Rail level crossing risk management and control method and device, computer equipment and storage medium

Abstract

The embodiment of the invention provides a track level crossing risk management and control method and device, computer equipment and a storage medium, and the method comprises the steps: monitoring whether the step change of signal data collected by a pressure sensor disposed on a track in advance occurs in real time, respectively deploying pressure sensors at the track level crossing along the track at a preset distance forwards and backwards; under the condition that the step change of the signal data occurs, first alarm information is sent to prompt pedestrians and vehicles to avoid; determining whether the corresponding train is overspeed according to the signal data; under the condition that the train is overspeed, second alarm information is sent to prompt the train to decelerate, whether the train arrives at the rail level crossing or not and the current speed of the train can be determined through real-time monitoring of the pressure sensor, and risks possibly occurring at the rail level crossing are managed and controlled in real time.

Description

Technical Field

[0001] This application relates to the field of railway track technology, specifically to a method, device, computer equipment, and storage medium for risk management at railway level crossings. Background Technology

[0002] Level crossings are intersections where railway tracks and highways meet, and these crossings present numerous operational scenarios in rail transit manufacturing enterprises due to manufacturing processes and shunting operations. During these operations, factors such as speed differences between railway and highway vehicles, unauthorized temporary stops and climbing, unauthorized crawling under trains, and blind spots can easily lead to safety accidents.

[0003] In related technologies, safety management is typically carried out through the following methods: First, traffic lights are installed at level crossings to indicate the direction of people and vehicles to different directions, alerting trains to approach and preventing traffic accidents. Second, fishtail markers are installed at level crossings to indicate to train drivers when they are changing tracks. Third, publicity and education, along with violation assessment and management, are strengthened. Publicity and education are conducted to raise awareness of the dangers of level crossings, and assessment and management are strengthened to ensure that vehicles and personnel comply with safe passage regulations at level crossings, reducing accidents. Fourth, surveillance cameras are used to track, monitor, and punish violations, while onboard dispatchers direct vehicle and personnel passage and take appropriate departure measures to maintain safety and order at level crossings.

[0004] However, the relevant technologies have the following problems: First, there are many blind spots at track level crossings, which can easily obstruct the view of dispatchers and train drivers, potentially leading to safety hazards; Second, in the absence of adequate supervision, relevant personnel are prone to taking chances and committing violations, resulting in safety risks; Third, safety management relies more on manual labor, and the control of violations and risks at the work site depends on on-site personnel, increasing costs without eliminating safety hazards; Fourth, monitoring real-time video or recording requires people to participate in the background to find violations, which is time-consuming, labor-intensive, and inefficient. Summary of the Invention

[0005] This application provides a method, apparatus, computer equipment, and storage medium for risk management at railway level crossings.

[0006] The first aspect of this application provides a risk management method for railway level crossings, including:

[0007] Real-time monitoring is conducted to check whether there are any abrupt changes in the signal data collected by pressure sensors pre-deployed on the track. Pressure sensors are deployed at predetermined distances forward and backward along the track at the track level crossing.

[0008] In the event of a step change in signal data, a first alarm message is sent to alert pedestrians and vehicles to give way.

[0009] Determine whether the corresponding train is speeding based on the signal data;

[0010] If the train is speeding, a second alarm message will be sent to prompt the train to slow down.

[0011] In an optional embodiment of this application, determining whether the corresponding train is speeding based on signal data includes:

[0012] The current train speed is determined based on the signal data using the following expression:

[0013] v=d / Δt

[0014] Where v is the current speed of the train, d is the distance between adjacent wheels of the train, and Δt is the time interval between adjacent pulse signals collected by the pressure sensor.

[0015] The train's current speed is compared with a preset speed threshold. If the train's current speed exceeds the preset speed threshold, the train is deemed to be speeding.

[0016] In an optional embodiment of this application, the method further includes:

[0017] Real-time monitoring to ensure that the infrared gratings deployed on both sides of the track at the level crossing are completely blocked by the train;

[0018] When the infrared grating is completely blocked by the train, the first camera device deployed at the track level crossing collects images of the pre-set danger zones on both sides of the train.

[0019] The image captured by the first camera device is input into the pre-trained human body recognition model, and the output is the result of whether a human body is recognized.

[0020] If the output shows that a human body has been detected, a third alarm message is sent to drive people away from the preset danger zone.

