You are on page 1of 6

Proceedings of the 2008 IEEE/ASME International Conference on Advanced Intelligent Mechatronics July 2 - 5, 2008, Xi'an, China

Rule-fused Technology for Typical Processes Representation and Automatic Process Decision
Lijiang Huang, Shunuan Liu, Xitian Tian and Haicheng Yang
Institute of CAPP & Manufacturing Engineering Softerware Northwestern Polytechnical University Xian 710072, P.R. China {Huanglj, Liusn, Tianxt, Yanghc}@nwpu.edu.cn
Abstract In order to represent process knowledge simply and efficiently, and enhance the automation and flexibility of process making-decision, a rule-fused method for representation of typical processes and automatic decision is presented. In the method, the relative information of a typical process is divided into invariable and variable parts. The invariable part is expressed by the united model of typical process, but the variable part is expressed by production rules embedded into the typical process. Because each typical process has an independent rule set, which keeps close relationship with the process, it is easy to conclude the correlative rules of the process and to realize process making-decision automatically. Using the cases of the typical processes of aircraft structure parts, the presentation of rule-fused typical process, the classification of production rules based on manufacturing features and the automatic process making-decision are discussed in detailed. The method was applied successfully to a computer aided process planning (CAPP) system of an aircraft manufacturing enterprise. The implementation proved that the method of process knowledge presentation was simple, efficient and maintained easily, and the process decision was realized automatically. Index Terms - Presentation of typical process; production rule; process decision; manufacturing feature; computer aided process planning.

as the sample and compared the features of the part with the ones of the sample. According to the result of comparison, the process decision was made. In the presentation of the sample, only the features are included in the relevant operation or step, but the tools, the cutting parameters and machining policy for machining those features are not done. Therefore the minute decision could not be made. In a manufacturing enterprise, there are few differences among the process routes of similar parts, but the sequence of operations, the tools and cutting parameter would be changed in the process. The character of similar parts is fit to plan the process using the typical process. This paper discusses a rulefused method to present the typical process and make the process decision considering the manufacturing feature, the typical process and decision rule. The numerical control (NC) machining process planning of Aircraft structure part is regarded as an example to validate the method. II. THE RULE-FUSED PROCESS DECISION METHOD The methods of the process decision are usually divided into two types, variant one and generative one. The rule-fused process decision method synthesizes the variant one and generative one. In this method, the parts are classified according to the similarity of the parts and the processes. The typical processes are found corresponding the different types of parts while the information about manufacturing method, operation, cutting tool and cutting parameter is expressed as the rule and embedded in the typical process. The process decision is made by utilizing the similar typical process and the rule-based reasoning. The process of the rule-fused process decision method is shown as Fig.1. The key steps are as followed: Step 1: classify parts and conclude typical process. Parts are classified according to the features of parts and the character of processes so that the part groups are formed. Then, the typical processes are concluded in every parts group while typical process routes, steps, operations and machining methods of features are done and included in the typical process. Part group and typical process are helpful to summarize the decision rules. Step 2: present the typical process based on rule-fused technology.

I. INTRODUCTION The convenient and effective methods for automatic process decision have been explored and researched in the field of Computer Aided Process Planning (CAPP). Many methods and technologies are applied such as Expert System, Artificial Neural Network, Case-based Reasoning. However, the most of the achievements are not in practice but academic. [3] Zhenming Zhang developed a feature-based CAPP expert system for aircraft parts and built the rule-based process repository to implement the automatic process decision. In the system, because the repository need satisfy all the aircraft parts and their process planning, a mass of rules were needed and the redundancy and conflict would happen among rules. [5] And it was difficult to obtain the rules. Yajun Pan adopted [6] the case-based process decision method. Xiangyun Sun adopted the process decision method based on typical process model. Both of their methods obtained the features of the part from CAD firstly, selected the typical process or process case

978-1-4244-2495-5/08/$25.00 2008 IEEE.

