Archive | 2019
PREDICTIVE MAINTENANCE OF SINGLE PHASE AC MOTOR USING IOT SENSOR DATA AND MACHINE LEARNING (SIMPLE LINEAR REGRESSION AND MULTIPLE LINEAR REGRESSION ALGORITHMS)
Abstract
Predictive maintenance techniques are proposed to find out the condition of operating equipment in order to predict when failure will take place or maintenance will be essential. The crucial aim is to provide cost savings over schedule based preventative maintenance or unexpected reactive maintenance, which could result in machinery being unavailable throughout the critical periods. There are many approaches for predicting failures, in this project current analysis approach is used where the voltage and current of electricity is measured which is supplied to the motor. But these analyses can t situate by themselves. Architecture is desirable to sustain the end-to-end workflow that allows for processing at the edge and moves the data from the integrated network to a central depository for analysis and monitoring. To do predictive maintenance, firstly we have to build an integrated network of PZEM-004T meter with a Node MCU ESP8266 -12E Module that will monitor and collect data about the voltage, current, and active power of AC motor under test which are used here as key variables for determining the status of AC motor. Our approach consists of two stages. The first incorporates data gathering, data processing, and then applying supervised learning methods to gain insights from the data and the second follows the first to build predictive models which can work easily and efficiently as per our requirement. Using Regression ML (Machine Learning) an algorithm that is Simple Linear Regression (SLR) And Multiple Linear Regression (MLR) machine under test (motor 1 and motor 2) performance can be optimize and provide new statistic patterns which make the backbone of prediction analysis. The machine learning algorithm is applied to the collected and processed data based on which results are obtained. The obtained test result is compared with obtained predicted result based on which the prediction accuracy of each machine learning algorithm is decided. Accuracy can be further improved by using much more data while applying the machine learning algorithms, as the number of data streams increases prediction accuracy increases. These models are trained for long time to respond to new data or new values and then it delivers the results we need with more accuracy.