Data Driven Multivariate Technique for Fault Detection of Waste Water Treatment Plant

Abstract

Classification of data originating from sensor in to two distinct categories (good and bad) is a challenging job and has a wide spread application. A lot of research is going on in this particular area. Because of the enhanced memory capacity of the present day computers, data logging has reached to a new level. The analyst has to classify the data according to their traits from the offline logged data. The whole task of collection of raw data, classification of data according to their traits involves different statistical as well as soft computational techniques. There are two types of classification algorithms like supervised classification and unsupervised classification. In supervised classification, the classification is done using neural network and in unsupervised classification the classification is done using different clustering algorithm. This dissertation studies and evaluates the performance of different classification algorithm in a waste water treatment plant. First of all waste water treatment plant is taken in to consideration and data driven classification techniques are implemented to find out the healthy data and faulty data. The healthy and faulty data is classified using supervised and unsupervised classification.