1367

The information of typical process can be separated into


Process Planning KB Part feature model

features can describe the shape of a part and decide the

Grouping Part

C A D

3D designing Feature Recognizing

Process planning model Defining typical process planning Typical process Typical routes Typical operations Production rules

Getting part features Searching typical process Automatic Process Decision Interactive modifying

Process planning decision

Outputting process planning Other knowledge C A M Calculating Tool path NC Programming

Fig. 1 The process of the process decision based on rule-fused technology

invariable one and variable one. Invariable information is presented based on the uniform typical process model. Variable information is presented in the way of production rule. Therefore each typical process is associated with a set of rules. The information model of features of a part must be built to describe the information of a part and search the typical process. Step 3: build the typical process repository. In the typical process repository, the data about part group, typical process and production rule is stored. Step 4: make the process decision. After obtain the information of the part and the feature, the most similar typical process with the one of the part is found by searching in the typical process repository. So the process route and manufacturing steps of the part can be known. And, the details about the process of the part can be generated by automatic reasoning in terms of the decision rules associated with the typical process. Step 5: modify the process plan. The details about the process of the part can be still amended in the interactive way if they are not enough full. As above discussion, it is known that the information model of the part and the features is a basic part, the rule-fused presentation of typical process is a key part, and the process decision algorithm is an important part. In the following sections, the NC machining process of aircraft structure parts will be taken as an example to expound these parts of this method. III. THE INFORMATION MODEL OF THE PART AND THE FEATURES The information model of the part and the features is the unified, formalized presentation of all the datum of the part and their relation to be recognized and deal by computer. The

manufacturing process so they are used to define the production rule in typical process. Therefore, the information model is the base of process decision. The Object-oriented technology is adopted in building the information model of the part and the features. One part can be regarded as an entity object that consists of the basic part information and some manufacturing features. The part and the features are
part -part No. : char -part name : char -product : char -... +selecting typical process() : bool +getting part features() +auto-coding features() +...()

1 general feature -feature No. : char = G -feature name : char -code length -... +coding general feature()

0..* feature A -feature No. : char = A -feature name : char -... +coding feature() +selecting Manu.way() 1..* 1..* feature attribute -attribute name : char -value type : char -attribute code lenth : int +coding attribute() 1..*

0..* feature B -feature No. : char = B -feature name : char -... +feature No.() +selecting Manu. way()

0..*

feature ... -feature No. : char = N -feature name : char -... +coding feature() +selecting Manu. way()

1..*

Fig. 2 Object Oriented model of the part and the features

abstracted into class. Part class is composed of the basic part information class (BPIC) and manufacturing feature classes (MFC). Each manufacturing feature class has own member variables and functions. Member variable describes the property of the feature. Member function can describe the feature code and manufacturing method. The property of the feature can be abstracted into a class which has own member

1368

variables and functions, too. The Object-oriented model of the part and the features is shown as Fig.2. Aircraft structure parts, including girder, frame, joint, wainscot, rib and so on, are the key parts in the aircraft development. They are generally manufactured by NC machining method so that the workload is great to make the process planning and NC program that occupies 15 to 20 % of the whole aircraft production. The information model of the aircraft structure parts is built shown as Fig.3. One BPIC and six MFC are concluded and defined. Six MFC are the pocket class, the rib class, the figure class, the convexity class, the process convexity class and the hole class. And the property and functions of the slot class are expounded in Fig.3.

Definition 2: Step. It is an ordinal set of all operations done continuously for machining a workpiece in a machine and one set-up. Definition 3: Operation. It is an ordinal set of all operations done for machining a MC on the condition that the cutting tool is not changed. A. Build the model of rule-fused typical process Rule-fused typical process is a kind of special process in that rules are added. The model of rule-fused typical process is built based on the process information model of the part, shown as Fig.4. On the upper section of Fig.4 is the objectoriented process information model of the part. The process of the part is abstracted into process basic information object,
Pocket bottom Attr. name: BOTTOM Attr. value: Enumeration Code length: 1 Attr. type Attr.object Attr.object Attr.object Attr.object Attr.object CHAR INT FLOAT FLOAT FLOAT FLOAT FLOAT FLOAT FLOAT The way of coding for attribute IF value=flat THEN code=0 IF value=pitched THEN code=1 IF value=curved THEN code=2

