer programs that capture the specific


4/8

130 Kyung Sam Park, Soung Hie Kimknowledge of a particular domain and mimic theproblem-solving strategies of human experts to providerecommendations?8-3o They represent a new problem-solving paradigm that utilizes many techniques developedfrom AI research. The KBES can capture causal andinferential knowledge about machining processes toprovide expert-level recommendations during decision-making processes and hence are valuable aids tomachining operators who face increasingly complex tasks.With the KBES technique, machining control deci-sions using the sensor information can be made tomaintain the machining parameters within criticalconstraints. Strictly speaking, on-line control with theKBES is an adaptive control of satisfying machiningconstraints, simply stated ACC, rather than an adaptivecontrol with optimization, ACO. Whereas AC0 systemsseek to adjust machining parameters in a direction thatoptimize a predefined performance index, i.e., objectivesuch as MRR, the aim of ACC systems is that themachining parameters are adjusted to their maximumpossible values given the constraints of the machiningprocess. Recent research on machining control usingKBES techniques has been found in Refs 31, 32. In thenext subsection, we will describe a KBES framework formachining control and present an example of simpleproduction rules for the determination of machiningparameters.3.2 Structure of KBES for adaptive controlA structure of the KBES approach for machiningcontrol is shown in Fig. 2. It consists of three modules:a knowledge base, an inference engine, and a sensor dataacquisition and processing module. The inference enginedrives the system and interfaces with the knowledgebase and hence supplies advice to the user and anexplanation to justify the systems line of reasoning. Theknowledge base can provide near-optimal machiningcontrol with experimental data. The methods forinference can be modeled as rules, e.g., IF (antecedent)THEN (consequence).To achieve the near-optimal machining parametersand machining control such as tool change, machining

Knowledge Base

stop, and so on, it is necessary that an adaptive controlalgorithm that uses the recursive adaptive model andthe constraint rules is developed. For example, theconstraint rules can be expressed as shown in Fig. 3.In KBES, many techniques for knowledge representa-tion have been developed, for instance, productionrules, semantic nets, frames, etc. The type of knowledgerepresentation that is appropriate in a given situationdepends on what sort of knowledge is being representedand how it is to be applied. In time-critical machiningcontrol applications, it is imperative the knowledgerepresentation scheme is efficient. Among the KBESapproaches to machining control, in Ref. 31 a frame-based scheme is used, and in Ref. 32 a production rulerepresentation is applied such as shown in the aboveparagraph.

4 NEURAL NETWORKS APPROACH4.1 An overviewNeural networks differ in various ways from con-ventional expert systems to traditional computing. Thereasons are as follows. First, unlike traditional expertsystems where knowledge is made explicit, neural netsgenerate their own knowledge by learning from domainexamples. This means that neural nets can easilymake the knowledge base by learning, and they do notrequire additional knowledge acquisition processeswhich require enormous time and efforts in the expertsystems. Supervised learning is achieved through thelearning rule which adapts the connection weights of thenetwork in response to the inputs and the desired outputpairs. Many other network learning rules have beeninverted also in Ref. 34.Second, neural computing is both distributed andassociative in knowledge representation.33 The dis-tributed and associative nature of neural net leads toa reasonable response even when presented withincomplete or previously unseen input. In particular,multi-layer neural nets which register in their hiddenlayers important features of the knowledge domain,

Fig. 2. A KBES structure for machining control.


5/8

Determination of CNC m achining parameters 131

Rules or Finding Machi ningSituat ion:

Rule 1: IF (NOT (A&c,,?,,, : Acstic c AcsticJ)/* Acstic = Intensity of Acoustic etmss~?n from sensor */THEN (Stop the machining operatmn AND Change tool If necess ary)

Rule 2: IF (NOT (Temp nil < Temp - Temp,> ,,, )I* Temp : Temperature from sensor *I

feedforward neural net with one hidden layer is shownin Fig. 4. Each node or processing element (PE) in everylayer is fully connected to other PE in the proceedinglayer, and ever PE sums its weighted inputs and passesthrough some kind of transfer function such as linearor sigmoid functions. The learning parameters are theconnection weights and the PEs parameters, i.e.,threshold values.

