Li-Man Liao

Improved MILP models for two-machine flowshop with batch processing machines

In this paper, we consider the problem of scheduling jobs in a flowshop with two batch processing machines such that the makespan is minimized. Batch processing machines are frequently encountered in many industrial environments such as heat treatment operations in a steel foundry and chemical processes performed in tanks or kilns. Improved Mixed Integer Linear Programming (MILP) models are presented for the flowshop problem with unlimited or zero intermediate storage. An MILP-based heuristic is also developed for the problem. Computational experiments show that the new MILP models can significantly improve the original ones. Also, the heuristic can obtain the optimal solutions for all the test problem instances.

Tabu search for non-permutation flowshop scheduling problem with minimizing total tardiness

Abstract This paper deals with the non-permutation flowshop problem which means that the job sequences are allowed to be different on machines. The objective function is minimizing the total tardiness. Firstly, three mixed-integer linear programming (MILP) models for non-permutation flowshop problems are described, and then are analyzed and assessed their relative effectiveness. Secondly, two Tabu search based algorithms, denoted by LH1 and LH2, are proposed to solve the complicated non-permutation flowshop problems. The algorithms are constructed on specific neighborhood structures which enable the searching method effective. Finally, the performance is evaluated against Taillard’s famous benchmark. Computational experiments show that the proposed algorithms, LH1 and LH2, are significantly superior to the L_TS algorithm. And the percentages of improved permutation flowshop instances by LH1 and LH2 algorithms are about 87.8% and 98.3% with respect to total tardiness, respectively. The non-permutation schedules also have achieved significant improvement in four different scenarios of due dates. Consequently, average percentage improvement (API) is 14.52% for flowshop problems with low tardiness factors. It is evident that exploring non-permutation schedule is effective and necessary for low tardiness factors.

Heuristic algorithms for two-machine flowshop with availability constraints

The majority of the scheduling studies carry a common assumption that machines are available all the time. However, machines may not always be available in the scheduling period due to breakdown or preventive maintenance. Taking preventive maintenance activity into consideration, we dealt with the two-machine flowshop scheduling problem with makespan objective. The preventive maintenance policy in this paper was dependent on the number of finished jobs. The integer programming model was proposed. We combined two recent constructive heuristics, HI algorithm and H algorithm, with Johnsons algorithm, and named the combined heuristic H&J algorithm. We also developed a constructive heuristic, HD, with time complexities O(n^2). Based on the difference in job processing times on two machines, both H&J and HD showed good performance, and the latter was slightly better. The HD algorithm was able to obtain the optimality in 98.88% of cases. We also employed the branch and bound (B&B) algorithm to obtain the optimum. With a good upper bound and a modified lower bound, the proposed B&B algorithm performed significantly effectively.

Single machine scheduling problem with fuzzy due date and processing time

Abstract This paper considers a single machine scheduling problem with fuzzy due date and fuzzy processing time. There are n jobs J 1, J 2, …, Jn to be processed on a single machine. Associated with each job Jj , there is a membership function of due date which describes the degree of satisfaction with respect to the completion time of Jj . The due date is structured similar to a trapezoidal fuzzy set, and the processing time is described as a triangular fuzzy set. The objective is to maximize the minimum grade of satisfaction over given jobs. In this paper, polynomial‐time algorithms with O(n2 logG) are proposed to solve the scheduling problem and are illustrated by means of an example.

Single facility scheduling with major and minor setups

This article considers the problem of scheduling a given set of jobs at a single facility, where jobs can be grouped into different classes and these classes can be further grouped into different families. A major setup and a minor setup are required when jobs are switched from one family to another while only a minor setup is required when jobs are switched from one class to another within the same family. The minimum mean flow time schedule of this problem can be solved optimally by dynamic programming (DP), but the exponential behaviour of the DP solution precludes its use to solve problems with a large number of classes. An efficient heuristic is thus developed in which a sequence of two-class problems is solved. Computational results show that the heuristic produces solutions that deviate 0.118% on average from the optimum.