Part

Basic Info. Pocket Face Rib


Process Convexity

Pocket Attr. alias Attr. name Pocket bottom BOTTOM Additional element AddElement Border upon face BorderUponFace Pocket wall PocketWall Hatched Hatched Feature name FEATNAME Feature vector VECNO Pocket height Height Out tolerance OutTolerance In tolerance InTolerance Roughness Roughness Max. allowing cutter MaxDia diameter Min. corner radius MinCornerRadius Fillet radius FilletRadius Auto-coding for pocket feature Process Planning for pocket feature

Additional element Attr. name: AddElement Attr. value: Enumeration Code length: 2 The way of coding for attribute IF value=Nothing THEN code=00 IF value=caved THEN code=01 IF value=flanged THEN code=02 Border upon face

Convexity

Hole

Fig. 3 The manufacturing feature information model of aircraft structure part

IV. THE PRESENTATION OF RULE-FUSED TYPICAL PROCESS To illuminate the rule-fused presentation of typical process better, the concepts of machining cell, step and operation are defined as followed. Definition 1: Machining Cell (MC). It is called Machining Cell that the process information set is formed by continuously machining a feature. For example, in the operation of milling the bottom of a slot, a slot is a feature so machining the slot can be regarded as a MC. However, in different operation, machining the same feature should be regarded as different MC. If Milling the bottom of a slot and milling the corner of a slot are two operations on the same slot, they should be done as two MC.

step object, operation object and MC object etc.. And the relation between objects is expressed by the special type of the objects property, object and object table. On the lower section is the model of rule-fused typical process. It is consist of two parts, typical process and rules. Typical process confirms the typical machining method, the arrangement of the process route and the process information including the steps, the operations, the equipments and the tools etc.. Rules are used to express the variable information that is relative with concrete feature of the part. In planning the process, the process content can be generated by applying rule-base reasoning and the feature information. The meaning of the symbols is respectively explained. Real line circle represents the typical process information. Broken line circle represents the rules. Shadow part represents MC, which is gotten by rulebase reasoning.

1369

Part Process

1:1

Attributes of operation object

Basic Info.
1:1

Attribute name

Attribute alias Operation No. Operation name Operation content Operation type Cutting tool Measurement MC

Attribute type Int. Char(64) text Char(32) Object Object Object List

1:n

Equipment Step
1:1 Clamping Tool 1:1 1:n

OPER_NO OPER_NAME OPER_CONT OPER_TYPE

Cutting tool Operation


1:1

CUTTOOL MEASURE

Measurement

M_C 1:1

the object-oriented process information model of the part

1:n

M.C.

Cutting parameter

Choosing typical process rule

MC generating rule Step rule MC sequencing rule

Basic info. of Process Fusedrule typical process

Basic info. of operation Equipment

Operation rule Basic info. of operation Cutter

MC evaluating rule Cutting parameter selection rule Machining policy selection rule

Typical step 1 Clamping Tool Typical step 2 Typical operation 1

MC 1 Typical operation 2 Typical step n Typical operation n MC n

Feature ID Machining policy

MC 2 Machining way Cutter parameter

the model of rule-fused typical process

Fig.4 The model of process information and the rule-fused typical process

For representing typical process, process route as same as step sequence is not changed. For the typical steps, we can get the next possible step by composite method of process according to characteristics of part-group and process. Then an ordered set made up of typical steps can be gotten by machining sequence. In order to simplify typical process representation and improve efficiency of process decision, typical operations will be separated into doubtless cutting-tool steps and selective cutting-tool steps at the moment of representing typical steps. Doubtless cutting-tool steps have doubtless machining method, cutting-tool type and specification; selective cutting-tool steps have doubtless machining method, cutting-tool type, but not doubtless cutting-tool specification. The typical process doesnt provide content, quantity and sequence of MC. Content and quantity of MC are created by rule reasoning and sequence of MC are created by the optimizing algorithm of machining path. B. The feature-based classification and presentation of decision rules

In this paper, the rules are mainly classified into three kinds that are respectively choosing typical process rule, step rule and operation rule. Choosing typical process rule is to confirm on which condition the typical process should be chosen. Step rule is to restrict the machine tool, the equipment and the fixture. Operation rule is to control the content of the operation including MC formation, sequence, and evaluation, cutting parameter selection and machining policy selection. Table I is the feature-based classification of decision rules. The decision rule is represented in the form of production rule. A kind of process knowledge presentation method is formed that is mixing typical process and production rules. In this method fusing rules in typical process, each typical process has own a set of independent rules that is closely related with itself. Thereforethis method is simple, comprehensible and in favor of concluding rules conveniently, increasing the pertinence of rules, and maintaining rules easily.
TABLE I
THE FEATURE-BASED CLASSIFICATION OF DECISION RULES

