Intelligent Factory Process Monitoring and Scheduling in Industry 4.0
Lead Research Organisation:
University of Nottingham
Department Name: Faculty of Engineering
Abstract
This research is in partnership with Siemens Congleton and aims to help the factory inform future technological choices with the goal of becoming an Industry 4.0 environment. A priority for Siemens is minimising the disruption to existing processes. As such, two research approaches have been explored: scheduling and machine health monitoring. A novel scheduling algorithm has been developed for generic factory environments that include manufacturing setup times. Alongside this, a wave soldering machine has been chosen as a use case scenario for health monitoring. A test bench analogue has been created to test a vibration monitoring Change Point Detection algorithm, which is being developed further into an online health monitoring method.
People |
ORCID iD |
| Svetan Ratchev (Primary Supervisor) | |
| Emil Tochev (Student) |
Publications
Tochev E
(2018)
A Comparison of Centralised and Decentralised Scheduling Methods Using a Simple Benchmark System
in IFAC-PapersOnLine
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509309/1 | 01/10/2015 | 30/03/2021 | |||
| 1788239 | Studentship | EP/N509309/1 | 01/10/2016 | 03/02/2021 | Emil Tochev |
| Description | I have developed an algorithm for production scheduling in factories with multiple product types. This is based on an existing algorithm that I used in a conference paper which compared different scheduling methods.The algorithm is meant to be easy to understand and implement, and is the basis of a journal paper currently being written. Three variants are presented, to provide schedules for single, serial and parallel machines. To aid understanding, the single machine variant is tutorialised. I have also designed a test bench to simulate a factory machine. This has been used to assess a change point detection algorithm, which can be used to monitor machine health with a minimum amount of setup. I have formulated a second, live change point detection algorithm based on existing literature. It can detect change points in live data and has been tested on the test bench mentioned above. |
| Exploitation Route | The algorithm has been tested against a simple representation of a factory, and has performed better than an approximation of the current factory scheduling system. There is also scope to improve the algorithm by making it more efficient. It can also be expanded to function for a greater variety of machine setups. The change point detection algorithm is intended to be generic enough to be applicable across a range of machines in a factory. The live change point detection algorithm can be used to detect machine faults shortly after they occur. |
| Sectors | Manufacturing, including Industrial Biotechology |