Electronic Equipment Failure Prediction

Abstract

Modern industrial equipment carries multiple sensors for providing automation of production process. These sensors are providing real time information of different parameters during production. It assumed that these sensors are providing true data. When sensors provide false data, it shall be replaced by new sensor. Periodic maintenance of equipment is the time to replace various parts; in case of sensor it is called “Mean Time to Replace Sensor”. For equipment operating 24/7 or performing critical task need to be highly available. The availability can be calculated on the basis of mean time to fail (MTTF) and mean time to repair (MTTR). This is an average time for periodic maintenance. The periodic maintenance leads early replacement of healthy sensors. It observed that sensors performance is affected by various external parameters such as temperature, humidity and pressure around the equipment. The vibrations produced by the equipment also show effect on sensor’s normal working. Hence it is necessary to generate a model that calculates effect of above-mentioned parameters on sensors performance. After proper learning, classification and validation of data, a prediction model is built to provide early warning of sensor’s failure. The aim of this paper is to suggest a predictive warning system for sensors using data representing ground reality and predictive model.