Railway vehicle ice and snow foreign object hitting identification and evaluation method and system based on multi-modal fusion

Abstract

This application discloses a method and system for identifying and assessing foreign object impacts on rail vehicles based on multimodal fusion. It constructs a multi-source monitoring data and operating environment parameter dataset by acquiring image information, structural vibration response information, and impact acoustic information of multiple key structural parts of the rail vehicle, combined with operating parameters, ambient temperature, and track snow cover information obtained from the onboard positioning system. Based on this, a multimodal fusion analysis method is used to identify foreign object impact events, determining the time of foreign object occurrence, impact time, and impact characteristic parameters. Furthermore, the impact load is calculated based on the foreign object's size and material parameters, and the foreign object's entrainment capacity and critical size range are analyzed using an aerodynamic adsorption model. Finally, statistical analysis of the impact events is performed, and the risk level of key structural parts is assessed. This application enables accurate identification and quantitative analysis of foreign object impact behavior on rail vehicles during winter ice and snow, improving operational safety.

Description

Technical Field

[0001] This application relates to the field of foreign object impact analysis technology for rail vehicles, specifically to a method and system for identifying and evaluating foreign object impacts from ice and snow on rail vehicles based on multimodal fusion. Background Technology

[0002] With the rapid development of high-speed rail transit, the operational safety of rail vehicles in cold regions and winter environments has received increasing attention. Under conditions of low temperature, snowfall, and icy tracks, rail vehicles are prone to being struck by ice and snow debris during operation. For example, ice may fall off the vehicle, snow on the track may be sucked up by airflow to form flying ice, and ice and snow may scrape against the underside of the vehicle. These phenomena can impact key structural components such as bogies, gearboxes, braking systems, and auxiliary equipment, thereby affecting the structural reliability and operational safety of the vehicle.

[0003] In the existing technology, research on the adaptability of rail vehicles to ice and snow environments mainly focuses on structural protection design and de-icing measures. The means of monitoring and analyzing the impact behavior of ice and snow foreign objects during operation are relatively limited. Some studies use a single sensor (such as a vibration sensor or video device) to record impact events. However, due to the characteristics of ice and snow foreign object impacts, such as strong randomness, short occurrence time, and complex action process, a single data source is difficult to fully reflect the occurrence process and action mechanism of impact events, which can easily lead to impact identification errors.

[0004] Meanwhile, during the high-speed operation of rail vehicles, there is a complex aerodynamic flow field around the vehicle body. Ice and snow particles on the track surface may be entrained and participate in the impact process under the action of aerodynamic negative pressure. However, the analysis of the foreign object entrainment ability under aerodynamic adsorption in the existing technology is insufficient, and there is a lack of systematic research on the critical size and mass of foreign objects with different densities and shapes, making it difficult to accurately define the range of foreign objects involved in the impact.

[0005] Therefore, there is an urgent need for an analytical method that can integrate multi-source monitoring data to accurately identify ice and snow foreign object impacts on rail vehicles during winter operation, analyze the movement characteristics of the foreign objects, and achieve impact frequency statistics and risk level assessment. Summary of the Invention

[0006] Purpose of the invention: In order to overcome the above shortcomings, the purpose of this application is to provide a method and system for identifying and evaluating ice and snow foreign object impacts on rail vehicles based on multimodal fusion.

[0007] To address the aforementioned technical problems, this application provides a method for identifying and assessing ice and snow foreign object impacts on rail vehicles based on multimodal fusion, including: S1: Acquire multi-source monitoring data of the rail vehicle under operating conditions collected by the monitoring units of multiple key structural parts of the rail vehicle and perform time synchronization marking on the multi-source monitoring data, wherein the multi-source monitoring data includes at least image information, structural vibration response information and impact acoustic information. S2: Obtain the real-time operating parameters of the on-board positioning system of the rail vehicle and obtain information on ambient temperature and snow cover status of the line, and construct an operating environment parameter dataset that matches the multi-source monitoring data; S3: Based on the multi-source monitoring data and operating environment parameter dataset, a multi-modal fusion analysis method is used to identify foreign object impact events, and to determine the time of foreign object appearance, impact time, and corresponding impact characteristic parameters; S4: Based on the size, material parameters and impact characteristic parameters of the foreign object, calculate the impact load of the foreign object on the hit component, and based on the aerodynamic adsorption model established according to the aerodynamic pressure distribution law during the operation of the rail vehicle, analyze the entrainment capacity of the track surface particles under the action of the train's aerodynamic negative pressure, and calculate the critical size and critical mass of foreign objects with different densities and shapes to determine the range of foreign objects involved in the impact. S5: Classify and statistically analyze all identified impact events based on operating section, ambient temperature, snow cover on the track, and vehicle location. Obtain impact frequency distribution characteristics and high-incidence section distribution patterns. Combine impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence section distribution to evaluate the impact on key parts of the rail vehicle and output comprehensive analysis results.

[0008] In a preferred embodiment of this application, step S1 includes: S11: Acquire image information of the rail vehicle under operating conditions collected by the monitoring unit of multiple key structural parts of the rail vehicle; S12: Obtain vibration response information of the rail vehicle structure under the action of foreign object impact from the monitoring unit of multiple key structural parts of the rail vehicle. S13: Acquire the impact acoustic information of the rail vehicle structure under the impact of foreign objects, collected by the monitoring unit of multiple key structural parts of the rail vehicle. S14: The image information, structural vibration response information, and impact acoustic information are uniformly time-stamped to associate them under the same time reference.

[0009] In a preferred embodiment of this application, step S2 includes: S21: Obtain real-time operating parameters collected by the on-board positioning system of the rail vehicle during train operation, wherein the operating parameters include train speed, mileage location in the operating section, and operating trajectory information; S22: Obtain ambient temperature information corresponding to the train operation process and obtain snow cover status information of the track line based on on-site observation or operation records, wherein the snow cover status information includes different degrees of snow cover such as track not covered, sleepers covered, partially covered rail height or fully covered. S23: Perform time correlation and data matching on the real-time operating parameters, ambient temperature information, and line snow cover status information to form an operating environment parameter dataset that matches the multi-source monitoring data.

[0010] In a preferred embodiment of this application, step S3 includes: S31: Process the image information frame by frame, use a preset detection method to identify moving foreign objects in the image and extract the time node, spatial location and motion trajectory information of the foreign object, and at the same time identify the image change features at the moment of impact to generate image recognition data. S32: By detecting the transient peak value, impact pulse width and frequency response characteristics in the structural vibration response information, the time point and impact intensity of the external impact on the structure are determined, and the corresponding peak acceleration and impact response energy are calculated to generate impact vibration identification data. S33: Use short-time Fourier transform or wavelet transform to extract the characteristic spectrum of impact sound information, identify preset impact sound events and extract their main frequency range, energy distribution and duration characteristics, and generate impact audio recognition data; S34: Perform time alignment processing on image recognition data, impact vibration recognition data and impact audio recognition data, and associate and match events in the three types of data by setting a time window. When at least two types of data simultaneously meet the preset impact characteristic conditions within the same time window, it is determined to be a valid foreign object impact event and the corresponding foreign object appearance time, impact time and corresponding impact characteristic parameters are output.

[0011] As a preferred embodiment of this application, in S31, the method includes: S311: Based on the trajectory information and spatial location of the foreign object, combined with the running speed and snow cover status information of the track in the operating environment parameter dataset, analyze the direction of motion and initial position of the foreign object relative to the rail vehicle. S312: Classify and label different types of impact events based on the characteristics of the foreign object's movement and the duration of the impact. When the foreign object detaches from the surface of the vehicle body and moves downward or backward, it is determined to be an impact from ice falling off the vehicle body. When the foreign object is sucked up from the surface of the track and moves upward or towards the vehicle body, it is determined to be an impact from flying ice on the track. When the foreign object makes continuous contact or slides along the bottom structure of the vehicle, it is determined to be an impact from snow scraping.

[0012] As a preferred embodiment of this application, in step S4, the formula for calculating the impact load of the foreign object on the struck component based on the foreign object's size, material parameters, and impact characteristic parameters is as follows: ,in For the maximum impact load, Foreign matter density, These are the length, width, and height dimensions of the foreign object. Let be the train speed, and the min function be the function that takes the minimum value.

[0013] As a preferred embodiment of this application, in S4, the method includes: The foreign object particle is simplified to a sphere, and its calculation formula is as follows: Therefore, the critical foreign object particle size under stable maximum negative pressure is calculated as follows: ,in The negative pressure exerted on the roadbed by the aerodynamic effect of the train is due to adsorption. Let be the radius of the sphere. It is the acceleration due to gravity; The formula for calculating the size and weight of adsorbable cubic foreign objects is as follows: Furthermore, the side length of the critical cubic foreign object under stable maximum negative pressure was calculated as follows: ,in The dimensions of the cubic-shaped foreign object are: length, width, and height. The area of ​​its top and bottom surfaces is: .