1370

Rule specification The rule decides which typical process is chosen for choosing typical process rule the part. The rule restricts the machine tool, the equipment step rule and the fixture. The rule decides which features are planned into the operation MC generating rule operation and generates the MC of the feature. rule MC sequencing rule The rule sequences the MCs in an operation. MC evaluating rule The rule evaluates the attribute values of a MC. cutting parameter The cutting parameter of a MC can be selected by selection rule the rule. machining policy The machining policy of a MC can be selected by selection rule the rule. .AUTOMATIC PROCESS DECISION BASED ON RULE-FUSED

Rule classify

Rules example IF part.Type = frame part and part.MachinedFaceNum = 2 THEN true. IF equipment. MaxWorkingsize > part.length and equipment. MaxWorkingangle > part.Maxangle THEN true. IF feature.Type = Pocket and feature.Height < 30 and feature.
MaxAllowedDia < 20 and feature. FilletRadius > 5 THEN true.

SequenceMCby(rib) IF feature.Type = rib THEN MC.MachiningWay = mill rib SelectMCPara(Material, Equipment, CuttingTool, Manchining Way, FeatrueType) SelectMCPolicy(ManchiningWay, FeatrueType)
N

typical process Automatic process decision based on rule-fused typical process mainly follows these steps: 1) Obtain the features of the part from CAD system. Analyze and deal the information. Finally form the list of features. 2) Traverse the typical processes in the typical process repository. According to the feature information, the typical process rules are invoked for choosing the similar typical process with the part. 3) Save the invariable information of the typical process as the process of this part. Confirm the equipment, the fixture in the steps by the reasoning based on the step rules. And confirm the operations, MC, MC sequence by the reasoning based on the operation rules. 4) In the interactive way, modify the process information generated in automatic decision. Fig.5 shows the process of automatic process decision based on rule-fused typical process. In the process, the steps, choosing the typical process, the equipment, the fixture, the feature, the cutting character, sequencing MCs, and evaluating MC, are realized by rule-based reasoning. Operation decision is important in the whole process decision, which includes the arrangement of operations and the generation of MC information. In typical process, operation sequence has been confirm, the operation decision just generates the concrete detail and decide whether adopts the typical operation. The process of operation decision is followed as below. 1) Select feature. Reasoning based on the MC generation rules in the operation rule. Then, in the feature list, select the feature that could be machined in this operation. If no fit feature in this operation, the operation should be delete in the operation chained list. This operation reasoning is end. 2) Generate MC. If it is a doubtless cutting tool operation, the MC can be generated directly. If it is a selective cutting tool operation, the new operation should be generated. And it replaces the corresponding operation in the typical process and insert into the operation chained list on the corresponding opposition. Furthermore, the specification of the cutting tool is gotten by the feature-based reasoning such as the size of a hole decides the diameter of the drill.At last, MC is generated.

Have steps circulated

Y Interactive modifying

process planning Saving process planning


N

Emd

Have operations circulated

Start Selecting machining policy Obtaining features Selecting cutting parameters

Have MC.s circulated

Choosing the similar typical process

Evaluating M.C. Circulating M.C.s Ensuring cutter spec Sequencing M.C.s

Saving as the process of the part

Circulating steps

Generating operation
selective cutting tool operation

Generating M.C.
doubtless cutting tool operation

Selecting equipment and tools

Have existed
N

Delete the operation

Select demand features Circulating operations

Fig.5 The rule-fused process decision flow

3) Sequence MCs. After all MC are generated in one operation, MCs must be sequenced according to the MC sequence rules in operation rules. 4) MC reasoning. MC information in every MC is obtained by reasoning based on the MC evaluation rules, the cutting parameter selection rules and the machining policy rules. 5) Save the operation information. The rule-fused technology for typical processes representation and automatic process decision is applied successfully into process decision of NC manufacturing for aircrafts parts. According to the characteristics of parts and

