A method for automatic inspection and early warning of track intrusion objects
By setting up cameras and dynamically allocating image data in rail transit, combining weather and patrol mode optimization processing, efficient identification and early warning of track invasion limits is achieved, and the problems of low detection efficiency and poor accuracy in the existing technology are solved, ensuring that The safety of railway operation and the efficiency of system resource utilization.
Patent Information
- Application Number
- CN202411146105.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-20
AI Technical Summary
The discovery and treatment of track invasions in rail transit in the prior art have problems such as long manual inspection cycle, difficulty in real-time monitoring, and uneven allocation of computing resources, resulting in low detection efficiency and poor accuracy, especially in bad weather conditions, image quality declines. Impact detection effect.
Design an automatic inspection and early warning method for track invasion and object limits. By setting cameras on both sides of the track, image preprocessing is performed in combination with weather types, image data is dynamically allocated to the site server, and the inspection scope and order are adjusted according to the inspection mode. , and use convolutional neural network to identify and warning invasion limits to ensure efficient utilization of system resources.
It improves image quality and detection accuracy, dynamically adapts to different work needs, avoids server overload, ensures that the inspection process is systematic and comprehensive, responds to potential risks in a timely manner, and ensures the safety of railway operation.
Smart Images

Figure CN118675279B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of automatic inspection of track restricted objects, and in particular to an automatic inspection and early warning method for track restricted objects. Background Art
[0002] Urban rail transit plays an increasingly important role in my country's public transportation industry with its advantages of large capacity, high speed, punctuality and reliability, strong safety, energy saving and environmental protection. However, with the rapid development of urban rail transit, the main reasons affecting the safety of urban rail transit operations are: track factors, vehicle factors, power supply factors, signal system factors and human factors. Among the track factors, obstacles in the three-dimensional driving space on the track can easily cause destructive collisions with trains, and even cause serious accidents such as derailment, causing major disasters and immeasurable losses. In order to ensure the driving safety of the track area, it is necessary to conduct effective inspections in advance and promptly discover abnormal conditions in the track area (track intrusions, etc.), and take practical measures to ensure the driving safety of rail trains.
[0003] At present, the discovery and treatment of train intrusions in the field of rail transit are partly based on manual inspections, which require the participation of a large number of personnel, have a long inspection cycle, and the inspection time is generally during the vehicle operation window at night, making it difficult to find problems in the first place. Another part is to use track rollers to inspect abnormalities in the track area. In addition, foreign body intrusion events are sudden and unpredictable, which can only solve part of the problem and cannot monitor foreign body intrusion in real time. Another part is to install fixed cameras near the track for automatic inspection. Each camera can only monitor a certain range, and a large number of cameras need to be arranged along the track. Although this type of inspection can save a lot of manpower, it has poor adaptability to the inspection environment. The performance of the camera will be affected under severe weather conditions (such as heavy rain, heavy snow, haze, etc.) and large changes in light, resulting in a decrease in image quality and affecting the detection effect. At the same time, the detection of intrusive objects requires real-time processing and analysis of image data, so a large amount of computing resources is needed to ensure the timeliness and accuracy of inspections. In actual operation, due to different inspection modes, different server deployments at rail stations and station intervals, and not all servers are dedicated to inspections, existing servers are generally fixedly allocated, resulting in uneven distribution of computing resources, which reduces efficiency when large-scale computing is required.
[0004] The purpose of the present invention is to design an automatic inspection and early warning method for track intrusions in view of the above-mentioned problems in the prior art. Summary of the invention
[0005] In view of this, an object of the present invention is to propose an automatic inspection and early warning method for track intrusion objects, which can solve the above-mentioned problems.
