Wei Weng

Dynamic routing strategies for JIT production in hybrid flow shops

In production processes, just-in-time (JIT) completion of jobs helps reduce both the inventory and late delivery of finished products. Previous research which aims to achieve JIT job completion mainly worked on static scheduling problems, in which all jobs are available from time zero or the available time of each job is known beforehand. In contrast, dynamic scheduling problems which involve continual arrival of new jobs are not much researched and dispatching rules remain the most frequently used method for such problems. However, dispatching rules are not high-performing for the JIT objective. This study proposes several routing strategies which can help dispatching rules realize JIT completion for jobs arriving dynamically in hybrid flow shops. The routing strategies are based on distributed computing which makes realtime forecast of completion times of unfinished jobs. The advantages include short computing time, quick response and robustness against disturbance. Computer simulations show that the performance of dispatching rules combined with the proposed routing strategies is significantly higher than that of dispatching rules only and that of dispatching rules combined with the previous routing methods.

Online Scheduling of Flexible Flow Shop Manufacturing

The flexible flow shop refers to such a manufacturing environment in which jobs are to be processed through serial stages, with one or multiple machines available at each stage. It is usually a complex task when specific objective is demanded such as minimum cost, minimum time, etc. Static scheduling of such problems has been much researched, however, little efforts have been made on realtime scheduling when the release time of each job is unknown. In this paper, some online scheduling methods are presented to deal with the tough problem of realtime Just-In-Time manufacturing. In addition to applicable dispatching rules, agent-based approaches are also proposed featuring feedback, learning, and realtime prediction. The simulation result reveals that the presented distributed learning approach, especially when combined with realtime prediction, delivers a high performance.

Study and application of scheduling method for just-in-time production in flexible job shops

The unsmooth job flow between the production shops along a manufacturing chain is a problem commonly seen in industries due to the inconsistency in processing speed and delivery between the chained production shops. With many factors involved in the coordination between production shops, the problem is complex and few solutions have been provided to date. This study presents an approach able to solve this problem by implementing time-based manufacturing that enables the speed and timing of each shops job outflow to match those of its successor shops job inflow. The proposed method is composed of offline schedule making and online job processing control. It aims to complete each job in a just-in-time (JIT) manner at the time the job is wanted by the next production shop. Designed upon a flexible job shop environment, which is easy to be transformed into other shops with similar characters, the proposed method is expected to be widely applicable to JIT scheduling problems. An industrial case study is made and results show that the proposed method has a strong ability in JIT job completion, tardy job prevention and makespan reduction.

Distributed Feedback Mechanism for Just-In-Time Scheduling Problem

The problem considered in this research is the just-in-time scheduling of a manufacturing environment that is able to produce several different products. New jobs come randomly into the system, expected to become one of the products. Each job must go through multiple stages before it can be finished as a product. There are multiple machines at each stage, and the processing time of each product on each machine is different. There exits delivery time between stages. Previous researches did much on the static scheduling of such problem by using mixed integer linear programming. However, little efforts have been made on realtime scheduling, which means the release time of jobs is unknown. But such scheduling is becoming more and more important under the increasingly competitive manufacturing market. In this paper, two distributed feedback mechanisms are proposed to solve the realtime scheduling problem of minimizing earliness and tardiness penalties of all jobs. The simulation shows that the proposed distributed feedback mechanisms deliver quite competitive performance for the targeted problem.

Self evolution algorithm for common due date scheduling problem

Inventory cost and delay penalty are two kinds of annoying spendings in manufactory industry. Accordingly, earliness and tardiness penalties are proposed to simulate such scheduling problems where the popular just-in-time (JIT) concept is considered to be of significant importance. In this paper, a self evolution algorithm is proposed to solve the problem of single machine total earliness and tardiness penalties with a common due date. Up to now, such problem has been solved without specific consideration of straddling V-shaped schedules, which may be better than pure V-shaped schedules for early due date problems; without specific discussions on g improving, where g refers to the idle time before the start of the first job; and the many individuals in all so far proposed GA-like algorithms become the bottleneck of execution time reduction. Therefore, in this research, efforts have been made on digging out the straddling V-shaped schedules, improving the efficiency of g improving, and reducing the execution time. In addition, a new RHRM approach is proposed to create the initial solution for evolution, which helps achieve the fast contingency of the algorithm. The performance of the proposed algorithm has been tested on 280 benchmark instances ranging from 10 to 1000 jobs from the OR Library, the results showing that the proposed self evolution algorithm delivers much higher efficiency in finding optimal or near-optimal solutions with both better results in total penalties and significant execution time reduction.

