Intelligent freight yard train passing protection control system

Abstract

The invention relates to the technical field of freight yard train passage protection, and discloses an intelligent freight yard train passage protection control system, which comprises a monitoring unit, a control unit and a control unit, the control unit also comprises the following steps: constructing a level crossing judgment model based on the historical data of the railway level crossing; judging whether the train can enter the station or not by combining the obtained first type of data, the obtained second type of data and a flat aisle judgment model; if the train can enter the station, a priority order of the freight yard train is generated; generating a scheduling strategy of the flat aisle area by combining the priority ranking of the freight yard trains, and generating a scheduling instruction based on the scheduling strategy; and generating a second early warning instruction in combination with the obtained first type of data and the second type of data. The method has many innovation points in the aspects of safety guarantee, operation efficiency improvement, data utilization, intelligent decision making and the like, and the safety and efficiency of freight yard train passing are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of freight yard train passage protection, and in particular to an intelligent freight yard train passage protection control system. Background Art

[0002] The level aisle area is a special area in railway transportation, which involves the intersection of trains, pedestrians, vehicles, etc. In order to ensure the safety of the level aisle area, it is necessary to adopt a variety of technical means.

[0003] On the one hand, a gate system is required to control the passage of trains and pedestrians. Gate control needs to be integrated with the train's operating status, which requires linkage between the gate control system and the train operation control system. On the other hand, video surveillance technology is used to monitor the level aisle area in real time to detect abnormal situations (such as pedestrian intrusion) and issue warnings. This ensures both train scheduling and safety. Summary of the Invention

[0004] The purpose of the present invention is to make a judgment by comparing the first arrival time T1 of the level crossing target and the second arrival time T2 of the train, where T1 and T2 are respectively obtained based on the target analysis of historical video data and the historical status data of the train. The train priority ranking comprehensively considers factors such as train failure, delay time, and expected passage time to generate a priority evaluation value D0. When generating the dispatch instruction, whether the train enters the station or not is considered. When entering the station, the train’s planned passing number is determined based on the priority ranking, pedestrian waiting time, etc. If the train does not enter the station, it is notified to wait for pedestrians to pass. The system has many innovations in terms of safety assurance, operational efficiency improvement, data utilization and intelligent decision-making, which effectively improves the safety and efficiency of train passage in the freight yard.

[0005] In order to achieve the above objectives, the present invention provides an intelligent freight yard train traffic protection control system, comprising: The monitoring unit obtains and classifies the railway level crossing data, generates first-category data based on the level crossing area data, and generates second-category data based on the track train data near the level crossing; A control unit generates a dispatch instruction for the flat aisle area based on the acquired first-category data and second-category data; The control unit further includes: The first processing module: builds a level crossing judgment model based on historical data of railway level crossings; The second processing module: combines the first type of data, the second type of data and the level aisle judgment model to determine whether the train can enter the station; If the train can enter the station, generate the priority ranking of the freight yard trains; Generate a dispatching strategy for the level aisle area based on the priority ranking of freight yard trains, and generate dispatching instructions based on the dispatching strategy; The third processing module: generates a second warning instruction based on the acquired first type of data and second type of data.

[0006] In some embodiments of the present invention, the monitoring unit is further configured to: A video surveillance device is set up based on the flat corridor area, and a monitoring mode of the video surveillance device is set. The monitoring mode includes: a daytime monitoring mode and a nighttime monitoring mode. Video data under the corresponding monitoring mode is obtained to generate the first type of data; Multiple monitoring devices are arranged on the track near the level aisle, and the arrangement spacing of each monitoring device is determined based on historical data. The status data of the track train is generated by combining the data of all monitoring devices, and the second type of data is generated based on the status data of the track train.

[0007] In some embodiments of the present invention, the first processing module is further configured to: Obtain historical video data of the flat aisle area, perform target detection based on the historical video data, generate multiple types of targets, and perform behavior prediction on each type of target to generate the corresponding target behavior prediction results; Obtain the first arrival time T1 based on the behavior prediction results of all targets; generating a second arrival time T2 of the rail train based on historical status data of the rail train; Generate a monitoring information history value PL by combining the first arrival time T1 and the second arrival time T2; PL=T1-T2; If PL>0, it is judged that the train can enter the station; If PL≤0, it is determined that the train cannot enter the station; If it is determined that the train can enter the station, a first warning instruction is generated and the gate of the flat aisle area is closed.

