To solve the problem of parking accuracy of urban rail trains, we designed a closed-loop braking system based on deceleration feedback. Based on this control system, a new function that can monitor the braking factors has been developed to monitor the braking status in real-time, to remind the speed limit, braking distance and braking time. The deceleration closed-loop control system is designed to solve the problem of precise braking. Generally, under the open-loop brake control device, the brake effect cannot be fed back in time. For urban rail trains that need to accurately stop at the platform, the braking force at low speed is too large or too small, which may lead to anti-skid or dislocation. Under the study of the brake system dynamics model, it is considered that the friction coefficient will fluctuate under different temperatures of the brake disc. Therefore, the braking force correction of the dynamic friction coefficient is adopted, and the actual braking effect is monitored by the deceleration sensor. The original dynamic friction coefficient set comes from the test bench. Use machine learning to predict the friction coefficient and obtain the change process of friction coefficient during braking at full speed. By comparing the feedback deceleration in the braking process with the expected deceleration calculated by BCU, the difference between the actual deceleration and the target deceleration can be obtained and adjusted using the dynamic friction coefficient set. In the process of repetition, the dynamic friction set is constantly revised. According to these data, the attenuation curve of the braking friction coefficient of the train can be analyzed. Through the dynamic model, we know that in addition to the friction coefficient between the brake disc and the brake pad, the adhesion between the wheel and rail will also affect the braking deceleration. This process is solved by fuzzy processing. However, relevant data can be recorded. Through the cloud system, we can collect the data of all vehicles passing in the same section. They all have the same adhesion relationship between the wheel and rail. This part of the information can be extracted through mathematical analysis. So far, we have obtained two most uncertain and complex coefficient information. Thus, the braking force calculation and braking control of the braking system have been further improved. In this technology, it is necessary to consider the operational capability of the BCU and the response speed of the braking system, which is the boundary condition of the system. Under the limited operational capability and time, data analysis and processing should be completed as much as possible, and accuracy should be guaranteed. At the same time, it shall not affect the operation of the braking system. This pile of data processing analysis and transmission has high requirements, which largely depend on the development of hardware technology and the continuous upgrading of algorithms. Compared with the first generation of closed-loop control system based on deceleration feedback, the latest system can take into account the influence of multiple dynamic factors, namely, the adhesion relationship between wheel and rail, the damaging trend of brake pads, the temperature change and vehicle weight during braking, accurately calculate the required braking force, accurately implement the braking process and supervise it. The final deceleration closed-loop control system has the functions of data collection, analysis and real-time adjustment, further monitoring the braking status in real-time, and reminding the speed limit, braking distance and braking time.

Dr. Yuzhu Wang, Assistant research fellow, CHINA ACADEMY OF RAILWAY SCIENCES; Dr. Xiang Zhang, Assistant Research Fellow, China Academy of Railway Sciences