Xinchang Hao

A novel bi-vector encoding genetic algorithm for the simultaneous multiple resources scheduling problem

To improve capital effectiveness in light of demand fluctuation, it is increasingly important for high-tech companies to develop effective solutions for managing multiple resources involved in the production. To model and solve the simultaneous multiple resources scheduling problem in general, this study aims to develop a genetic algorithm (bvGA) incorporating with a novel bi-vector encoding method representing the chromosomes of operation sequence and seizing rules for resource assignment in tandem. The proposed model captured the crucial characteristics that the machines were dynamic configuration among multiple resources with limited availability and sequence-dependent setup times of machine configurations between operations would eventually affect performance of a scheduling plan. With the flexibility and computational intelligence that GA empowers, schedule planners can make advanced decisions on integrated machine configuration and job scheduling. According to a number of experiments with simulated data on the basis of a real semiconductor final testing facility, the proposed bvGA has shown practical viability in terms of solution quality as well as computation time.

The cooperative estimation of distribution algorithm: a novel approach for semiconductor final test scheduling problems

A large number of studies have been conducted in the area of semiconductor final test scheduling (SFTS) problems. As a specific example of the simultaneous multiple resources scheduling problem, intelligent manufacturing planning and scheduling based on meta-heuristic methods, such as the genetic algorithm (GA), simulated annealing, and particle swarm optimization, have become common tools for finding satisfactory solutions within reasonable computational times in real settings. However, only a few studies have analyzed the effects of interdependent relations during group decision-making activities. Moreover, for complex and large problems, local constraints and objectives from each managerial entity and their contributions toward global objectives cannot be effectively represented in a single model. This paper proposes a novel cooperative estimation of distribution algorithm (CEDA) to overcome these challenges. The CEDA extends a co-evolutionary framework incorporating a divide-and-conquer strategy. Numerous experiments have been conducted, and the results confirmed that CEDA outperforms hybrid GAs for several SFTS problems.

Network modeling and evolutionary optimization for scheduling in manufacturing

Scheduling is one of the most important fields in Advanced Planning and Scheduling or a manufacturing optimization. In this paper, we propose a network modeling technique to formulate the complex scheduling problems in manufacturing, and focus on how to model the scheduling problems to mathematical formulation. We propose a multi-section evolutionary algorithm for the scheduling models formulated by network modeling. Through a combination of the network modeling and this multi-section evolutionary algorithm, we can implement the auto-scheduling in the manufacturing system. The effectiveness and efficiency of proposed approach are investigated with various scales of scheduling problems by comparing with recent related researches. Lastly, we introduced service-oriented evolutionary computation architecture software. It help improved the evolutionary computation’s availability in the variable practical scheduling in manufacturing.

Effective multiobjective EDA for bi-criteria stochastic job-shop scheduling problem

This paper proposes an effective multiobjective estimation of distribution algorithm (MoEDA) which solves the bi-criteria stochastic job-shop scheduling problem with the uncertainty of processing time. The MoEDA proposal minimizes the expected average makespan and the expected total tardiness within a reasonable amount of computational time. With the framework of proposed MoEDA, the probability model of the operation sequence is estimated firstly. For sampling the processing time of each operation with the Monte Carlo methods, allocation method is used to decide the operation sequence, and then the expected makespan and total tardiness of each sampling are evaluated. Subsequently, updating mechanism of the probability models is proposed according to the best solutions to obtain. Finally, for comparing with some existing algorithms by numerical experiments on the benchmark problems, we demonstrate the proposed effective estimation of distribution algorithm can obtain an acceptable solution in the aspects of schedule quality and computational efficiency.

An Effective Multi-objective EDA for Robust Resource Constrained Project Scheduling with Uncertain Durations☆

Abstract Project scheduling is a complex process involving many resource types and activities that require optimizing. The resource- constrained project scheduling problem (rcPSP) is one of well-known NP-hard problems where activities of a project must be scheduled to minimize the project duration. This paper presents a stochastic multiple mode resource constrained project scheduling problem (S-mrcPSP) with the uncertainty of durations. An effective multi-objective estimation distribution algorithm (moEDA) is proposed to solve S-mrcPSP to minimize its robustness and expected makespan. The proposed moEDA employs Markov network modelling activity assignment where the effects between decision variables are represented as an undirected graph model. Furthermore, slack-based metric based assessing algorithm is used to measure the robustness, where a free slack based heuristic method is adopted to achieve better performance. We demonstrate an empirical validation for the proposed method by applying it to solve various benchmark resource constrained project scheduling problems.

