Train mileage data correction method and device

Abstract

The invention discloses a train mileage data correction method and device. The method comprises the following steps: acquiring an inspection image acquired by a linear array camera in a train inspection process; inputting the inspection image into a target equipment identification model, and obtaining target equipment information and actual mileage data of the train from the inspection image through the target equipment identification model; taking the target equipment information and the actual mileage data of the train as retrieval conditions, and retrieving from a pre-established ledger information database to obtain ledger standard mileage data of the train; carrying out difference processing on the actual mileage data of the train and the standing book standard mileage data to obtain deviation data; and correcting the actual mileage data of the train according to the deviation data to obtain corrected train mileage data. According to the invention, the accuracy of train mileage data can be improved, and accurate positioning of line equipment is realized.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method and device for correcting train mileage data. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention recited in the claims. No admission is made that the description herein is prior art by inclusion in this section.

[0003] The train mileage data is one of the key data for railway transportation management and inspection, which directly affects the positioning of line equipment, the calibration of inspection results, and the accurate implementation of maintenance plans. However, since trains have been running in a complex railway environment for a long time, their mileage data is easily interfered by various factors, which will lead to deviations in mileage information. The collection of line equipment information is closely related to train mileage data. The deviation of train mileage data will directly affect the accuracy of equipment information inspection results, resulting in inaccurate positioning of line equipment. Summary of the invention

[0004] An embodiment of the present invention provides a method for correcting train mileage data, which is used to correct the train mileage data, improve the accuracy of the train mileage data, and achieve accurate positioning of line equipment. The method includes:

[0005] Obtain inspection images captured by the linear array camera during train inspection;

[0006] The inspection image is input into the target device recognition model, and the target device information and the actual mileage data of the train are obtained from the inspection image through the target device recognition model;

[0007] Using the target equipment information and the actual mileage data of the train as search conditions, the standard mileage data of the train is retrieved from the pre-established ledger information database;

[0008] The actual mileage data of the train is subtracted from the standard mileage data in the ledger to obtain the deviation data;

[0009] Correcting the actual mileage data of the train according to the deviation data to obtain corrected train mileage data;

[0010] Among them, the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

[0011] The embodiment of the present invention further provides a train mileage data correction device, which is used to correct the train mileage data, improve the accuracy of the train mileage data, and achieve accurate positioning of line equipment. The device includes:

[0012] An inspection image acquisition module is used to acquire inspection images collected by a linear array camera during train inspection;

[0013] The actual mileage data acquisition module is used to input the inspection image into the target device recognition model, and obtain the target device information and the actual mileage data of the train from the inspection image through the target device recognition model;

[0014] The ledger standard mileage data acquisition module is used to retrieve the train's ledger standard mileage data from a pre-established ledger information database using the target equipment information and the train's actual mileage data as search conditions;

[0015] Deviation data calculation module, used to perform difference processing between the actual mileage data of the train and the standard mileage data of the ledger to obtain deviation data;

[0016] The train mileage data correction module is used to correct the actual mileage data of the train according to the deviation data to obtain the corrected train mileage data;

[0017] Among them, the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

[0018] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for correcting train mileage data when executing the computer program.

[0019] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for correcting the train mileage data is implemented.

[0020] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method for correcting the train mileage data is implemented.

[0021] In an embodiment of the present invention, an inspection image captured by a linear array camera during a train inspection is acquired; the inspection image is input into a target device recognition model, and the target device information and the actual mileage data of the train are obtained from the inspection image through the target device recognition model; the target device information and the actual mileage data of the train are used as retrieval conditions, and the ledger standard mileage data of the train is retrieved from a pre-established ledger information database; the actual mileage data of the train is subtracted from the ledger standard mileage data to obtain deviation data; the actual mileage data of the train is corrected according to the deviation data to obtain the corrected train mileage data; wherein the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8. In the above process, the embodiment of the present invention uses the target device recognition model constructed by the YOLOv8 network structure to identify the inspection image, obtain the actual mileage data of the train, compare the actual mileage data of the train with the ledger standard mileage data recorded in the ledger information database, and calculate the deviation data between the actual mileage data and the ledger standard mileage data, so as to correct the mileage data of the train based on the deviation data, thereby improving the accuracy of the train mileage data and realizing the precise positioning of the line equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0023] Figure 1 Flow chart of a method for correcting train mileage data in an embodiment of the present invention;

