Search

It is difficult to detect the faulty feeder if single phase-earth faults occur in a resonant grounding system, mainly due to its very low-current fault. Since the similarity of transient zero-sequence current waveforms between faulty feeder and sound feeder is lower than that between two sound feeders, this paper presents a protection scheme for faulty feeder detection based on the grey relation degree that can characterize the similarity among curves. Slope relation degree, one kind of grey relation degree, is chosen to characterize the similarity of transient zero-sequence current waveforms between one feeder and other feeders, and the slope relation matrix that represents the relationship among all feeders can be established. The protection criterion established by using the slope relation matrix can be applied to detect the faulty feeder of single phase-earth faults. It is noted that the proposed method is applicable to the system where at least three or more feeders are presented. Electromagnetic Transients Program (EMTP) simulation results and adaptability analysis have confirmed the effectiveness and reliability of the proposed scheme.

With the development of a substation automation system, it is now feasible for the intelligent electronic devices-based protection system to collect its measurements from electronic instrument transformers through a communication network such as an Ethernet-based local-area network. The application of a wide-area broadband network in power utilities introduces underlying danger for protection systems to maloperate when receiving fake data packages from an intruder. This paper proposes identifying a vicious fault by using context information, such as voltage and current, of the same substation. When the protection system detects a fault based on measurements transmitted through a network, it collects all measurements of the substation and feeds these data to a probabilistic neural network. Thereafter, the fault caused by fake data that differs from the known fault pattern can be identified and blocked.

This paper proposes an agent-based paradigm for self-healing protection systems. Numerical relays implemented with intelligent electronic devices are designed as a relay agent to perform a protective relaying function in cooperation with other relay agents. A graph-theory-based expert system, which can be integrated with supervisory control and a data acquisition system, has been developed to divide the power grid into primary and backup protection zones online and all relay agents are assigned to specific zones according to system topological configuration. In order to facilitate a more robust, less vulnerable protection system, predictive and corrective self-healing strategies are implemented as guideline regulations of the relay agent, and the relay agents within the same protection zone communicate and cooperate to detect, locate, and trip fault precisely with primary and backup protection. Performance of the proposed protection system has been simulated with cascading fault, failures in communication and protection units, and compared with a coordinated directional overcurrent protection system.