[0014] As a preferred embodiment of this application, in S5, all identified impact events are classified and statistically analyzed based on operating range, ambient temperature, snow cover level on the line, and vehicle location, including: S51: Statistically analyze the identified foreign object impact events according to the preset time window, calculate the number of impacts per unit time, and perform segmented statistical analysis of the impact events according to the rail vehicle operating section to obtain the impact frequency distribution of different sections. S52: Correlation analysis is performed between the impact events and the corresponding operating speed and ambient temperature to form impact frequency distribution characteristics; S53: By matching the impact events with the mileage location of the rail vehicle operating section, the high-incidence sections of impact are determined. Combined with the snow cover status information of the line, the distribution of impact events under different snow cover conditions is compared and analyzed to obtain the distribution pattern of high-incidence sections of impact events.

[0015] As a preferred embodiment of this application, in step S5, a risk level assessment of key components of the rail vehicle is performed by combining impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence interval distribution, including: S54: Based on the impact frequency distribution characteristics, count the number of foreign object impacts on each key structural part per unit time and determine the impact frequency level; and based on the impact characteristic parameters and impact load, determine the impact intensity level of each key structural part under the impact of foreign object. S55: Combining the aforementioned range of foreign objects and the distribution pattern of high-incidence areas, a comprehensive analysis is conducted on the impact risk of each key structural component under different operating conditions and different snow cover levels, generating an impact spatial distribution. S56: Based on the impact frequency level and impact intensity level, classify the key structural parts of the rail vehicle into different risk levels and identify high-risk parts. S57: Output includes a comprehensive analysis of impact type distribution, impact spatial distribution, impact frequency level, impact intensity level, and risk level.

[0016] This application also provides a multimodal fusion-based system for identifying and assessing ice and snow foreign object impacts on rail vehicles, which implements the aforementioned method, comprising: The multi-source monitoring unit is distributed in multiple critical structural parts of the rail vehicle that are susceptible to impact from foreign objects. The critical structural parts include at least one of the following: bogie area, wheel axle area, gearbox area, braking system components, snow skid, sand spreading device, air spring, car body floor, and external auxiliary equipment area. A control unit, connected to a multi-source monitoring unit, is used to perform and implement multi-modal fusion-based identification and assessment of ice and snow foreign object impacts on rail vehicles. The control unit includes: The multi-source data module is used to acquire multi-source monitoring data of the rail vehicle under the operating state collected by the monitoring units of multiple key structural parts of the rail vehicle and to perform time synchronization marking on the multi-source monitoring data. The multi-source monitoring data includes at least image information, structural vibration response information and impact acoustic information. The environmental data module is used to acquire real-time operating parameters of the on-board positioning system of the rail vehicle and to acquire information on ambient temperature and snow cover status of the line, and to construct an operating environment parameter dataset that matches the multi-source monitoring data. The foreign object detection module is used to identify foreign object impact events based on the multi-source monitoring data and operating environment parameter dataset, using a multi-modal fusion analysis method, and to determine the time of foreign object appearance, impact time, and corresponding impact characteristic parameters. The foreign object analysis module is used to calculate the impact load of foreign objects on the impacted parts based on the size, material parameters and impact characteristic parameters of foreign objects. Based on the aerodynamic adsorption model established according to the aerodynamic pressure distribution law during the operation of rail vehicles, it analyzes the entrainment capacity of track surface particles under the action of train aerodynamic negative pressure and calculates the critical size and critical mass of foreign objects with different densities and shapes to determine the range of foreign objects involved in the impact. The statistical evaluation module is used to classify and statistically analyze all identified impact events based on the operating section, ambient temperature, snow cover on the track, and vehicle parts. It obtains the impact frequency distribution characteristics and high-incidence interval distribution patterns, and combines impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence interval distribution to evaluate the impact on key parts of the rail vehicle, and outputs comprehensive analysis results.

[0017] The technical solution described in this application has the following advantages over the prior art: 1. This application achieves collaborative analysis of multi-source monitoring data by integrating image information, structural vibration response information and impact acoustic information. Compared with single sensor detection methods, it can more comprehensively reflect the foreign object impact process and improve the accuracy of impact event recognition.

[0018] 2. Based on the identification of foreign object impact events, this application introduces factors such as foreign object size, material parameters, and train speed to establish an impact load calculation method, thereby realizing a quantitative analysis of the impact effect of foreign objects. Compared with traditional experience-based judgment methods, it can more accurately characterize the impact intensity.

[0019] 3. This application combines the aerodynamic pressure distribution law during the operation of rail vehicles to construct an aerodynamic adsorption model, analyze the entrainment capacity of particles on the track surface, and calculate the critical size and mass of foreign objects with different densities and shapes, thereby clarifying the range of foreign objects involved in the impact and making up for the lack of analysis of the aerodynamic entrainment mechanism in the prior art.

[0020] 4. This application, through statistical analysis of impact events under operating conditions, ambient temperature, and snow-covered track conditions, can identify high-incidence impact sections and critical structural parts that are easily affected, and further conduct risk level assessments, providing a basis for the structural optimization design and operation and maintenance of rail vehicles. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0022] Figure 1 This is a flowchart of a method for identifying and evaluating ice and snow foreign object impacts on rail vehicles based on multimodal fusion, provided in an embodiment of this application.

[0023] Figure 2 This is a physical schematic diagram of the arrangement of video monitoring points on the target EMU car 1 provided in the embodiments of this application.

[0024] Figure 3 This is a physical schematic diagram of the arrangement of video monitoring points for each of the three cars of the target EMU provided in this application embodiment.

[0025] Figure 4 This is a schematic diagram of the installation of monitoring points for the snow protection plate of the target EMU car 1 provided in the embodiment of this application, wherein (a) is the installation of the observation camera for the snow protection plate of the target EMU car 1, and (b) is the arrangement of vibration acceleration measuring points 1#, 2#, and 3# for the snow protection plate of the target EMU car 1.

[0026] Figure 5 This is a schematic diagram of the camera on the undercarriage of the target EMU 1 falling after being hit, provided in an embodiment of this application. (a) is before it falls, (b) is after it falls, (c) is audio analysis, and (d) is a diagram of the camera falling.

[0027] Figure 6 This is a schematic diagram of the snow cover around the measuring point of the target EMU 3 car and the track provided in the embodiment of this application, wherein (a) shows the snow and ice around the measuring point, (b) shows the snow and ice around the measuring point, and (c) shows the snow cover on the track.

[0028] Figure 7 This is a schematic diagram of the snow protection plate of the target EMU car 1 being struck according to the embodiments of this application, wherein (a) is before the whole snow bale is struck, (b) is after the whole snow bale is struck, (c) is before the large ice block is struck, (d) is after the large ice block is struck, (e) is before the snow protection plate is covered with ice, (f) is after the snow protection plate is struck and the ice falls off, (g) is before the snow protection plate is covered with ice over a large area, and (h) is after the snow protection plate is struck over a large area and the ice falls off.

[0029] Figure 8 This is a schematic diagram of the lower right corner of the snowproof board being struck by flying ice, provided in an embodiment of this application. (a) shows the moment of impact, (b) shows the moment after impact, and (c) shows the acceleration response of the snowproof board.

[0030] Figure 9 This is a schematic diagram of ice splashing in the middle of the snowproof board provided in the embodiment of this application, where (a) is before the impact, (b) is after the impact, and (c) is the acceleration response of the snowproof board.

[0031] Figure 10 This is a schematic diagram of the snowproof board being struck by a cluster of flying ice in the middle, provided in an embodiment of this application. (a) is before the impact, (b) is after the impact, and (c) is the acceleration response of the snowproof board.

[0032] Figure 11 This is a schematic diagram of the right side of the lower baffle of the snowproof board being hit by a cluster of flying ice, provided in the embodiment of this application, where (a) is before the impact, (b) is after the impact, and (c) is the acceleration response of the snowproof board.

[0033] Figure 12This is a schematic diagram of the bottom of the snowproof board being hit by flying ice in an embodiment of this application, where (a) is the moment of impact, (b) is after the impact, and (c) is the acceleration response of the snowproof board.

[0034] Figure 13 This is a schematic diagram of the target EMU 1 car drive shaft being struck by flying ice, provided in the embodiment of this application, where (a) is before the impact, (b) is during the impact, and (c) is after the impact.

[0035] Figure 14 This is a schematic diagram of the target EMU 3 cars scraping snow on the track, provided in the embodiment of this application. (a) is before the right GFX antenna scrapes snow, (b) is when the right GFX antenna scrapes snow, (c) is before the left GFX antenna scrapes snow, and (d) is when the left GFX antenna scrapes snow.

[0036] Figure 15 This is a schematic diagram of a sample of foreign objects in ice and snow provided in an embodiment of this application.

