Materials Science Forum Vols.

626-627 (2009) pp 605-610 Online available since 2009/Aug/31 at © (2009) Trans Tech Publications, Switzerland doi:10.4028/

Study on Intelligent Tool Optimal Selection System in Milling* X.F.Fang1,a, T.X.Lan1,b, S.W.Zhang1,c,W.Jia2,d, T.Y.Wang3,e
1 Modern Manufacturing Technology Institute of Jiangsu University of Science and Technology, Zhengjiang, China 212003 2 Engineering Technology Center of HHM, Shanghai, China 200129 3 Department of Mechanical Engineering, Huaiyin Institute of Technology, Huaiyin 223001
a,, Selection, Rule-Based Reasoning, Weight

Keywords: Milling Process, Tool Optimal Decision-Making, Analytic Hierarchy Process

Abstract. Rule-based reasoning (RBR) and weight decision making have been widely used in a lot of decision support systems. According to the waste phenomenon of manufacturing corporations due to the reasonless use of tools during production process, and considering the characteristics of former tool selection systems, develop a tool optimal selection system with Visual Basic 6.0 as development tool and SQL Server 2000 as database developing platform. The system adopts the rule-based reasoning and weight decision-making theory, combines fuzzy theory, artificial intelligent technology and production conditions of corporations. According to the information of workpiece and processing conditions imported by users, the system can select the reasonable tools for users. The system overcomes the limitation of single theory tool selection system, simplifies reasoning mechanism and structure of knowledge base, makes programmer easy to realize, promotes operation efficiency of system, and raises the accuracy and efficiency of tool selection in actual production. A developed prototype system and an example have verified some presented techniques and the research results are the basis of the future development. Introduction In recent years, with the continuous innovation of workpiece, lots of new tools come forth. As the important implement during production process, tools play a very important role in manufacturing system. Meanwhile, the problems about tool such as high price, short life and unreasonable usage hamper the development of corporations, so tool optimal selection is very important during production process, and corporations urgent need to develop their own tool optimal selection system. The study on tool selection system has experienced almost 20 years since the 1990s, and has made some results[1-6]. Some universities, scientific research institutions and corporations have developed a series of tool selection system successively, and improved the production automation degree of corporations. By now the theories used in tool selection systems include rule-based reasoning, case-based reasoning, feature-based, weight decision-making theory, etc. The paper combines the present situation of certain marine diesel engine corporation, adopts the method of combining of rule-based reasoning and weight decision-making, and establishes a tool optimal selection system in milling for the marine diesel engine corporation. The developed system


Project supported by the National Natural Science Foundation of Jiangsu Province Education Department(Grant No.07KJB460008)
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP, (ID:, Indian Institute of Kharagpur, Kharagpur, India-04/10/13,12:50:53)

so this system adopts forward reasoning. cutting speed database and rule database. uses weight decision-making and combines various influence factors to select the best tool on the basis of selected production target. The rules of this system are very definite. Rule database contains the knowledge about reasonable check of workpiece material.606 Advances in Materials Manufacturing Science and Technology XIII Volume I contains key parts of marine diesel engine and the tools used during processing. So far predicate logic is the most accurate language to express human thought activity. the system classifies workpiece into grey cast iron. reasoning can be classified into three types: forward reasoning. and structures of knowledge bases are simple. process type database. predicate. According to different order of condition and conclusion. Tool Optimal System General Design System Function. frame-type. For key parts of marine diesel engine.1 Structure of tool optimal system for marine diesel engine Representation of Knowledge and Establishment of Knowledge Base. the first part uses the rule reasoning to select tools from tool database. Predicate logic is usually expressed by implication. The system global structure chart is shown as Fig. User Choose machine tool Choose/input workpiece Workpiece exists? N Input information of workpiece Rule-based reasoning knowledge base Tool exists? N Inform user the requisite tool information Y Weight decision making Rank with weight value End store Optimization results database Manufacturing information Obtain knowledge form of workpiece Y Read information from database Machine tool database Workpiece database store Tool database store Fig. Realization of Reasoning Mechanism. tool database. and selects the best tool for user with workpiece information and processing information. etc. and it can be stored in computer for precise treatment. Process type database contains the process types such as milling flat. Reasoning of system is on basis of knowledge stored in knowledge bases. and there are several ways for the expression of knowledge including production-rule. nodular cast iron and alloy structural steel by their material to store in workpiece database. The second part takes the tools selected in first part as alternative set. Cutting speed database contains the recommended speed when one tool cuts named material. etc. such as workpiece database. store knowledge about metal cutting. According to workpiece information and manufacturing information chosen by users. reverse reasoning and hybrid reasoning[2][6].The technology of rule-based reasoning is to conclude knowledge and experience of experts to the rules of reasoning.1. processing requirements. This system adopts predicate as the expression way of knowledge[7]. and reason with these rules. which means reasoning starts from initial condition and uses .The knowledge used in expert system has the character of multi-form and multi-level.This system is composed of two parts. tool and cutting speed. milling flutes. Knowledge bases.

