Modeling And Verifying Method Research On High-speed Train Braking Model

Train control onboard equipment is one of the core subsystems that ensures the safe operation of trains in CTCS-3train control system. To ensure the correctness of the various functions of onboard equipment and strict performance of overspeed protection to train’s status, we need build a safe and effective train braking model. Parameters in braking model are prerequisites to ensure the safety of the results. Some EMU Parameters can’t be obtained neither by direct method nor by measure or calculation. How to get these Parameters become an important issue. Meanwhile, model algorithm is the core of braking model, and to verify the reliability of the brake models become the basis and key to ensure the safe operation of trains. At the same time parameter random effect analysis of the model becomes a important content to ensure the safety of train running.Based on analyzing the demand of domestic high-speed onboard equipment on the braking model, this paper has carried on deep research to the modeling and verification method of braking model. Firstly, according to the reality of the situation of high-speed railway in China, relevant parameters of braking model were summarize, the format was described. Then the paper describes in detail the thinking of braking model, the two outputs of the model which contain the monitoring data to supervise brake and introduce the driver, and its calculation method and process. The paper also analyses some special circumstances that may arise in the scenario, then improves the model.This paper adopts the method of parameter identification based on neural network, and analyzes its possible error. By using the Monte Carlo method, the random parameters analysis on the deceleration parameter model and braking model is made, and model simplification process of the deceleration parameter model is finished. Finally, in order to verify the efficiency and usability of braking model on actual situation, we develop a braking model calculation and compare software, design a series of evaluation indexes, get the evaluation of the model depended on the same situation contrast experiments on real and model results.