Survey Paper on Fault Detection in Sensor Data

Process of data classification of data suffers with the increasing dimensionality of data. Fault Detection becomes important and critical in several industries. For organizations, it is essential to continuously improve the productivity. In semiconductor manufacturing it is crucial to detect faults at initial stages. So, quick identification of abnormal results is primary objective. Data classification possesses some issues because of unbounded size of data and imbalance nature of the data. Data imbalance means the number of instances in one class greatly outnumbers the number of instances in the other class. In classification, the available standard algorithms tend to favor the majority class and produce low detection of minority class as a result when the class sizes are highly imbalanced. This results in inaccurate classifier generation and wrong prediction of data. In literature many fault detection algorithms are available to address these issues. An online fault detection algorithm based on incremental clustering performs well and efficiently process the data.