Soft Computational Technique to Estimate Inflow for Fault Detection in Control Valve

Abstract

A sensor less flow measurement technique is proposed in this paper. The objective of the proposed work is to design an observer which can estimate the inflow to the control valve by soft computation techniques, so as to detect the faults in the control valve. Orifice flow meter is placed at the outlet of the control valve for measurement of outflow. By analyzing the behavior of control valve for different input signal, in flow to the control valve is estimated. Control valve works on the principle of regulating the inflow based on the control signal. Performance of the control valve can be established by measuring outflow, by knowing input signal and estimation of inflow to the control valve. To estimate the inflow, neural network based model is incorporated in the proposed work. Designed estimator is compared with the standard signal to evaluate the performance. The root mean square of percentage error of 0.91 shows successful design of estimator.