Research On High-speed Railway Wind Prediction And Warning Methods

As the miles of high-speed railways in China continues to increase,the coverage area is also expanding,with many high-speed railways passing through windy areas.During operation,crosswinds can cause significant surface impact and uncontrolled changes in air pressure,resulting in trains swaying beyond their limits and ultimately causing safety incidents.In recent years,derailments and overturning accidents caused by strong winds have occurred both at home and abroad,making it necessary to establish a high wind early warning system for high-speed railways.In this paper,based on wind speed monitoring data from a wind speed collection point of a typical high-speed railway,SSA-BPNN model,KF-BPNN-DS evidence theoretical model and CNN-LSTM-AT hybrid model incorporating CEEMDAN are established respectively for wind speed prediction along the railway line.And the design of the strong wind monitoring and warning system is investigated by combining the early warning rules for the operation of high-speed trains in strong wind conditions.The main research in this paper is as follows:(1)We describe the types of high winds that affect the safety of railway operations and the impact of strong winds on the safety of high-speed train operations,elaborate on the functional requirements of the high-speed railway monitoring and warning system,analyze the basic architecture of the high-speed railway monitoring and warning system.and briefly explain the role of each structural layer in this system.(2)According to the actual situation of the environment along the railway line,the SSA-BPNN model is established for the non-linearity and non-smoothness of wind speed and other characteristics(using sparrow search algorithm to optimize the threshold and weight values of the BP neural network)、KF-BPNN-DS Evidence Theory model(the results of the pre-processing of the Kalman filter algorithm are first predicted using BP neural network,and then the DS evidence theory is used to fuse the two levels of predictions to obtain the final wind speed)and a hybrid CNN-LSTM-AT model incorporating CEEMDAN(firstly,the original wind speed sequence is decomposed using CEEMDAN to obtain several modal components,a CNN-LSTM model is built separately for each sub-information,and the attention mechanism is used to efficiently assign weight values to different parts of the neural network,and the wind speed prediction results are finally superimposed)to predict the measured wind speed and compare the prediction performance of different models.(3)Based on the established wind speed combination model to predict wind speed,combined with warning rules and conducted on the design of various functional modules of the strong wind monitoring and warning system.(4)This paper summarizes the workflow of the strong wind monitoring and warning system for trains and operating specifications in windy conditions,and elaborates the key technologies for strong wind monitoring and warning on high-speed railways,and finally introduces the methods for adjusting operation after the warning,including stopping the train,running at reduced speed and running around the line.