[0024] Figure 2 A flowchart of establishing a target device identification model in an embodiment of the present invention;

[0025] Figure 3 This is a flowchart of obtaining standard mileage data of the ledger in an embodiment of the present invention;

[0026] Figure 4 Schematic diagram of a device for correcting train mileage data in an embodiment of the present invention;

[0027] Figure 5 Schematic diagram of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0029] Figure 1 Flow chart of a method for correcting train mileage data in an embodiment of the present invention, the method comprising:

[0030] Step 101, obtaining an inspection image captured by a linear array camera during a train inspection process;

[0031] Step 102, input the inspection image into the target device recognition model, and obtain the target device information and the actual mileage data of the train from the inspection image through the target device recognition model;

[0032] Step 103, using the target device information and the actual mileage data of the train as search conditions, retrieve the log standard mileage data of the train from the pre-established log information database;

[0033] Step 104, performing a difference process between the actual mileage data of the train and the standard mileage data of the ledger to obtain deviation data;

[0034] Step 105, correcting the actual mileage data of the train according to the deviation data to obtain corrected train mileage data;

[0035] Among them, the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

[0036] Each step is described in detail below.

[0037] In step 101, an inspection image captured by a linear array camera during a train inspection is obtained.

[0038] In a specific embodiment, a high-resolution industrial camera is installed on the inspection train to collect continuous image data along the track. To ensure the quality of the collected data, sampling is required under different lighting conditions, weather environments and running speeds, covering all types of line equipment.

[0039] In step 102, the inspection image is input into the target device recognition model, and the target device information and the actual mileage data of the train are obtained from the inspection image through the target device recognition model. The target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

[0040] In a specific embodiment, the backbone feature extraction network uses the MobileNet v3 network. MobileNet v3 is an efficient lightweight network that uses deep separable convolution and Squeeze-and-Excitation (SE) modules, and is optimized by neural architecture search (NAS). Among them, the deep separable convolution significantly reduces the number of parameters and calculations, which is suitable for embedded devices and real-time scenarios; the Squeeze-and-Excitation (SE) module enhances the feature relationship between channels and improves the feature expression capability. The hard Swish activation function has both nonlinear expression capability and computational efficiency; while maintaining high feature extraction capabilities, the network greatly reduces computational complexity, so that the network achieves a good balance between reasoning speed and accuracy, which is suitable for fast positioning scenarios.

[0041] Figure 2 This is a flow chart of establishing a target device identification model in an embodiment of the present invention. In one embodiment, the target device identification model is established based on the optimized YOLOv8 network structure, and further includes:

[0042] Step 201, input the inspection image into the backbone feature extraction network, perform multiple sampling on the input inspection image through the backbone feature extraction network, and output multiple feature maps of different scales; the YOLOv8 network structure includes a backbone feature extraction network, a feature pyramid network, and a detection head network;

[0043] Step 202, embed the ASFF network into the feature pyramid network of YOLOv8, perform adaptive feature fusion on feature maps of multiple scales through the ASFF network, and obtain a target device recognition model based on YOLOv8 multi-scale fusion.

[0044] In a specific embodiment, Adaptive Spatial Feature Fusion (ASFF) is used to replace the feature fusion module of the original YOLOv8. It improves the multi-scale target detection capability by dynamically adjusting the feature fusion weights of different levels. It has the following characteristics: Dynamic feature fusion dynamically allocates the importance of features at each layer through an adaptive weight allocation mechanism to improve feature fusion efficiency. Multi-scale information enhancement efficiently captures key information in small targets, occluded targets and complex backgrounds. Lightweight design reduces redundant calculations in the feature fusion process and maintains high efficiency. Context-aware capabilities enhance the modeling of global information, enabling the network to better understand the semantic relationship between targets. The introduction of ASFF significantly enhances the detection performance of small targets and multi-scale targets, and is suitable for complex scenarios of facility positioning.

[0045] In one embodiment, the MPDIoU loss function is used to replace the original loss function of YOLOv8, including:

[0046] Calculate the loss value of YOLOv8 based on the MPDIoU loss function:

[0047]

[0048] Among them, ρ 2 (b,b gt ) represents the square of the Euclidean distance between the center point of the predicted box and the true box; c 2 represents the square of the diagonal of the bounding box; Δw and Δh represent the width difference and height difference between the predicted box and the real box respectively; ΔA represents the area difference between the predicted box and the real box; w gt ,h gt , A gt Represents the width, height, and area of ​​the real box; α, β, and γ represent weight coefficients.