[0006] The present invention provides a method for automatic inspection and early warning of track restricted objects, based on an automatic inspection and early warning system for track restricted objects, the system comprising:
[0007] A plurality of cameras are respectively arranged on both sides of the track, the cameras are facing the travel direction of both sides of the track, and are used to acquire and pre-process image data in the travel direction of both sides of the track;
[0008] a plurality of station servers, communicatively connected to the camera, for analyzing image data in the direction of track travel;
[0009] A central server, communicatively connected to the site servers, for allocating computing resources of the site servers;
[0010] The method comprises:
[0011] Acquire the first image data captured by the camera and the corresponding weather type, perform image quality preprocessing on the first image data corresponding to the current weather type, and obtain second image data;
[0012] Obtain the current inspection mode, and obtain the inspection range and inspection order of the corresponding camera according to the current inspection mode;
[0013] Dynamically distribute the second image data within the inspection range to the site server according to the type of the camera and the real-time load of the site server;
[0014] The second image data is sequentially subjected to image recognition of restricted objects according to the inspection sequence of the camera, and the information of the identified restricted objects is sent to the central server for early warning.
[0015] Further, the acquiring of the first image data and the corresponding weather type, performing image quality preprocessing according to the first image data corresponding to the current weather type, and obtaining the second image data comprises:
[0016] Performing corresponding weather detection on the first image data to obtain the weather type existing in the first image data;
[0017] A corresponding weather type algorithm is selected to preprocess the first image data to obtain second image data.
[0018] Further, the performing corresponding weather detection on the first image data to obtain the weather type existing in the first image data includes:
[0019] Collect orbital images under different weather conditions, annotate the weather classification of orbital images, and construct orbital weather image training samples;
[0020] Constructing a convolutional neural network model, and using orbital weather image samples to train the convolutional neural network model to obtain a weather classification model;
[0021] The first image data is input into the weather classification model to obtain the weather type.
[0022] Further, the obtaining of the current inspection mode includes:
[0023] Determine whether the current time is skylight time, if yes, enter station inspection mode, otherwise determine whether the current time is train departure time;
[0024] If the current time is the train departure time, it enters the section following mode, otherwise it enters the line inspection mode.
[0025] Further, obtaining the inspection range and inspection order of the corresponding camera according to the current inspection mode includes:
[0026] If the inspection mode is a line inspection mode, the inspection range is the entire line where the train runs, and the inspection sequence is a cycle of the train travel sequence of the entire line where the train runs;
[0027] If the inspection mode is a station inspection mode, the inspection range is each station, and the inspection sequence is a cycle of the train travel sequence within the range of each station;
[0028] If the inspection mode is the section following inspection mode, the inspection range is the station section ahead of the train, and the inspection order is the forward order of the station section ahead of the train.
[0029] Furthermore, the camera comprises:
[0030] A station camera is used to obtain image data in the travel direction on both sides of the track at the preprocessing station;
[0031] The interval camera is used to acquire and pre-process image data in the direction of track travel between stations.
[0032] Further, the dynamically allocating the second image data within the inspection range to the site server according to the type of the camera and the real-time load of the site server includes:
[0033] Determine the real-time load of the site server, and if the real-time load of the site server does not exceed the load threshold, allocate the second image data acquired by the site camera to the corresponding site server, otherwise classify the second image data acquired by the site camera and the second image data acquired by the interval camera as second image data to be allocated;
[0034] Calculate the priority of the site server according to the connection time and the real-time load of the site server, and generate a priority list;
[0035] Dynamically poll the second image data to be allocated in the priority list of the site server.
[0036] Further, calculating the priority of the site server according to the connection time and the real-time load of the site server to generate a priority list includes:
[0037] Obtaining the CPU occupancy rate and memory usage rate of the site server, and calculating the load weight of the site server;
[0038] The connection time between the camera and the site server is obtained, and the site servers are prioritized in combination with the load weight of the site server to generate a priority list.
[0039] Further, the step of sequentially performing image recognition of restricted objects on the second image data according to the inspection order of the camera and sending the recognized restricted object information to the central server for early warning includes:
[0040] Using a pre-trained intrusion object image recognition model, the second image data is sequentially subjected to intrusion object image recognition to obtain an intrusion object image and its category;
[0041] The camera position corresponding to the image of the restricted object and its category is reported to the central server for early warning and secondary image recognition of the restricted object.
