Fault Detection in LV Distribution Networks Based on Augmented Complex Kalman Filter

Low voltage distribution networks in the presence of distributed energy resources introduce new challenges in fault detection and protection system. While conventional protection schemes such as over current relays are not reliable in active distribution networks with low-frequency measurement devices, the monitoring and protection of active distribution networks with high tech measurements is currently not cost-effective. Moreover, the unpredictable customer loads behaviour decreases the accuracy of the protection schemes with fixed fault current thresholds. In order to overcome these issues, a new protection scheme based on the corrective error in augmented complex Kalman filter state estimation is proposed to detect shallow faults in low voltage distribution networks. Our empirical investigation shows a normal distribution with zero mean while the standard deviation is fitted on the corrective error in distribution network state estimator, where the calculated standard deviation determines a threshold for fault detection. An unexpected rise in the corrective error over the fault threshold can be considered as a fault in the distribution network. In order to evaluate the performance of the proposed method in this paper, one thousand fault scenarios in a radial distribution network with six-buses is examined, using real measured customer loads data.

I. Introduction

By emerging distributed energy resources (DERs), distribution networks have experienced new challenges in control, monitoring and protection [1]. Hence, new methods are required to develop a reliable protection algorithm considering the low voltage (LV) distribution networks limitations. In the presence of the low-frequency measurement devices in LV distribution networks, the conventional overcurrent protection schemes are not reliable enough for fault detection [2]. The current protection schemes in distribution networks are implemented in high voltage (HV)/medium voltage (MV) substations. As a result, a fault in a small zone of distribution networks isolates entire downstream networks. In near future, by implementation of peer-peer markets, customer loads with rooftop PVs can sell their generated power [3], while disconnecting the network by circuit breakers in HV/MV substations might decrease the customers expected profit. Implementing the overcurrent protection scheme in LV distribution networks with fixed fault current thresholds is not reliable due to the high uncertainty in customer loads behaviour. As a result, a measured current at 12 pm of a summer day can be categorized as a fault condition, while a rise in the measured current as a result of a shallow fault at 12 am is not detected by over current protection scheme with a fixed fault current threshold. Different articles have addressed a similar issue by implementing communication platforms, such as [4]–[6]. Although communication-based protection schemes like line differential relays are reliable in distribution networks, implementing communication platforms are not cost-effective in widespread LV distribution networks.