Constructive Search Algorithm Research Articles

Tailoring customer order scheduling search algorithms

Customer Order Scheduling Problem (COSP) with minimisation of the total completion time as the objective is NP-Hard. COSP has many applications that include the pharmaceutical and the paper industries. However, most existing COSP algorithms struggle to find very good solutions in large-sized problems. One key reason behind is that those algorithms are based on generic templates and as such lack problem specific structural knowledge. In this paper, we capture such knowledge in the form of heuristics and then embed those heuristics within constructive and perturbative search algorithms. In the proposed deterministic constructive search algorithm, we use processing times in various ways to obtain initial dispatching sequences that are later used in prioritising customer orders during search. We also augment the construction process with solution exploration. In the proposed stochastic perturbative search, we intensify its diversification phase by prioritising rescheduling of customer orders that are affected badly. Our tailoring of the search in this case is to make informed decisions when the search has lost its direction. On the contrary to that, in the intensification phase, we then take diversifying measures and use multiple neighbourhood operators randomly so that the search does not get stuck very quickly. Our experimental results show that the proposed algorithms outperform existing state-of-the-art COSP algorithms.

Dynamic scheduling of parallel heat treatment furnaces: A case study at a manufacturing system

In the present work, a case study focusing on online and dynamic scheduling of parallel heat treatment furnaces at a real manufacturing company is presented. The problem under consideration in this study involves release times, eligibility constraints, due dates, sequence-dependent setup times due to heating up or cooling down the furnaces, breakdowns and maintenance periods. Moreover, dynamic events, for instance, arrivals of rush orders, cancellations of already handled jobs, changes in due dates or in lot sizes exist in the nature of the problem considered here. Such dynamic factors make the introduced problem more challenging. One more difficulty is the vast variety of products at the firm. Generating a large-scaled setup matrix is neither practical nor reliable. Therefore, based on the distinctive attributes of the products, an implicit clustering is employed. As a solution approach, a multi-start and constructive search algorithm is proposed for this problem. Finally, as a dynamic scheduling module for the heat treatment furnaces, the proposed algorithm is embedded into the enterprise resource planning system of the company that is built on JAVA. The current system contributes to the firm by online scheduling of the heat treatment furnaces, generating online Gantt Charts for managers, online transmission of work orders and digitalized performance analysis reports. Thus, as demonstrated by numeric data presented in the present paper, a decrease in energy consumption and an increase in annual income are expected.

The birth unit practice development project

Customer Order Scheduling Problem (COSP) with minimisation of the total completion time as the objective is NP-Hard. COSP has many applications that include the pharmaceutical and the paper industries. However, most existing COSP algorithms struggle to find very good solutions in large-sized problems. One key reason behind is that those algorithms are based on generic templates and as such lack problem specific structural knowledge. In this paper, we capture such knowledge in the form of heuristics and then embed those heuristics within constructive and perturbative search algorithms. In the proposed deterministic constructive search algorithm, we use processing times in various ways to obtain initial dispatching sequences that are later used in prioritising customer orders during search. We also augment the construction process with solution exploration. In the proposed stochastic perturbative search, we intensify its diversification phase by prioritising rescheduling of customer orders that are affected badly. Our tailoring of the search in this case is to make informed decisions when the search has lost its direction. On the contrary to that, in the intensification phase, we then take diversifying measures and use multiple neighbourhood operators randomly so that the search does not get stuck very quickly. Our experimental results show that the proposed algorithms outperform existing state-of-the-art COSP algorithms.

Modeling and solving assembly line design problems by considering human factors with a real‐life application

AbstractErgonomics has been playing an important role in assembly system design (ASD) that contains not only the main assembly line balancing problem but also the subassembly line balancing and assembly layout problem. The ergonomics in ASD has an impact both on productivity and on workers’ health, especially when frequent changes in the product mix occur. In this study, we propose a systematic approach in order to handle ASD, which consists of three subproblems, while considering ergonomic risk factors. The first two subproblems are solved simultaneously using the proposed rule‐based constructive search algorithm, where ergonomic risks are evaluated by OCRA method. Later, layout problem is solved under transportation constraints using local search methods with various neighborhood structures. To provide the applicability and evaluate the performance of the proposed systematic approach, a real‐life case study in a harness manufacturing company is solved and prototype productions are performed.

Solving integrated process planning and scheduling problem with constructive meta-heuristics

For product manufacturing, process planning is to select a series of manufacturing processes according to the product design specification, and scheduling is to allocate manufacturing resources such as machines and tools to these processes. It is a common problem that the process plan and the schedule are not able to cope with the changes in real time manufacturing. Integrated process planning and scheduling (IPPS) is to conduct the process planning and scheduling functions concurrently, with the aim to improve the dynamic responsiveness of the production schedule. This paper investigates the formulation and implementation of constructive meta-heuristics for solving IPPS problems. To begin with, a model representation is established to express IPPS problems with AND/OR graphs. With this model representation, a generic framework is proposed for implementing constructive meta-heuristics in the solution model. The generic framework provides a common procedure for the constructive meta-heuristics, which encapsulates the calculation of the search frontier and state transitions, and provides two interfaces for accommodating different constructive search algorithms. Ant colony optimization (ACO), a commonly-used algorithm which possesses all typical characteristics of constructive meta-heuristics, is adopted as a representative example for illustrating the implementation. Experiments and tests are conducted to validate the proposed system. The single objective minimizing the makespan is set for evaluating the performance of the proposed system. Experimental results of the benchmark problems have shown the effectiveness and high performance of the proposed approach based on the integration of the generic framework and ACO strategy.