Distributed multi agent network scheduling in factory 4.0 environment based on job shop

نوع: Type: thesis

مقطع: Segment: PHD

عنوان: Title: Distributed multi agent network scheduling in factory 4.0 environment based on job shop

ارائه دهنده: Provider: Naeeme Bagheri Rad

اساتید راهنما: Supervisors: Dr. Javad Behnamian

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Dr. Amir Saman Kheirkhah-Dr. Parviz Fattahi-Dr. Mostafa Zandieh

زمان و تاریخ ارائه: Time and date of presentation: 11June 2023-11.30

مکان ارائه: Place of presentation: Department of Engineering

چکیده: Abstract: For a quick response to the needs of customers and in time exchange of information between the agents in the industries, data should be provided to the production unit immediately after receiving. Such communication at factory level has led to the birth of smart factories, which compared to traditional production methods, produce higher quality products faster. In addition, nowadays, many industries have moved from a centralized to a decentralized structure. In a decentralized structure, factories are located in different geographical locations. In this structure, factories with independent ownerships are placed in a network of several factories, called a virtual production network, and each factory focuses on its own interests. In this thesis, each factory is considered as an agent in such network. The cooperation of factories in these networks provides the possibility of very fast diffusion of technology and immediate reaction to changes for each of the members. Due to the fact that the factories are equipped with online equipment, it is assumed that there is a possibility of transferring job between the factories in this network. Also, real-world production environments are associated with many unpredictable events. Among these events, the arrival of new jobs and the breakdown of machines happen more often. Due to the virtual communication of factories, these events are communicated between different production units through RFID systems. This makes scheduling and planning to be done intelligently, accurately and optimally. Due to the importance of this issue, in the current research, the issue of real-time scheduling of distributed multi-agent production network in the environment of smart factories has been studied. Considering that the investigated problem is a combination of static and real-time schedulings, First, a mixed integer two-objective linear programming model is presented; And then a dynamic approach has been proposed to solve the real-time scheduling problem. Considering the successful applications of the Lagrangian Relaxation algorithm in solving various problems, in this research, the improved Lagrangian Relaxation(LR) algorithm has been used to solve small and medium-sized problems. To check the performance of the proposed Lagrangian Relaxation algorithm, its results have been compared to solving the model by the Augmented constraint epsilon(AEC) method. The results showed that the proposed Lagrangian Relaxation algorithm has a better performance than the Augmented constraint epsilon(AEC) method. Also, according to the complex structure of the investigated problem, a learning-based Memetic algorithm was proposed. In order to evaluate the performance of the proposed algorithm, the results of this algorithm were compared with the closest research. The calculation results showed that, in the large sizes of the problem, the proposed algorithm performs better than the competing algorithm

فایل: ّFile: Download فایل