1371

processes, the technology can classify aircrafts parts and build the database of typical processes, such as the typical processes of frame parts including double-face frame, singleface frame, half frame, C-shape frame, and so on. We develop a CAPP system for NC manufacturing for aircrafts parts. This system includes many functions: information model management, typical process management, process decision, process management and so on. The function for information model management can manage and maintain the feature model and process information model of parts. The function for typical process management can manage and maintain typical process classifying, foundation and rules definition of parts and process knowledge; it is a function module managing the database of process knowledge. The function for process decision can get feature information of part and complete process decision reasoning of part. The function for process management can interactively modify and output process. The technology has many excellences: simple, agility, convenient for managing and maintaining the database of process knowledge. It is very easy to realize automatic process decision by this technology. .CONCLUSION This paper investigated a rule-fused technology for typical processes representation and automatic process decision. And we mainly researched the fundamental and applied method of this technology. It is applied successfully to process decision of CNC manufacturing for aircrafts parts. In this application, we focused on part information model based on manufacturing characteristics, a method of typical processes representation based on rule-fused technology, classifying production rules based on manufacturing characteristics of parts and the algorithm of process decision. The typical process decision based on rule-fused technology has many specialties: 1) it is a method based on the decision methods, such as variant method, generative method, and expert system; 2) Invariable process information is decided by typical processes, but variable process information is decided by production rules; 3) the process information of unsure decision is decided by driving-rules of parts characteristics. The method was applied successfully to a CAPP system of an aircraft manufacturing enterprise. The implementation proved that the method of process knowledge presentation was simple, efficient and maintained easily, and the process decision was realized automatically. The technology especially adapts to process design of parts with good comparability, manufacturing complexity, and large workload of designing NC. REFERENCES
[1] H. B. Marri, A. Gunasekaran and R. J. Grieve. Computer-Aided Process Planning: A State of Art. Int J Adv Manuf Technol, vol.14, pp.261-268, 1998 [2] Pratt M.J and Wilson P.R Requirements for support of form features in a solid modelling systems R-85-ASPP-01 CAM-I Inc. Arlington, Texas, USA, 1985. [3] ZHANG Zhengming, HUANG Naikang, XU Jianxin. ME Based Decision_Making Model and Its Application for Feature_Based Machining Process Planing of Mechanical Parts. AERONAUTICAL MANUFACTURING TECHNOLOGY, pp.17~20, May 2000. (in Chinese)

[4] JIA Xiaoliang, XU Jianxin, ZHANG Zhengming etc. The Research and Application of CAPP Expert System Based on CAPPFramework. Computer Engineering and Applications, pp. 5~7,36, June 2003. (in Chinese) [5] PAN Yajun, PAN Xiaohong, LI Shanchun. On the Architecture of Casebased Integrated CAPP Systems. Mechanical science and technology, pp. 96-499, May 2000. (in Chinese). [6] SUN Xiangyun, ZHENG Lianyu, WEI Li. Semi - generative CAPP system for automobile chassis based on typical process model. Modular Machine Tool & Automatic Manufacturing Technique, pp. 70-73, April 2004. (in Chinese) [7] XU Jianxin, KONG Xianguang, ZHANG Zhenming etc. A Development Platform for Process Planning in P.R China. Journal of Northwestern Polytechnical University, pp. 132~136, February 2002. (in Chinese) [8] Ali Oral, M.Cemal Cakir. Automated cutting tool selection and cutting tool sequence optimization for rotational parts. Robotics and ComputerIntegrated Manufacturing, vol. 20, no.2, pp.127141, April 2004 [9] ZHANG Zhengming, HUAN Yongxin, XU Jianxin etc. An ObjectOriented modeling System for CAPPFramework. Mechanical science and technology, pp. 1019~1021, November 2000. (in Chinese) [10]M.R. Alam et al. Automated process planning for the manufacture of sliders. Computers in Industry, vol. 43, pp.249-262, 2000. [11]K. S. Lee, M. R. Alam, M. Rahman et al. Automated Process Planning for the Manufacture of Lifters. Int J Adv Manuf Technol, vol. 17, no.10, pp.727734, May 2001.

1372

You might also like