Low-voltage direct-current arc fault detection method

The invention discloses a low-voltage direct-current arc fault detection method. The method comprises the following steps: acquiring a current signal of a low-voltage direct-current loop; constructing a fault arc characteristic parameter extraction model based on LMD-PCA and sample entropy, and extracting a current signal characteristic vector; training an LSSVM (Least Square Support Vector Machine) through the current signal feature vector, performing parameter optimization on the LSSVM through a WOA whale optimization algorithm, and constructing a WOA-LSSVM arc fault recognition model; and based on the WOA-LSSVM arc fault identification model, carrying out arc fault detection on the low-voltage DC loop. According to the method, low-voltage direct-current arc fault diagnosis is realized, the problem that a proper fault feature extraction and identification diagnosis method for small sample data of current parameters does not exist is solved, the operation state of an electricity utilization line can be monit