[0037] Figure 16 This is a schematic diagram of the vehicle coordinate system provided in the embodiments of this application, wherein (a) is a side view, (b) is a rear view, (c) is a top view, and (d) is a velocity vector diagram.

[0038] Figure 17 This is a schematic diagram of the geometric model of the target EMU provided in the embodiments of this application, wherein (a) is the geometric model of the train formation, (b) is the geometric model of the head shape, and (c) is the simplified geometric model of the bogie.

[0039] Figure 18 This is a schematic diagram of the target EMU provided in the embodiments of this application, wherein (a) is the surface grid of the entire vehicle body, (b) is the surface grid of the front of the car, and (c) is the fairing grid of the bogie end of the middle car.

[0040] Figure 19 This is a schematic diagram of pressure fluctuation data at various ground monitoring points provided in the embodiments of this application.

[0041] Figure 20 This is a schematic diagram of the pressure distribution of the target EMU on the roadbed surface provided in the embodiments of this application.

[0042] Figure 21 This is a schematic diagram of the snow cover degree of different lines on which the target EMU runs, provided in the embodiments of this application. (a) is an uncovered sleeper, (b) is a sleeper covered only, (c) is covered with 1 / 6-1 / 4 of the rail height, (d) is covered with 1 / 3-1 / 2 of the rail height, (e) is covered with 2 / 3-5 / 6 of the rail height, and (f) is a covered track.

[0043] Figure 22This is a schematic diagram of the module connections of the multimodal fusion-based rail vehicle ice and snow foreign object impact identification and evaluation system provided in the embodiments of this application. Detailed Implementation

[0044] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0045] This application utilizes video monitoring of various components of the EMU, including the ship-shaped fairing and obstacle remover, snow deflector, height adjustment valve, bogie frame crossbeam, brake cylinder and brake air pipe, wheel axles, gearbox, brake pads and brake calipers, gearbox oscillation temperature sensor, air spring and air spring pressure switch, axle box, upper / lower tie rod, universal joint bracket bolts, BTM guard plate, floor plate, and GFX antenna, to visualize the surrounding environment when a strike occurs. Combined with vibration acceleration and aerodynamic load testing, GPS speed measurement, and trajectory playback tracking and monitoring technologies, the application analyzes vulnerable strike zones and carriages, providing a basis for subsequent handling and optimization.

[0046] Therefore, for reference Figure 1 As shown in some embodiments, a method for identifying and assessing ice and snow foreign object impacts on rail vehicles based on multimodal fusion is involved, the method comprising: S1: Acquire multi-source monitoring data of the rail vehicle under operating conditions collected by monitoring units of multiple key structural parts of the rail vehicle and perform time synchronization marking on the multi-source monitoring data, wherein the multi-source monitoring data includes at least image information, structural vibration response information and impact acoustic information.

[0047] Specifically, in step S1, the method includes: S11: Acquire image information of the rail vehicle under operating conditions collected by the monitoring unit of multiple key structural parts of the rail vehicle, wherein the key structural parts include at least one of the following: bogie area, wheel and axle area, gearbox area, braking system components, snow skid, sand spreading device, air spring, car body floor, and external auxiliary equipment area.

[0048] Specifically, a camera device can be installed near the key structural parts. The camera device is preferably an industrial camera with waterproof, dustproof and low temperature resistance properties and is installed on the corresponding part by a fixed bracket so that its field of view covers the area where the foreign object may move. The camera device continuously collects video data or image sequences to record the location of the foreign object, its movement trajectory and the state at the moment of impact.

[0049] S12: Obtain vibration response information of the rail vehicle structure under the action of foreign object impact from the monitoring unit of multiple key structural parts of the rail vehicle, and record the acceleration change during the impact.

[0050] Specifically, a vibration acceleration sensor can be installed on the surface of the structure at the key structural part. The vibration acceleration sensor is preferably a single-axis or multi-axis acceleration sensor and is fixedly installed at a position with high structural rigidity and susceptible to impact. The vibration acceleration sensor collects the transient vibration signal generated by the structure under the action of foreign object and records the acceleration change and vibration response process during the impact.

[0051] S13: Acquire the impact acoustic information of the rail vehicle structure under the action of foreign object impact from the monitoring unit of multiple key structural parts of the rail vehicle, so as to record the time node and frequency of impact.

[0052] Specifically, an audio acquisition element can be installed inside the vehicle body floor or in a relatively enclosed space to collect the impact sound signal generated during the impact of a foreign object. The time domain signal of the impact sound is obtained through the audio acquisition element, and the time node and frequency information of the impact are recorded.

[0053] S14: The image information, structural vibration response information, and impact acoustic information are uniformly time-stamped to associate them under the same time reference.

[0054] Specifically, each monitoring unit can synchronize its time using a unified clock source or perform synchronization calibration based on the time signal provided by the vehicle positioning system, so that all types of data have a unified timestamp when they are collected and the image information, structural vibration response information and impact acoustic information are aligned in chronological order to establish the correspondence between multi-source monitoring data.

[0055] S2: Obtain the real-time operating parameters of the on-board positioning system of the rail vehicle and obtain information on ambient temperature and snow cover status of the line, and construct an operating environment parameter dataset that matches the multi-source monitoring data, wherein the real-time operating parameters include at least the rail vehicle operating speed, operating section mileage location, operating trajectory information and ambient temperature information.

[0056] Specifically, in step S2, the method includes: S21: Obtain real-time operating parameters collected by the on-board positioning system of the rail vehicle during train operation, wherein the operating parameters include train speed, mileage location in the operating section, and operating trajectory information.

[0057] Specifically, the on-board positioning system can use a satellite positioning device or a track circuit-based positioning device to acquire the position coordinate information of the rail vehicle in real time during operation and combine it with the speed information output by the train control system to form the running speed and running trajectory data at the corresponding time. The mileage position of the running section can be obtained by mapping the positioning coordinates to the line mileage marker system, thereby realizing the accurate positioning of the train in the section.

[0058] S22: Obtain ambient temperature information corresponding to the train operation process and obtain snow cover status information of the track line based on on-site observation or operation records. The snow cover status information includes different degrees of snow cover such as no track cover, covered sleepers, partially covered track height, or full coverage. The ambient temperature information is used to characterize the external ambient temperature when the rail vehicle is running.

[0059] Specifically, the ambient temperature information can be obtained through a temperature sensor installed on the outside of the vehicle or through the temperature data of the corresponding operating range obtained through the on-board environmental monitoring system; the snow cover status information of the line can be obtained through line inspection records, manual observation records or existing operating data and classified and marked according to the preset snow cover level standard to characterize the snow or ice cover of the line under different operating conditions.

[0060] S23: Perform time correlation and data matching on the real-time operating parameters, ambient temperature information, and line snow cover status information to form an operating environment parameter dataset that matches the multi-source monitoring data.

[0061] Specifically, based on a unified time reference, various types of data are timestamped and the operational parameter data and multi-source monitoring data are aligned according to the time series, so that the operational parameters and corresponding monitoring data within the same time period are correlated. At the same time, the data is segmented according to a preset time window, and the operational parameters, ambient temperature and snow cover status information within each time period are combined to form the corresponding operational environment parameter dataset.

[0062] S3: Based on the multi-source monitoring data and operating environment parameter dataset, a multimodal fusion analysis method is used to identify foreign object impact events, and to determine the time of foreign object appearance, impact time, and corresponding impact characteristic parameters.

[0063] Specifically, in step S3, the method includes: S31: The image information is processed frame by frame, and a preset detection method is used to identify moving foreign objects in the image and extract the time node, spatial location and motion trajectory information of the foreign object. At the same time, the image change features at the moment of impact (local occlusion, deformation or splashing phenomenon caused by the contact between the foreign object and the structure) are identified, and image recognition data is generated. The preset detection method includes at least one of background subtraction method, optical flow analysis, target detection or change detection method.

[0064] Specifically, continuous frame difference calculation or optical flow field calculation is performed on video image sequences to identify target regions that change relative to the background and to determine the spatial location of foreign objects through connected component analysis or bounding box marking; at the same time, the trajectory of foreign objects is obtained by tracking the changes in the target position in adjacent frames; for the moment of impact, transient phenomena caused by contact between foreign objects and structures are identified by detecting features such as sudden changes in brightness, rapid changes in local areas or changes in occlusion areas in the image.

[0065] In S31, the method includes: S311: Based on the foreign object's trajectory information and spatial location in the image recognition data, and combined with the running speed and track snow cover status information in the operating environment parameter dataset, analyze the foreign object's direction of motion and initial position relative to the rail vehicle; wherein, the direction of the foreign object's origin can be determined by comparing the relative relationship between the foreign object's trajectory and the vehicle's direction of motion.

