• Tidak ada hasil yang ditemukan

2.3 Approaches based on Artificial Intelligence and Soft Computing

2.3.4 Fuzzy Logic based Approaches

Table 2.8. Major Setup planning approaches using evolutionary algorithms

References Type of part Main constraints considered Vancza and Markus [1996]

Zhang et al. [1997]

Dereli and Filiz [1999]

Shunmugam et al. [2000]

Shunmugam et al. [2002]

Li et al. [2005]

Li et al. [2005]

Bo et al. [2006]

Mohemmed et al. [2008]

Guo et al. [2009]

Guo et al. [2009]

Vijaykumar et al. [2003]

Lv and Zhu [2005]

Krishna and Rao [2006]

Prismatic Prismatic Prismatic Prismatic Rotational

Prismatic Prismatic Prismatic

Both rotational and prismatic Prismatic Prismatic

Rotational Prismatic Prismatic

Feature relations, resources, cost Precedence relations, time, cost

Feature interaction, TAD, part geometry Cutting force, surface finish, tool capacity Tolerance, precedence relations, surface finish

Resources, machining cost Resources, machining cost Precedence relations, optimality

Shortest route

TAD, manufacturing resources, cost TAD, manufacturing resources, time

Process parameters, cost

Workshop resources, time and cost Precedence relations, tolerance

Chapter 4. Fuzzy logic have been used in CAPP for automating process planning, setup planning, selection of cutting parameters for machining, etc.

Setup planning and feature relations have been the focus of extensive research by Ong and Nee [1994, 1995, 1996, and 1997]. Precedence relation between two features may depend on many uncertain factors. This fact was realized by Ong and Nee as early as 1994. They represented the imprecise feature relations using fuzzy sets and fuzzy relations to generate setup plans for prismatic parts fulfilling the requirements of process planning and fixture planning. The authors applied the concept of feature dependency grades to deal with uncertain feature relations.

Feature relations and their dependency grades generate the feature precedence relations for grouping features into setups. Different feature relations such as geometric relation, tolerance relation, datum and fixturing relations, and heuristic relations are considered. Setup planning knowledge is represented as production rules in these approaches. The same authors [1998] stressed that fixturability analysis of the different features of a part and the part as a whole is an inevitable part of setup planning, neglecting which may lead to infeasible setup plans. They proposed a fixturability evaluation method for a part in two levels using fuzzy logic.

At the micro level, each feature is accessed for its suitability to be used as a fixturing element considering its surface area, surface finish, face-feature relationship, orientation, intricacy, symmetry, etc. At the global level, the overall fixturability of the part is accessed after setup plans are made for the part based on stability of the part and the cost estimation of the fixturing system. Even if the methodology is for prismatic parts, it is applicable to castings and non-prismatic parts too.

Zhang and Huang [1994] solved the process plan selection problem with a fuzzy set based approach by evaluating the process plans quantitatively based on maximum contribution to the shop floor. First, a set of process plans with maximum contribution to the shop floor is identified, and then a progressive refinement strategy is applied to the set to reduce the manufacturing resources needed. The objectives considered are minimum number of setups, machining steps, and machining time. Zhao [1995] presented a similar approach of optimal process plan selection using fuzzy sets and fuzzy decision making strategy. Initially, a set of alternative process plans are generated using alternative machining operations,

machine tools, and cutting tools. Fuzzy decision making strategy is applied to find the optimal process plan with the objectives of minimum machining time, minimum number of setups, and minimum dissimilarity among the machine tools and cutting tools.

Gu et al. [1997] emphasized on feature prioritization for simplifying operation sequencing in machining of prismatic parts. The authors presented a combined fuzzy logic and neural network model to evaluate feature priorities based on manufacturability of the features and identify the important features. Tool capability, symmetry, orientation, accessibility, intricacy, etc. are some of the parameters for evaluating manufacturability of features. Criteria used for operation sequencing are minimum number of setups and tool changes and non-violation of feature precedence relations.

