Predictive Maintenance and Monitoring of Industrial Machine with Electronic Communication and Machine Learning

Abstract

This project targets the performance monitoring and providing predictive maintenance of a dedicated industrial machine. It aims to use machine learning algorithms to do predictive maintenance and predict the future faults that might occur based on machine learning model formed where the energy meter readings are given as input data. Apart from the energy meter reading we also include some parameter which is related to particular machine such as vibration, temperature and grising level also plays vital role for machine behaviour. The project also constructed to provide Real-time meter reading's running hour time through the continous fetched per minute data.Also via this project, the concerned dicipline gets power report, where power consumed is automatically calculated and sent via mail at a dedicated period of time. Here power report include some important, effective and crucial to predict machine behaviour which are Power Fator, Average Current, Average voltage, Current phase and KWH. To get real time data on LAN, modbus enabled TCP-IP device is used. The data collected from EMs via ModBus communication, is then later on bifurcated and analayzed to build appropriate ML model. To analysis the collected real time data, mean average value will be taken and according to that particular value prediction of normal and abnormal condition of dedicated machine will be reveal.The resultant conditions shall produce the critical-non critical value and accordingly alarm will be generate and actions shall be undertaken. According to the current status of project, system is able to get real time data on coustmize software of 33 energy meter and reporting of that data is done regularly to the respected authorities by mail. Aprt from that efficiency of the particular machine drive is predict from the current scenario.