[0066] S312: Different types of impact events are classified and labeled according to the characteristics of the foreign object's movement and the duration of the impact. When the foreign object detaches from the surface of the vehicle body and moves downward or backward, it is judged as an impact from ice falling off the vehicle body; when the foreign object is sucked up from the surface of the track and moves upward or towards the vehicle body, it is judged as an impact from flying ice on the track; when the foreign object makes continuous contact or slides along the bottom structure of the vehicle, it is judged as an impact from snow scraping. Specifically, the classification can be made according to the changes in trajectory direction, the trend of speed change, and the duration threshold.

[0067] S32: By detecting the transient peak value, impact pulse width, and frequency response characteristics in the structural vibration response information, the time point and impact intensity of the external impact on the structure are determined, and the corresponding peak acceleration and impact response energy are calculated to generate impact vibration identification data; specifically, the vibration signal can be filtered to remove low-frequency background vibration and the sudden impact signal can be identified through the peak detection algorithm. At the same time, the impact pulse width is determined according to the peak duration and the impact response energy is calculated by integrating the square of the signal.

[0068] S33: Short-time Fourier transform or wavelet transform is used to extract the characteristic spectrum of impact sound information, identify impact sound events with suddenness and high energy concentration, and extract their main frequency range, energy distribution and duration characteristics to generate impact audio recognition data, which is used to distinguish foreign object impact from environmental noise; the audio signal can be processed in frames and the energy of each frame can be calculated. When the energy of a certain frame exceeds a preset threshold, it is determined as a candidate impact sound event, and the foreign object impact sound is distinguished from background noise by combining the spectral distribution characteristics.

[0069] S34: Perform time alignment processing on image recognition data, impact vibration recognition data and impact audio recognition data, and associate and match events in the three types of data by setting a time window. When at least two types of data simultaneously meet the preset impact characteristic conditions within the same time window, it is determined as a valid foreign object impact event and the event recognition result is output, which includes the corresponding foreign object appearance time, impact time, impact location, initial judgment of impact type and corresponding impact characteristic parameters.

[0070] Specifically, step S34 synchronizes and aligns events from different data sources according to time order based on the timestamp of the unified time stamp in step S1 and sets a preset time window. When the time difference between events from different data sources is less than the threshold of the time window, they are determined to be the same impact event. When at least two types of data simultaneously meet the preset impact characteristic conditions within the same time window, they are determined to be a valid foreign object impact event, and an event identification result including the corresponding foreign object appearance time, impact time, impact location, preliminary determination of impact type, and corresponding impact characteristic parameters is output.

[0071] S4: Based on the size, material parameters, and impact characteristic parameters of the foreign object, calculate the impact load of the foreign object on the impacted component. Based on the aerodynamic adsorption model established according to the aerodynamic pressure distribution law during the operation of the rail vehicle, analyze the entrainment capacity of the track surface particles under the action of the train's aerodynamic negative pressure and calculate the critical size and critical mass of foreign objects with different densities and shapes to determine the range of foreign objects involved in the impact.

[0072] Specifically, in step S4, the geometric dimensions and motion state of the foreign object involved in the impact are determined based on the size of the foreign object, material parameters, and impact characteristic parameters.

[0073] Specifically, the feature dimensions of the foreign object in the image are obtained based on image recognition data, and proportional conversion is performed using known structural dimensions to obtain the length L, width W, and height H of the foreign object. The density ρ of the foreign object can be selected based on the typical density range of ice and snow materials. Then, based on the size of the foreign object, material parameters, and train speed, the impact load of the foreign object on the struck component is calculated using the following formula: ,in For the maximum impact load, Foreign matter density, These are the length, width, and height dimensions of the foreign object. Let be the train speed, and the min function be the function that takes the minimum value.

[0074] The above formula is based on the principle of momentum change and contact time estimation. It uses the minimum characteristic size of the foreign object as the impact contact scale to characterize the impact time, thereby obtaining an approximate calculation result of the impact load.

[0075] Based on the calculation of impact load, an aerodynamic adsorption model is established to analyze the entrainment capacity of track surface particles under the aerodynamic negative pressure of the train. In the specific implementation process, based on the aerodynamic negative pressure formed in the bottom area of ​​the train during the operation of the rail vehicle, the aerodynamic force acting on the foreign object is compared and analyzed with its own gravity to determine whether the foreign object can be entrained into the airflow.

[0076] Specifically, in S4, the method includes: The foreign object particle is simplified to a sphere, and its calculation formula is as follows: Therefore, the critical foreign object particle size under stable maximum negative pressure is calculated as follows: ,in The negative pressure exerted on the roadbed by the aerodynamic effect of the train is due to adsorption. Let be the radius of the sphere. The acceleration due to gravity is used to determine whether an object of a certain size can be drawn in by the airflow when the aerodynamic force is greater than or equal to the gravity.

[0077] For non-spherical foreign objects, further simplifying the foreign object into a cubic structure, the formulas for calculating the size and weight of the adsorbable cubic foreign object are as follows: Furthermore, the side length of the critical cubic foreign object under stable maximum negative pressure was calculated as follows: ,in The dimensions of the cubic-shaped foreign object are: length, width, and height. The area of ​​its top and bottom surfaces is: .

[0078] Therefore, the critical size of foreign objects under different density conditions is obtained through the above calculations, thereby determining the range of foreign objects that can be entrained under a given operating speed. Based on the impact load calculation results and the critical size range of foreign objects calculated by the aerodynamic adsorption model, the foreign objects involved in the impact are screened to determine the effective range of foreign objects that may have an impact effect on the rail vehicle, providing a basis for subsequent risk assessment.

[0079] S5: Classify and statistically analyze all identified impact events based on operating section, ambient temperature, snow cover on the track, and vehicle location. Obtain impact frequency distribution characteristics and high-incidence section distribution patterns. Combine impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence section distribution to evaluate the impact on key parts of the rail vehicle and output comprehensive analysis results.

[0080] Specifically, in S5, all identified impact events are categorized and statistically analyzed based on operating area, ambient temperature, snow cover level on the line, and vehicle location, including: S51: Based on the preset time window, the identified foreign object impact events are statistically analyzed, and the number of impacts per unit time is calculated. In the specific implementation process, the continuous time can be divided into equally spaced time periods, and the number of impact events in each time period can be accumulated and statistically analyzed. At the same time, the impact events are segmented and statistically analyzed according to the rail vehicle operating section, and the impact events are classified according to the corresponding mileage interval to obtain the impact frequency distribution of different intervals.

[0081] S52: Correlate the impact events with the corresponding operating speed and ambient temperature to form impact frequency distribution characteristics.

[0082] Specifically, the impact events are grouped and statistically analyzed according to different speed and temperature ranges to obtain the changes in impact frequency under different operating conditions, thereby characterizing the relationship between impact behavior and operating conditions.

[0083] S53: By matching the impact events with the mileage location of the rail vehicle operating section, the high-incidence sections of impact are determined. Combined with the snow cover status information of the line, the distribution of impact events under different snow cover conditions is compared and analyzed to obtain the distribution pattern of high-incidence sections of impact events.

[0084] Specifically, the number of hitting events in each interval is counted and sorted or filtered by threshold. Intervals with a hitting frequency higher than a preset threshold are identified as high-incidence intervals, thereby obtaining the distribution pattern of high-incidence intervals of hitting events.

[0085] Specifically, in S5, risk level assessments of key components of rail vehicles are conducted by combining impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence interval distribution, including: S54: Based on the impact frequency distribution characteristics, count the number of foreign object impacts on each key structural part per unit time and determine the impact frequency level; and based on the impact characteristic parameters and impact load, determine the impact intensity level of each key structural part under the impact of foreign object.

[0086] Specifically, based on the grading standards or segmented intervals set in advance by the designers, the frequency of impact and the intensity of impact are divided into different levels to represent different degrees of influence.

[0087] S55: Combining the aforementioned range of foreign objects and the distribution pattern of high-incidence areas, a comprehensive analysis is conducted on the impact risk of each key structural component under different operating ranges and different snow cover conditions, generating an impact spatial distribution.

[0088] Specifically, the frequency and intensity of impacts on each key structural component, as well as the frequency of impacts in their respective zones, are analyzed in combination to generate corresponding spatial distribution information of impacts.

[0089] S56: Based on the impact frequency level and impact intensity level, classify the key structural parts of the rail vehicle into different risk levels and identify high-risk parts.

[0090] Specifically, a multi-level classification method is adopted to divide risks into low-risk, medium-risk, and high-risk levels, and areas with both high frequency of impact and high impact intensity are identified as high-risk areas.

[0091] S57: Output includes a comprehensive analysis of impact type distribution, impact spatial distribution, impact frequency level, impact intensity level, and risk level.

[0092] Specifically, the analysis results can be output in the form of charts, distribution maps or data reports and categorized according to operating sections or vehicle parts for use in rail vehicle operation safety assessment and structural optimization analysis.

[0093] Thus, by way of example, this application monitors key parts of the EMU by means of video surveillance, audio pickup and vibration testing, visualizes the surrounding environment where the impact fault occurs, and analyzes the impact section and carriage.

