Modeling of cable fault system

Modeling is the essential part of implementing the prediction and location of three-phase cable fault. To predict and locate cable fault, a model of three-phase cable fault system is constructed based on a great deal of measured validation data by choosing BP neural network that has nonlinear characteristic and using the unproved BP algorithm, Levenberg-Marquardt data-optimized method. It is shown by the simulation using MATLAB software that the parameters of the model converge rapidly, and the simulated output of the neural network model and the measured output of cable fault system are approximately equal, and the mean value of the relatively predictive error of the fault distance is smaller than 0.3%, so that the model quality is reliable.