and whether the alternative set is built reasonably influence the accuracy of weight decision-making..2 Forward reasoning process Weight Decision Making Method about Tool Optimal Selection During establishment of mathematic model of weight decision-making. use the factors of layer B to evaluate the tools of tool alternative set. There are a lot of factors which influence tool selection. recommendatory cutting speed of tool.. tool rake angle.   r62 .    Rn  r61 r 12 . Establishment of Tool Alternative Set and Determination of Influencing Factors Set. production cost. . and obtain the evaluation matrix R: R 1  r 11 R  r R =  2  =  21  . tool rake angle. considering all factors which can influence result has the same importance as a good method for weight decision-making[8-12].Materials Science Forum Vols.. it will obtain workpiece information and processing conditions evaluation information from corresponding knowledge bases by the workpiece and processing conditions chosen by users. 626-627 607 knowledge of knowledge bases with certain strategies to obtain conclusion. r 1n  r22 . tool life. its process is shown by Fig. production efficiency.. includes surface quality.. tool relief angle. q3 …qn represents each alternative tool.. and use these information to match the rules of rule database. tool edge angle. tool relief angle. r2n   . production efficiency. Suppose Q as tool alternative set: Q = {q1 q2 q3 qn } (1) Thereinto. such as demand of surface quality. These factors are divided into two layers[5]. q2. and tool helix angle.. which is seen as processes target layer. rij represents the membership degree of tool number j under the factor i of layer B. Determination of Matrix Model between Sets..  .2... The first layer (layer A). the set is . q1.. and the second layer (layer B) includes tool cost... . production cost. tool life.. recommendatory cutting speed of tool. tool cost.. etc. Alternative set is the basic unit of weight decision-making.. Build weight distribution set of factors of layer B to processes target x of layer A. When system is running. r6n  (2) Thereinto. Start Input information of workpiece and manufacture Obtain knowledge from knowledge base Confirm tool information according to rule Tool exists? Y Reasoning succeed End N Output tool information and stop reasoning Change correlative restriction Fig. The tool alternative set of the system is composed of the tools selected by rule reasoning.After the processes target is confirmed.

Normalize the importance ranking values and obtain single rank weight values of factors of this layer to corresponding factor of the layer above. .. RI as same-order average random consistency index and λmax as maximal eigenvalue. 1  (8) Table 1 Scale and its meaning of judgment matrix from order 1 to 9 Scale Meaning 1 two factors are same importance 3 ui is ratherish important than uj 5 ui is obviously important than uj 7 ui is mightily important than uj 9 ui is extremely important than uj 2. Sc represents the evaluation set when the processes target is production cost.8 intermediate value between adjacent two judgments above reciprocal value opposite meaning of above Single Hierarchical Arrangement and Consistency Check.6.. The judgment matrix U is as follows:  1 u12 u 1 U =  21  . CI=(λmax-n)/(n-1) and CR=CI/RI.   . CI as single ranking consistency index. .Obtain the maximal eigenvalue and its eigenvectors of judgment matrix.1. The reference values of RI are shown in table 2.Set uij as the relative importance value of ui to uj.. and set the sum of all the weight values is 1... or adjust judgment matrix. and T=[t1 t2 t3]. table 1 shows the reference values of uij is. the rank weight values will be received. u1n  .. Confirm the weight value with the method of analytic hierarchy process (AHP)..Use weight value to show the influence degree of each factor. Sq represents the evaluation set when the processes target is surface quality. then the total evaluation set of tool alternative set named S is Sq   S =T • sn ] (7) Sc  = [ s1 s2 s3  Se   Determination of Weight[13]. Regard CR as consistency ratio. The method and process of confirming weight value is shown as follows... the eigenvectors are the importance ranking values of influencing factors.608 Advances in Materials Manufacturing Science and Technology XIII Volume I named Wx: WX = [ w 1x w 2x w 3x w 4x w 5x w 6x ] (3) Obtain the evaluation set of tool alternative set with the formula Sx=Wx·R when the processes target is x. (4) Sq = W q •R =  snq  s1q s2q s3q  Sc = W c •R = [ s1c s2c s3c Se = W e •R = [ s1e s2e s3e snc ] sne ] (5) (6) Set T as the relative weight set of the processes targets of layer A. Build Judgment Matrix. Thererinto....  un1 uu 2 ..4. Se represents the evaluation set when the processes target is production efficiency. . If CR<0.. u2 n   .