[0008] In some embodiments of the present invention, when obtaining the first arrival time T1 based on the behavior prediction results of all targets, the method further includes: Build target recognition models based on historical video data; Combined with the target recognition model to obtain the target type in the historical video; Behavior prediction is performed for each target type to obtain the movement direction and speed change of the corresponding target type; The time when the current target type arrives at the level crossing gate is predicted based on the moving direction and speed change of the current target type, and the first arrival time T1 is generated based on the time when the current target type arrives at the gate of the level crossing area.

[0009] In some embodiments of the present invention, when generating the second arrival time T2 of the rail train based on the status data of the rail train, the method further includes: Number all trains on the track based on the historical status data of the track trains; Obtain the travel time M2 of all trains passing through the level aisle area; And generate a second arrival time T2 based on the travel time of all trains passing through the level aisle area; T2=M2*k1; Among them, k1 is the proportional coefficient, which is determined based on actual usage.

[0010] In some embodiments of the present invention, generating a priority ranking of freight yard trains further includes: generating a plurality of judgment indicators based on the acquired operation data of the current numbered train; The process includes: generating a first evaluation value D1 based on whether the current numbered train has a fault; Obtain the scheduled time of all numbered trains, calculate the delay time of each numbered train in sequence based on the current time and the scheduled time, and obtain the delay time of all numbered trains to generate a second evaluation value D2; Generate a third evaluation value D3 based on the estimated train passing time of the current numbered train; Combining the first evaluation value, the second evaluation value, and the third evaluation value to generate a priority evaluation value D0 of the current numbered train; Obtain the priority evaluation value D0 of the numbered train to generate the priority ranking of the freight yard trains.

[0011] In some embodiments of the present invention, the step of generating the priority evaluation value D0 of the current numbered train further includes: If the current numbered train has a fault, obtain the fault type of the current numbered train; If the fault type of the current numbered train belongs to the first type of fault, D0=1; If the fault type of the current numbered train belongs to the second type of fault, then D1=1, D0=D1*(w1*D2+w2*D3); Among them, the first type of fault type is the fault type that affects the train deceleration; the second type of fault type is the fault type that does not affect the train deceleration; w1 is the weight of the second evaluation value D2, and w2 is the weight of the third evaluation value D3.

[0012] In some embodiments of the present invention, the generating of the third evaluation value D3 further includes: The first passing time of the currently numbered train is generated by combining the obtained running speed of the currently numbered train and the distance to the level aisle area; generating a second passing time by combining the running speed of the current numbered train and the length of the current numbered train; Adding the first passing time and the second passing time of the current numbered train to generate the passing time of the current numbered train; Obtaining the passing time of all numbered trains and defining a passing time gradient based on the passing time range of all numbered trains; A third evaluation value D3 is set for each time gradient; The third evaluation value D3 is negatively correlated with the train's passing time; A third evaluation value D3 of the current numbered train is generated based on the passing time gradient.

[0013] In some embodiments of the present invention, the generating of the scheduling instruction further includes: If it is determined that the train can enter the station, the gates in the level aisle area will be closed; Determine the numbering order of passing freight yard trains based on the priority ranking of freight yard trains; The passing time of freight yard trains is accumulated sequentially based on the train number sequence of the freight yard; Preset pedestrian waiting time and generate the planned passing times of freight yard trains by comparing the passing time of freight yard trains and pedestrian waiting time; At the same time, based on the planned passing trains of the freight yard, trains are dispatched to pass through the level crossing area in sequence. After all the planned passing trains have been detected to have passed through the level crossing area, the gates are opened and the first alarm instruction is cancelled. If it is determined that the train cannot enter the station, the gate in the level aisle area will be opened, and all freight trains will be notified to wait for pedestrians to pass.

[0014] In some embodiments of the present invention, the third processing module is further configured to: Performing target detection on the gate area in the monitoring video to generate target detection results for the gate area; Obtaining the target type of the area within the gate based on the target detection result of the area within the gate; Determine whether there is a preset target type in the area within the gate based on the target type of the area within the gate; If it exists, a second warning instruction is generated; It includes: generating alarm information to notify management personnel and the train currently passing, and playing alarm voice in the level crossing area.

[0015] Compared with the prior art, the intelligent freight yard train traffic protection control system provided by the embodiment of the present invention has the following advantages: By combining the target behavior prediction of the flat aisle area and the arrival time of the rail train to determine whether the train can enter the station, the dynamic situation of people and other targets in the flat aisle area as well as the operating status of the train are fully considered. This can effectively avoid the risk of collision between the train and people or vehicles in the flat aisle area, and improve the safety of train passage in the freight yard.

[0016] When generating the priority ranking of freight yard trains, whether the train has a fault is taken into consideration, and different priority assessment methods are given to different types of faults. This helps to give priority to trains with higher fault risks during the scheduling process and avoid safety accidents caused by train faults in the flat aisle area.