[0021] In an optional embodiment of this application, the human body recognition model is trained through the following steps:

[0022] The track level crossing is divided into zones, and dangerous areas that need to be monitored are selected from the divided zones. The scope and boundaries of human detection in the dangerous areas are then determined.

[0023] Collect monitored images and annotate them to obtain the annotation results of human bodies in the images;

[0024] The monitored images are used as input to the model, and the annotation results of the human bodies in the images are used as output to train the human body recognition model, thus obtaining a trained human body recognition model.

[0025] In an optional embodiment of this application, the inference speed of the human body recognition model is improved by optimizing the floating-point operation, wherein the floating-point operation is determined by the following expression:

[0026]

[0027] Where FLOPs represents floating-point operations, L represents the number of layers in the human body recognition model, and C represents the number of layers in the model. in C represents the number of channels in the input feature map. out K is the number of channels in the output feature map, H is the convolution kernel, and H and W are the height and width of the output feature map, respectively.

[0028] In an optional embodiment of this application, the loss function of the human body recognition model is expressed as follows:

[0029] L box =1-IoU+αd 2 +βv

[0030] Among them, L box Here, represents the loss function value of the human body recognition model, IoU represents the overlap between the target human body and the predicted bounding box, and d represents the loss function value of the human body recognition model. 2 denoted as the center offset between the target human body and the prediction box, v is the aspect ratio deviation between the target human body and the prediction box, and α and β are the weights of the center offset and aspect ratio deviation, respectively.

[0031] In an optional embodiment of this application, the method further includes:

[0032] Second cameras are deployed at track switch locations and in various hazardous areas along the track within the factory area. The images captured by the second cameras are displayed in real time on wireless display terminals deployed on the train. These cameras are used to display the status of the fishtail markers at the track switch locations and images of various hazardous areas along the track, providing the train driver with a complete field of vision.

[0033] A second aspect of this application provides a risk management device for railway level crossings, comprising:

[0034] The monitoring module is used to monitor in real time whether there is a step change in the signal data collected by the pressure sensors pre-deployed on the track. The pressure sensors are deployed at preset distances forward and backward along the track at the track level crossing.

[0035] The first transmitting module is used to send a first alarm message when the signal data undergoes a step change, in order to prompt pedestrians and vehicles to give way.

[0036] The determination module is used to determine whether the corresponding train is speeding based on the signal data.

[0037] The second sending module is used to send a second alarm message when the train is speeding, to prompt the train to slow down.

[0038] A third aspect of this application provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of any of the above-mentioned track level crossing risk management methods.

[0039] A fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when executed by a processor, the computer program implements the steps of the track level crossing risk management method as described in any of the preceding claims.

[0040] Compared with the prior art, the technical solutions provided in this application have at least some or all of the following advantages:

[0041] The risk management method for track level crossings described in this application involves real-time monitoring of signal data collected by pressure sensors pre-deployed on the track to determine if there are any abrupt changes. Pressure sensors are deployed at predetermined distances forward and backward along the track at the track level crossing. When a abrupt change occurs in the signal data, a first alarm message is sent to alert pedestrians and vehicles to avoid the crossing. The method also determines whether the corresponding train is speeding based on the signal data. If the train is speeding, a second alarm message is sent to prompt the train to slow down. This method enables real-time monitoring of potential risks at track level crossings by determining whether a train has arrived at the level crossing and its current speed. Attached Figure Description

[0042] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0043] Figure 1 A flowchart illustrating a risk management method for track level crossings during train travel, provided in one embodiment of this application;

[0044] Figure 2 A flowchart illustrating a risk management method for track level crossings during train travel, provided in another embodiment of this application;

[0045] Figure 3 This is a schematic diagram of wireless coverage along a track provided in one embodiment of this application;

[0046] Figure 4 This is a schematic diagram of the deployment of track level crossing equipment according to one embodiment of this application;

[0047] Figure 5A flowchart illustrating a risk management method for track level crossings when the train is stationary, provided in one embodiment of this application;

[0048] Figure 6 A schematic diagram of a risk control device for a track level crossing provided in one embodiment of this application;

[0049] Figure 7 This is a schematic diagram of a computer device structure provided in one embodiment of this application. Detailed Implementation

[0050] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0051] Please see Figure 1 and Figure 2 The risk management method for railway level crossings provided in this application includes the following steps S100 to S400:

[0052] S100 monitors in real time whether the signal data collected by the pressure sensors pre-deployed on the track undergoes a step change. Pressure sensors are deployed at preset distances forward and backward along the track at the track level crossing.

