A Multi-Perspective Vehicle Dynamic Weighing Method and Device Based on Vehicle-Road-Cloud-Road-End Collaboration

Abstract

This application provides a multi-view dynamic vehicle weighing method and apparatus based on vehicle-road-cloud-edge collaboration, relating to the field of cloud-edge collaborative computing. It includes: in response to a dynamic weighing operation targeting a target vehicle, acquiring license plate and width information through first-view monitoring of the target vehicle; acquiring vehicle type and target speed information through second-view monitoring of the target vehicle; acquiring multi-feature information through third-view monitoring of the target vehicle; constructing a multi-view graph structure corresponding to the target vehicle based on the license plate, width, vehicle type, target speed, and multi-feature information; acquiring strain signals generated when the target vehicle passes through a fiber optic grating sensor; and performing time-series analysis on the strain signals through the multi-view graph structure to calculate the actual weight of the target vehicle. This application solves the problem that existing static weighing or low-speed dynamic weighing methods are insufficient to meet the weighing requirements of vehicles passing at high speeds in highway scenarios.

Description

Technical Field

[0001] This application relates to the field of cloud-edge collaborative computing, and in particular to a multi-view dynamic vehicle weighing method and device based on vehicle-road-cloud-road-device collaboration. Background Technology

[0002] With the rapid development of intelligent transportation systems, vehicle-road-cloud integration is gradually becoming the core model for achieving efficient traffic management and information sharing. However, most toll station weighing systems currently operate independently, lacking data sharing and system linkage capabilities, which limits their integrated application in intelligent transportation systems and restricts the potential for overall efficiency improvement.

[0003] Traditional weighing systems typically rely on a single type of sensor for vehicle weight detection, making it difficult to maintain high-precision measurements in complex traffic environments. Because factors such as the number of axles, driving speed, and dynamic load distribution significantly affect sensor measurement results, existing static or low-speed dynamic weighing methods are insufficient to meet the weighing requirements of vehicles traveling at high speeds on highways.

[0004] Therefore, there is an urgent need for a multi-perspective vehicle dynamic weighing method and device based on vehicle-road-cloud-road-end collaboration. Summary of the Invention

[0005] This application provides a multi-view dynamic vehicle weighing method and device based on vehicle-road-cloud-road-device collaboration, which solves the problem that existing static weighing or low-speed dynamic weighing methods are difficult to meet the weighing needs of vehicles passing through quickly in highway scenarios.

[0006] The first aspect of this application provides a multi-view vehicle dynamic weighing method based on vehicle-road-cloud-road-device collaboration. The method includes: in response to a dynamic weighing operation on a target vehicle, performing multi-dimensional dynamic monitoring of the target vehicle, including first-view monitoring, second-view monitoring, and third-view monitoring; acquiring license plate information and width information of the target vehicle through first-view monitoring; acquiring vehicle model information and target speed information of the target vehicle through second-view monitoring; acquiring multi-feature information of the target vehicle through third-view monitoring, including contour information, position information, axle information, and wheelbase information; constructing a multi-view image structure corresponding to the target vehicle based on the license plate information, width information, vehicle model information, target speed information, and multi-feature information; acquiring strain signals generated when the target vehicle passes by using a fiber optic grating sensor; and performing time-series analysis on the strain signals using the multi-view image structure to calculate the actual weight of the target vehicle.

[0007] Optionally, the license plate information and width information of the target vehicle are obtained, specifically including: first-view monitoring through optical character recognition technology and a first-view model, wherein the first-view model is a convolutional neural network model; license plate information is obtained through optical character recognition technology; and width information is obtained through the first-view model.

[0008] Optionally, the vehicle model information and target speed information of the target vehicle are obtained, specifically including: second-view monitoring through dual radar speed detectors and a second-view model. The calculation formulas for the forward and backward propagation of the second-view model are as follows:

[0009] ;

[0010] in, For the second-person perspective model during forward propagation, the first The intermediate layer features corresponding to the convolutional structure For the second-person perspective model during forward propagation, the first The first convolution weights corresponding to the layer convolution structure, To perform convolution operations, Indicates the first The intermediate layer features corresponding to the convolutional structure are smoothed during computation. Indicates the fusion process. For the second-view model during backpropagation, the first The second convolution weights corresponding to the layer convolution structure For the first Each weight update function The gradient accumulated during the backpropagation of the second-view model is used to transmit the vehicle model information and the first vehicle speed information corresponding to the target vehicle through the second-view model, and the second vehicle speed information when the target vehicle passes by is obtained through dual radar speed detectors; the target vehicle speed information is calculated based on the first vehicle speed information and the second vehicle speed information.