Wu and Zhang [1998] proposed an object oriented approach for generating alternative setup plans and fuzzy set theory is applied to find the optimal setup plan for machining of prismatic parts. The authors are of the view that a rigid optimal setup plan may not be optimal in a changed manufacturing environment, so there should be alternative optimal plans to adapt to the dynamic manufacturing resources.

Three types of classes are defined to represent the setup planning knowledge in object oriented approach, viz. process control class, setup generation class and setup evaluation class. A setup evaluation function is formulated to evaluate the generated setup plans using fuzzy set approach considering dimensional and geometric tolerances, precision of the part, satisfaction of the basic setup planning rules, etc.

After evaluating a set of candidate setup plans, an optimal setup plan with the highest evaluation function value can be obtained.

Wong et al. [2003] used a hybrid of fuzzy expert system FuzzyCLIPS and GA for machining operation selection and operation sequencing under uncertainty for prismatic parts. Optimal operation sequence is found based on manufacturing resource and cost. First, a local optimal operation sequence is generated satisfying feature precedence relations and then a global optimal operation sequence is searched based on manufacturing resources and cost.

Gaoliang et al. [2005] used a hybrid of fuzzy set theory and Hopfield neural network for setup planning of prismatic parts. The setup planning problem is divided into setup formation, operation sequencing and setup sequencing sub-problems. The setups are formed based on the optimal TAD of the features. A fuzzy set based algorithm is used to find the optimal TAD of features. Using production rules and fuzzy set theory, the feature precedence relationships matrix (FPR) is formed considering feature geometry, tolerances, datum relationship, heuristic rules and manufacturing cost. Operation sequencing and setup sequencing is done based on the FRP. A similar setup planning approach is developed by Wenjian and Gaoliang [2005] which can be integrated with the Internet. Java and Web technologies coupled with XML (eXtensible Modeling Language) file format provide means for the transfer of information between various manufacturing systems. A fuzzy set theory based approach for setup planning is introduced considering TAD, tolerance requirements, precedence relations, manufacturing cost, and good manufacturing practice. Another Internet-based integrated setup planning and fixturing system can be found in Gaoliang et al. [2005]. The proposed system has client/server architecture comprising an information server, a database server, and a number of setup planning clients. The use of Java and XML (eXtensible Modeling Language) file format adds flexibility to the system and operable under different platforms.

Yuru and Gaoliang [2005] developed an integrated setup planning and fixture design method using a hybrid of knowledge-based and fuzzy set based approach. Tolerance among the features, feature precedence relations, fixturing requirements, and manufacturing cost are the main constraints considered. Table 2.9 shows some fuzzy logic based setup planning efforts.

The main weaknesses of fuzzy logic based methods are that they are restricted to the fields where expert knowledge is available and are unable to automatically acquire knowledge. The problem of finding appropriate membership functions for the fuzzy variables also poses a challenge to the researchers.

Table 2.9. Major Setup planning approaches using fuzzy logic

References Type of part Main constraints considered Ong and Nee [1994, 1995,

1996, and 1997]

Zhang and Huang [1994]

Zhao [1995]

Gu et al. [1997]

Ong and Nee [1998]

Wu and Zhang [1998]

Wong et al. [2003]

Gaoliang et al. [2005]

Wenjian and Gaoliang [2005]

Gaoliang et al. [2005]

Yuru and Gaoliang [2005]

Amaitik and Kilic [2007]

Prismatic

Prismatic Prismatic Prismatic Prismatic Prismatic Prismatic Prismatic Prismatic Prismatic Prismatic Prismatic

Precedence relations due to feature dependency, tolerance, fixtures, Resources, uncertainty, machining time Resources, uncertainty

Feature prioritization, part geometry, TAD Fixturability, precedence relations, tolerance Precedence relations, TAD, tolerance Precedence relations, uncertainty, resources Precedence relations, TAD, tolerance Precedence relations, TAD, tolerance, cost Precedence relations, TAD, tolerance

Precedence relations, tolerance, cost, fixturing TAD, tolerance, surface finish, part geometry