Tian Xiang Cai

Pareto-based grouping discrete harmony search algorithm for multi-objective flexible job shop scheduling

Abstract This paper proposes a Pareto-based grouping discrete harmony search algorithm (PGDHS) to solve the multi-objective flexible job shop scheduling problem (FJSP). Two objectives, namely the maximum completion time (makespan) and the mean of earliness and tardiness, are considered simultaneously. Firstly, two novel heuristics and several existing heuristics are employed to initialize the harmony memory. Secondly, multiple harmony generation strategies are proposed to improve the performance of harmony search algorithm. The operation sequence in a new harmony is produced based on the encoding method and the characteristics of FJSP. Thirdly, two local search methods based on critical path and due date are embedded to enhance the exploitation capability. Finally, extensive computational experiments are carried out using well-known benchmark instances. Three widely used performance measures, number of non-dominated solutions, diversification metric and quality metric, are employed to test the performance of PGDHS algorithm. Computational results and comparisons show the efficiency and effectiveness of the proposed PGDHS algorithm for solving multi-objective flexible job-shop scheduling problem.

Discrete harmony search algorithm for flexible job shop scheduling problem with multiple objectives

Flexible job-shop scheduling problem (FJSP) is a practically useful extension of the classical job shop scheduling problem. This paper proposes an effective discrete harmony search (DHS) algorithm to solve FJSP. The objectives are the weighted combination of two minimization criteria namely, the maximum of the completion time (Makespan) and the mean of earliness and tardiness. Firstly, we develop a new method for the initial machine assignment task. Some existing heuristics are also employed for initializing the harmony memory with discrete machine permutation for machine assignment and job permutation for operation sequencing. Secondly, we develop a new rule for the improvisation to produce a new harmony for FJSP incorporating machine assignment and operation sequencing. Thirdly, several local search methods are embedded to enhance the algorithm’s local exploitation ability. Finally, extensive computational experiments are carried out using well-known benchmark instances. Computational results and comparisons show the efficiency and effectiveness of the proposed DHS algorithm for solving the FJSP with weighted combination of two objectives.

A two-stage artificial bee colony algorithm scheduling flexible job-shop scheduling problem with new job insertion

A heuristic is proposed for initializing ABC population.An ensemble local search method is proposed to improve the convergence of TABC.Three re-scheduling strategies are proposed and evaluated.TABC is tested using benchmark instances and real cases from re-manufacturing.TABC compared against several state-of-the-art algorithms. This study addresses the scheduling problem in remanufacturing engineering. The purpose of this paper is to model effectively to solve remanufacturing scheduling problem. The problem is modeled as flexible job-shop scheduling problem (FJSP) and is divided into two stages: scheduling and re-scheduling when new job arrives. The uncertainty in timing of returns in remanufacturing is modeled as new job inserting constraint in FJSP. A two-stage artificial bee colony (TABC) algorithm is proposed for scheduling and re-scheduling with new job(s) inserting. The objective is to minimize makespan (maximum complete time). A new rule is proposed to initialize bee colony population. An ensemble local search is proposed to improve algorithm performance. Three re-scheduling strategies are proposed and compared. Extensive computational experiments are carried out using fifteen well-known benchmark instances with eight instances from remanufacturing. For scheduling performance, TABC is compared to five existing algorithms. For re-scheduling performance, TABC is compared to six simple heuristics and proposed hybrid heuristics. The results and comparisons show that TABC is effective in both scheduling stage and rescheduling stage.

An improved artificial bee colony algorithm for flexible job-shop scheduling problem with fuzzy processing time

Improved ABC algorithm is proposed for FJSP with fuzzy processing time.A heuristic, named MInEnd, is proposed to initialize population.New strategies are proposed to generate new solutions.The objectives are fuzzy maximum completion time and maximum machine workload.Benchmarks and realistic remanufacturing instances are solved by IABC. This study addresses flexible job-shop scheduling problem (FJSP) with fuzzy processing time. An improved artificial bee colony (IABC) algorithm is proposed for FJSP cases defined in existing literature and realistic instances in remanufacturing where the uncertainty of the processing time is modeled as fuzzy processing time. The objectives are to minimize the maximum fuzzy completion time and the maximum fuzzy machine workload, respectively. The goal is to make the scheduling algorithm as part of expert and intelligent scheduling system for remanufacturing decision support. A simple and effective heuristic rule is developed to initialize population. Extensive computational experiments are carried out using five benchmark cases and eight realistic instances in remanufacturing. The proposed heuristic rule is evaluated using five benchmark cases for minimizing the maximum fuzzy completion time and the maximum fuzzy machine workload objectives, respectively. IABC algorithm is compared to six meta-heuristics for maximum fuzzy completion time criterion. For maximum fuzzy machine workload, IABC algorithm is compared to six heuristics. The results and comparisons show that IABC algorithm can solve FJSP with fuzzy processing time effectively, both benchmark cases and real-life remanufacturing instances. For practical remanufacturing problem, the schedules by IABC algorithm can satisfy the requirement in real-life shop floor. The IABC algorithm can be as part of expert and intelligent scheduling system to supply decision support for remanufacturing scheduling and management.

