Proficiency Analysis of Unscented Kalman Filter for Bad Data Detection During State Estimation

The accurate and reliable estimation of system states is paramount in various engineering problems. The state estimation process is critical for real-time monitoring and control in many control and power system problems. However, the state estimation process may give erroneous results when there is the presence of bad, erroneous or noisy data. Thus, this paper investigates the proficiency of the Unscented Kalman Filter (UKF) as a robust tool for detecting the impact of bad data during the state estimation process. The UKF, an extension of the traditional Kalman Filter, exhibits superior performance in handling nonlinear and non-Gaussian systems. It accommodates the uncertainties associated with bad data. The ability of the UKF to detect the bad data points while maintaining accuracy during state estimation have been demonstrated in this paper. The detection of bad data may further helps to generate an alarm to the operator alerting possible mal-function of any controller. A second order nonlinear system has been taken into consideration to test the bad data detection capabilities under its different parameter values and settings of UKF.