THEN (Stop the machining operation AND Change tool if necessary)

Rule 5: IF ((FC[k] J FC[k]J AND (feedm c feed))/* FC[k]-Cuttmg Force at axis k. k=x,y, from sensor. feed=Feedrate 1THEN (feed = feed 0 01 until FC[k] L FC[kJ,J

Rule6: IF ((FC[k] -z FC[k],J AND (teed I feed,,,\) AND (speed speed,,,J)THEN (speed = speed + 0 02 until FC[k] < FC[k],,,,,b)

Rule 7: IF ((FC[k] - F([k],,,,\) AND ( feed < feed,,,J)THEN (feed = feed + 0 01 unti! FC[k] : FC[k]J

As a learning rule, Rumelhart and McClelland33 havedeveloped the generalized delta rule called back-propagation algorithm that is basically a gradientmethod. This rule aims at minimizing the global errorof the system by adjusting the learning parameters. Thebackpropagation algorithm does not always find globalminimum but may stop at a local minimum. However,in most cases, the system can usually be driven to theglobal minimum or to the desired accuracy with anappropriate choice of hidden PEs. The number ofhidden PEs must be large enough to form a decisionregion that is as complex as required by the givenproblem, and on the other hand is small enough that thegeneralization ability remains good.

Rule n: IF ((FC[k] : FC[k],J AND (feed 1 feed,,,\) AND (speed ? speed,,>J) 4.3 Optimal control phaseTHEN (Sto p machinmg operatmn)

Fig. 3. An example of rules for adapting machining parameters.can use this hidden knowledge to generate non-trivialgeneralizations.

Assuming that a neural network has been trained by theprocedure mentioned previously, then the objective ofoptimal control phase is to determine an appropriatemachining parameter that optimizes a performanceindex, given machining constraints on the networkoutputs.In machining domain, neural nets can possess abilitiesto learn from experience and to use the knowledgegathered during the learning process to optimize themachining control. Experience is represented by input-output data, where input variables are machiningparameters such as feedrate and cutting speed andoutput variables are signals from sensors such as cuttingforce, tool wear, temperature and acoustic emission. Theaim of learning is to establish a generalized mappingbetween the input and output, where note that this

statement is a supervised learning. This section dealswith a supervised learning approach (see Refs 35,36). Forunsupervised learning approaches refer to Refs 37, 38.

Let us consider a trained neural net with n input PEsand m output PEs. Let ai and dip respectively, be actualand desired output of the ith output PE. For k of the moutput nodes, the di represent the desired outputs of thePEs, whereas for the remaining m - k PEs, the di is themaximum allowable outputs of the PEs (e.g., cuttingforce and horsepower). Then, a performance index PIis defined by PI = w 1- ERR - w2. MRR , where MRRrepresents metal removal rate (see Section 2.1)ERR = 5 (di - ai)2/2,

i=l

In addition, there are several techniques for repre-senting the input-output relationship: multipleregression (see Section 2.1), the group method of datahandling, and the neural network (see next subsection).A study4 has reported that the neural network is themost effective method for tool wear identificationthrough a comparative analysis among the abovetechniques. Furthermore, neural net knowledge-baseacquired during the learning phase can subsequently beused for determining optimal machining parameters.4.2 Learning phaseFor building machining knowledge base, a two-layer Fig. 4. A neural net structure for representing machiningknowledge.

outputlayer

Hiddenlayer

Inputlayer


6/8

132 Kyung Sam Park, Soung Hie Kimand w1 and w2 are constants that represent the relativeimportance of ERR and MRR.Machining optimization problem is to find the ninputs, denoted by pi, that minimize PI subject to thefollowing constraints:

Pmini I Pi I Pmaqr i= 1,...,n,di - ai > 0, i=k+l,...,m.

The solution method of this constrained minimizationproblem can be found in Ref. 35.