[0049] In a specific embodiment, the optimized YOLOv8 network structure combines efficient lightweight design and enhanced multi-scale feature extraction and fusion capabilities. At the same time, the MPDIoU (Modified Proportional Distance IoU) loss function is used to improve the traditional IoU loss function, thereby improving the detection performance of the network model.

[0050] In step 103, the target device information and the actual mileage data of the train are used as search conditions to retrieve the standard mileage data of the train from the pre-established ledger information database.

[0051] In a specific embodiment, during the process of collecting inspection images, the mileage information of each frame of the image is recorded to ensure the synchronization between the camera acquisition frequency and the train running speed. Each frame of the image is accompanied by an acquisition timestamp and a frame number to form a time-serialized image data stream. The train mileage information is obtained from the train integrated system in real time, and the correspondence between the mileage and the timestamp is recorded to form a time-serialized mileage data stream. A unified high-precision clock is used to synchronize the linear array camera and the train integrated system to ensure that the inspection image and mileage data share a consistent time reference. By associating the inspection image acquisition frame sequence with the train movement timeline, a preliminary correspondence between the image data and the train running position can be achieved, forming a time-position mapping relationship, laying the foundation for subsequent mileage correction.

[0052] According to the timestamp of the inspection image, find the data point with the closest timestamp in the mileage data stream and pair the two.

[0053] ΔT=|T i，mage -T mile |

[0054] Among them, T image is the inspection image acquisition timestamp, T mile is the timestamp of the actual mileage data of the train. The point with the smallest ΔT is selected as the matching target.

[0055] If there is a gap between the inspection image timestamp and the actual mileage data timestamp, the mileage value corresponding to the inspection image frame is estimated by linear interpolation. A mapping relationship is established between the serial number and timestamp of each inspection image frame and the actual mileage data value to establish an account information database.

[0056] Figure 3 The flowchart of obtaining the standard mileage data of the ledger in an embodiment of the present invention is as follows. In one embodiment, the standard mileage data of the ledger of the train is retrieved from the pre-established ledger information database using the target device information and the actual mileage data of the train as the retrieval conditions, including:

[0057] Step 301, obtaining the train running speed and the acquisition frequency of the linear array camera;

[0058] Step 302, determining a search radius starting from the actual mileage data of the train according to the running speed of the train and the acquisition frequency of the linear array camera;

[0059] Step 303, using the search radius and target equipment information as search conditions, retrieve the train's standard mileage data from a pre-established ledger information database.

[0060] In a specific embodiment, due to the time difference or offset between the acquisition of the inspection image and the recording of the actual mileage data of the train, the positioning of the target device needs to be searched within a certain mileage range. Suppose the actual mileage data of the train corresponding to the current frame of the inspection image is S raw , then the search scope can be defined as [S raw -Δr, S raw +Δr], where Δr is the search radius, which is determined comprehensively based on the train running speed and the camera acquisition frequency. Within this range, based on the target equipment information, the ledger standard mileage data with the smallest absolute value of mileage difference is selected from the ledger information database.

[0061] In step 104, the actual mileage data of the train is subtracted from the standard mileage data in the ledger to obtain deviation data.

[0062] In a specific embodiment, the deviation data is calculated according to the following formula:

[0063] ΔS=S raw -S standard

[0064] Among them, S raw is the actual mileage data of the train, ΔS is the deviation data, S standard This is the standard mileage data in the ledger.

[0065] In step 105, the actual mileage data of the train is corrected according to the deviation data to obtain corrected train mileage data.

[0066] In a specific embodiment, the actual mileage data of the train is corrected according to the calculated deviation data, and the train mileage information of the current frame inspection image is updated. The corrected train mileage data can be expressed as:

[0067] S corrected =S rnw -ΔS

[0068] Among them, S correcte d is the corrected train mileage data, S raw is the actual mileage data of the train, and ΔS is the deviation data.

[0069] The present invention also provides a train mileage data correction device, as described in the following embodiments. Since the principle of the device to solve the problem is similar to the train mileage data correction method, the implementation of the device can refer to the implementation of the train mileage data correction method, and the repeated parts will not be repeated.

