Anomaly Detection in Estimation of Load and Prediction of Load in Networked Control System Using Correlation and Regression Data Analysis

Abstract

Balancing of load on a networked control system is an important operation in continuous process control. An unbalanced load causes overloading of a network, which results in loss of data or interruption in accessing the real-time control data. This paper is involved in the identification of factors and their dependency on load using correlation and further the detection of an anomaly in load on the network. Also, the mathematical relationship between load and the devices on the network is developed for the prediction of the load with the help of statistical regression data analysis. Foundation Fieldbus network segment, a popularly used communication protocol in process control applications is taken as an example for numerical analysis. The results of this approach found that the average anomaly in load estimation is about 0.482% due to a residual load of 1.5417% for segment configuration.