[0053] S200 sends a first alarm message when there is a step change in signal data to alert pedestrians and vehicles to avoid the area.

[0054] S300 determines whether the corresponding train is speeding based on signal data;

[0055] S400 sends a second alarm message when the train is speeding, to prompt the train to slow down.

[0056] In one optional embodiment of this application, pressure sensors can be deployed 300 meters before and after the track level crossing. The pressure sensors detect whether a train is approaching and triggers a pressure sensor signal. If a pressure sensor signal is received, it means that a train is about to approach the track level crossing, and an audible and visual alarm is immediately triggered to warn pedestrians and vehicles at the track level crossing to give way. At the same time, the algorithm server receives the pressure sensor pulse signal data and calculates whether the train is speeding. If the train is speeding, the background system notifies the safety officer to intervene and conduct an assessment.

[0057] In an optional embodiment of this application, for hazards in the train shunting operation scenario at a track level crossing, such as pedestrians or vehicles entering the railway in the direction of train travel during shunting operations, causing collisions; trains are not allowed to exceed speed limits on shunting tracks within the factory area, resulting in a certain braking distance if the train exceeds the speed limit, and excessive braking distance can easily cause emergency risks; personnel illegally climbing onto the train body and crawling under it when the train temporarily occupies the track, which can easily lead to the risk of being run over due to blind spots for the driver and commander; and the risk of the train not changing its lane correctly or even derailing and overturning if the track switch in the direction of train travel is not switched to the correct position, the signal data collected by the pressure sensor on the track is used to determine whether the train is about to arrive at the track level crossing and the current speed of the train, thus constructing an integrated risk prevention and control mechanism.

[0058] In one optional embodiment of this application, before real-time monitoring of whether the signal data collected by pressure sensors pre-deployed on the track undergoes a step change, a 5G basic network is constructed. 5G equipment is deployed on the shunting test line within the factory area to build a private 5G dedicated network, providing basic network support for video acquisition, signal control, and service data flow. The 5G network devices along the track are deployed as follows: Figure 3 and Figure 4 As shown, after building the 5G basic network, front-end equipment (monitoring video, infrared laser, pressure sensors, audible and visual alarms, programmable logic controllers) and linkage computing modules are deployed. Cameras are deployed in high-risk areas to provide omnidirectional coverage of level crossings. Pressure sensors are deployed 300 meters before and after the track at the level crossing intersection. When a train is about to arrive at the level crossing, the audible and visual alarm is triggered to warn vehicles and pedestrians near the crossing. Infrared gratings are installed on both sides of the track at the level crossing to detect whether a train is temporarily blocking the track at the level crossing. If a train stops and someone illegally enters the risk area, the audible and visual alarm is triggered to drive them away, and the background alarm records the violation. Cameras are deployed at the track switches, and wireless display terminals are deployed in the train driver's cab to display the status of the fishtail markers at the switches in real time. At the same time, the monitoring images of dangerous areas along the route can be displayed in real time, providing the train driver with a omnidirectional view of blind spots.

[0059] In an optional embodiment of this application, step S300, determining whether the corresponding train is speeding based on the signal data, includes:

[0060] The current train speed is determined based on the signal data using the following expression:

[0061] v = d / Δt

[0062] Where v is the current speed of the train, d is the distance between adjacent wheels of the train, and Δt is the time interval between adjacent pulse signals collected by the pressure sensor.

[0063] The train's current speed is compared with a preset speed threshold. If the train's current speed exceeds the preset speed threshold, the train is deemed to be speeding.

[0064] In an optional embodiment of this application, when a train passes the pressure sensor on the track at a constant speed, it is determined whether the train is speeding. When a train passes the pressure sensor above the track at a constant speed v, the sensor detects periodic pressure pulses as the wheels pass. The relationship of uniform motion is as follows: the train speed is v, the wheel spacing is d, and the interval Δt between adjacent wheels passing the sensor satisfies d = v·Δt, which simplifies to v = d / Δt. The regularity of the pressure signal is as follows: the pressure signal P(t) output by the sensor is a periodic pulse, with each pulse corresponding to the passing of one wheel. If the duration of wheel contact with the sensor is τ, then a single pulse can be represented as a rectangular function:

[0065]

[0066] Where u(t) is the unit step function, P is the pressure of a single wheel, and N is the total number of wheels.

