A Line Fault Detection Method Based on Feature-Driven Extreme Learning Machine

With the expansion of the scale and complexity of the active distribution network, the risk of fault is also rising significantly. The line fault detection has become a key to ensure the safe and stable operation of the system. The existing researches do not fully consider the impact of unmeasured branches, and the fault feature mining is insufficient. For improvement, this paper proposes a new fault detection method based on feature-driven extreme learning machine (ELM). This method collects voltage and current signals, and deeply excavates four key characteristic parameters: differential modulus feature, current change modulus feature, voltage modulus feature and voltage and current phase difference feature. Then, by analyzing the line fault feature data, a feature-driven ELM model was constructed, so as to significantly improve the accuracy of fault detection and effectively deal with undetectable branches. The simulation test results show that the ELM model trained by fault features can accurately detect faults of lines with undetectable branches. The proposed method is expected to provide a more solid guarantee for the safe and stable operation of active distribution network.