[0042] Further, the identified restricted object information is sent to the central server for restricted object warning and then the following is performed:
[0043] Confirm that the restricted object information corresponds to the station corresponding to the camera, and switch the inspection mode of the station to the station inspection mode;
[0044] If the intrusion warning is eliminated, the corresponding inspection mode will be entered according to the current time.
[0045] Beneficial effects of the present invention:
[0046] First, preprocess the image data according to the current weather type, and perform defogging and raindrop removal on foggy and rainy days. These preprocessing steps can significantly improve the clarity and quality of the image. Image preprocessing can improve image quality, reduce image noise and blur at the source, and reduce the possibility of false detection and missed detection, thereby improving the accuracy of subsequent intrusion detection.
[0047] Second, by obtaining the current inspection mode, the system can adjust the inspection scope and sequence according to different work requirements. This flexibility ensures that the system can dynamically adapt to different work requirements and improve the efficiency of inspection. Dynamically adjusting the inspection scope according to the inspection mode can flexibly respond to various inspection needs, making the system more adaptable.
[0048] Third, by classifying and processing according to camera types, the best processing method can be used for different types of cameras to improve processing efficiency. Image data is dynamically allocated according to camera types and the real-time load of site servers to avoid overloading of individual servers, ensure that each server can efficiently process the allocated data, and maximize the use of system resources.
[0049] Fourth, the second image data is sequentially identified for intrusions in the order of the camera inspections to ensure that the inspection process is systematic and comprehensive and to reduce omissions. After identifying intrusions, the information is immediately sent to the central server for early warning, which can respond to potential risks in a timely manner and ensure the safety of railway operations. The information on intrusions is centrally reported to the central server for unified management and secondary identification, which improves the centralization of data management and the overall coordination of the system. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0051] Figure 1 It is a system module diagram of the first embodiment.
[0052] Figure 2 It is a flow chart of the method of embodiment 2.
[0053] Figure 3 It is a schematic diagram of the line inspection mode of the second embodiment.
[0054] Figure 4 It is a schematic diagram of the site inspection mode of the second embodiment.
[0055] Figure 5 It is a schematic diagram of the interval following inspection mode of the second embodiment. DETAILED DESCRIPTION
[0056] To facilitate understanding by those skilled in the art, the structure of the present invention is now further described in detail with reference to the embodiments and the accompanying drawings. It should be understood that the steps mentioned in the present embodiment, unless otherwise specified in their order, can be adjusted in order according to actual needs, and can even be executed simultaneously or partially simultaneously.
[0057] Embodiment 1
[0058] like Figure 1 As shown, an embodiment of the present invention provides a track intrusion automatic inspection and warning system, comprising:
[0059] A plurality of cameras are respectively arranged on both sides of the track, the cameras are facing the travel direction of both sides of the track, and are used to acquire and pre-process image data in the travel direction of both sides of the track;
[0060] The camera comprises:
[0061] A station camera is used to obtain image data in the travel direction on both sides of the track at the preprocessing station;
[0062] The interval camera is used to acquire and pre-process image data in the direction of track travel between stations.
[0063] a plurality of station servers, communicatively connected to the camera, for analyzing image data in the direction of track travel;
[0064] A central server, communicatively connected to the site servers, for allocating computing resources of the site servers;
[0065] In this embodiment, the basic working principle of the system is to set up N bidirectional cameras in the track area for on-site monitoring, which can comprehensively monitor the track in real time. The camera can pre-process the acquired image data and quickly transmit it to the site server for analysis of intrusions. The computing resources of several site servers are allocated through the central server to avoid overloading of a single server and improve the efficiency and stability of the system. The distributed architecture composed of several cameras and servers improves the redundancy and reliability of the system. Even if a camera or server fails, it will not cause the entire system to fail. The camera with edge computing function has a pre-processing function, which can preliminarily filter and process image interference data, reduce the transmission amount of irrelevant data, and improve the overall processing efficiency of the system.