[0067] If Δt = 0.5s is measured, and d = 10m is known, then the velocity is:

[0068]

[0069] In an optional embodiment of this application, see [link to relevant documentation]. Figure 5 The method further includes:

[0070] Real-time monitoring to ensure that the infrared gratings deployed on both sides of the track at the level crossing are completely blocked by the train;

[0071] When the infrared grating is completely blocked by the train, the first camera device deployed at the track level crossing collects images of the pre-set danger zones on both sides of the train.

[0072] The image captured by the first camera device is input into the pre-trained human body recognition model, and the output is the result of whether a human body is recognized.

[0073] If the output shows that a human body has been detected, a third alarm message is sent to drive people away from the preset danger zone.

[0074] In an optional embodiment of this application, the human body recognition model is trained through the following steps:

[0075] The track level crossing is divided into zones, and dangerous areas that need to be monitored are selected from the divided zones. The scope and boundaries of human detection in the dangerous areas are then determined.

[0076] Collect monitored images and annotate them to obtain the annotation results of human bodies in the images;

[0077] The monitored images are used as input to the model, and the annotation results of the human bodies in the images are used as output to train the human body recognition model, thus obtaining a trained human body recognition model.

[0078] In one optional embodiment of this application, human body detection is performed on a designated area of ​​the level crossing, and an alarm is triggered when a human body is detected in the image, with the alarm information reported to the management platform. The specific implementation steps are as follows:

[0079] First, the work area is set as a scene for object detection: The level crossing area is first divided into scene sections, key areas requiring monitoring are marked, and the scope and boundaries of human detection are clearly defined. Human targets in the monitoring footage are then labeled using annotation tools to form a training dataset.

[0080] Second, complete the training of the algorithm model: Based on the deep learning framework (YOLOv8), use the labeled dataset to train the model and optimize the model's recognition accuracy and detection speed for human targets.

[0081] Third, test and optimize the trained model: validate the model's performance on the test set, evaluating its accuracy, recall, and false positive rate. Based on the test results, fine-tune the model to ensure it can still run stably in complex environments (such as changes in lighting, occlusion, etc.).

[0082] Fourth, complete the interface development of the algorithm model and return the detection results: encapsulate the trained model into a callable API interface, receive monitoring screen data in real time and return the detection results, including the location, number and alarm status of human targets.

[0083] In an optional embodiment of this application, the inference speed of the human body recognition model is improved by optimizing the floating-point computation, wherein the floating-point computation is determined by the following expression:

[0084]

[0085] Where FLOPs represents floating-point operations, L represents the number of layers in the human body recognition model, and C represents the number of layers in the model. in C represents the number of channels in the input feature map. out K is the number of channels in the output feature map, H is the convolution kernel, and H and W are the height and width of the output feature map, respectively.

[0086] In an optional embodiment of this application, the loss function of the human body recognition model is expressed as follows:

[0087] L box =1-IoU+αd2 +βv

[0088] Among them, L box Here, represents the loss function value of the human body recognition model, IoU represents the overlap between the target human body and the predicted bounding box, and d represents the loss function value of the human body recognition model. 2 denoted as the center offset between the target human body and the prediction box, v is the aspect ratio deviation between the target human body and the prediction box, and α and β are the weights of the center offset and aspect ratio deviation, respectively.

[0089] In this application, YOLOv8 supports edge devices, can run on low-power devices, and is suitable for industrial inspection.

[0090] In an optional embodiment of this application, the method further includes:

[0091] Second cameras are deployed at track switch locations and in various hazardous areas along the track within the factory area. The images captured by the second cameras are displayed in real time on wireless display terminals deployed on the train. These cameras are used to display the status of the fishtail markers at the track switch locations and images of various hazardous areas along the track, providing the train driver with a complete field of vision.