[0070] Figure 4 Schematic diagram of a train mileage data correction device in an embodiment of the present invention, the device comprises:

[0071] The inspection image acquisition module 401 is used to acquire the inspection images collected by the linear array camera during the train inspection process;

[0072] The actual mileage data acquisition module 402 is used to input the inspection image into the target device recognition model, and obtain the target device information and the actual mileage data of the train from the inspection image through the target device recognition model;

[0073] The ledger standard mileage data acquisition module 403 is used to retrieve the ledger standard mileage data of the train from the pre-established ledger information database using the target equipment information and the actual mileage data of the train as the retrieval conditions;

[0074] The deviation data calculation module 404 is used to perform a difference process between the actual mileage data of the train and the standard mileage data of the ledger to obtain deviation data;

[0075] The train mileage data correction module 405 is used to correct the actual mileage data of the train according to the deviation data to obtain the corrected train mileage data;

[0076] Among them, the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

[0077] In one embodiment, a target device identification model building module is further included, which is specifically used to:

[0078] The inspection image is input into the backbone feature extraction network, and the input inspection image is sampled multiple times through the backbone feature extraction network to output multiple feature maps of different scales; the YOLOv8 network structure includes the backbone feature extraction network, the feature pyramid network and the detection head network;

[0079] The ASFF network is embedded into the feature pyramid network of YOLOv8, and the feature maps of multiple scales are adaptively fused through the ASFF network to obtain a target device recognition model based on YOLOv8 multi-scale fusion.

[0080] In one embodiment, a YOLOv8 network structure optimization module is also included, which is specifically used to:

[0081] Calculate the loss value of YOLOv8 based on the MPDIoU loss function:

[0082]

[0083] Among them, ρ 2 (b,b gt) represents the square of the Euclidean distance between the center point of the predicted box and the true box; c 2 represents the square of the diagonal of the bounding box; Δw and Δh represent the width difference and height difference between the predicted box and the real box respectively; ΔA represents the area difference between the predicted box and the real box; w gt ,h gt , A gt Represents the width, height, and area of ​​the real box; α, β, and γ represent weight coefficients.

[0084] In one embodiment, the ledger standard mileage data acquisition module 403 is specifically used to:

[0085] Obtain the train running speed and the acquisition frequency of the linear array camera;

[0086] According to the running speed of the train and the acquisition frequency of the linear array camera, the search radius starting from the actual mileage data of the train is determined;

[0087] Using the search radius and target equipment information as search conditions, the train's standard mileage data is retrieved from a pre-established ledger information database.

[0088] An embodiment of the present invention further provides a computer device, Figure 5 It is a schematic diagram of a computer device in an embodiment of the present invention. The computer device 500 includes a memory 510, a processor 520, and a computer program 530 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 530, the above-mentioned train mileage data correction method is implemented.

[0089] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for correcting the train mileage data is implemented.

[0090] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method for correcting the train mileage data is implemented.

[0091] In an embodiment of the present invention, an inspection image captured by a linear array camera during a train inspection is acquired; the inspection image is input into a target device recognition model, and the target device information and the actual mileage data of the train are obtained from the inspection image through the target device recognition model; the target device information and the actual mileage data of the train are used as retrieval conditions, and the ledger standard mileage data of the train is retrieved from a pre-established ledger information database; the actual mileage data of the train is subtracted from the ledger standard mileage data to obtain deviation data; the actual mileage data of the train is corrected according to the deviation data to obtain the corrected train mileage data; wherein the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8. In the above process, the embodiment of the present invention uses the target device recognition model constructed by the YOLOv8 network structure to identify the inspection image, obtain the actual mileage data of the train, compare the actual mileage data of the train with the ledger standard mileage data recorded in the ledger information database, and calculate the deviation data between the actual mileage data and the ledger standard mileage data, so as to correct the mileage data of the train based on the deviation data, thereby improving the accuracy of the train mileage data and realizing the precise positioning of the line equipment.

[0092] Those skilled in the art will appreciate 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. Moreover, 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.

[0093] 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.

[0094] 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.

[0095] 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.