[0066] Embodiment 2
[0067] like Figure 2 As shown, an embodiment of the present invention provides a method for automatic inspection and early warning of track intrusion objects, comprising:
[0068] S1 obtains first image data captured by the camera and the corresponding weather type, performs image quality preprocessing on the first image data corresponding to the current weather type, and obtains second image data;
[0069] S101 performs corresponding weather detection on the first image data to obtain the weather type existing in the first image data;
[0070] S1011 collects orbital images under different weather conditions, annotates the weather classification of orbital images, and constructs orbital weather image training samples;
[0071] S1012 constructs a convolutional neural network model, and uses the orbital weather image samples to train the convolutional neural network model to obtain a weather classification model;
[0072] S1013: Input the first image data into the weather classification model to obtain the weather type.
[0073] S102 selects a corresponding weather type algorithm to pre-process the first image data to obtain second image data.
[0074] In this step, the current weather type is obtained by directly analyzing and identifying the image, and the image is identified through a pre-trained big data model to ensure the accuracy of the identification. Then, the appropriate algorithm is selected according to the corresponding weather type, for example: defogging algorithm: use a defogging algorithm (such as a dark channel prior algorithm, a variational defogging method, etc.) to enhance the image contrast and clarity; deraining algorithm: use a deraining algorithm (such as a deraining network based on deep learning) to remove raindrop interference in the image; according to the specific weather type that has been detected, select the best preprocessing algorithm, which can significantly improve the image quality. This targeted preprocessing can better remove noise, enhance contrast, adjust brightness, etc., making the image clearer and the details more obvious.
[0075] S2 obtains the current inspection mode, and obtains the inspection range and inspection order of the corresponding camera according to the current inspection mode;
[0076] S201 determines whether the current time is skylight time, if yes, enters station inspection mode, otherwise determines whether the current time is train departure time;
[0077] S202 If the current time is the train departure time, enter the section following mode, otherwise enter the line inspection mode.
[0078] In this step, the track inspection window period (or track maintenance "window time") refers to a time window specially set in the operation of railways or rail transit for necessary maintenance and repair work. During this time period, the track line partially or completely suspends train operations so that maintenance personnel can safely inspect and repair on the track. Therefore, the entire line can be inspected during the window time, and station inspections or train follow-up inspections can be carried out outside the window time according to the train departure requirements.
[0079] Specifically, obtaining the inspection range and inspection order of the corresponding camera according to the current inspection mode includes:
[0080] like Figure 3 As shown, if the inspection mode is a line inspection mode, the inspection range is the entire line where the train runs, and the inspection sequence is a cycle of the train travel sequence of the entire line where the train runs;
[0081] In this step, this mode is to divide all the trackside cameras of the entire line into two inspection units according to the running direction. Each inspection unit will patrol and capture analysis in turn according to the set inspection cycle. Each camera will capture the trackside image at a fixed time (set by the inspection cycle) and upload it to the site server for video analysis. By dividing all the trackside cameras into two inspection units according to the direction, the full coverage of the entire line is ensured to avoid monitoring blind spots.
[0082] like Figure 4 As shown, if the inspection mode is a station inspection mode, the inspection range is each station, and the inspection sequence is a cycle of the train travel sequence within the range of each station;
[0083] In this step, this mode is based on the station as a unit, and all cameras around the station are divided into two inspection units according to the running direction. Each inspection unit performs patrol capture analysis in turn according to the set inspection cycle. In this mode, each station can perform automatic periodic inspections at the same time, which can shorten the inspection cycle, improve inspection efficiency, and facilitate the rapid discovery of intrusions. Each inspection unit performs patrol capture analysis according to the set cycle, which can significantly shorten the overall inspection cycle. The cameras at each station work simultaneously, with clear division of labor, to ensure that the station and surrounding areas can be inspected more frequently.
[0084] like Figure 5 As shown, if the inspection mode is the section following inspection mode, the inspection range is the station section ahead of the train, and the inspection order is the forward order of the station section ahead of the train.
[0085] In this step, this mode is an operation mode that inspects the section line that the train is about to enter in units of station sections. The section following inspection mode is a one-time inspection (non-periodic). Only when there is a train entering the station will it trigger the line inspection of the section between stations ahead. The duration of a single inspection in the section following inspection mode is generally less than 30 seconds, and the inspection can be completed without significantly affecting train scheduling. The inspection is only triggered when the train is about to enter a specific section, and it can specifically check potential problems in the current direction of the train's travel. Inspection resources are concentrated on key moments of train operation to ensure that resources can be used most effectively and reduce unnecessary waste.