Improvements to genetic algorithm for flexible job shop scheduling with overlapping in operations

Flexible job-shop scheduling problem (FJSP) is an extended job-shop scheduling problem. FJSP allows an operation to be processed by several different machines. FJSP with overlapping in operations means that each operation is divided into several sublots. Sublots are processed and transferred separately without waiting for the entire operation to be processed. In previous research, a mathematical model was developed and a genetic algorithm proposed to solve this problem. In this study, we try to improve the procedure of previous research to achieve better results. The proposed improvements were tested on some benchmark problems and compared with the results obtained by previous research.

A control method to save machine energy in production

Nowadays, energy saving is one of the most talked about issues in our life, it is also increasingly important in the manufacturing industry. This research considers the dynamic flexible flow shop scheduling (DFFS) problem, which is an extended version of the classical flow-shop scheduling problem. A flexible flow shop has multiple stages with multiple machines at each stage for processing multiple products. Previous research on DFFS aimed to achieve just-in-time production, or reducing difference between the actual completion time and the due date of each job. However, little research has been made on energy saving of machines in production. To address such a need, this paper proposes a method that dynamically turns on and off machines so as to reduce energy consumption while achieving JIT production. The proposed method has been tested on different environments, and the results show that it is high performing for both JIT production and energy saving.

A genetic algorithm for unrelated parallel machine scheduling minimizing makespan cost and electricity cost under time-of-use (TOU) tariffs with job delay mechanism

Unrelated parallel machine scheduling under time-of-use electricity price is addressed in this paper. In this setting, price of electricity can be different among various periods of the day. The objective is to minimize total cost consisting of makespan cost and electricity cost. Genetic algorithm (GA) is used to solve the unrelated parallel machine scheduling under time varying tariffs. Chromosome decoding, inspired by greedy total cost, is proposed to transform individual into feasible schedule. Furthermore, generated schedule from the individual is improved by job delay mechanism that shifts jobs to other periods to avoid high electricity cost. Finally, numerical experiment is conducted to implement the approach. Preliminary result shows that our proposed approach is effective and efficient to solve the corresponding problem.

Abnormal data analysis in process industries using deep-learning method

This research is mainly about the abnormal data analysis in factories of process industries. In the processing factory, there are many sensors which transmit the values to each other. Workers in process factory need to be alerted when the values of some sensors are abnormal values. In our research, the main target is to detect the potential abnormal value from different sensors of process industries. Since the value is filled with noise and delays, we first use the cross-correlation and wavelet transformation to remove them. Then, use deep-learning method to train the model with processed data and use the model to detect potential abnormal value. Finally, we evaluate the model we trained by the data extracted from a real process factory. The result shows that our model performs well.

An improved teaching-learning-based optimization algorithm to solve job shop scheduling problems

Job shop scheduling problem (JSP) is a strongly NP-hard combinatorial optimization problem. It is difficult to solve the problem to the optimum in a reasonable time. Teaching-learning-based optimization (TLBO) algorithm is a novel population oriented meta-heuristic algorithm. It has been proved that TLBO has a considerable potential when compared to the best-known heuristic algorithms for scheduling problems. In this paper, the traditional TLBO is improved to enhance diversification and intensification when exploring solutions for JSP. The improvements include changing the coding method, increasing number of teachers, introducing new learners and performing local search around potentially optimal solutions. To show effectiveness of the improved TLBO algorithm, the simulation results obtained by the improved TLBO for benchmark problems are compared with results obtained by the traditional TLBO and the best known lower bounds.