[0017] The third processing module performs target detection on the area inside the gate. When a preset target type is detected, it generates a second warning instruction, notifies management personnel and trains, and plays an alarm voice. This helps to promptly detect abnormal situations in the flat aisle area and further ensure the safety of train traffic.

[0018] Prioritize trains in the freight yard based on multiple criteria (such as train failure status, delay time, and estimated train transit time). By numbering the aisle areas, trains can be routed through the aisles in a reasonable order, reducing train waiting times and improving overall freight yard operational efficiency.

[0019] When generating dispatch instructions, pedestrian waiting times are preset, and planned train passes are generated by comparing train transit times with pedestrian waiting times. This approach considers both the needs of trains and the rights of pedestrians. While ensuring safety, it rationally arranges the passage order of trains and pedestrians, avoids unnecessary waiting time, and improves the overall operational efficiency of the freight yard.

[0020] Target recognition models are built to identify target types in historical video and predict their behavior. These models help the system more intelligently analyze the situation in the level aisle area, providing strong technical support for operations such as train entry determination, priority sorting, and dispatch instruction generation, thereby improving the system's intelligence and decision-making efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a structural diagram of an intelligent freight yard train traffic protection control system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following embodiments are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0023] In the description of the present invention, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0024] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0025] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0026] Example 1: The embodiment of the present invention provides an intelligent freight yard train traffic protection control system, such as Figure 1 As shown, including: The monitoring unit obtains and classifies the railway level crossing data, generates first-category data based on the level crossing area data, and generates second-category data based on the track train data near the level crossing; A control unit generates a dispatch instruction for the flat aisle area based on the acquired first-category data and second-category data; The control unit further includes: The first processing module: builds a level crossing judgment model based on historical data of railway level crossings; The second processing module: combines the first type of data, the second type of data and the level aisle judgment model to determine whether the train can enter the station; If the train can enter the station, generate the priority ranking of the freight yard trains; Generate a dispatching strategy for the level aisle area based on the priority ranking of freight yard trains, and generate dispatching instructions based on the dispatching strategy; The third processing module: generates a second warning instruction based on the acquired first type of data and second type of data.

[0027] Embodiment 2: The monitoring unit is further used for: A video surveillance device is set up based on the flat corridor area, and a monitoring mode of the video surveillance device is set. The monitoring mode includes: a daytime monitoring mode and a nighttime monitoring mode. Video data under the corresponding monitoring mode is obtained to generate the first type of data; Multiple monitoring devices are arranged on the track near the level aisle, and the arrangement spacing of each monitoring device is determined based on historical data. The status data of the track train is generated by combining the data of all monitoring devices, and the second type of data is generated based on the status data of the track train.

[0028] In this embodiment, the aisle area is monitored with video around the clock to record the on-site situation. A high-resolution camera (such as 4K resolution) is installed to cover the aisle and surrounding areas. The camera is equipped with infrared night vision function to ensure the night monitoring effect. AI visual algorithms are introduced to perform real-time analysis of the monitoring images: Pedestrian detection: Identify the number and location of pedestrians through deep learning models. Vehicle detection: Distinguish different types of vehicles (cars, motorcycles, etc.) and count the number. Abnormal behavior identification: such as pedestrians breaking in or vehicles illegally parked. The monitoring data is stored in a local NAS device or cloud server. The recording supports playback and query by date and time period. Set up a data encryption mechanism to protect privacy information.

[0029] Several monitoring devices are arranged on the track near the level aisle, including: Detect when a freight train approaches a level aisle and trigger the protection mechanism in time.

[0030] Electromagnetic sensors or acoustic wave detectors are installed on both sides of the track. Electromagnetic sensors determine the train's position by detecting changes in the track's magnetic field. Acoustic wave detectors sense train movement through the principle of ultrasonic wave reflection.

[0031] The sensor spacing is set to one group every 50 meters to ensure coverage of the entire flat aisle.

[0032] Signal transmission: Industrial-grade wireless communication modules (such as LoRa or NB-IoT) transmit train location information to the intelligent control system in real time. Signal transmission latency is kept to less than 10 milliseconds to ensure high real-time performance. The system triggers the protection mechanism when the train is at least 30 seconds from the level aisle. The protection activation time is dynamically adjusted by calculating the train's speed and position.

[0033] Embodiment 3: The first processing module is further configured to: Obtain historical video data of the flat aisle area, perform target detection based on the historical video data, generate multiple types of targets, and perform behavior prediction on each type of target to generate the corresponding target behavior prediction results; Obtain the first arrival time T1 based on the behavior prediction results of all targets; generating a second arrival time T2 of the rail train based on historical status data of the rail train; Generate a monitoring information history value PL by combining the first arrival time T1 and the second arrival time T2; PL=T1-T2; If PL>0, it is judged that the train can enter the station; If PL≤0, it is determined that the train cannot enter the station; If it is determined that the train can enter the station, a first warning instruction is generated and the gate of the flat aisle area is closed.