[0086] Further, if Figure 5As shown, for skipped stops, when the skipped stop information is received from the central server, the section following inspection range is automatically adjusted to the station section in front of the stop point, which can reduce the number of system operations to extend the system life and improve availability.
[0087] S3 dynamically distributes the second image data within the inspection range to the site server according to the type of the camera and the real-time load of the site server;
[0088] S301 determines the real-time load of the site server. If the real-time load of the site server does not exceed the load threshold, the second image data acquired by the site camera is allocated to the corresponding site server; otherwise, the second image data acquired by the site camera and the second image data acquired by the interval camera are classified as second image data to be allocated;
[0089] S302 calculates the priority of the site server according to the connection time and the real-time load of the site server, and generates a priority list;
[0090] S3021 obtains the CPU occupancy rate and memory usage rate of the site server, and calculates the load weight of the site server;
[0091] S3022 obtains the connection time between the camera and the site server, and prioritizes the site servers in combination with the load weight of the site server to generate a priority list.
[0092] S303 dynamically polls the priority list of the site server for the second image data to be divided.
[0093] In this step, each site server is physically set up in each site to facilitate maintenance by staff. In the actual working process, in addition to calculating and analyzing the image data of the camera, the site server also needs to perform other tasks, such as: station train information broadcasting, section signal control, etc. Therefore, the computing resources of the site server are not fixed. In order to ensure the efficiency of inspection image calculation and analysis, it is necessary to dynamically allocate according to the actual load of the site server, while avoiding the problem of the site server and the camera being too far away. The priority list and dynamic polling mechanism ensure the uniform distribution of data load, optimize the use of server resources, reduce the overload of a single server, and improve the operating efficiency of the overall system. At the same time, it can effectively avoid the failure of a site server, and can be promptly allocated to a server with a lower load and a closer distance to ensure the efficiency of the inspection.
[0094] S4 performs image recognition of restricted objects on the second image data in sequence according to the inspection order of the camera, and sends the recognized restricted object information to the central server for early warning.
[0095] S401, using a pre-trained intrusion object image recognition model, sequentially performs intrusion object image recognition on the second image data to obtain an intrusion object image and its category;
[0096] S402 reports the image of the restricted object and the camera position corresponding to its category to the central server for early warning and secondary image recognition of the restricted object.
[0097] After the identified restricted object information is sent to the central server for an intrusion warning, the following is performed:
[0098] Confirm that the restricted object information corresponds to the station corresponding to the camera, and switch the inspection mode of the station to the station inspection mode;
[0099] If the intrusion warning is eliminated, the corresponding inspection mode will be entered according to the current time.
[0100] In this step, the image data is identified in sequence according to the inspection order of the camera to ensure that no camera data is missed. The pre-trained intrusion image recognition model can quickly and accurately identify intrusions such as ordinary paper boxes, hanging objects, people, falling stones, displacement of trackside equipment, falling of trackside equipment, and items left behind in the second image data. The central server performs secondary intrusion image recognition to further verify and confirm the intrusion information, thereby improving the accuracy and reliability of the warning. When the central server confirms that there is an intrusion warning, the staff enters the track to remove the intrusion. During the intrusion processing, in the inspection mode, the intrusion warning information generated by the entry of personnel into the track can be marked as a processing operation, and no warning will be issued. Once an intrusion is detected, the system will automatically switch to the station inspection mode until the intrusion warning is eliminated, strengthen key monitoring of problem areas, concentrate resources to handle emergencies, and ensure the safety of the station area.
[0101] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented 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.
[0102] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0103] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0104] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0105] It should be noted that in the claims, any reference signs placed between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claim. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.
[0106] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0107] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
[0108] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0109] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.
