Wind Turbine Blade Multibranch Lightning Protection Down Conductors Fault Detection and Localization Method

Lightning strikes causing faults in wind turbine blade lightning protection down conductors pose serious threats to the safe and stable operation of large wind farms. Existing methods are difficult to address the need for fault detection and accurate localization in complex topologies of down conductors. To this end, this article investigates the propagation law of injected pulses in multibranch lightning protection down conductors, clarifies the correlation between the time–frequency characteristics of the fault difference wave and the location of the fault point, and then proposes a new method of fault detection and localization based on a convolutional neural network (CNN). The proposed method combines rich time–frequency fault information and realizes accurate detection and localization of poor contact fault types and short branch fault points based on the traditional time-domain method, while getting rid of the three major limitations of the traditional time-domain method, which relies on single-parameter characteristics, the pulse propagation speed is difficult to be measured practically, and the fault-reflected wavehead is easy to be misjudged. The validation results show that the fault detection and localization method proposed in this article has good localization accuracy in broken conductor faults and poor contact faults, and the maximum measured error is only 0.8 and 1.1 m in the multiple localization experiments of broken conductor and poor contact faults at 50 and 55 m, respectively, in the main down conductor. Method comparison experiments and CNN model ablation experiments demonstrate the effectiveness of the proposed method.