[0034] Embodiment 4: When the first arrival time T1 is obtained based on the behavior prediction results of all targets, the method further includes: Build target recognition models based on historical video data; Combined with the target recognition model to obtain the target type in the historical video; Behavior prediction is performed for each target type to obtain the movement direction and speed change of the corresponding target type; The time when the current target type arrives at the level crossing gate is predicted based on the moving direction and speed change of the current target type, and the first arrival time T1 is generated based on the time when the current target type arrives at the gate of the level crossing area.

[0035] In this embodiment, building a target recognition model based on historical video data includes: counting information such as the flow, direction, vehicle type, and speed of vehicles on the road to understand the traffic load of the road, and providing a basis for determining the time periods and sections for freight trains to have priority passage. For example, during peak traffic periods, the priority right of freight trains can be appropriately reduced to ensure the basic traffic capacity of the road.

[0036] Understand information such as pedestrian travel patterns, traffic distribution, peak hours, etc., especially at railway crossings, stations and other areas, so as to take targeted safety measures, such as setting up reasonable pedestrian waiting areas and adjusting traffic light durations.

[0037] First, preprocess the historical video data. This includes standardizing the video format, adjusting the frame rate, and enhancing the image. For example, converting videos from different formats (such as MP4 and AVI) to a unified format, adjusting the frame rate to a fixed value (such as 25 frames per second), and enhancing the image's contrast and brightness through methods such as histogram equalization for subsequent processing.

[0038] Extract key frames from historical video data. Due to the large amount of video data, directly processing all frames increases the computational burden. Using specific algorithms (such as those based on time intervals or scene change detection), representative key frames are extracted. These key frames can reflect the state of the flat aisle area at different times.

[0039] For the extracted keyframes, various feature extraction methods are used. For example, the convolutional layer of a convolutional neural network (CNN) is used to extract local image features, such as edges and textures. Traditional feature extraction methods can also be combined, such as the scale-invariant feature transform (SIFT) algorithm to extract feature points and the histogram of oriented gradients (HOG) algorithm to extract the shape features of the target.

[0040] Fusion of features extracted by different methods. For example, deep features extracted by CNN are combined with SIFT and HOG features to form a comprehensive feature vector that can more comprehensively describe the target in the flat aisle area.

[0041] Select an appropriate machine learning or deep learning algorithm to build a target recognition model. For example, you can use traditional machine learning algorithms such as support vector machines (SVM) and random forests (RF), or deep neural network algorithms such as Faster R-CNN and YOLO to build a target recognition model.

[0042] The annotated historical video data (labeled with object types in the level aisle area, such as pedestrians and vehicles) is divided into a training set, a validation set, and a test set. The model is trained using the training set. During training, the model parameters are adjusted based on a predefined loss function (such as the cross-entropy loss function) to ensure accurate object identification. The validation set is also used to evaluate the model's performance. Based on the evaluation results, the model's hyperparameters (such as the learning rate and number of network layers) are adjusted to prevent overfitting. Finally, the trained model is tested on the test set to ensure good generalization.

[0043] Input historical video data (which can be pre-processed video or keyframes) into the trained object recognition model. The model scans the input video or image and identifies possible target areas based on previously learned feature patterns.

[0044] For each identified target area, the model matches its features with predefined target type features to determine the approximate type of the target. For example, if the target area's features are similar to predefined pedestrian features, the target is considered a pedestrian; if they are similar to vehicle features, the target is considered a vehicle, and so on.

[0045] After initially determining the target type as pedestrian or vehicle, further refine the classification. For pedestrians, further subdivision can be performed based on clothing characteristics (e.g., work clothes, casual clothes), and items carried (e.g., tools, packages). For vehicles, subdivision can be performed based on vehicle type (e.g., truck, forklift), color, and other information. This helps to more accurately understand the specific situation of targets in the level aisle area.

[0046] For each identified target type, trajectory tracking is performed within the historical video data. The target's motion trajectory is determined by observing changes in its position between frames. For example, an object detection algorithm is used to locate the target's center position in consecutive keyframes. Connecting these locations forms the target's motion trajectory.

[0047] During the trajectory tracking process, algorithms such as Kalman filtering are used to optimize the target position, reduce the impact of noise and error on the trajectory, and make the trajectory smoother and more accurate.