Claims
1. A method for automatic inspection and early warning of track intrusion objects, characterized in that: Based on an automatic inspection and early warning system for track intrusion objects, the system includes: A plurality of cameras are respectively arranged on both sides of the track, the cameras are facing the travel direction of both sides of the track, and are used to acquire and pre-process image data in the travel direction of both sides of the track; Furthermore, the camera comprises: A station camera is used to obtain image data in the travel direction on both sides of the track at the preprocessing station; The interval camera is used to acquire and pre-process image data in the direction of track travel between stations; a plurality of station servers, communicatively connected to the camera, for analyzing image data in the direction of track travel; A central server, communicatively connected to the site servers, for allocating computing resources of the site servers; The method comprises the following steps: Acquire the first image data captured by the camera and the corresponding weather type, perform image quality preprocessing on the first image data corresponding to the current weather type, and obtain second image data; Get the current inspection mode, and obtain the inspection range and inspection order of the corresponding camera according to the current inspection mode, specifically: Determine whether the current time is skylight time, if yes, enter station inspection mode, otherwise determine whether the current time is train departure time; If the current time is the train departure time, it will enter the section following mode, otherwise it will enter the line inspection mode; If the inspection mode is a line inspection mode, the inspection range is the entire line where the train runs, and the inspection sequence is a cycle of the train travel sequence of the entire line where the train runs; If the inspection mode is a station inspection mode, the inspection range is each station, and the inspection sequence is a cycle of the train travel sequence within the range of each station; If the inspection mode is the section following inspection mode, the inspection range is the section of the station ahead of the train, and the inspection order is the forward order of the section of the station ahead of the train; The second image data within the inspection range is dynamically distributed to the site server according to the type of the camera and the real-time load of the site server, specifically: Determine the real-time load of the site server, and if the real-time load of the site server does not exceed the load threshold, allocate the second image data acquired by the site camera to the corresponding site server, otherwise classify the second image data acquired by the site camera and the second image data acquired by the interval camera as second image data to be allocated; The priority of the site server is calculated according to the connection time and the real-time load of the site server, and a priority list is generated. Specifically: Obtaining the CPU occupancy rate and memory usage rate of the site server, and calculating the load weight of the site server; Obtaining the connection time between the camera and the site server, and prioritizing the site servers in combination with the load weight of the site server to generate a priority list; Dynamically polling the second image data to be allocated in the priority list of the site server; The second image data is sequentially subjected to image recognition of restricted objects according to the inspection sequence of the camera, and the information of the identified restricted objects is sent to the central server for early warning.
2. The method for automatic inspection and early warning of track intrusion according to claim 1 is characterized in that: The obtaining of the first image data and the corresponding weather type, and performing image quality preprocessing according to the first image data corresponding to the current weather type to obtain the second image data comprises: Performing corresponding weather detection on the first image data to obtain the weather type existing in the first image data; A corresponding weather type algorithm is selected to preprocess the first image data to obtain second image data.
3. The method for automatic inspection and early warning of track intrusion according to claim 2 is characterized in that: The performing corresponding weather detection on the first image data to obtain the weather type existing in the first image data includes: Collect orbital images under different weather conditions, annotate the weather classification of orbital images, and construct orbital weather image training samples; Constructing a convolutional neural network model, and using orbital weather image samples to train the convolutional neural network model to obtain a weather classification model; The first image data is input into the weather classification model to obtain the weather type.
4. The method for automatic inspection and early warning of track intrusion according to claim 1 is characterized in that: The step of sequentially performing image recognition of restricted objects on the second image data according to the inspection order of the camera, and sending the recognized restricted object information to the central server for early warning includes: Using a pre-trained intrusion object image recognition model, the second image data is sequentially subjected to intrusion object image recognition to obtain an intrusion object image and its category; The camera position corresponding to the image of the restricted object and its category is reported to the central server for early warning and secondary image recognition of the restricted object.
5. The method for automatic inspection and early warning of track intrusion according to claim 1 is characterized in that: After the identified restricted object information is sent to the central server for an intrusion warning, the following is performed: Confirm that the restricted object information corresponds to the station corresponding to the camera, and switch the inspection mode of the station to the station inspection mode; If the intrusion warning is eliminated, the corresponding inspection mode will be entered according to the current time.
