Characteristics and GWO–ANN Model Based on Prior Knowledge for Contact Resistance in a Pantograph–Catenary

Abstract

The current-collection performance of a pantograph–catenary directly influences the safe operation of high-speed electric locomotives. The contact resistance is a primary indicator to evaluate the stability and reliability of the current-collection performance. The effects of sliding speed and electric current on the contact resistance between the carbon strip and the contact wire under a dynamic pressure load were studied using a sliding electrical contact testing machine. To predict the contact resistance under various working conditions, an artificial neural network (ANN) regression model, based on the gray wolf optimizer (GWO) algorithm and prior knowledge, was developed. The GWO algorithm was used to optimize the initial biases and weights of the ANN to avoid becoming trapped in local optima. The performance of the GWO–ANN can be further enhanced by the use of prior knowledge. The results showed that the GWO–ANN model that included prior knowledge can be introduced as a new technique for estimating the contact resistance in pantograph–catenary systems.