5 PROBABILISTIC INFERENCE APPROACHAgogino et a1.43have proposed an influence diagramas a framework for integrating machining operatorsexpertise, first-principle knowledge and experimentaldata for the wide range of sensors possible for in-processmonitoring and control. The use of multiple sensorsreduces the sensitivity of the system to any specificsensors drawbacks. The non-deterministic or prob-abilistic nature of the inference problem and noisysensor data is handled by operations with Bayesianprobability.Influence diagram has been developed for represent-ing complex decision problems based on incomplete anduncertain information from a variety of sources.39~44Knowledge of the interrelationships between variables isrepresented in a compact graphical and numericalframework which identifies the critical variables andexplicitly reveals any conditional independence betweenthem.The knowledge representation using influence dia-grams can be viewed from three hierarchical levels:topological, functional and numerical level.39 At thetopological or relational level, the nodes in the diagramrepresent the key variables in the system being modeled,and the arcs or arrows identifies conditional influencesor functional relations between the nodes. In the CNCmachining, examples of key variables are machiningparameters and machining responses from sensor

information. The nature of the influences is specifiedat the functional level and further quantified at thenumerical level.From the discussion in the preceding two paragraphs,influence diagram is defined by an acyclic directed-graphG = (N, A) with A c N x N: It contains three types ofnodes in the node set N. The chance node, which is

circular shape, represents uncertain or certain states(e.g., cutting force, tool wear, acoustic emission), therectangular-shaped decision node (e.g., feedrate, cuttingspeed) reveals a variables whose value is chosen by thedecision maker, and the diamond-shaped value node(e.g., metal removal rate) represents the objective to bemaximized in expectation by the decision analysis. Itshould be noted that influence diagrams on thetopological level do not need a mathematical orprobabilistic basis to justify themselves. Their influencesare justified by mathematical or probabilistic repre-sentation at the functional level. At the final level,numerical level, utilities of the decision maker, andprobability distributions from prior information byexperiments are assessed numerically for each node.Shown in Fig. 5 is a simple example of an influencediagram for machining optimization.Once a complete influence diagram is generated, thediagram is manipulated and evaluated for determiningthe optimal decision strategy. A direct solution proce-dure to automate influence diagrams has been proposedin Refs 40-42. This algorithm consists of the value-preserving translations, node removal and arc reversals,which correspond to the rollback procedure in deci-sion tree models.45 For more detailed description onapplied influence diagrams to machining monitoringand control, see Ref. 43.

6 CONCLUDING REMARKSThis paper presented a survey of prior studies ondetermining an optimal machining parameter andmachining control. We particularly focused on reviewing

Fig. 5. An influence diagram for determining machining parameters.


7/8

Determination of CNC m achining parameters 133AI-based methods for on-line adaptive control, i.e.,KBES, neural networks and probabilistic inferenceapproaches. These approaches are commonly based onthe sensor information of machining responses in realtime, and based upon the prior knowledge frommachining experiments in field and laboratory, suchas machining constraints, machinability data (i.e.,mechanical and material data of cutting tool andworkpiece), etc. The prior knowledge can be elicitedfrom data of machining handbooks (e.g., see Ref. 14)or the experience and knowledge of the partprogrammer.Observe that the difference that underlies AIapproaches is only the style of knowledge base beingrepresented in the system and its reasoning (or inference)for machining optimization and control. That is: TheKBES approach uses a knowledge base such asproduction rules (see Ref. 32) and frame-based tech-niques (see Ref. 31), with an inference engine. Theneural networks approach uses a feedforward networkmodel (see Refs 35, 36) based on supervised learningrules or self-organization network model (see Refs 37,38) based on unsupervised or competitive learningrules, without an additional inference engine. The prob-abilistic approach uses an influence diagram model (seeRef. 43) with a probabilistic reasoning engine.In building an intelligent machining control system, aprinciple problem is how to represent knowledge in away that is suitable for a particular machining domain.There have not been a specified guideline for thechoice of a knowledge representation scheme. Thegeneral or conceptual guidelines are efficiency andsufficiency, where sufficiency means that the precisionof the predictions we obtain meets our requirements,and efficiency refers to their practical applicability. AllAI approaches mentioned in this paper may be sufficientfor representing a machining knowledge, although theirknowledge representation styles are different. If theabove statement is true, the most important criteriabecomes efficiency.There may be some factors for measuring efficiency inquantitative or qualitative manner, for example, easy fornaive user and reaction time. Among them, reactiontime would be the most important factor in real-timemachining control systems. Namely, reaction tomachining failures and unfavorable machining has tobe made within seconds or milliseconds to guarantee thesafety and reliability of the machining process. In thispoint of view, we prefer a knowledge representationscheme in which reaction time is less sensitive as the sizeof the knowledge base increases.Apart from the sufhciency, what kind of knowledgerepresentation technique is better than the others in atime-critical machining control system? This paper cannot give an answer for this question, because therehave not been a comparative analysis among the AIapproaches. The problem in order to address thisquestion is a promising further research issue.