[0092] In one optional embodiment of this application, a camera and two pairs of infrared gratings are deployed at the track level crossing. It is determined whether the two pairs of infrared gratings are completely blocked by a train. If they are completely blocked, it means that the train is temporarily working and parked at the track level crossing. At this time, personnel are not allowed to pass through or climb over the gratings. If both pairs of infrared gratings are blocked at the same time, the dangerous area intrusion algorithm of the level crossing is activated. If illegal crossing or climbing behavior is detected, the sound and light alarm is activated and the on-site command personnel are notified to stop and drive away the illegal behavior in a timely manner, and the background security is notified to handle the illegal behavior.

[0093] In one optional embodiment of this application, the camera parameters are: maximum pixel count: 4 megapixels, sensor size: 1 / 1.8", focal length: 2.8–12mm, zoom method: motorized zoom, aperture: F1.2, audio input: 1 input, audio output: 1 output, alarm input: 2 inputs, alarm output: 1 output, and waterproof / dustproof rating: IP67. The audible and visual alarm parameters are: switchable alarm volume and tone, MODBUS-RTU protocol, default baud rate of 9600, and built-in RJ45. 45 network ports, supports USB flash drive replacement. Operating voltage: DC12V / DC24V; AC220V. Rated power: 25W. Sound level: 120dB (adjustable). Light source type and method: LED illumination, strobe illumination. Tone: Customizable special voice. Programmable logic controller parameters: Cortex-A9 1GHz RAM, 1GEMMC hard drive, 4G interface, 1 EtherCAT, 2 x 100M Ethernet, 1 USB 2.0 port (TXPEG), 1 HDMI port (32DI32D0). Display terminal parameters: 0.2 inches... The device features a 1.5-inch capacitive touchscreen supporting 2-point touch and simultaneous output of 1920*1080P@60Hz video. It supports network access to Uniview IPCs and NVRs, with a maximum of 500 channels (IPC / NVRs). It also supports ONVIF protocol access to third-party IPCs and NVRs, with one-click search and one-click addition of IPCs and NVRs. It supports H.264 and H.265 decoding formats, with a maximum decoding capability of 1 channel of 4K, 4 channels of 1080P, and 8 channels of 720P. The infrared grating has a parameter of 6 beams at 108cm. The algorithm server parameter is SIMATIC. S7-1200, CPU 1214C, compact CPU, DC / DC / RLY, onboard I / O: 14 24V DC digital inputs; 10 24V DC digital outputs; 2AI 0-10V DC, power supply: DC 20.4-28.8V, 100KB program / data memory, SM1231 analog input module 8AI 13-bit resolution; 8 analog outputs, ±10V, ±5V, ±2.5V or 0-20mA range.

[0094] In this application, a control display terminal deployed in the train driver's cab provides real-time video previews of key areas on the factory track and offers an early warning function. If someone illegally intrudes, an alarm will be triggered to remind the driver, providing coverage of dangerous areas in the work scenario. Multiple pressure sensors are installed on the track. By triggering the sensors at frequencies during train operation, the calculation formula of this invention can be used to determine whether the train is speeding. Pressure sensors are deployed at specific locations on the track. The signals from the pressure sensors are used to sense the distance between the train and the dangerous area at the level crossing, triggering an audible and visual alarm to warn pedestrians and vehicles. The dangerous area at the level crossing is fully covered by surveillance video, and a human intrusion algorithm is deployed. When someone appears in the dangerous area on either side of the train while it is stopped, an audible and visual alarm is triggered to warn and stop the violation, driving the person away from the dangerous area. This enables real-time video, train speed detection, train approach warning, and level crossing intrusion alarms in the train dispatching operation scenario at the level crossing.

[0095] In an optional embodiment of this application, the method further includes:

[0096] The system associates train speed with train dispatching parameters at track level crossings, sets access permissions for the associated train speed and track level crossing parameters, and modifies the set access permissions in the event of a train speeding.

[0097] In an optional embodiment of this application, when a train exceeds its speed limit, the access permissions are changed, including:

[0098] Step 1: Collect the train number at the track level crossing as the train's fingerprint;

[0099] Step 2: Collect train dispatching parameters in advance, and analyze the sensitivity level of the train dispatching parameters using natural language processing technology to generate data sensitivity labels; the data sensitivity labels contain sensitivity level information;

[0100] Step 3: Construct a dynamic permission mapping model based on train fingerprints and data sensitivity tags; the dynamic permission mapping model includes the association between train fingerprints and data sensitivity tags;

[0101] Step 4: Generate a train behavior baseline based on historical train behavior monitoring using a self-learning detection model; the train behavior baseline includes the behavioral characteristics of the equipment under normal conditions;

[0102] Step 5: Based on the degree of deviation between the train behavior baseline and the real-time train behavior, trigger the permission change of the dynamic permission mapping model. Train behavior includes, but is not limited to, train speed.