[0094] like Figure 2As shown, four cameras (1#, 2#, 4#, and 8#) were installed to observe the sand spreading device on the EMU. Cameras 1#, 2#, and 4# were installed at the FSK mounting base behind the snow deflector of car 1, while camera 8# was installed on the side guard plate of car 1. These cameras, with a 120° field of view, were used to observe the sand spreading pipe, ambient temperature changes in the bogie area, and the impact of foreign objects on the brake pad support and brake caliper. For the curved floor plate, one camera (3#) was installed to observe it. Camera 3# was installed at the FSK mounting base behind the snow deflector of car 1. A 120° explosion-proof dome camera was used to observe the impact of foreign objects between the curved snow deflector and the obstacle remover. For the gearbox, two cameras (5# and 6#) are set up to observe the gearbox. Measurement points 5# and 6# are installed on the side of the universal joint safety bracket of vehicle 1. Explosion-proof hemispherical cameras with a 120° field of view are used to observe the axle, wheels, gearbox, brake pad support, brake caliper, and vibration temperature sensor for foreign object impact. For the air spring, two cameras (7# and 9#) are set up to observe the air spring. Measurement points 7# and 9# are installed on the side guard plate of vehicle 1. Explosion-proof hemispherical cameras with a 120° field of view are used to observe the air spring and other parts for foreign object impact. Measurement point 10# is set inside the floor plate to pick up audio information of the floor plate being hit by foreign objects.

[0095] like Figure 3 As shown, for the GFX antenna of car 3, one camera (12#) is set up to observe the GFX antenna. The 12# measuring point is installed in the middle of the upper protective plate of car 3. A 120° explosion-proof hemispherical camera is used to observe the GFX antenna, brake pad support, and brake caliper foreign object impact. The 11# measuring point is set inside the bottom plate to pick up the audio information of the bottom plate being hit by foreign objects. The 13# measuring point is installed on the charger hoisting partition to observe the foreign object impact on the side of the train skirt.

[0096] Later, a new high-frame-rate camera was installed on the structural beam behind the FSK mounting bracket on vehicle 1 to observe the impact of ice and snow debris on the snow protection plate. The installation location is as follows: Figure 4 As shown in (a).

[0097] Three uniaxial vibration acceleration measuring points (1#, 2#, and 3#) were installed at the center of the back of the upper and lower baffles of the snowproof board, as well as on the surface of the inner plate of the bottom plate. Figure 4 As shown in (b).

[0098] By installing high-frequency / high-resolution monitoring cameras on key parts of the bogies of the lead car and intermediate cars of the target EMU, the source, trajectory, and location of foreign object impacts are monitored. The GPS function of the main unit of the equipment is used to track and monitor the train's running speed, the location of the impact on the car body, and the ambient temperature at the time of the impact is observed by a thermal imager. All monitoring data is recorded for historical data analysis. Through functional modules such as 4G, BD, and WIFI, real-time remote viewing, vehicle positioning, voice dispatching, historical data download, and real-time data download can be achieved. Data can be transferred and analyzed using ground software by downloading via USB flash drive or by removing the onboard hard drive.

[0099] Since November 2023, the target EMU has been observed to experience phenomena such as ice falling onto the car body, snow scraping by the sand-spreading pipes, camera impacts, snow skid impacts, flying ice hitting the wheels, and snow scraping by the GXF antenna. Details are shown in the table below: Table: Statistics on Ice and Snow Object Strikes on Target EMU Trains At 13:11 on December 6, 2023, while the target EMU was running at 223 km / h between Station A West and Station B South, measuring point #5, installed on the side of the universal joint safety bracket, observed ice falling off the bottom of the gearbox of car 1. Analysis revealed that the turbulence caused by the negative pressure in the rear car was more pronounced. The high temperature in the gearbox melted the entrained snow, causing ice to accumulate. When track conditions and temperature changed, coupled with factors such as car body vibration and aerodynamic effects during operation, the ice at the bottom of the gearbox detached.

[0100] At 13:46 on December 10, 2023, while the target EMU was running at 112 km / h between Station A West and Station B South, measuring point #5 on the side of the universal joint safety bracket observed ice detaching from the bottom corner of the universal joint and striking the brake disc. Analysis indicated that the track environment was relatively good at the time of the impact. Due to factors such as train vibration and aerodynamic effects of the wind, the ice detached from the bottom corner plate of the universal joint and struck the brake disc. Because the ice was small and the speed was low, the impact was minimal.

[0101] At 13:00 on December 22, 2023, when the target EMU was running at 244 km / h between stations C and D, measuring point #5, installed on the side of the universal joint safety bracket of car 1, observed ice falling from the root of the universal joint and striking the wheels. Analysis showed that the ice accumulated on the gearbox broke off and struck the wheels due to factors such as high train speed, temperature changes, vibration, and wind.

[0102] At 12:57 PM on February 13, 2024, when the target EMU was running at 249 km / h between stations C and D, measuring point #9, installed on the side guard plate of car 1, observed ice detachment from the air spring. Analysis indicated that the track environment was relatively good at the time of the impact. Due to car body vibration and aerodynamic effects from the wind, ice detached from the suspension below the air spring and struck the train. Because the ice was small, the impact was minimal.

[0103] At 13:20 on February 13, 2024, when the target EMU was running at 200 km / h between stations C and D, measuring point #9, installed on the side guard plate of car 1, observed ice detachment from the air spring. Analysis indicated that the track environment was relatively good at the time of the impact. Due to car body vibration and aerodynamic effects of the wind, the ice and snow mixture attached to the side of the air spring detached and impacted the air spring. Because the ice and snow mixture was relatively soft, the impact was minimal.

[0104] At 16:06 on February 13, 2024, when the target EMU was running at 247 km / h near Station H-E, measuring point #3, located at the FSK mounting base on the rear side of the snowproof panel of car 1, observed ice detaching from the seams at the bottom of the snowproof panel. Analysis indicated that the track environment was relatively good at the time of the impact. Due to factors such as car body vibration and aerodynamic effects from the wind, the ice detached from the seams at the bottom of the snowproof panel and struck the track. Because the ice was small, the impact was minimal.

[0105] At 13:21 on February 18, 2024, when the target EMU was running at 50 km / h between stations R and F, measuring point #5, installed on the side of the universal joint safety bracket, observed ice falling off the bottom of the gearbox of car 1. Analysis revealed that the turbulence caused by the negative pressure in the rear car was more pronounced. The high temperature in the gearbox melted the entrained snow, forming ice deposits. When track conditions and temperature changed, coupled with factors such as car body vibration and aerodynamic effects during operation, the ice at the bottom of the gearbox detached.

[0106] At 13:09 on February 19, 2024, when the target EMU was running at 248 km / h between stations C and D, measuring point #5, installed on the side of the universal joint safety bracket, observed ice falling off the bottom of the gearbox of car 1. Analysis revealed that the turbulence caused by the negative pressure in the rear car was more pronounced. The high temperature in the gearbox melted the entrained snow, forming ice that accumulated. When track conditions and temperature changed, coupled with factors such as car body vibration and aerodynamic effects during operation, the ice at the bottom of the gearbox detached.

[0107] At 8:33 AM on November 17, 2023, when the target EMU was traveling at 39 km / h between Station I and Station J, camera at monitoring point #8 on car 1 captured a scraping incident between the car's sand-spreading pipe and snow accumulation on the switch track. Analysis revealed that ice and snow accumulate at the rail joints when the EMU passes the switch, making it prone to severe friction or collision with the sand-spreading pipe as the train passes. This results in significant stress on the components, potentially causing impact damage to the pipe and the lower-positioned sand-spreading nozzles.

[0108] At 6:52 AM on November 18, 2023, when the target EMU was running at 149 km / h between stations K and L, camera #8 on car 1 captured a continuous and severe scraping between the sand-spreading pipe of car 1 and the snow pile on the track. Analysis revealed that the sand-spreading pipe was positioned relatively low during the journey, and the bottom sand-spreading pipe and the snow pile beside the track were scraping continuously for a long time. The severe scraping stirred up a large amount of flying snow, covering the entire monitoring screen.

[0109] At 8:49 AM on January 12, 2024, when the target EMU was running at 85 km / h near station M, the camera at measuring point #2, installed on the FSK mounting base behind the snow protection plate of car 1, observed a relatively serious snow-scraping phenomenon in the sand-spreading pipe.

[0110] At 8:32 AM on January 15, 2024, when the target EMU was running at 250 km / h near station N, the camera at measuring point 2 on the back of the snow protection plate of car 1 observed a serious snow-scraping phenomenon caused by the sand spreading device.