[0048] Analyze the target's movement direction based on its trajectory. Calculate the trajectory's slope or direction vector to determine whether the target is moving toward or away from the road crossing gate, or laterally within the roadway area.

[0049] Calculate the target's velocity. Calculate the target's displacement over different time intervals and divide it by the time interval to obtain the velocity. Observe how the velocity changes over time to determine whether the target is accelerating, decelerating, or moving at a constant speed. For example, if the target's displacement increases gradually between two consecutive frames, the target is considered to be accelerating.

[0050] First, determine the distance between the current target type and the level crossing gate. If the level crossing area has a pre-set coordinate system, the Euclidean distance between the two can be calculated based on the coordinates of the target's current location and the coordinates of the level crossing gate.

[0051] If there is no coordinate system, the actual distance between the target and the gate can be obtained by converting the pixel distance between the target and the gate in the image into the known actual size of the flat aisle area.

[0052] According to the current moving direction and speed change of the target, the time of arrival at the flat road crossing gate is predicted, and the predicted time of the current target type arriving at the gate of the flat road crossing area is determined as the first arrival time T1.

[0053] Embodiment 5: When the second arrival time T2 of the rail train is generated based on the state data of the rail train, the method further includes: Number all trains on the track based on the historical status data of the track trains; Obtain the travel time M2 of all trains passing through the level aisle area; And generate a second arrival time T2 based on the travel time of all trains passing through the level aisle area; T2=M2*k1; Among them, k1 is the proportional coefficient, which is determined based on actual usage.

[0054] In this embodiment, data related to train identification and running track are first screened from the historical status data of the rail train. This data may include train number information, train departure and arrival stations, and train arrival and departure times at different stations.

[0055] Organize the filtered data and remove duplicate and invalid data items. For example, data from the same train number but with different record formats will be formatted uniformly to ensure data consistency and accuracy.

[0056] Develop a numbering scheme based on the track layout and the direction of train travel. For example, if the track is unidirectional, the tracks can be numbered in the order in which the trains pass on the track, starting from 1 and increasing in number. If the track is bidirectional, the tracks can be numbered based on the direction of the trains (e.g., odd numbers for upward travel and even numbers for downward travel).

[0057] Consider the type of train (e.g. freight train, passenger train, etc.) and distinguish them in the number. For example, you can add a specific identifier to the number, using different numbers or letters to represent different types of trains.

[0058] Number all trains on the track according to the established numbering rules. Associate the numbering information with other historical train status data to facilitate subsequent query and analysis. For example, add a "train number" field to the train history database and store the numbering results in this field.

[0059] Dedicated timing equipment can be installed in the level aisle area, or existing sensors on the track can be used to collect the time data of trains passing through the level aisle area. These sensors can be track sensors (such as axle counters), which can accurately detect the time when the front and rear of the train pass through the level aisle area.

[0060] The train travel time in historical data is extracted from the train operation records. These records may include the start time of the train entering the level aisle area and the end time of the train leaving the level aisle area. The train travel time can be obtained by subtracting the two.

[0061] The collected train travel time data is processed to remove outliers. Outliers may be caused by sensor failure, data recording errors, etc. For example, if a train's travel time is significantly shorter or longer than the normal range, it can be considered an outlier and corrected or removed.

[0062] The travel time of the same train through the level aisle area multiple times is averaged to obtain a more accurate train travel time M2. If the train operating conditions (such as load, speed, etc.) vary significantly, the travel time can be classified and averaged according to these conditions.

[0063] Determine the proportionality factor k1 based on actual usage. These factors include the length of the level aisle area, the track gradient, and the average train speed. For example, if the level aisle area is long and the track gradient is steep, the train's speed will vary significantly after passing through the level aisle area. A larger proportionality factor k1 may be required to accurately reflect the train's arrival time at the next station.

[0064] The value of k1 is determined through historical data analysis and field testing. Train operation data under different conditions can be collected to analyze the speed changes after the train passes through the level aisle area and the actual time it takes to reach the next station. The value of k1 can then be determined by establishing a mathematical model (such as a linear regression model).

[0065] Example 6: When generating the priority ranking of freight yard trains, the method further includes: generating a plurality of judgment indicators based on the acquired operation data of the current numbered train; The process includes: generating a first evaluation value D1 based on whether the current numbered train has a fault; Obtain the scheduled time of all numbered trains, calculate the delay time of each numbered train in sequence based on the current time and the scheduled time, and obtain the delay time of all numbered trains to generate a second evaluation value D2; Generate a third evaluation value D3 based on the estimated train passing time of the current numbered train; Combining the first evaluation value, the second evaluation value, and the third evaluation value to generate a priority evaluation value D0 of the current numbered train; Obtain the priority evaluation value D0 of the numbered train to generate the priority ranking of the freight yard trains.