[0011] Optionally, the target vehicle speed information is calculated based on the first vehicle speed information and the second vehicle speed information, specifically including: obtaining the first speed influencing factor and the second speed influencing factor, wherein the first speed influencing factor is the factor affecting the calculation of the second view model, and the second speed influencing factor is the factor affecting the calculation of the dual radar speed detector; and calculating the target vehicle speed information based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information.

[0012] Optionally, the target vehicle speed information is calculated based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information, specifically including: calculating the target vehicle speed information using the following formula:

[0013] ;

[0014] in, For target vehicle speed information, The number of factors influencing the first speed. The number of factors influencing the second speed. The first one calculated using the second-view model First vehicle speed information The first [speed sensor] obtained through dual radar speed detectors The second vehicle speed information, The first weight corresponds to the first vehicle speed information, and , The second weight corresponds to the second vehicle speed information, and .

[0015] Optionally, obtaining multi-feature information corresponding to the target vehicle specifically includes: capturing a vehicle image corresponding to the target vehicle using a preset camera; extracting features from the vehicle image to obtain a feature image corresponding to the target vehicle, the feature image including a first feature image, a second feature image, and a third feature image, wherein the third feature image contains more initial channels than the second feature image, and the second feature image contains more initial channels than the first feature image; and encoding and decoupling the first feature image, the second feature image, and the third feature image respectively to obtain multi-feature information corresponding to the target vehicle.

[0016] Optionally, the first feature image, the second feature image, and the third feature image are encoded and decoupled respectively, specifically including: encoding and decoupling the first feature image, the second feature image, and the third feature image respectively using the following formula:

[0017] ;

[0018] in, These are the first feature image, the second feature image, and the third feature image, respectively. These represent the transposes of the weight matrices corresponding to the first feature image, the second feature image, and the third feature image, respectively. These represent the first feature image, the second feature image, and the third feature image, respectively, and the output images after encoding and decoupling. The coordinates of the first pixel in the feature image. The coordinates of the second pixel in the output image are given; the correspondence between the coordinates of the first pixel and the coordinates of the second pixel is determined by the convolution operation in the above formula.

[0019] The second aspect of this application provides a multi-view vehicle dynamic weighing device based on vehicle-road-cloud-road-device collaboration. The device includes an acquisition module and a processing module, wherein...

[0020] The acquisition module is used to respond to the dynamic weighing operation of the target vehicle by performing multi-dimensional dynamic monitoring of the target vehicle, including first-view monitoring, second-view monitoring, and third-view monitoring. By performing first-view monitoring of the target vehicle, the module acquires the license plate information and width information of the target vehicle; by performing second-view monitoring of the target vehicle, the module acquires the vehicle model information and target speed information of the target vehicle; and by performing third-view monitoring of the target vehicle, the module acquires multi-feature information of the target vehicle, including contour information, position information, axle information, and wheelbase information.

[0021] The processing module is used to construct a multi-view image structure corresponding to the target vehicle based on license plate information, width information, vehicle type information, target vehicle speed information, and multiple feature information; acquire the strain signal generated when the target vehicle passes by through a fiber optic grating sensor; and perform time-series analysis on the strain signal through the multi-view image structure to calculate the actual weight of the target vehicle.

[0022] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described above.

[0023] A fourth aspect of this application provides a computer-readable storage medium storing a computer program, which is executed by a processor using the method described in any of the foregoing descriptions.

