A Novel Regression Model-Based Toolbox for Induced Voltage Prediction on Rail Tracks Due to AC Electromagnetic Interference of Adjacent Power Lines

Abstract

AC electromagnetic interference between rail tracks and adjacent power lines causes serious concerns about personnel and railway equipment safety. The existing AC interference analysis method uses the complex computer simulation software to estimate induced voltages on rail tracks, and such simulation becomes especially difficult at the transmission line routing stage when only limited information is available. To overcome this challenge, a novel regression model-based toolbox is developed in this paper to predict induced voltages on rail tracks due to AC interference. To develop this toolbox, the dataset acquisition is a critical step due to very limited research conducted in this area. A dataset is produced in this study using our newly developed AC interference analysis method, where variations of various factors are considered, including the power line's current, the separation distance between power lines and railway, the ballast resistance, and the length of rail tracks. To improve the accuracy, hyperparameters of regression algorithms are optimized by Bayesian optimization. Two models are eventually chosen to predict induced voltages on rail tracks: “Gaussian process regression” with “matern 3/2” kernel function; and a tri-layered “neural network” model with “sigmoid” activation function. The toolbox is accurate and easy-to-use for design engineers working on transmission line routing, and has been currently in use by Manitoba Hydro in Canada.