High Impedance Fault Detection Method Based on Traveling Wave Full Waveform Fault Characteristics

High impedance ground fault (HIF) occurs in a distribution network, the fault features are weak and susceptible to factors such as noise interference. Extracting and detecting the wavefronts in this case can be challenging, leading to low reliability in HIF detection methods based on waveforms. To address these issues, this paper proposes an HIF detection method based on self-recognition of fault features using traveling wave full waveform (TWFW) characteristics. Firstly, it analyzes the differences between HIF and normal transient disturbances wave signals in the time-frequency domain based on TWFW. Then, it utilizes a CBAM-CNN neural network model to explore fault features from TWFW grayscale images from multiple dimensions, enhancing the method's fault detection and interference resistance capabilities. Simulation results demonstrate that the proposed method achieves a high detection accuracy of up to 99.8%, reliably detects HIF, exhibits excellent noise resistance, and is unaffected by factors such as fault location and fault impedance.