Effective Estimation of Distribution Algorithm for Stochastic Job Shop Scheduling Problem

Abstract This paper propose an effective estimation of distribution algorithm (EDA), which solves the stochastic job-shop scheduling problem (S-JSP) with the uncertainty of processing time, to minimize the expected average makespan within a reasonable amount of calculation time. With the framework of proposed EDA, the probability model of operation sequence is estimated firstly. For sampling the processing time of each operation with the Monte Carlo methods, we use allocation method to decide the operation sequence then the expected makespan of each sampling is evaluated. Subsequently, updating mechanism of the probability models is proposed with the best solutions to obtain. Finally, for comparing with some existing algorithms by numerical experiments on the benchmark problems, we demonstrate the proposed effective estimation of distribution algorithm can obtain acceptable solution in the aspects of schedule quality and computational efficiency.

Rule driven multi objective dynamic scheduling by data envelopment analysis and reinforcement learning

This paper presents a rule driven method of developing composite dispatching rule for multi objective dynamic scheduling. Data envelopment analysis is adopted to select elementary dispatching rules, where each rule is justified as efficient for optimizing specific operational objectives of interest. The selected rules are subsequently combined into a single composite rule using the weighted aggregation manner. An intelligent agent is trained using reinforcement learning to acquire the scheduling knowledge of assigning the appropriate weighting values for building the composite rule to cope with the WIP fluctuation of a machine. Implementation of the proposed method in a two objective dynamic job shop scheduling problem is demonstrated and the results are satisfactory.

An effective Markov network based EDA for flexible job shop scheduling problems under uncertainty

This paper presents a min-max regret version programming model for the stochastic flexible job shop scheduling problem (S-FJSP) with the uncertainty of processing time. An effective Markov network based estimation of distribution algorithm (EDA) is proposed to solve S-FJSP to minimize its maximum regret. The proposal employs Markov network modeling machine assignment where the effects between decision variables are represented as an undirected graph model. Furthermore, min-max regret metric based assessing algorithm is used to measure the robustness, where a critical path-based local search method is adopted to achieve better performance. We present an empirical validation for the proposal by applying it to solve various benchmark flexible job shop problems.

Cooperative Bayesian Optimization Algorithm: A Novel Approach to Simultaneous Multiple Resources Scheduling Problem

During the past several years, there has been a significant amount of research conducted simultaneous multiple resources scheduling problem (SMRSP) Intelligence manufacturing based on meta-heuristics, such as genetic algorithms (GAs), simulated annealing (SA) particle swarm optimization(PSO), has become a common tool to find satisfactory solutions within reasonable computational times in real settings. However, there are few researches considering interdependent relation during the decision activities, moreover for complex and large problems, local constraints and objectives from each managerial entity cannot be effectively represented in a single model for complex and large problems. In this paper, we propose a novel cooperative Bayesian optimization algorithm (COBOA) undertaking divide-and-conquer strategy and co-evolutionary framework. Considerable experiments are conducted and the results confirmed that COBOA outperforms recent researches for the scheduling problem in FMS.

Advances in Hybrid Metaheuristics for Stochastic Manufacturing Scheduling: Part II Case Studies

Manufacturing Scheduling plays a very important role in the intelligent manufacturing system, where it can have a major impact on the productivity of a production process. However, it is very difficult to find an optimal solution for manufacturing scheduling problems since most of them fall into the class of NP-hard problem. Because real world manufacturing problems often contain nonlinearities, multiple objectives conflicting each other and also uncertainties that are too complex to be modeled analytically. In these environments, hybrid metaheuristic based optimization is a powerful tool to find optimal system settings to the stochastic manufacturing scheduling problems. Evolutionary algorithm (EA) in hybrid metaheuristics is a generic population-based metaheuristic, which can find compromised optimal solutions well for a complicated manufacturing scheduling problem. By using the hybrid sampling strategy-based EA (HSS-EA) and the multi-objective estimation of distribution algorithm (MoEDA), we survey several case studies such as stochastic multi-objective jobshop scheduling problem (S-MoJSP), stochastic multi-objective assembly line balancing (S-MoALB) problem and stochastic multi-objective resource-constrained project scheduling problem (S-MoRcPSP) with numerical experimental results to get the better efficacy and efficiency than existing NSGA-II, SPEA2 and awGA algorithms.