[0096] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for correcting train mileage data, characterized in that: include: Obtain inspection images captured by the linear array camera during train inspection; The inspection image is input into the target device recognition model, and the target device information and the actual mileage data of the train are obtained from the inspection image through the target device recognition model; Using the target equipment information and the actual mileage data of the train as search conditions, the standard mileage data of the train is retrieved from the pre-established ledger information database; The actual mileage data of the train is subtracted from the standard mileage data in the ledger to obtain the deviation data; Correcting the actual mileage data of the train according to the deviation data to obtain corrected train mileage data; Among them, the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

2. The method according to claim 1, characterized in that The MPDIoU loss function is used to replace the original loss function of YOLOv8, including: Calculate the loss value of YOLOv8 based on the MPDIoU loss function: Among them, ρ 2 (b,b gt ) represents the square of the Euclidean distance between the center point of the predicted box and the true box; c 2 represents the square of the diagonal of the bounding box; Δw and Δh represent the width difference and height difference between the predicted box and the real box respectively; ΔA represents the area difference between the predicted box and the real box; w gt ,h gt , A gt Represents the width, height, and area of ​​the real box; α, β, and γ represent weight coefficients.

3. The method according to claim 1, characterized in that Using the target equipment information and the actual mileage data of the train as the search conditions, the standard mileage data of the train is retrieved from the pre-established ledger information database, including: Obtain the train running speed and the acquisition frequency of the linear array camera; According to the running speed of the train and the acquisition frequency of the linear array camera, the search radius starting from the actual mileage data of the train is determined; Using the search radius and target equipment information as search conditions, the train's standard mileage data is retrieved from a pre-established ledger information database.

4. The method according to claim 1, characterized in that The target device recognition model is established based on the optimized YOLOv8 network structure, which also includes: The inspection image is input into the backbone feature extraction network, and the input inspection image is sampled multiple times through the backbone feature extraction network to output multiple feature maps of different scales; the YOLOv8 network structure includes the backbone feature extraction network, the feature pyramid network and the detection head network; The ASFF network is embedded into the feature pyramid network of YOLOv8, and the feature maps of multiple scales are adaptively fused through the ASFF network to obtain a target device recognition model based on YOLOv8 multi-scale fusion.

5. A train mileage data correction device, characterized in that: include: An inspection image acquisition module is used to acquire inspection images collected by a linear array camera during train inspection; The actual mileage data acquisition module is used to input the inspection image into the target device recognition model, and obtain the target device information and the actual mileage data of the train from the inspection image through the target device recognition model; The ledger standard mileage data acquisition module is used to retrieve the train's ledger standard mileage data from a pre-established ledger information database using the target equipment information and the train's actual mileage data as search conditions; Deviation data calculation module, used to perform difference processing between the actual mileage data of the train and the standard mileage data of the ledger to obtain deviation data; The train mileage data correction module is used to correct the actual mileage data of the train according to the deviation data to obtain the corrected train mileage data; Among them, the target device recognition model is established based on the optimized YOLOv8 network structure; in the process of optimizing the YOLOv8 network structure, the MobileNetv3 network is used as the backbone feature extraction network of YOLOv8, and the MPDIoU loss function is used to replace the original loss function of YOLOv8.

6. The device according to claim 5, characterized in that It also includes the YOLOv8 network structure optimization module, which is specifically used for: Calculate the loss value of YOLOv8 based on the MPDIoU loss function: Among them, ρ 2 (b,b gt ) represents the square of the Euclidean distance between the center point of the predicted box and the true box; c 2 represents the square of the diagonal of the bounding box; Δw and Δh represent the width difference and height difference between the predicted box and the real box respectively; ΔA represents the area difference between the predicted box and the real box; w gt ,h gt , A gt Represents the width, height, and area of ​​the real box; α, β, and γ represent weight coefficients.

7. The device according to claim 5, characterized in that The module for acquiring standard mileage data of the ledger is specifically used for: Obtain the train running speed and the acquisition frequency of the linear array camera; According to the running speed of the train and the acquisition frequency of the linear array camera, the search radius starting from the actual mileage data of the train is determined; Using the search radius and target equipment information as search conditions, the train's standard mileage data is retrieved from a pre-established ledger information database.

8. The device according to claim 5, characterized in that It also includes a target device identification model building module, which is specifically used to: The inspection image is input into the backbone feature extraction network, and the input inspection image is sampled multiple times through the backbone feature extraction network to output multiple feature maps of different scales; the YOLOv8 network structure includes the backbone feature extraction network, the feature pyramid network and the detection head network; The ASFF network is embedded into the feature pyramid network of YOLOv8, and the feature maps of multiple scales are adaptively fused through the ASFF network to obtain a target device recognition model based on YOLOv8 multi-scale fusion.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 4 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

11. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images