[0103] The generation of data sensitivity labels includes the following steps:

[0104] Step 21: Perform natural language processing on the original text information of the train dispatching parameters to extract key features;

[0105] The extraction of key features includes the following steps:

[0106] Step 211: Perform word segmentation, part-of-speech tagging, and stop word filtering on the original text information;

[0107] Step 212: Extract entity information from the original text information using named entity recognition technology, wherein the entity information includes, but is not limited to, parameter values;

[0108] Step 213: Use TF-IDF or TextRank algorithms to extract high-frequency or high-weight words from the original text information as keywords;

[0109] Step 214: Perform topic analysis on the original text information using the LDA model to extract potential topic distributions;

[0110] Step 22: Based on the key features, generate data sensitivity labels using preset sensitivity classification rules.

[0111] The construction of the dynamic permission mapping model includes the following steps:

[0112] Step 31: Input the train fingerprint and the data sensitivity tag, and establish a mapping relationship between the train fingerprint and the data sensitivity tag through association rule mining technology to generate a preliminary permission mapping table;

[0113] Step 32: Based on the preliminary permission mapping table, classify the train fingerprints using cluster analysis technology to generate train classification results;

[0114] Step 33: Based on the train classification results and the data sensitivity labels, construct the dynamic permission mapping model that outputs permission decision rules using the decision tree algorithm;

[0115] The process of generating a train behavior baseline includes the following steps:

[0116] Step 41: Collect historical train behavior data and construct train behavior feature vectors;

[0117] Step 42: Based on the train behavior feature vector, construct a train behavior clustering model using an unsupervised learning algorithm;

[0118] Step 43: Based on the train behavior clustering model, construct a probability distribution model of train behavior using density estimation techniques;

[0119] Step 44: Based on the probability distribution model, set a threshold and generate a train behavior baseline;

[0120] The train behavior baseline is a set of rules and thresholds used to distinguish between normal and abnormal behavior.

[0121] The permission change that triggers the dynamic permission mapping model includes the following steps:

[0122] Step 51: Collect train behavior data in real time and calculate the deviation from the train behavior baseline;

[0123] The deviation calculation includes the following sub-steps:

[0124] Step 511: Real-time behavioral feature extraction.

[0125] Using the same feature engineering method as in step 41, features are extracted from the real-time acquired train behavior data to generate real-time behavior feature vectors. This ensures that the feature space is completely consistent with the feature space of the behavior baseline model.

[0126] Step 512: Calculate the behavioral probability score based on the probability distribution model;

[0127] Step 513: Calculate the deviation index based on the probability score;

[0128] Step 514: Obtain the final deviation by weighted summation;

[0129] Step 52: Based on the deviation threshold, generate a permission change trigger signal, and trigger the dynamic permission mapping model according to the permission change trigger signal;

[0130] Step 523: In each sliding window, generate a structured trigger signal based on the multi-level deviation threshold;

[0131] Step 53: When it is determined that the dynamic permission mapping model has been triggered, the collected train fingerprints and data sensitivity tags of the devices are input into the dynamic permission mapping model to obtain the permission decision tags output by the dynamic permission mapping model.

[0132] It should be understood that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order constraint on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the diagram may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0133] Please see Figure 6 One embodiment of this application provides a risk control device 600 for railway level crossings, comprising:

[0134] Monitoring module 610 is used to monitor in real time whether the signal data collected by pressure sensors pre-deployed on the track undergoes a step change. Pressure sensors are deployed at preset distances forward and backward along the track at the track level crossing.

[0135] The first transmitting module 620 is used to send a first alarm message when the signal data undergoes a step change, in order to prompt pedestrians and vehicles to avoid the area.

[0136] The determination module 630 is used to determine whether the corresponding train is speeding based on the signal data.

[0137] The second sending module 640 is used to send a second alarm message when the train is speeding, to prompt the train to slow down.

