سیستم های پشتیبانی تصمیم گیری مبتنی بر GA برای برنامه ریزی تولید پیش ساخته

Appropriate production plans can produce effective resource utilization and minimize waste. However, most precast fabricators currently propose production plans depending on the rule of thumb, resulting in squandered resources and postponed delivery. Computerized scheduling techniques provide more precise outcomes than manual scheduling. The objective of this study is to develop GA-based Decision Support Systems (GA-DSS) to assist production managers in arranging production plans. This research first establishes a flowshop sequencing model based on the current production status by considering the buffer sizes between production stations. A multiple objective genetic algorithm is then applied to search for solutions with minimum makespan and tardiness penalties. The GA-DSS performance is verified using two examples. The results demonstrate that the proposed system can offer appropriate production plans. By taking buffer sizes into consideration more reasonable and feasible production sequences can be achieved.

The formwork method has been applied in building construction for a long period of time. However, this traditional construction method is not competitive as the labor costs and schedule inefficiency rise each year. Therefore, the formwork method has been gradually replaced by precast construction technologies to overcome the labor shortage and environmental uncertainty problems on building sites [1]. Precast construction is an enhanced method accomplished by erecting prefabricated concrete elements [2]. To support a construction schedule, precast fabricators deliver elements to a site according to an erection schedule. Making production plans is one of the most important tasks in manufacturing. Throughput, makespan, and waiting time are dominated by production plans. To enhance fabricator competitiveness, production planers face the challenges of satisfying multiple objectives in which one objective may conflict with the others [3]. Technology development in the manufacturing industry has focused attention on the inefficiencies of traditional scheduling in dealing with current complex production systems. As the scheduling problems become more complicated, the optimized solution cannot be achieved within a reasonable time. Therefore, studies on scheduling have shifted to applying heuristic algorithms to obtain near-optimal solutions within a short time. Leu and Hwang [4] regarded the three working zones in a precast factory as a flowshop sequencing model. A genetic algorithm was applied to achieve the solution for this model with minimum makespan. In another research [5] written by the same authors, a genetic algorithm was tested in several projects to minimize the time spent producing precast elements. Chan and Hu [6] applied the conventional flowshop sequencing scheduling method as the basic production model, which was adjusted depending on different precast production process features. The work activities in their model were redefined into two types, interruptible and uninterruptible activities. The work day was divided into two sections, working and non-working hours. A genetic algorithm was used in their study to search for better production sequences. In the research conducted by Benjaoran et al. [7], the Bespoke Precast Flowshop Scheduling Model (BP-FSSM) was established, in which the features of production methods in prefabrication factories were considered. A Multi-Objective Genetic Algorithm was also applied to achieve a solution in their work. However, precast fabrication requires a rather large manufacturing space. Previous studies ignored the buffer size between work stations, resulting in unrealistic production plans. The objective of this study is to develop a GA-Based Decision Support System (GA-DSS) to assist production managers in making appropriate production plans. A limited buffer size between stations is considered in this system. This paper first introduces multi-objective genetic algorithms. Current precast production practices are then discussed. Section 4 presents the precast production process using mathematical forms. A decision support system is developed in section 5 to facilitate decision making. A case study is used to validate the applicability of the developed system. Finally, conclusions induced from the experiment are documented.

This study established production models based on the practical production status in precast factories. A multi-objective genetic local search algorithm was applied to search for production plans with minimum makespan and tardiness penalty. At the end, GA-DSS, a production planning system with graphical user interface was developed using JAVA and C languages. Two examples were applied to validate the GA-DSS performance. The experimental results demonstrate that the multi-objective genetic local search algorithm developed in this study is efficient in offering solutions to complex production planning problems. The buffer size between stations is crucial for production planning and considered in the proposed method. The two outcomes mentioned above indicate that the proposed GA-DSS is capable of offering feasible solutions to precast production for enhancing decision making. By applying suitable algorithms in which real situations are involved, a set of Pareto-optimal solutions can be provided for production managers as a reference when creating production plans. Production managers can thus select the most advantageous plan based on the production constraints and objectives. By comparing current manually planning, the proposed GA-DSS effectively utilizes production resources. For future study, the derived Pareto-optimal solution accuracy could be tested using the ratio of the actual Pareto solution number over the derived solution number. Solution spread and spacing could also be future goals for testing the software performance. Steel molds play a key factor in the precast production process. Further studies can focus on how to determine the optimal number of molds by applying GA-DSS. The Pareto-optimal solution evaluation is not fully addressed in the related references, such as Zitzler et al. [30], Deb et al. [31], and Ishibuchi et al. [32]. Further research on evaluating the quality of derived Pareto-optimal solutions will be addressed in the future.