In this paper, a fast prediction model based on machine learning is established to predict the coupling coefficient between pantograph arcing and GSM-R antenna. Compared with the method of solving coupling coefficient based on electromagnetic simulation, the prediction model established in this paper can not only ensure the accuracy of coupling coefficient, but also avoid a lot of repeated simulation and reduce the simulation time. First, the data set is constructed by Latin hypercube sampling. Then, by analyzing the distribution characteristics of the data set, the Radial Based Function (RBF) and Generalized Regression Neural Network (GRNN) are selected for modeling. Furthermore, in order to improve the prediction precision, Modular Neural Network (MNN) is introduced, and the collaborative integration strategy of segmented prediction is proposed, which realizes the complementary advantages of RBF and GRNN, and completes the construction of prediction model. Finally, the precision and validity of the prediction model is verified by the analysis of the prediction results, relative prediction deviation and other related parameters.

In the positioning process of a high-speed train, cumulative error may result in a reduction in the positioning accuracy. The assisted positioning technology based on kilometer posts can be used as an effective method to correct the cumulative error. However, the traditional detection method of kilometer posts is time-consuming and complex, which greatly affects the correction efficiency. Therefore, in this paper, a kilometer post detection model based on deep learning is proposed. Firstly, the Deep Convolutional Generative Adversarial Networks (DCGAN) algorithm is introduced to construct an effective kilometer post data set. This greatly reduces the cost of real data acquisition and provides a prerequisite for the construction of the detection model. Then, by using the existing optimization as a reference and further simplifying the design of the Single Shot multibox Detector (SSD) model according to the specific application scenario of this paper, the kilometer post detection model based on an improved SSD algorithm is established. Finally, from the analysis of the experimental results, we know that the detection model established in this paper ensures both detection accuracy and efficiency. The accuracy of our model reached 98.92%, while the detection time was only 35.43 ms. Thus, our model realizes the rapid and accurate detection of kilometer posts and improves the assisted positioning technology based on kilometer posts by optimizing the detection method.

For high-speed train, balise transmission module (BTM) system is easily interfered with by other equipment of the train. This could cause the train to malfunction. Studying the electromagnetic susceptibility (EMS) of the BTM is very important for the performance and efficiency of the train. In this paper, a modular, system-level modeling method is proposed to predict the EMS of BTM systems. Based on object-oriented technology and a modular method, the BTM system is disassembled into several modules according to the electromagnetic characteristics of the whole system rather than the physical structure. All the modules are mutually independent, and the total EMS could be evaluated by the output of them. The modules of three key elements of electromagnetic compatibility (EMC), i.e., sources, coupling paths, and sensitive equipment, are established by the theoretical method, full-wave simulation method, and black-box test method, respectively, and put into different layers. According to the functions of the BTM system, the EMS of BTM is given by analyzing the interrelation of input and output of modules. Results of the proposed model were verified by measurement.