[0066] In this embodiment, the train operation data includes information such as the freight train number, operation time, operation speed, route, train length, train weight, etc., so as to reasonably arrange the priority passage sequence and time of the trains and avoid conflicts and congestion between trains.

[0067] Data related to equipment status is extracted from the operating data of the current train, including operating parameters of key train components (e.g., engine, brake system, signaling system, etc.). Thresholds are set to determine whether the train is faulty. For example, if the engine oil temperature exceeds the upper limit of the normal operating range, or the brake system pressure falls below the specified value, the train is considered faulty.

[0068] Faults are classified into different severity levels. For example, faults that seriously affect the safe operation of trains (such as brake failure) are classified into one category, while faults that affect operational efficiency but do not endanger safety (such as air conditioning failure) are classified into another category.

[0069] The first evaluation value D1 is set based on the fault type. For a train with a serious fault, D1 = 1, indicating that the train needs to be handled first. For a train with a minor fault, D1 can be set to a smaller value (such as 0.5), indicating that its priority is relatively low.

[0070] Get the scheduled times for all numbered trains. This data can be obtained from the train dispatch plan. Also get the current time, which can be obtained through the clock system or synchronization with a time server.

[0071] Calculate the delay time for each numbered train. Delay time = current time - train's scheduled time. A positive result indicates a train delay; a negative result indicates a train early arrival.

[0072] The second evaluation value D2 is determined according to the length of the delay time.

[0073] Compare the priority evaluation values D0 of all numbered trains and arrange the trains in order of D0 value from largest to smallest (or from smallest to largest, depending on the actual priority definition).

[0074] Finally, the priority ranking of freight yard trains is obtained. This ranking result will be used for train scheduling decisions, such as giving priority to high-priority trains passing through the flat aisle area and other operations.

[0075] Embodiment 7: When generating the priority evaluation value D0 of the current numbered train, the method further includes: If the current numbered train has a fault, obtain the fault type of the current numbered train; If the fault type of the current numbered train belongs to the first type of fault, D0=1; If the fault type of the current numbered train belongs to the second type of fault, then D1=1, D0=D1*(w1*D2+w2*D3); Among them, the first type of fault type is the fault type that affects the train deceleration; the second type of fault type is the fault type that does not affect the train deceleration; w1 is the weight of the second evaluation value D2, and w2 is the weight of the third evaluation value D3.

[0076] Embodiment 8: When generating the third evaluation value D3, the method further includes: The first passing time of the currently numbered train is generated by combining the obtained running speed of the currently numbered train and the distance to the level aisle area; generating a second passing time by combining the running speed of the current numbered train and the length of the current numbered train; Adding the first passing time and the second passing time of the current numbered train to generate the passing time of the current numbered train; Obtaining the passing time of all numbered trains and defining a passing time gradient based on the passing time range of all numbered trains; A third evaluation value D3 is set for each time gradient; The third evaluation value D3 is negatively correlated with the train's passing time; A third evaluation value D3 of the current numbered train is generated based on the passing time gradient.

[0077] Embodiment 9: When generating the scheduling instruction, the method further includes: If it is determined that the train can enter the station, the gates in the level aisle area will be closed; Determine the numbering order of passing freight yard trains based on the priority ranking of freight yard trains; The passing time of freight yard trains is accumulated sequentially based on the train number sequence of the freight yard; Preset pedestrian waiting time and generate the planned passing times of freight yard trains by comparing the passing time of freight yard trains and pedestrian waiting time; At the same time, based on the planned passing trains of the freight yard, trains are dispatched to pass through the level crossing area in sequence. After all the planned passing trains have been detected to have passed through the level crossing area, the gates are opened and the first alarm instruction is cancelled. If it is determined that the train cannot enter the station, the gate in the level aisle area will be opened, and all freight trains will be notified to wait for pedestrians to pass.

[0078] Embodiment 10: The third processing module is further configured to: Performing target detection on the gate area in the monitoring video to generate target detection results for the gate area; Obtaining the target type of the area within the gate based on the target detection result of the area within the gate; Determine whether there is a preset target type in the area within the gate based on the target type of the area within the gate; If it exists, a second warning instruction is generated; It includes: generating alarm information to notify management personnel and the train currently passing, and playing alarm voice in the level crossing area.

[0079] In this embodiment, the gate physically isolates the level aisle area, preventing pedestrians and vehicles from entering. It utilizes a dual-motor drive design to ensure rapid raising and lowering. It is linked to the control unit, which sends a command to close the gate upon receiving an incoming train signal.