and system will read relative information of the workpiece from workpiece database.10. or adjust judgment matrix. for example.26 0. Realization Process about Tool Optimal Selection When start the system. S=[0.Materials Science Forum Vols.1. Take rough machining as example which means the importance of processes targets from big to small is production efficiency. system according to the suited relationship of tool and workpiece.25. Take frame part as example. the rank weight values will be received.4~Eq. 626-627 609 Table 2 1 2 0. the maximum value of CR=0. If j =1 j =1 m m CR<0. Then system will calculate the membership degree of each tool under the factor i of layer B according to their information and weight distribution set of factors of layer B to each processes target of layer A.00 Reference values of RI about judgment matrix from order 1 to 9 3 4 5 6 7 8 9 0. By calculation of U.41 1. which belongs to ductile iron and is treated by anneal.42. obtain that T=[0.24 1. the result usually contains several tools.0. we can know that the material trademark of frame is QT500-7. Choose processing property which can confirm the weight value of each processes target. choose workpiece which needs processing from dialog box. Fig.12 1. So system will calculate the total weight value of each tool.4 Result of tool optimal selection .0345<0.58 0.0.The process of total hierarchical arrangement starts from the highest layer to the lowest layer. and when machining hardened material should use tools with high hardness. so the result is received.64].00 0. production cost and surface quality.10. The tool optimal select result is shown as Fig.90 1. when machining cast iron.10 0.17. the maximum value of CR=0.16].Then choose process type such as milling flat and input 450mm as the smallest width of machined surface and 200mm as the smallest fillet radius such as 200mm. and its hardness is HB170~230. Set CR = ∑ a j CI j / ∑ a j RI j .7.4. Click command of rule reasoning.0. Then regard these tools as tool alternative set and use weight decision-making to optimize the result. according to Eq. and during total hierarchy ranking consistency check.32 1. use corresponding rules to select the tools that suit for this workpiece and process type.045<0. can’t use the tools of cemented carbide P. Because there are fewer constraints in reasoning.During single hierarchy ranking consistency check.45 Total Hierarchical Arrangement and Consistency Check.

B.Wang and Z.S. p27 [7] S. 20(2005).K.L.C. Manuf. 26(2005).Y. S. P. Cemal Cakir: Robotics and Computer-Integrated Manufacturing 20(2004). The tool optimal selection system combines the benefits of rule-based reasoning and weight decision-making. H. Huang: Artificial Intelligence and Knowledge Engineering (China Railway Publishing House. p61 [11] K.J. p26 [3] G. References [1] Q. p32 [4] LD. J. Xu and Y. Li and S. Adv. And then selects the best tool from these tools for different processes target with weight decision-making theory.Asokan. Wang and Z.M. Manuf. Yang: Journal of Harbin University of Science and Technology 10(3) (2005). The developed prototype system and an example have verified some presented techniques and the research results are the basis of the future development. Mei. Maropoulos: Journal of Materials Processing Technology 170(2000). Carpenter and P. Adv. p28 [6] Y.Da: Journal of Manufacturing Processes 9(1) (2007). China 1986) .H.Z.Saravanan and G.G.W. China 1999) [8] Ali Oral and M. R. Z.X. He: Analytic Hierarchy Process (Science Press. improves the benefits. Technol.Keung.J. p1078 [13] H. gets over the shortage of the selection system based on a single theory.C. Lin.B. W. Technol.Suresh Kumar Reddy and P. Zhou and Q. J. Zhang: Aviation Engineering & Maintenance 2(1993). 18(2001). enhances the veracity and efficiency of selecting tool. p1202 [10] X. Hou: Tool Engineering 42(2) (2008).Prabhaharan: Int.Ip and T.H. Yu: Cemented Carbide 9 (2) (1992). Manuf. Technol. Zhao. p127 [9] N.C. Zhou. p106 [2] S. Xi: Journal of South-Central University for Nationalities (Natural Science Edition) 18(2) (1999).Lee: Int.C.610 Advances in Materials Manufacturing Science and Technology XIII Volume I Conclusion The system uses rule-based reasoning to select the tools used for the workpiece according to the information of the workpiece and tools. Xu and J. p143 [5] Y.Baskar.F. p771 [12] N. Tian and H. So this system has a very important applied value.Venkateswara Rao: Int. avoids the waste because of selecting wrong tool. H. J. Adv.

Saravanan and G. R.Asokan. 20(2005).Advances in Materials Manufacturing Science and Technology XIII Volume I 10. p1078 doi:10.4028/www.Baskar. Maropoulos: Journal of Materials Processing Technology 170(2000).4028/www.scientific.626-627 Study on Intelligent Tool Optimal Selection System in Milling 10.1007/s00170-003-1939-9 .626-627. Adv.1016/S0924-0136(00)00707-X [12] N.605 DOI References [4] LD. J. Carpenter and P.Prabhaharan: Int. Manuf. p143 doi:10.scientific.G.