Smart scheduling for industry 4.0

Abstract

Decision making is an important function of any industry, from the planning phase until the product is shipped, many important decisions are taken. A good decision is the one that is based on concepts and algorithms. With the onset of industry 4.0, the world is going towards AI and automation, which plays a crucial role in decision making. Industry 4.0 exploits the application of CyberPhysical Systems (CPS), Computer Optimization, and Industrial IoT Techniques in manufacturing enterprises to create a system capable of remote monitoring, intelligent planning, and optimal decision making with proactive feedback support. At its most fundamental level, it enables ‘non-human entities to autonomously interact with each other and with humans, to intelligently work towards their goals, and make data-centric decisions using cyber-physical transformations and Internet of Things.’ The world is moving in an era where the industrial assets take most of the decisions without any human interference; this is achieved by the help of cyber-twin and cyber-physical systems. In this project, the work is done towards making a CPS more useful in decision making; the main focus was on the sequencing of the jobs. In the coming future, the intelligence is to be embedded in every entity of the industry, keeping that thing in mind Agent-based job scheduling algorithm is designed for job scheduling in a flow shop environment. This algorithm is first of a kind that enables scheduling in a distributed manner, where every machine is included in the decision making for the final schedule. Due to the decentralized nature of the algorithm, there are many advantages it has, which are discussed later in this report. More specifically, the benefit of using a communication algorithm over conventional methods, such as Johnson’s rule, in minimizing the makespan for a production horizon and genetic algorithm is shown. It further demonstrates how the impact increases as i) the number of jobs to be scheduled increase, ii) the number of machines increases, and iii) the variability in the job-processing time increases. After developing the algorithm for the most common flow shop, the work is extended to the real industry sequencing problem. A simulation-based approach is tried to solve the problem, in which a part of the industry is simulated in Witness Horizon, considering all the real-time interruption and challenges. The result obtained by the witness optimizer is further compared with the results coming from the present method of job scheduling in the company.