[0111] At 10:33 AM on November 18, 2023, while the target EMU was traveling at 190 km / h between stations I and J, a camera on the undercarriage of car 1 was knocked off by a foreign object. Figure 5 As shown. Preliminary analysis indicates that due to the low position of the base plate, the view of the monitoring camera at this location was obstructed by a large amount of swirling snow. The incident could only be reconstructed through sound and subsequent inspection after the camera returned to the depot. After the camera was struck by an unknown foreign object, the sound time-domain signal abnormally increased, with the dominant frequency of the impact sound at 1400Hz. The camera did not immediately fall after the impact, remaining fixed to the base plate for 6 seconds. Afterward, it was suspended by the cable above the camera. The wind from the train acted on the suspended camera, with wind noise at a dominant frequency of 500Hz. The camera's view then changed to the sleepers on the rail surface.

[0112] At 19:08 on December 22, 2023, while the target EMU was traveling at 250 km / h between Station B South and Station A West, the camera at monitoring point #12, located in the middle of the upper protective plate of car 3, fell off. The train was traveling in the same direction as car 1, and at a high speed. Due to icing, no foreign object was detected. Audio analysis revealed no obvious abnormalities. Video footage showed significant abnormal camera shaking at 18:55 and 18:59, with the camera starting to wobble at 19:08, which was determined to be the time the camera fell. Analysis of other monitoring points revealed thick snow accumulation in that section and significant ice buildup around the monitoring points after the train entered the depot. Figure 6 As shown.

[0113] At 18:48 on December 14, 2023, when the target EMU was running at 246 km / h between Station B South and Station P East, a whole snow bale was observed hitting the snow protection plate at the FSK mounting point installed on the rear side of the snow protection plate of car 1. Figure 7 As shown in (a) and (b). Analysis shows that the snow accumulation on the track bed in this section is relatively thick, with obvious snow bulges and sharp edges. The train is running at a high speed and in strong winds. The train is in the last position in the direction of travel. The flying snow hits the middle part of the bottom baffle of the snow guard, causing it to be recessed inward by about 1-2 cm, resulting in the deformation of the baffle.

[0114] At 10:55 AM on December 26, 2023, when the target EMU was running at 250 km / h between Station O West and Station G, a foreign object struck the snowproof plate (the last car of car 1) at measuring point #3, which was installed at the FSK mounting seat behind the snowproof plate. Figure 7 As shown in (c) and (d). Due to the high speed of the EMU, the small observation area, and the insufficient image acquisition frame rate, even after analyzing the video at a maximum frame rate of 25 frames / s, the shape and size of the impacting foreign object were still not captured. However, based on the splashing situation, it was initially judged to be ice, and there was obvious damage to the bottom baffle of the snowproof board.

[0115] At 22:09 on December 30, 2023, when the target EMU was running at 99 km / h between H station and Q station, the lower left part of the snow protection plate was hit by flying ice, causing the surface ice to fall off. Figure 7 (e) and (f). One of the ice shards flew towards the camera, but due to the low speed of the train, it did not damage any parts or pose a safety hazard to the train.

[0116] At 2:57 PM on December 31, 2023, while the target EMU was traveling at 153 km / h between Station B South and Station P East, a large chunk of ice detached from the snow protection panel of one car after being struck. Figure 7(g) and (h). It was determined that a large piece of ice struck the lower left side of the snow guard, at a certain angle along the width of the vehicle. After the impact, the ice attached to the snow guard fell off as a whole. Due to the low vehicle speed, no damage was caused after the impact.

[0117] At 20:12 on January 25, 2024, while the target EMU was traveling at 248 km / h between Station B South and Station P East, the lower right corner of the snow guard plate on car 1 was struck by ice. Figure 8 As shown, after the impact, there were obvious ice marks from the splashing, and the maximum normal vibration acceleration of the snow guard was 3.09g.

[0118] At 7:38 AM on January 27, 2024, when the target EMU was traveling at 250 km / h near Station N, a high-frequency high-definition camera captured footage of ice hitting the middle of the snow guard of a car. Figure 9 As shown, the high-frequency camera can clearly observe the splashing moment after the ice block foreign object hits, and the maximum normal vibration acceleration of the snowproof plate is 2.98g.

[0119] At 12:52 PM on January 27, 2024, when the target EMU was running at 200 km / h between P station East and R station, images were observed of the lower left corner of the snow guard of car 1 before and after it was struck by a cluster of flying ice. Figure 10 As shown, the high-frequency camera can capture images and the shape of the foreign object before and after the ice block is hit relatively clearly, and the maximum normal vibration acceleration of the snowproof board reaches 3.20g.

[0120] At 12:56 PM on January 27, 2024, when the target EMU was running at 200 km / h between P Station East and R Station, images were observed of the right side of the snow guard of car 1 before and after it was struck by a cluster of flying ice. Figure 11 As shown, the maximum normal vibration acceleration of the snowproof board reaches 3.11g.

[0121] At 19:17 on January 27, 2024, when the target EMU was traveling at 250 km / h near the south of Station B, the bottom of the snow protection panel of one car was struck by flying ice, with a normal acceleration of 3.43g (e.g., Figure 12 (As shown).

[0122] At 12:39 PM on December 6, 2023, while the target EMU was running at 223 km / h between Station A West and Station B South, measuring point #5, installed on the side of the universal joint safety bracket, observed ice floes striking the drive shaft (impact velocity approximately 4 m / s). Figure 13 As shown. Preliminary analysis suggests that after the ice on the car body melted, factors such as car body vibration and train aerodynamic effects caused the ice on the end plate or bottom plate to break off and strike the bottom drive shaft of the bogie of car 1 along the width of the car.

[0123] At 10:00 AM on January 13, 2024, when the target EMU entered Station S at a speed of 5 km / h, measuring point #12, located in the middle of the upper protective plate of car 3, observed continuous deep scraping of snow from the right-side GFX antenna on the track. Figure 14 As shown in (a) and (b).

[0124] At 10:28 AM on January 15, 2024, when the target EMU was traveling at 100 km / h near the west side of Station A, the left-side GFX antenna was observed scraping against the snow accumulated on the track at measuring point #12 in car 3. Figure 14 As shown in (c) and (d).

[0125] Based on the 36cm selected from the on-site route survey L )×6cm( W )×6cm( H Conical ice blocks (such as) Figure 15 As shown), the analysis of the EMU (Electric Multiple Unit) is based on... The magnitude of the impact load on the ice block during operation. In the calculation, the ice density is taken as... The volume of ice is The mass of the ice cube is approximately m = ρ·V =1.2kg, the ice block's impact velocity relative to the train is .

[0126] Longitudinal impact along the length direction L =0.36m required time is According to the momentum theorem The calculated longitudinal impact load is: , Longitudinal impact in the width direction L =0.06m required time is According to the momentum theorem The calculated transverse impact load is: When impacted from a relatively small angle, the foreign object can generate a maximum impact force of nearly 100kN, equivalent to an impact force of 10 tons.

[0127] The generalized formula for calculating impact load, taking into account the size of the foreign object, material parameters, and impact velocity, is as follows: In the formula, For the maximum impact load, ρ The density of the ice cube L , W , H These are the length, width, and height dimensions of the ice object. Let be the train speed, and the min function be the function that takes the minimum value.

[0128] Aerodynamic simulation analysis was performed on the CRH5 vehicle. The coordinate systems were defined as follows: the positive direction of the X-axis (corresponding to the longitudinal axis of the vehicle body) is the direction of motion; the positive direction of the Z-axis (corresponding to the vertical axis of the vehicle body) is vertically upward; and the Y-axis (corresponding to the transverse axis of the vehicle body) lies on the horizontal plane. Figure 16 As shown.

[0129] The vehicle body geometry model was created using CATIA software, including the geometry model, nose shape, and bogie model, such as... Figure 17 As shown. The vehicle body surface mesh was generated using STARCCM+ software, with reasonable simplifications made for local areas. This application selected 8 vehicle sections for analysis and research. Figure 18 The finite element model consists of the overall mesh and local meshes of the vehicle body surface, the mesh of the tunnel wall, and the mesh of the open windows. The finite element model simplifies local areas such as the bogie to a certain extent, and such simplification has little impact on the flow field at the end.

[0130] The extreme value of the pressure distribution on the roadbed surface by the high-speed train at 250km / h is P =-1306.08Pa (adsorption lift), occurred when the wheel behind the first-position end bogie of the lead car passed the measuring point, but the duration of this pressure peak was extremely short, such as Figure 19 As shown.

[0131] Considering the presence of snow, ice, and ice covering the ballast on the tracks during winter, particles of varying densities will be generated: ballast, ice, and ballast encased in ice. The density of the ballast is... The density of ice is The volume ratio of ballast to ice in ice-covered ballast is defined as follows: r ( r = V 石 / V 冰 ), for different ballast ice ratios r Ice-covered ballast with compositions of 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, and 90% has the following densities: ρ =1128, 1336, 1544, 1752, 1960, 2168, 2376, 2584, 2792 .

