یک الگوریتم تکاملی برای تولید کارگاهی مونتاژ همراه با اشتراک گذاری بخش
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|19020||2009||11 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Computers & Industrial Engineering, Volume 57, Issue 3, October 2009, Pages 641–651
Assembly job shop problem (AJSP) is an extension of classical job shop problem (JSP). AJSP first starts with a JSP and appends an assembly stage after job completion. Lot Streaming (LS) technique is defined as the process of splitting lots into sub-lots such that successive operation can be overlapped. In this paper, the previous study of LS to AJSP is extended by allowing part sharing among distinct products. In addition to the use of simple dispatching rules (SDRs), an evolutionary approach with genetic algorithm (GA) is proposed to solve the research problem. A number of test problems were conducted to examine the performance of the proposed algorithm. Computational results suggested that the proposed algorithm can outperform the previous one, and can work well with respect to the objective function. Also, the inherent conflicting relationship between the primary objective and the system measurements can be addressed.
For classical job shop problem (JSP), there are m machines and n jobs. Each machine can process only one operation. A job (or lot), which is defined as a batch of identical items, should be processed on all machines until all of its operations are completed. Also, each job can only visit each machine once and the processing sequence of jobs should be strictly followed. Lot streaming (LS) technique depicts a process of splitting lots into sub-lots such that successive operations of the same lot can be processed in parallel on different machines. Over the past few decades, there has been an increasing emphasis on the application of LS to JSP, and the results are promising. In this paper, assembly job shop problem (AJSP), which is an extension of JSP, is investigated. AJSP first starts with JSP and appends an assembly stage after job completion. Therefore, the completed jobs (after the JSP stage) must be assembled if they belong to the bill-of-material (BOM) of the same product. The product assembly can start once all jobs of the same BOM are completed or available after the JSP stage. In the previous study ( Chan, Wong, & Chan, 2008), part sharing is not allowed such that completed jobs from distinct BOMs cannot be assembled. To allow part sharing, jobs are classified in two types: Unique and Standard. Unique job type is specific to only one product and only standard job type can be shared among distinct products. Suppose the BOM of Product 1 or P1 contains Job 1 or J1 (unique type) and J2 (standard type one) while the BOM of P2 includes J3 (unique type) and J4 (standard type one). Since both J2 and J4 are of the same standard job type, J2 can substitute J4 for the assembly of P2 or J4 can replace J2 for the assembly of P1. In this connection, the assembly of each product may involve jobs from other BOMs. Obviously, part sharing has enhanced the complexity of the research problem. To solve this problem efficiently, a genetic algorithm (GA) approach is developed with dedicated crossover and mutation operators. The paper is structured as follows: the literature review is discussed in the next section. In Section 3, the problem background and formulations are presented. Section 4 depicts the development of the proposed approach. Computation results are reported in Section 5. Section 6 discusses the results, and concludes the paper together with future research works.
نتیجه گیری انگلیسی
In this paper, a GA-based approach has been developed and proposed to solve AJSP with LS technique. Two experiments were presented to examine the dominances of the current algorithm (NAL) over our previous algorithm (PAL) for a number of smaller problems, and the performances of NAL on a number of bigger problems. Computational results suggest that NAL outperforms PAL significantly, and NAL with ES mode overwhelms NAL with No and VS modes. Moreover, NAL has been proved to work well with respect to the single objective function (Z: lateness cost). The study of the four system measurements can be used to examine the positive and negative impacts of LS on AJSP. Specifically, the conflicting relationships between Z–SWIP (WIP) and Z–SSetup (setup cost) have been explicitly addressed. Since there is no similar GA-based approach to AJSP with LS technique, this study may provide some useful insights about the application of GA to solve LS and AJSP simultaneously. To enhance the usefulness of the model, resource constraints such as fixtures and operation tools will be studied. Also, we will consider multi-objective function among several conflicting system objectives addressed in this study. Last but not least, continuous effort will be made to improve the operational efficiency of the proposed model.