[0080] After the train has completely passed, the intelligent control system sends a command to open the gate.

[0081] Supports manual control mode and can be operated by remote control in emergency.

[0082] The third processing module performs object detection on the gated area in the monitoring video. In this process, it may use a pre-trained object detection algorithm, such as the Faster R-CNN or YOLO algorithm based on deep learning.

[0083] The surveillance video is processed frame by frame, and each frame is used as input to the target detection algorithm. The algorithm searches for possible targets in the image and determines the target's location and approximate range by identifying its features (such as shape, texture, and color).

[0084] The target detection results include the location information (such as coordinate range) of the detected target within the gated area, the target confidence level (indicating the probability that the detected target is a real target), and other information. Based on the target detection results within the gated area, the target type is further determined by matching the detected target features with a pre-established target type feature library.

[0085] The target type feature library contains typical features of various possible target types (such as pedestrians, vehicles, animals, etc.). For example, if the detected target has two legs and walks upright, it may be classified as a person; if it has four wheels and a rectangular shape, it may be classified as a vehicle.

[0086] Based on the feature matching results, the target type in the gate area is determined. For example, the target detection result shows that the detected target is classified as a pedestrian.

[0087] Some specific target types are pre-set as the types that need attention. For example, pedestrians are not allowed to appear in the gate area in some cases (when a train is about to pass), so pedestrians are one of the preset target types.

[0088] The acquired target type of the gate area is compared with the preset target type. If a matching target type exists, for example, a pedestrian is detected in the gate area and the pedestrian is a preset target type, then it is determined that the preset target type exists in the gate area.

[0089] If a preset target type is detected, a second warning instruction is generated. First, an alarm message is generated, which contains relevant information about the presence of an abnormal target type in the area within the gate, such as the target type (pedestrian) and the location (coordinate range determined by the target detection results).

[0090] Notify management personnel of the alarm information. The alarm information can be sent to the personnel responsible for the level crossing area through the message push function in the management system, SMS or internal communication system.

[0091] At the same time, the train currently passing through is notified. The alarm information is sent to the train driver through the train communication system (such as a wireless communication network) to inform the train that there is an abnormality in the area inside the gate in front of the train.

[0092] Play an alarm message at the crossing. A pre-recorded alarm message, such as "There is an unusual object inside the gate, please pay attention," is played through the speaker system installed at the crossing. The alarm message is intended to alert nearby personnel to safety and take appropriate measures.

[0093] Finally, it should be noted that it is apparent that those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, to the extent such modifications and variations fall within the scope of the present invention and its equivalents, the present invention is intended to include such modifications and variations.

[0094] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention.

Claims

1. An intelligent freight yard train traffic protection control system, characterized in that: include: The monitoring unit obtains and classifies the railway level crossing data, generates first-category data based on the level crossing area data, and generates second-category data based on the track train data near the level crossing; A control unit generates a dispatch instruction for the flat aisle area based on the acquired first-category data and second-category data; The control unit further includes: The first processing module: builds a level crossing judgment model based on historical data of railway level crossings; The second processing module: combines the first type of data, the second type of data and the level aisle judgment model to determine whether the train can enter the station; If the train can enter the station, generate the priority ranking of the freight yard trains; Generate a dispatching strategy for the level aisle area based on the priority ranking of freight yard trains, and generate dispatching instructions based on the dispatching strategy; The third processing module: generates a second warning instruction based on the acquired first type of data and second type of data.

2. The intelligent freight yard train traffic protection control system according to claim 1 is characterized in that: The monitoring unit is further configured to: A video surveillance device is set up based on the flat corridor area, and a monitoring mode of the video surveillance device is set. The monitoring mode includes: a daytime monitoring mode and a nighttime monitoring mode. Video data under the corresponding monitoring mode is obtained to generate the first type of data; Multiple monitoring devices are arranged on the track near the level aisle, and the arrangement spacing of each monitoring device is determined based on historical data. The status data of the track train is generated by combining the data of all monitoring devices, and the second type of data is generated based on the status data of the track train.

3. The intelligent freight yard train traffic protection control system according to claim 2 is characterized in that: The first processing module is further configured to: Obtain historical video data of the flat aisle area, perform target detection based on the historical video data, generate multiple types of targets, and perform behavior prediction on each type of target to generate the corresponding target behavior prediction results; Obtain the first arrival time T1 based on the behavior prediction results of all targets; generating a second arrival time T2 of the rail train based on historical status data of the rail train; Generate a monitoring information history value PL by combining the first arrival time T1 and the second arrival time T2; PL=T1-T2; If PL>0, it is judged that the train can enter the station; If PL≤0, it is determined that the train cannot enter the station; If it is determined that the train can enter the station, a first warning instruction is generated and the gate of the flat aisle area is closed.