[0024] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0025] 1. When performing dynamic weighing on a target vehicle, the system performs multi-dimensional dynamic monitoring of the target vehicle. Specifically, it acquires the license plate and width information from a first-view perspective, the vehicle type and target speed from a second-view perspective, and multiple feature information from a third-view perspective. Based on these information, a multi-view image structure is constructed. Furthermore, a fiber optic grating sensor acquires the strain signal generated when the target vehicle passes by. The strain signal is then analyzed over time using the multi-view image structure to calculate the actual weight of the target vehicle. This significantly improves the accuracy and efficiency of dynamic vehicle weighing, enhances the system's adaptability to complex highway traffic scenarios, and enables real-time linkage and sharing of vehicle information and weighing data. This solves the problem that existing static weighing or low-speed dynamic weighing methods are insufficient to meet the weighing needs of vehicles traveling at high speeds on highways.

[0026] 2. By capturing vehicle images corresponding to the target vehicle using a preset camera, and extracting features from the vehicle images to obtain a first feature image, a second feature image, and a third feature image corresponding to the target vehicle, the first feature image, the second feature image, and the third feature image are encoded and decoupled respectively to obtain the multi-feature information corresponding to the target vehicle. This achieves accurate extraction and decoupling of various features of the target vehicle, optimizes the hierarchical representation of the feature images, and improves the model's comprehensive analysis capabilities of vehicle appearance, size, and dynamic state, thereby providing more accurate multi-dimensional information support for subsequent vehicle recognition, speed calculation, and weight assessment.

[0027] 3. Obtain the first and second speed influencing factors, and calculate the target vehicle speed information based on the first and second speed influencing factors, the first vehicle speed information, and the second vehicle speed information. This improves the accuracy and reliability of vehicle speed information calculation, fully considers the influence of different factors on speed measurement, and combines multi-source data for fusion to ensure that the actual speed of the target vehicle can still be accurately reflected in complex traffic environments, especially during high-speed driving or speed change. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating the multi-view vehicle dynamic weighing method based on vehicle-road-cloud-road-end collaboration provided in this application embodiment;

[0029] Figure 2 This is a schematic diagram of a multi-view vehicle dynamic weighing device based on vehicle-road-cloud-road-end collaboration provided in an embodiment of this application;

[0030] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0031] Explanation of reference numerals in the attached figures: 21. Acquisition module; 22. Processing module; 301. Processor; 302. Communication bus; 303. User interface; 304. Network interface; 305. Memory. Detailed Implementation

[0032] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0033] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items.

[0034] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0035] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0036] Please refer to Figure 1 The flowchart illustrates a multi-view vehicle dynamic weighing method based on vehicle-road-cloud-road-end collaboration provided in this application embodiment. The flowchart mainly includes the following steps: S101 to S107.

[0037] Step S101: In response to the dynamic weighing operation of the target vehicle, multi-dimensional dynamic monitoring is performed on the target vehicle, including first-view monitoring, second-view monitoring and third-view monitoring.

[0038] Specifically, the target vehicle can be any vehicle to be detected. When the target vehicle enters the tollbooth, the OBU (On-Board Unit) monitors the target vehicle's position changes through a high-precision positioning system to ensure the target vehicle's driving safety and position accuracy within the tollbooth area. The OBU receives precise location information from the high-precision positioning system and detects traffic conditions around the target vehicle, uploading this information, along with the target vehicle's data, to the cloud control platform for more accurate data support for subsequent lane guidance, speed control, and other operations. The OBU continuously transmits signals at a specific frequency through its built-in radio frequency communication module, further enhancing its sensing capability against the RSU (Road-Side Unit) through technologies such as the Doppler effect. When the vehicle enters the effective communication range of the RSU, the RSU receives and analyzes the signal. Simultaneously, the RSU also sends a response signal to the OBU; the two confirm each other's presence and establish a communication connection through signal interaction. After the RSU and OBU establish a communication connection, a dynamic weighing operation is performed on the target vehicle. At this time, the target vehicle is first subjected to multi-dimensional dynamic monitoring to obtain vehicle images corresponding to the target vehicle from multiple different perspectives. The multi-dimensional dynamic monitoring includes first-view monitoring, second-view monitoring and third-view monitoring.

[0039] Step S102: By performing first-person perspective monitoring on the target vehicle, obtain the license plate information and width information corresponding to the target vehicle.

[0040] Specifically, multi-dimensional dynamic monitoring includes first-person perspective monitoring, which acquires vehicle images of the target vehicle from the corresponding perspective, thereby obtaining the license plate information and width information of the target vehicle.