[0138] Specific limitations regarding the aforementioned device 600 can be found in the above description of the risk management method for railway level crossings, and will not be repeated here. Each module in the aforementioned device 600 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0139] In one embodiment, a computer device is provided, the internal structure of which can be as shown in the figure. Figure 7 As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method for risk management of level crossings. It includes: a memory and a processor; the memory stores a computer program; and the processor executes the computer program to implement any step of the above-described method for risk management of level crossings.

[0140] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, can perform any of the steps in the above-mentioned risk management method for level crossings.

[0141] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0142] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0143] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0144] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0145] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0146] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for risk management at railway level crossings, characterized in that, include: Real-time monitoring is conducted to check whether there are any abrupt changes in the signal data collected by pressure sensors pre-deployed on the track. Pressure sensors are deployed at predetermined distances forward and backward along the track at the track level crossing. In the event of a step change in signal data, a first alarm message is sent to alert pedestrians and vehicles to give way. Determine whether the corresponding train is speeding based on the signal data; If the train is speeding, a second alarm message will be sent to prompt the train to slow down.

2. The method according to claim 1, characterized in that, The step of determining whether a train is speeding based on signal data includes: The current train speed is determined based on the signal data using the following expression: v=d / Δt Where v is the current speed of the train, d is the distance between adjacent wheels of the train, and Δt is the time interval between adjacent pulse signals collected by the pressure sensor. The train's current speed is compared with a preset speed threshold. If the train's current speed exceeds the preset speed threshold, the train is deemed to be speeding.

3. The method according to claim 1, characterized in that, The method further includes: Real-time monitoring to ensure that the infrared gratings deployed on both sides of the track at the level crossing are completely blocked by the train; When the infrared grating is completely blocked by the train, the first camera device deployed at the track level crossing collects images of the pre-set danger zones on both sides of the train. The image captured by the first camera device is input into the pre-trained human body recognition model, and the output is the result of whether a human body is recognized. If the output shows that a human body has been detected, a third alarm message is sent to drive people away from the preset danger zone.

4. The method according to claim 3, characterized in that, The human body recognition model is trained through the following steps: The track level crossing is divided into zones, and dangerous areas that need to be monitored are selected from the divided zones. The scope and boundaries of human detection in the dangerous areas are then determined. Collect monitored images and annotate them to obtain the annotation results of human bodies in the images; The monitored images are used as input to the model, and the annotation results of the human bodies in the images are used as output to train the human body recognition model, thus obtaining a trained human body recognition model.

5. The method according to claim 3, characterized in that, The inference speed of the human body recognition model is improved by optimizing floating-point operations, wherein the floating-point operations are determined by the following expression: Where FLOPs represents floating-point operations, L represents the number of layers in the human recognition model, and C represents the number of layers in the model. in C represents the number of channels in the input feature map. out K is the number of channels in the output feature map, H is the convolution kernel, and H and W are the height and width of the output feature map, respectively.

6. The method according to claim 3, characterized in that, The loss function of the human body recognition model is expressed as follows: L box =1-IoU+ad 2 +βv Among them, L box Here, represents the loss function value of the human body recognition model, IoU represents the overlap between the target human body and the predicted bounding box, and d represents the loss function value of the human body recognition model. 2 denoted as the center offset between the target human body and the prediction box, v is the aspect ratio deviation between the target human body and the prediction box, and α and β are the weights of the center offset and aspect ratio deviation, respectively.

7. The method according to claim 1, characterized in that, The method further includes: Second cameras are deployed at track switch locations and in various hazardous areas along the track within the factory area. The images captured by the second cameras are displayed in real time on wireless display terminals deployed on the train. These cameras are used to display the status of the fishtail markers at the track switch locations and images of various hazardous areas along the track, providing the train driver with a complete field of vision.

8. A risk control device for track level crossings, characterized in that, include: The monitoring module is used to monitor in real time whether there is a step change in the signal data collected by the pressure sensors pre-deployed on the track. The pressure sensors are deployed at preset distances forward and backward along the track at the track level crossing. The first transmitting module is used to send a first alarm message when the signal data undergoes a step change, in order to prompt pedestrians and vehicles to give way. The determination module is used to determine whether the corresponding train is speeding based on the signal data. The second sending module is used to send a second alarm message when the train is speeding, to prompt the train to slow down.

9. A computer device, comprising: A memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the risk management method for track level crossings according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the risk management method for track level crossings as described in any one of claims 1 to 7.

Images