4. The intelligent freight yard train traffic protection control system according to claim 3 is characterized in that: When the first arrival time T1 is obtained based on the behavior prediction results of all targets, the method further includes: Build target recognition models based on historical video data; Combined with the target recognition model to obtain the target type in the historical video; Behavior prediction is performed for each target type to obtain the movement direction and speed change of the corresponding target type; The time when the current target type arrives at the level crossing gate is predicted based on the moving direction and speed change of the current target type, and the first arrival time T1 is generated based on the time when the current target type arrives at the gate of the level crossing area.

5. The intelligent freight yard train traffic protection control system according to claim 4 is characterized in that: When the second arrival time T2 of the rail train is generated based on the state data of the rail train, the method further includes: Number all trains on the track based on the historical status data of the track trains; Obtain the travel time M2 of all trains passing through the level aisle area; And generate a second arrival time T2 based on the travel time of all trains passing through the level aisle area; T2=M2*k1; Among them, k1 is the proportional coefficient, which is determined based on actual usage.

6. The intelligent freight yard train traffic protection control system according to claim 5 is characterized in that: The step of generating the priority ranking of freight yard trains further includes: generating a plurality of judgment indicators based on the acquired operation data of the current numbered train; The process includes: generating a first evaluation value D1 based on whether the current numbered train has a fault; Obtain the scheduled time of all numbered trains, calculate the delay time of each numbered train in sequence based on the current time and the scheduled time, and obtain the delay time of all numbered trains to generate a second evaluation value D2; Generate a third evaluation value D3 based on the estimated train passing time of the current numbered train; Combining the first evaluation value, the second evaluation value, and the third evaluation value to generate a priority evaluation value D0 of the current numbered train; Obtain the priority evaluation value D0 of the numbered train to generate the priority ranking of the freight yard trains.

7. The intelligent freight yard train traffic protection control system according to claim 6 is characterized in that: When generating the priority evaluation value D0 of the current numbered train, the method further includes: If the current numbered train has a fault, obtain the fault type of the current numbered train; If the fault type of the current numbered train belongs to the first type of fault, D0=1; If the fault type of the current numbered train belongs to the second type of fault, then D1=1, D0=D1*(w1*D2+w2*D3); Among them, the first type of fault type is the fault type that affects the train deceleration; the second type of fault type is the fault type that does not affect the train deceleration; w1 is the weight of the second evaluation value D2, and w2 is the weight of the third evaluation value D3.

8. The intelligent freight yard train traffic protection control system according to claim 7 is characterized in that: The generating of the third evaluation value D3 further includes: The first passing time of the currently numbered train is generated by combining the obtained running speed of the currently numbered train and the distance to the level aisle area; generating a second passing time by combining the running speed of the current numbered train and the length of the current numbered train; Adding the first passing time and the second passing time of the current numbered train to generate the passing time of the current numbered train; Obtaining the passing time of all numbered trains and defining a passing time gradient based on the passing time range of all numbered trains; A third evaluation value D3 is set for each time gradient; The third evaluation value D3 is negatively correlated with the train's passing time; A third evaluation value D3 of the current numbered train is generated based on the passing time gradient.

9. The intelligent freight yard train traffic protection control system according to claim 8, characterized in that: The generation of the scheduling instruction further includes: If it is determined that the train can enter the station, the gates in the level aisle area will be closed; Determine the numbering order of passing freight yard trains based on the priority ranking of freight yard trains; The passing time of freight yard trains is accumulated sequentially based on the train number sequence of the freight yard; Preset pedestrian waiting time and generate the planned passing times of freight yard trains by comparing the passing time of freight yard trains and pedestrian waiting time; At the same time, based on the planned passing trains of the freight yard, trains are dispatched to pass through the level crossing area in sequence. After all the planned passing trains have been detected to have passed through the level crossing area, the gates are opened and the first alarm instruction is cancelled. If it is determined that the train cannot enter the station, the gate in the level aisle area will be opened, and all freight trains will be notified to wait for pedestrians to pass.

10. The intelligent freight yard train traffic protection control system according to claim 9, characterized in that: The third processing module is further configured to: Performing target detection on the gate area in the monitoring video to generate target detection results for the gate area; Obtaining the target type of the area within the gate based on the target detection result of the area within the gate; Determine whether there is a preset target type in the area within the gate based on the target type of the area within the gate; If it exists, a second warning instruction is generated; It includes: generating alarm information to notify management personnel and the train currently passing, and playing alarm voice in the level crossing area.

Images