1.

2.

3.

4.

5.

6.

7.

8.

9.10.

11.

12.

13.

14.

15.

16.

17.18.19.

20.

Balakrishnan, P. and Dervries, M. F., Sequential estima-tion of machinability parameters for adaptive optimizationof machinability data base systems. ASME Journal ofEngineering for Industry, 1985, 107, 159- 166.Bhattacharya, A. R., Faria-Gonzalez and Ham, 1. Y.,Regression analysis for predicting surface finish and itsapplication in the determination of optimum machiningconditions. ASME Journal of Engineering for Industry,1970,92,711-714.Chandiramani, K. L. and Cook, N. H., Investigation onthe nature of surface finish and its variation with cuttingspeed. ASME Jounal of Engineering for Industry, 1964,86,134-140.Chryssolouris, G. and Domroese, M., An empirical studyof strategies for integrating sensor information inmachining. Annals of CZRP, 1989, St?(l), 425-428.DeGarmo, E. P., Black, J. T. and Kohser, R. A., Materialsand Processes in Manufacturing. MacMillan PublishingCo., New York, 1984, pp. 429-629.Fusell, B. K. and Srinivasan, K., An investigation of theend milling process under varying machining conditions.ASME Journal of Engineering for Industry, 1989, 111,27-36.Kline, W. A., DeVor, R. E. and Lindberg, J. R., Theprediction of cutting forces in end milling with applicationto cornering cut. International Journal of Machine ToolDesign and Research, 1982, 22(l), 7-22.Kline, W. A., DeVor, R. E. and Shareef, I. E., Theprediction of surface accuracy in end milling. ASMEJournal of Engineering for Zndustry, 1982, 104,272-278.Koren, Y., Computer Control of Manufacturing Systems.McGraw-Hill Co., New York, 1983, pp. 193-220.Martellotti, M. E., An analysis of the milling process.Transactions of the American Society of MechanicalEngineers, 1941, 63, 677-700.Neter, J., Wasserman, W. and Kutner, M. H., AppliedLinear Statistical Models: Regression, Analysis of Variance,and Experimental Design. Irwin Inc., Homewood, Illinois,1985.Park, K. S., Analysis of Cutting Force in Ball-End MillingProcess (in Korean). M.E. thesis, Dept. of IndustrialEngineering, Korea Advanced Institute of Science andTechnology, Korea, 1991.Park, K. S. and Kim, S. H., Use of AI-based approachesfor determining CNC machining parameters in manu-facturing. In Encyclopedia of Microcomputers, eds A. Kentand J. G. Williams. Marcel Dekker, Inc., New York,1995.Thompson, R. W., Tipnis, V. A. and Kegg, R. L. (eds),Machinability Testing and Utilization of Machining Data.Materials/Metalworking Technology Series, AmericanSociety for Metals, Metals Park, Ohio, 1979.Tipnis, V. A., Buescher, S. C. and Garrison, R. C.,Mathematically modeled machining data for adaptivecontrol of end milling operations. Proceedings of theFourth NAMRC, 1976, 279-286.Tlusty, J. and Andrews, G., A critical review of sensors forunmanded machining, Annals of the CZRP, 1983, 32(2),563-572.Tlusty, J . and MacNeil, P., Dynamics of cutting forces inend milling. Annals of the CZRP, 1975, 24(l).Mantyla, M. An Introduction to Solid Modeling, ComputerScience Press Inc., 1988.Wang, W. P. and Wang, K. K., Geometric modeling forswept volume of moving solids. ZEEE Computer Graphicsand Applications, 1986, 6, 8- 17.Wang, W. P., Solid modeling for optimizing metal