[0132] The following analysis examines the size and weight of adsorbable foreign objects for both spherical and cubic shapes: 1) Spherical foreign object analysis: The foreign object particle is simplified as a sphere. The critical point at which the pressure on the sphere equals its own weight is the point at which it becomes unstable and flies off, as shown in the following formula: In the formula, P The negative pressure exerted on the roadbed by the aerodynamic effect of the train is due to adsorption. RLet be the radius of the sphere. ρ Density of the spherical foreign object g Let gravitational acceleration be (take) Therefore, the critical foreign object particle size under stable maximum negative pressure is calculated to be: .

[0133] The critical particle sizes for different types of foreign matter calculated accordingly are shown in the table below: Table: Size and weight of spherical foreign objects that can be adsorbed by the aerodynamic effect of the target EMU (particle diameter is twice the radius). 2) The length, width, and height of the cubical foreign object are all recorded as follows: a (Unit: m), the area of ​​the top and bottom surfaces is S = a 2 The size and weight of the adsorbable cubic foreign object are determined by using the condition that the adsorption force acting on that area is greater than or equal to the weight of the cubic foreign object. Furthermore, the side length of the critical cubic foreign object under stable maximum negative pressure was calculated as follows: .

[0134] The critical particle sizes for different types of foreign matter calculated accordingly are shown in the table below: Table: Size and weight of spherical foreign objects (side length of cubic foreign objects) that can be adsorbed by the aerodynamic effect of the target EMU The above analysis shows that when the target EMU runs at 250 km / h, the adsorbable spherical ballast particles have a diameter of 67 mm and a weight of 0.465 kg, ice ball particles have a diameter of 217 mm and a weight of 4.942 kg, and 10%~90% of the spherical ice-covered ballast particles have a diameter of 72~177 mm and a weight of 0.537~3.288 kg; the adsorbable cubic ballast particles have a diameter of 44 mm and a weight of 0.263 kg, ice ball particles have a diameter of 145 mm and a weight of 2.797 kg, and 10%~90% of the spherical ice-covered ballast particles have a diameter of 48~118 mm and a weight of 0.304~1.860 kg.

[0135] The statistics cover the period from December 22, 2023 to January 9, 2024. Based on the degree of snow cover on different lines, the data is categorized as follows: uncovered sleepers, sleepers only covered, covered 1 / 6-1 / 4 of the track height, covered 1 / 3-1 / 2 of the track height, covered 2 / 3-5 / 6 of the track height, and fully covered. Figure 21 As shown.

[0136] The following table summarizes the impact data on vulnerable parts of the target EMU, including the bottom drive shaft, sand spreading device, snow shield, wheels, and GXF antenna, under different operating temperatures, track snow cover levels, and different track sections: Table: Statistics and Intersection Distribution of Foreign Object Strikes on Target EMU Trains Statistical analysis shows that snowplows are the primary targets of foreign object (FOB) attacks, with a high frequency of impacts and diverse types of FOB objects, including large snow bales and small ice pieces. Most FOB attacks occur on tracks where snow only covers the sleepers, while snow-scraping by sand-spreading pipes, sand-spreading devices, and GXF antennas occurs on tracks where snow has buried the rails to a certain depth. Furthermore, the distribution of FOB attacks shows that nearly 80% of the attacks occur on the Mudanjiang-Jiamusi Railway, specifically in the sections between A Station West and B Station South, B Station South and P Station East, K Station and L Station West, and L Station West and D Station.

[0137] Therefore, from November 2023 to the present, the monitoring of the target EMU line operation has revealed phenomena such as ice falling on the car body, cameras falling off due to impact, snow scraping by sand spreading pipes, and impacts from snow protection plates and flying ice. Among them, there were two cases of cameras falling off due to impact, which occurred in the middle of the underboard of car 1 and the upper protective plate of car 3, respectively, caused by an unknown foreign object. There were four cases of snow scraping by sand spreading devices, which were manifested as short-term or continuous scraping between the sand spreading pipes and sand spreading devices and the thick layer of snow on the line. The impact of snow protection plates was the most prominent. Because the snow protection plates face the direction of operation and have a large wind-exposed area, they are the hardest hit areas for foreign objects such as ice and snow. The types of foreign objects include large snow bales, ice blocks and small ice blocks. No impacts caused by ballast stones on the line have been found so far. Analysis of the adsorption force of the aerodynamic effect of high-speed train operation on the roadbed shows that when the target EMU runs at 250km / h, the adsorbable spherical ballast particles have a diameter of 67mm and a weight of 0.465kg, ice ball particles have a diameter of 217mm and a weight of 4.942kg, and 10%~90% of the spherical ice-covered ballast particles have a diameter of 72~177mm and a weight of 0.537~3.288kg. The adsorbable cubic ballast particles have a diameter of 44mm and a weight of 0.263kg, ice ball particles have a diameter of 145mm and a weight of 2.797kg, and 10%~90% of the spherical ice-covered ballast particles have a diameter of 48~118mm and a weight of 0.304~1.860kg.

[0138] refer to Figure 22 As shown. Therefore, this application also relates to a multimodal fusion-based system for identifying and assessing ice and snow foreign object impacts on rail vehicles, which implements the above method, comprising: The multi-source monitoring unit 20 is distributed in multiple critical structural parts of the rail vehicle that are susceptible to impact from foreign objects. The critical structural parts include at least one of the following: bogie area, wheel axle area, gearbox area, braking system components, snow skid, sand spreading device, air spring, car body floor, and external auxiliary equipment area.

[0139] The multi-source monitoring unit 20 includes at least a monitoring camera, a thermal imager, a vibration acceleration sensor, and a microphone.

[0140] Control unit 21, connected to the multi-source monitoring unit, is used to perform and realize the identification and assessment of ice and snow foreign object impacts on rail vehicles based on multimodal fusion, wherein the control unit 21 includes: The multi-source data module 211 is used to acquire multi-source monitoring data of the rail vehicle under the operating state collected by the monitoring units of multiple key structural parts of the rail vehicle and to perform time synchronization marking on the multi-source monitoring data, wherein the multi-source monitoring data includes at least image information, structural vibration response information and impact acoustic information. The environmental data module 212 is used to acquire real-time operating parameters of the on-board positioning system of the rail vehicle and acquire information on ambient temperature and snow cover status of the line, and to construct an operating environment parameter dataset that matches the multi-source monitoring data. The foreign object detection module 213 is used to identify foreign object impact events by using a multimodal fusion analysis method based on the multi-source monitoring data and operating environment parameter dataset, and to determine the time of foreign object appearance, impact time and corresponding impact characteristic parameters. The foreign object analysis module 214 is used to calculate the impact load of the foreign object on the hit component based on the size, material parameters and impact characteristic parameters of the foreign object, and to analyze the entrainment capacity of track surface particles under the action of train aerodynamic negative pressure based on the aerodynamic pressure distribution law during the operation of the rail vehicle, and to calculate the critical size and critical mass of foreign objects with different densities and shapes, so as to determine the range of foreign objects involved in the impact. The statistical evaluation module 215 is used to classify and statistically analyze all identified impact events based on the operating section, ambient temperature, snow cover degree of the line and vehicle parts, obtain the impact frequency distribution characteristics and high-incidence section distribution patterns, and combine the impact characteristic parameters, foreign object range, impact frequency distribution characteristics and high-incidence section distribution to evaluate the impact on key parts of the rail vehicle, and output comprehensive analysis results.

[0141] In some embodiments of this application, this application also relates to a computer medium storing a computer program that is executed by a processor to implement the method described thereon.

[0142] In some embodiments of this application, this application also relates to a computer, including the aforementioned computer medium.

[0143] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0144] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for identifying and assessing ice and snow foreign object impacts on rail vehicles based on multimodal fusion, characterized in that, Includes the following steps: S1: Acquire multi-source monitoring data of the rail vehicle under operating conditions collected by the monitoring units of multiple key structural parts of the rail vehicle and perform time synchronization marking on the multi-source monitoring data, wherein the multi-source monitoring data includes at least image information, structural vibration response information and impact acoustic information. S2: Obtain the real-time operating parameters of the on-board positioning system of the rail vehicle and obtain information on ambient temperature and snow cover status of the line, and construct an operating environment parameter dataset that matches the multi-source monitoring data; S3: Based on the multi-source monitoring data and operating environment parameter dataset, a multi-modal fusion analysis method is used to identify foreign object impact events, and to determine the time of foreign object appearance, impact time, and corresponding impact characteristic parameters; S4: Based on the size, material parameters and impact characteristic parameters of the foreign object, calculate the impact load of the foreign object on the hit component, and based on the aerodynamic adsorption model established according to the aerodynamic pressure distribution law during the operation of the rail vehicle, analyze the entrainment capacity of the track surface particles under the action of the train's aerodynamic negative pressure, and calculate the critical size and critical mass of foreign objects with different densities and shapes to determine the range of foreign objects involved in the impact. S5: Classify and statistically analyze all identified impact events based on operating section, ambient temperature, snow cover on the track, and vehicle location. Obtain impact frequency distribution characteristics and high-incidence section distribution patterns. Combine impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence section distribution to evaluate the impact on key parts of the rail vehicle and output comprehensive analysis results.

