I. Introduction
Now-a-days most of the process industries (such as auto mobile manufacturing, food processing industries, power plants, bio-chemical and pharmacy industries etc), are focusing on two major things to reduce the maintenance, failure and insurance costs. First one is pperformance and later one is Anomaly Detection[3]. If the point of failure or period of failure is pre-identified, it will reduce the above mentioned costs with great extent. For that purpose most of the process industries equipped with various sensors at various stages in the process flow. The number of sensors equipped is depended on various factors such as monitoring equipment, size of the plant, sensitivity, level of monitoring etc. In general this number starts from 1000 to > 10000. These sensors continuously monitoring the equipment and send the readings to Distributed Control Systems (DCS). But this data is not stored at DCS. It has to be redirected to any historian to store such high volume of data. So every industry has to maintain Data historian to store sensor data for applications such as data analytics. From this point onwards big data analytics and predictive analytics will come into picture. This paper focuses on Independent Sensor Data Analysis and Equipment Anomaly detection by analysing sensor data in process flows.