Unscented Kalman Filter for Distribution Networks with Quantization Effects and False Data Injection Attacks

An improvement of the unscented Kalman filter algorithm is designed for the problem of distribution network state estimation under the influence of quantization effects and false data injection attacks. Firstly, the measured data are quantified using a probabilistic uniform quantizer. Secondly, the random variables obeying Bernoulli distribution are used to express whether the attack occurs or not, and the measurement equations of the system are reconstructed. Then, the upper bound of the estimated error covariance is derived by using the quantization error and the norm boundedness of the attack vector and minimising the upper bound of the estimated error covariance at each moment to obtain the optimal filtering gain. Finally, simulations based on the IEEE-69 distribution network test system were carried out to verify that the improved unscented Kalman filter algorithm can effectively ensure the estimation performance of the system against the quantization effect and false data injection attacks.