2. The method for identifying and evaluating foreign object impacts on rail vehicles based on multimodal fusion according to claim 1, characterized in that, In step S1, the method includes: S11: Acquire image information of the rail vehicle under operating conditions collected by the monitoring unit of multiple key structural parts of the rail vehicle; S12: Obtain vibration response information of the rail vehicle structure under the action of foreign object impact from the monitoring unit of multiple key structural parts of the rail vehicle. S13: Acquire the impact acoustic information of the rail vehicle structure under the impact of foreign objects, collected by the monitoring unit of multiple key structural parts of the rail vehicle. S14: The image information, structural vibration response information, and impact acoustic information are uniformly time-stamped to associate them under the same time reference.

3. A method for identifying and assessing ice and snow foreign object impacts on rail vehicles based on multimodal fusion, as described in claim 1 or 2, characterized in that... In step S2, the method includes: S21: Obtain real-time operating parameters collected by the on-board positioning system of the rail vehicle during train operation, wherein the operating parameters include train speed, mileage location in the operating section, and operating trajectory information; S22: Obtain ambient temperature information corresponding to the train operation process and obtain snow cover status information of the track line based on on-site observation or operation records, wherein the snow cover status information includes different degrees of snow cover such as track not covered, sleepers covered, partially covered rail height or fully covered. S23: Perform time correlation and data matching on the real-time operating parameters, ambient temperature information, and line snow cover status information to form an operating environment parameter dataset that matches the multi-source monitoring data.

4. The method for identifying and evaluating ice and snow foreign object impacts on rail vehicles based on multimodal fusion according to claim 3, characterized in that, In step S3, the method includes: S31: Process the image information frame by frame, use a preset detection method to identify moving foreign objects in the image and extract the time node, spatial location and motion trajectory information of the foreign object, and at the same time identify the image change features at the moment of impact to generate image recognition data. S32: By detecting the transient peak value, impact pulse width and frequency response characteristics in the structural vibration response information, the time point and impact intensity of the external impact on the structure are determined, and the corresponding peak acceleration and impact response energy are calculated to generate impact vibration identification data. S33: Use short-time Fourier transform or wavelet transform to extract the characteristic spectrum of impact sound information, identify preset impact sound events and extract their main frequency range, energy distribution and duration characteristics, and generate impact audio recognition data; S34: Perform time alignment processing on image recognition data, impact vibration recognition data and impact audio recognition data, and associate and match events in the three types of data by setting a time window. When at least two types of data simultaneously meet the preset impact characteristic conditions within the same time window, it is determined to be a valid foreign object impact event and the corresponding foreign object appearance time, impact time and corresponding impact characteristic parameters are output.

5. The method for identifying and evaluating foreign object impacts on rail vehicles based on multimodal fusion according to claim 4, characterized in that, In S31, the method includes: S311: Based on the trajectory information and spatial location of the foreign object, combined with the running speed and snow cover status information of the track in the operating environment parameter dataset, analyze the direction of motion and initial position of the foreign object relative to the rail vehicle. S312: Classify and label different types of impact events based on the characteristics of the foreign object's movement and the duration of the impact. When the foreign object detaches from the surface of the vehicle body and moves downward or backward, it is determined to be an impact from ice falling off the vehicle body. When the foreign object is sucked up from the surface of the track and moves upward or towards the vehicle body, it is determined to be an impact from flying ice on the track. When the foreign object makes continuous contact or slides along the bottom structure of the vehicle, it is determined to be an impact from snow scraping.

6. The method for identifying and evaluating ice and snow foreign object impacts on rail vehicles based on multimodal fusion according to claim 4, characterized in that, In S4, the formula for calculating the impact load of the foreign object on the struck component, based on the foreign object's size, material parameters, and impact characteristic parameters, is as follows: ,in For the maximum impact load, Foreign matter density, These are the length, width, and height dimensions of the foreign object. Let be the train speed, and the min function be the function that takes the minimum value.

7. The method for identifying and evaluating foreign object impacts on rail vehicles based on multimodal fusion according to claim 6, characterized in that, In S4, the method includes: The foreign object particle is simplified to a sphere, and its calculation formula is as follows: Therefore, the critical foreign object particle size under stable maximum negative pressure is calculated as follows: ,in The negative pressure exerted on the roadbed by the aerodynamic effect of the train is due to adsorption. Let be the radius of the sphere. It is the acceleration due to gravity; The formula for calculating the size and weight of adsorbable cubic foreign objects is as follows: Furthermore, the side length of the critical cubic foreign object under stable maximum negative pressure was calculated as follows: ,in The dimensions of the cubic-shaped foreign object are: length, width, and height. The area of ​​its top and bottom surfaces is: .

8. The method for identifying and evaluating foreign object impacts on rail vehicles based on multimodal fusion according to claim 6, characterized in that, In S5, all identified impact events are categorized and statistically analyzed based on operating section, ambient temperature, snow cover level on the line, and vehicle location, including: S51: Statistically analyze the identified foreign object impact events according to the preset time window, calculate the number of impacts per unit time, and perform segmented statistical analysis of the impact events according to the rail vehicle operating section to obtain the impact frequency distribution of different sections. S52: Correlation analysis is performed between the impact events and the corresponding operating speed and ambient temperature to form impact frequency distribution characteristics; S53: By matching the impact events with the mileage location of the rail vehicle operating section, the high-incidence sections of impact are determined. Combined with the snow cover status information of the line, the distribution of impact events under different snow cover conditions is compared and analyzed to obtain the distribution pattern of high-incidence sections of impact events.

9. The method for identifying and evaluating foreign object impacts on rail vehicles based on multimodal fusion according to claim 8, characterized in that, In step S5, a risk level assessment is conducted on key components of the rail vehicle, taking into account impact characteristic parameters, the range of foreign objects, impact frequency distribution characteristics, and high-incidence interval distribution. This includes: S54: Based on the impact frequency distribution characteristics, count the number of foreign object impacts on each key structural part per unit time and determine the impact frequency level; and based on the impact characteristic parameters and impact load, determine the impact intensity level of each key structural part under the impact of foreign object. S55: Combining the aforementioned range of foreign objects and the distribution pattern of high-incidence areas, a comprehensive analysis is conducted on the impact risk of each key structural component under different operating conditions and different snow cover levels, generating an impact spatial distribution. S56: Based on the impact frequency level and impact intensity level, classify the key structural parts of the rail vehicle into different risk levels and identify high-risk parts. S57: Output includes a comprehensive analysis of impact type distribution, impact spatial distribution, impact frequency level, impact intensity level, and risk level.

10. A multimodal fusion-based system for identifying and assessing ice and snow foreign object impacts on rail vehicles, implementing the method of any one of claims 1-9, characterized in that, include: The multi-source monitoring unit is distributed in multiple critical structural parts of the rail vehicle that are susceptible to impact from foreign objects. The critical structural parts include at least one of the following: bogie area, wheel axle area, gearbox area, braking system components, snow skid, sand spreading device, air spring, car body floor, and external auxiliary equipment area. A control unit, connected to a multi-source monitoring unit, is used to perform and implement multi-modal fusion-based identification and assessment of ice and snow foreign object impacts on rail vehicles. The control unit includes: The multi-source data module is used to acquire multi-source monitoring data of the rail vehicle under the operating state collected by the monitoring units of multiple key structural parts of the rail vehicle and to perform time synchronization marking on the multi-source monitoring data. The multi-source monitoring data includes at least image information, structural vibration response information and impact acoustic information. The environmental data module is used to acquire real-time operating parameters of the on-board positioning system of the rail vehicle and to acquire information on ambient temperature and snow cover status of the line, and to construct an operating environment parameter dataset that matches the multi-source monitoring data. The foreign object detection module is used to identify foreign object impact events based on the multi-source monitoring data and operating environment parameter dataset, using a multi-modal fusion analysis method, and to determine the time of foreign object appearance, impact time, and corresponding impact characteristic parameters. The foreign object analysis module is used to calculate the impact load of foreign objects on the impacted parts based on the size, material parameters and impact characteristic parameters of foreign objects. Based on the aerodynamic adsorption model established according to the aerodynamic pressure distribution law during the operation of rail vehicles, it analyzes the entrainment capacity of track surface particles under the action of train aerodynamic negative pressure and calculates the critical size and critical mass of foreign objects with different densities and shapes to determine the range of foreign objects involved in the impact. The statistical evaluation module is used to classify and statistically analyze all identified impact events based on the operating section, ambient temperature, snow cover on the track, and vehicle parts. It obtains the impact frequency distribution characteristics and high-incidence interval distribution patterns, and combines impact characteristic parameters, foreign object range, impact frequency distribution characteristics, and high-incidence interval distribution to evaluate the impact on key parts of the rail vehicle, and outputs comprehensive analysis results.

Images