8/8

Kyung Sam Park, Soung Hie Kim

removal of three-dimensional NC end milling. Journal ofManufacturing Systems, 1988, 7(l), 57-65.21. Hook, T. V., Real-time shaded NC milling display.Computer Graphics, 1986, 20(4), 15-20.22. Jerard, R. B. and Drysdale, R. L., Geometric simulationof numerical control machining. Proceedings of ASMEInternat ional Com put ers in Engineering Conference, 1988,129-136.23. Takata, S., Tsai, M. D., Inui, M. and Sata, T., A cuttingsimulation system for machinability evaluation using aworkpiece model. Annuls of the CZRP, 1989, 38(l),417-420.24. Taha, H. A., An Introduction to Operations Research.MacMillan Publishing Co., New York, 1982, 15-157.25. Chankong, V. and Haimes, Y. Y., Multiobjective DecisionMak ing: Theory and Met hodology. North Holland, NewYork, 1983.26. Steuer, R. E., Multiple Criteria Optimization: Theory,Computation, and Application. John Wiley & Sons, NewYork, 1986.27. Malakooti, B. and Deviprasad, J., An interactive multiplecriteria approach for parameter selection in metal cutting.Operations Research, 37(5), 805-8 18.28. Hayes-Roth, F., Waterman, D. A. and Lenat, T. J.,Building Expert Systems. Addison-Wesley, Reading MA,1983.29. Buchanan, B. G. and Shortliffe, E. H., Rule-Based ExpertSyst ems: The MYCZN Experiments of the StanfordHeuristic Programming Project. Addison-Wesley, ReadingMA, 1984.30. Lu, S. C-Y. and Komanduri, R. (eds), Knowledge-BasedExpert Syst ems for Manufacturing. The American Societyof Mechanical Engineers, New York, 1986.31. Lingarkar, R., Liu, L., Elbestawi, M. A. and Sinha, N. K.,Knowledge-based adaptive computer control inmanufacturing systems: a case study. IEEETransactions on Systems, Man, and Cybernetics, 1990,20(3), 606-618.32. Billatos, S. B. and Tseng, P. C., Knowledge-based

optimization for intelligent machining. Journal ofManufacturing Systems, 1991, 10(6), 464475.

33. Rumelhart, D. and McClelland, J., Parallel DistributedProcessing, Vol. 1. MIT Press, Cambridge, 1986.34. Lippmann, R., An introduction to computing with neuralnets. IEEE Transactions on ASSP, 1987, 4(2), 4-22.

35. Rangwala, S. S. and Dornfeld, D. A., Learning andoptimization of machining operations using computingabilities of neural networks. IEEE Transactions onSystems, Man, and Cybernetics, 1989, 19(2), 299-314.36. Dornfeld, D. A., Neural network sensor fusion for toolcondition monitoring. Annuls of CZRP, 1990, 39(l),101-105.37. Burke, L. I., Competitive learning based approaches totool-wear identification. IEEE Transactions on Sys tem s,Man, and Cybernetics, 1992, 22(3), 559-563.

38. Burke, L. I., An unsupervised approach to tool wearidentification. ZIE Transactions, 1993.39. Howard, R. A. and Matheson, J. E., Influence diagrams.In The Principles and Applications of Decision Analysis,Vol. 2, eds R. A. Howard and J. E. Matheson. MenloPark, CA, Strategic Decision Group, 1984.40. Shachter, R., Evaluating influence diagrams. OperationsResearch, 1986, 34, 871-882.41. Shachter, R., Probabilistic inference and influencediagrams. Operations Research, 1988, 36, 589-604.42. Rege, A. and Agogino, A. M., Topological framework forrepresenting and solving probabilistic inference problemsin expert systems. IEEE Transactions on Sys tem s, Man,

and Cybernetics, 1988, 18(3), 402-414.43. Agogino, A. M., Srinivas, S. and Schneider, K. M.,Multiple sensor expert system for diagnostic reasoning,monitoring and control of mechanical systems. MechanicalSystems and Signal Processing, 1988, 2(2), 165-185.

44. Oliver, R. M. and Smith, J. Q. (eds), Influence Diagrams,Belief Nets and Decision Analysis. John Wiley & Sons, NewYork, 1990.

45. Bunn, D. W., Applied Decision Analysis. McGraw-HillInc., New York, 1984.

AI in CNC machining

Documents

Transcript of AI in CNC machining