APS, ERP and MES systems integration for semiconductor backend assembly

In this paper, the system integration of an Advanced Planning and Scheduling (APS) system with Enterprise Resource Planning (ERP) and Manufacturing Execution System (MES) for a Semiconductor Backend Assembly environment is described. The company is one of the worldwide market leaders in semiconductor packaging technology. SIMTech was responsible for the implementation of the Planning & Scheduling functions through its APS application which is called Gintic Scheduling System (GSS). GSS system has been integrated with the other two key manufacturing software systems, namely ERP and MES, with the CIM framework. This paper emphasizes the system integration between the GSS and the customers existing ERP and MES systems. The paper starts with a broader overview of the APS, ERP and MES systems with the focus on the linkage of APS with ERP and APS with MES. It then introduces the brief background information of the implementation and the integration methodology. Detailed work of the integration is laid out in the fond of integration model structure and illustrated by the system integration mechanism. Various data types that need to be integrated are also elaborated. Further discussions continue with the practical implementation issue like the required frequency of data integration and also the two approaches of data transfer. Finally the paper shares the experiences and lessons the team has learned through the implementation process and concludes the essence of the system integration work.

Practical lot release methodology for semiconductor back-end manufacturing

This paper presents a new method and system that has been developed to solve production lot release problems in a discrete semiconductor back-end manufacturing environment, wherein there is always a huge product mix and a multitude of capacity constraints. The methodology is a multi-constraint based finite capacity production control mechanism to plan lot release of the desired mix of products for the semiconductor assembly and test operations. Practical lot prioritization considerations, order release policies, finite capacity constraints and a novel technique of multi-level loading pattern for minimizing machine conversion are discussed in detail. The system and methodology presented in this paper has been successfully implemented in a semiconductor back-end factory in Asia.

A Heuristics-based Advanced Planning and Scheduling System with Bottleneck Scheduling Algorithm

This paper presents a heuristics-based advanced planning and scheduling (APS) system with bottleneck scheduling algorithm. It has been designed to solve production scheduling problems in discrete manufacturing industry. The proposed APS system can be configured to be deployed in different production environments, including make-to-stock, make-to-order, bottleneck-driven shop floor, through its forward, backward and bottleneck scheduling algorithms. It allows users to specify heuristic rules at each operation based on the scheduling policy of the operation. The embedded scheduling techniques facilitates the generation of feasible and practical schedule to achieve a fine balance among the conflicting production goals of maximizing resource utilization, minimizing work-in-process (WIP), and reduction of cycle time. In addition, the system can be easily reconfigured to address them various requirements imposed by the physical and operational constraints of the production environment. The APS system deploys two layers of heuristic algorithms intertwined within the scheduling engine. The two layers of heuristic algorithms are job prioritization (JP) rules and machine selection (MS) rules. JP heuristics rules are designed to prioritize orders at each operation, while machine selection (MS) algorithm selects the best-fit machines and other optional resources to generate the dispatching list. The modular and configurable approach adopted in the design and development of the scheduling engine allows the reconfiguration of basic core JP and MS modules for different industry-specific requirements. The proposed APS system has been successfully implemented to fulfil the daily production scheduling needs of a few semiconductor backend assembly companies.

An integrated modeling framework for capacity planning and production scheduling

Efficient system modeling can eliminate capacity discrepancy between mid-term capacity planning and by consistently delivering promised capacity. This paper proposes an integrated modeling framework which consists of capacity constraints and configurable constraints for capacity planning and production scheduling to address the issue of capacity discrepancy. The capacity constraints that derived from machine timeline, production rate and machine allocation preference matrix can resolve the conflicting objectives in capacity planning and production scheduling; whereas the configurable constraints that are designed and implemented for special concerns in planning and scheduling functions will facilitate the pursuing of optimized production plans and schedules. The paper depicts an implementation of the proposed framework in the semiconductor back-end assembly environment.

Hybrid discrete harmony search algorithm for scheduling re-processing problem in remanufacturing

This paper proposes a hybrid discrete harmony search algorithm for solving the re-processing scheduling problem in pump remanufacturing. The process of pump remanufacturing and the scheduling problem of re-processing for pump subassembly are modeled. An experience based strategy is proposed for solving the unpredictability of subassembly re-processing time in remanufacturing. Hybrid discrete harmony search algorithm and local search are employed for scheduling re-processing of pump subassembly. The objectives of pump subassembly re-processing scheduling are minimization of the maximum completion time (makespan), and the mean of earliness and tardiness (E/T). These objectives are considered individually as well as together as a multi-objective problem. Computational experiments are carried out using real data from a pump remanufacturing enterprise. Computational results show that the objectives makespan and E/T can be optimized and the resulting schedules can be used in practice.

Integrated Production Planning, Scheduling and Shop Floor Tracking System for High-Mix Low-Volume Manufacturing - A Consortium Approach

In Singapore, the Precision Engineering (PE) industry is the backbone of the manufacturing sector; it supports a large number of manufacturing industries such as Electronics, Chemicals, Pharmaceuticals, Biotechnology, Aerospace, Oil & Gas and Automotive. From our recent visits to companies in the PE industry, it was observed that given such a dynamic and stochastic manufacturing environment, most of these PE companies are still predominantly using Microsoft Excel as their planning & scheduling tool, shop floor tracking is normally performed manually through paper traveler, and production status is only updated at the end of the day or as and when demanded by customer. The common problems highlighted include inability to provide realistic delivery commitment, machine and manpower resources are not optimally utilized to fulfill customer orders, unacceptable order cycle time and lack of production visibility. Therefore, there is an increasing demand and impending need for a computerized integrated production planning/scheduling and shop floor tracking system. In this paper, a consortium approach for the development of an integrated production planning/scheduling and shop floor tracking system for the High-Mix Low-Volume (HMLV) PE environment is presented. An illustration through the implementation of the proposed system in a PE company is highlighted. With this system, companies would be able to improve their delivery performance through effective handling of ad-hoc customer orders and the ability to react to deviations or unplanned events in production.