Modeling the measurement error of energy meter using NARX model

Abstract

The paper presents a methodology to model the nonlinearity associated with the error in measurement of energy metering system using nonlinear autoregressive with exogenous input (NARX) model. The experimental results are obtained for active and reactive power under both sinusoidal and non-sinusoidal conditions. The system identification technique is presented to understand the basics of nonlinear modeling structures. The prevalent operating condition is identified which is contributing most of the error to the system and its nonlinear modeling is carried out. The comparison is performed for experimental and simulated output of different models, to decide the best modeling structure. This method plays a vital role in determining nonlinear model which reflects high fidelity for continuously varying error pattern of energy metering system under different power quality phenomenon. Thus, the behavioral pattern of energy meter for different parametric variations becomes a prior available and this information can be utilized for enhanced performance of energy meter.