Development of Mobile Monitoring System for Manufacturing Industry

Wong, Qiu Yu (2020) Development of Mobile Monitoring System for Manufacturing Industry. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text WongQiuYu_Full Text.pdf Restricted to Registered users only Download (4MB)

Abstract

This project aims to develop a monitoring system for manufacturing process and provide guidance on the system framework which comprises general functions of data collection, communication, storage, analysis and visualization. To simulate the data collection process in industry, a DHT 22 sensor integrated with WeMos board and two independently running PCs are employed to collect and upload both environmental and production data to the dedicated storage for future process. The wireless communication between the front-end devices to the central server are realised with the implementation of HTTP protocols. Then, the data collected are centralised and stored in a MySQL database sever hosted on a microcomputer board, Raspberry Pi 3. Typical process control techniques are employed in the system, which are OEE index computation and process control chart. Besides, an expert system is developed to recognise the production trend from the control chart, to reduce the effort for manual analysis and improve the overall monitoring process. The last layer of the developed system is completed with an Android-based mobile application, which visualises the collected data and provides analysed output as contextual–aware feedback to the end users. Experiments are conducted on the developed monitoring system for investigation of (1) time latency for data transmission and (2) optimum design of Control Chart Pattern Recognition (CCPR) expert system. For first part, it had found that the average latencies for data transmission with HTTP protocol are 61 ms and 38.78 ms, for data uploading and downloading respectively. Time latency is not significant when all devices are operating within 2 metres radius from the server and network access point in the developed monitoring system. For second part, the CCPR expert system is developed using CART-based systematic approach suggested by (Gauri and Cgakraborty, 2009), where the six mathematical-based shape features are utilised to discriminate among nine type of chart patterns. Three expert system logical structures are proposed and tested for accuracy. Eventually, a logical structure with 10 decision nodes had yield in the highest accuracy of 86%, and thus it is implemented into the proposed monitoring system. In short, this project had developed a five-layer remote monitoring system for production industry, in which comprises general functions of data collection, communication, storage, analysis and visualization.