Parallel machines scheduling with machine preference via agent-based approach

Abstract This paper deals with parallel machines scheduling for electro-etching in the manufacturing processes of aluminum foil. Due to the attributes of the machines and the fitness of jobs to each, the processing time and setup time for a job depends on the machine on which the job is processed. In this paper, the application of agent-based multi-negotiation mechanism is developed for this manufacturing environment and for further scheduling uncertainty research. To support the dynamic negotiation mechanism between jobs and machines, three categories of agents are currently employed by the system. They are job agents, machine agents and supervisor agent. To establish job allocations, job agents and machine agents have to bid interactively in accordance with the decision model embedded in. Supervisor agent is designed to manage negotiation process between job agents and machine agents to ensure the global bi-objective of system are being observed. Experiments were conducted to evaluate the performance of the proposed architecture. Computational experiences demonstrate that the proposed architecture can work more efficiently and is capable of obtaining better solutions to the heuristic based on earliest-due-date rule applied by enterprise practice on case company.

Multi-agent System for Balancing Mixed-Model Assembly Lines with Bi-objective

Due to increasing competition and differentiated demand, assembly line is no longer sufficient to offer only standardized products. To satisfy the requirement of customers with the advantages of an efficient flow line, some product variants are allowed to produce simultaneously on the same assembly line. This paper proposes a multi-agent based algorithm for a mixed-assembly line balancing problem (MALBP). The model is constructed by two-level agent architecture. In the first level, a planning agent is built to determine line cycle time based on customer demand. According to the cycle time, the ideal number of workstation is determined. In the second level, there are a balancing agent and multiple machine agents collaborating to balance the workloads of all workstations. The balancing agent records the task precedence constraints and the workload of each workstation, and calculates the efficiency of line balancing. Tabu search algorithm is applied as a communication mechanism between machine agents to adjust the workloads of all workstations. The primary objective is to minimize the number of workstation for a given cycle time, and secondary objective is to minimize the variance of the workload of workstations. In this paper, a workstation is viewed as an agent. Adjacent swap and insert mechanism is applied to search the neighborhood, and, furthermore, tabu list is used to avoid the solution trapped into local optimal.

An intelligent agent-based collaborative workflow for inter-enterprise PCB product design

Currently, there is tremendous pressure to design and develop products in a short period of time to reduce time-to-market and projects life cycle. Researchers often propose workflow management system as a mechanism to facilitate teamwork in a collaborative product development environment. In this paper, we develop an agent-based collaborative workflow platform (ACWP), in which a group of intelligent agents work autonomously and collaboratively to perform workflow tasks and access data from a distributed environment. By the aid of intelligent agents, our system can execute customized workflow tasks for inter-enterprise users based on their individual requests, and conquer the complexity of product design in distributed, heterogeneous platform. The application of multi-agent systems ACWP for collaborative workflow management is investigated through an inter-enterprise PCB product design. ACWP facilitates PCB product design and team interaction in a collaborative but distributed product development environment and moves up the whole performance of design process.

An artificial immune based algorithm for parallel-machine scheduling with preference of machines

In this paper, an artificial immune based algorithm (AIA) is proposed for production scheduling of aluminum foil factory. The objective function is to minimize the total tardiness. The AIA algorithm has a systematic immune mechanism which mainly is built on the clonal selection and affinity maturation principles of the immune system. Machine-based encoding method is used and earliest due date (EDD) rule is applied to find the better initial antibodies. The antibodies are then separated into memory or suppressive cells by affinity, and the suppressive cells will be eliminated based on the concentration level. Further, crossover and mutation techniques are used to construct antibody proliferation mechanism to obtain superior antibodies. To avoid falling into local optimal solutions, receptor editing mechanism is applied to suppress the worse antibodies and accelerate solution convergence.

Multi-agent-Based Negotiation Approach for Scheduling Jobs on Parallel Machines with Ready Times and Machine Preference

The manufacturing processes of aluminum foil often included seven main steps and the electro-etching is critical step which usually has many parallel machines with three distinct voltage equipments. Therefore, the production scheduling is parallel-machine and its choices of eligibility. In this paper, an agent-based architecture is proposed to develop a parallel-machine scheduling application for this manufacturing environment. There are three types of agents in the proposed system. Each job/machine is viewed as a agent, named job/machine agent. To establish job allocations, job agents and machine agents have to bid interactively. Management agent plays as a system supervisor and coordinator to manage negotiation processes between job agents and machine agents to ensure the global objectives such as minimizing total jobs’ tardiness and flow time, balancing machines’ loading, and maximizing total revenue can be achieved. Eventually, experiments are conducted to evaluate the performance of the proposed system. The result shows that the proposed architecture provides a parallel eligible machines scheduling solutions with a better performance.