[0041] In one possible implementation, step S102 further includes: monitoring the first perspective using optical character recognition technology and a first perspective model, wherein the first perspective model is a convolutional neural network model; obtaining the license plate information using the optical character recognition technology; and obtaining the width information using the first perspective model.

[0042] Specifically, vehicle images of the target vehicle are acquired from a first-person perspective, and clear images containing license plate and width information are selected. The acquired vehicle images undergo preprocessing, including but not limited to grayscale conversion, binarization, and denoising, to improve the accuracy and efficiency of subsequent processing. Optical Character Recognition (OCR) technology is used to identify the license plate number: the license plate area is located, the region of interest is extracted, character segmentation and recognition are performed on the license plate area, and the license plate number is output. A pre-trained first-person perspective model is used to extract vehicle width information: the preprocessed image is input into the first-person perspective model, and the model outputs the vehicle width value through feature extraction and regression calculation. Afterwards, the acquired license plate and width information are transmitted to a cloud control platform for further data analysis and application: data augmentation techniques are used to improve the robustness of the first-person perspective model to adapt to different lighting conditions and vehicle models; the first-person perspective model is fine-tuned to optimize its performance in real-world scenarios; and post-processing of the width information, such as smoothing and outlier removal, is performed to ensure data accuracy.

[0043] Step S103: By performing first-person perspective monitoring of the target vehicle, obtain the license plate information and width information corresponding to the target vehicle.

[0044] Specifically, multi-dimensional dynamic monitoring includes second-view monitoring, which acquires vehicle images of the target vehicle from the corresponding viewpoint, thereby obtaining the vehicle model information and target speed information of the target vehicle.

[0045] In one possible implementation, step S103 further includes: performing second-view monitoring using dual radar speed detectors and the second-view model, wherein the calculation formulas for the forward and backward propagation of the second-view model are as follows:

[0046] ;

[0047] in, For the second perspective model during forward propagation, the first The intermediate layer features corresponding to the convolutional structure For the second perspective model during forward propagation, the first The first convolution weights corresponding to the layer convolution structure, To perform convolution operations, Indicates the first The intermediate layer features corresponding to the convolutional structure are smoothed using computation. Indicates the fusion process. For the second perspective model during backpropagation, the first The second convolution weights corresponding to the layer convolution structure For the first Each weight update function The gradient accumulated during the backpropagation of the second perspective model is used to obtain the vehicle type information and first vehicle speed information corresponding to the target vehicle through the second perspective model, and the second vehicle speed information when the target vehicle passes by is obtained through the dual radar speed detector; the target vehicle speed information is calculated based on the first vehicle speed information and the second vehicle speed information.

[0048] Specifically, the process of acquiring the vehicle model and speed information of the target vehicle through dual radar speed detectors and a second-view model includes the following steps: The second-view model uses a convolutional neural network, performing forward propagation through a multi-layer convolutional structure. Input data includes relevant monitoring image sequences of the vehicle and previous feature data. In each convolutional structure, the intermediate layer features... Convolution is performed using the output features from the previous layer. The sigmoid activation function is then used. The intermediate feature layers are smoothed to enhance the model's robustness and detail capture ability, and features from each layer are accumulated through fusion to gradually generate high-dimensional semantic features. Furthermore, the convolutional weights of each layer in the second-view model... During backpropagation, using the formula The update is performed, including the weight update function. The adaptive learning model is optimized by adjusting the weights to improve its ability to predict the target vehicle's model and speed information. The speed information predicted by the second-view model is recorded as the first speed information. Dual radar speedometers acquire the vehicle's speed before and after passing the speedometers and record the second speed information. The target vehicle speed information is calculated based on the first and second speed information. The model features output by the second-view model are then matched against a vehicle database to verify the model information.

[0049] In one possible implementation, step S103 further includes: obtaining a first speed influencing factor and a second speed influencing factor, wherein the first speed influencing factor is a factor affecting the calculation of the second viewpoint model, and the second speed influencing factor is a factor affecting the calculation of the dual radar speed detector; and calculating the target vehicle speed information based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information.

[0050] Specifically, the factors influencing first and second speed are identified. The first speed influencing factors include: the complexity of the vehicle's trajectory, the deviation between the monitoring viewpoint and the vehicle's direction of travel, ambient lighting conditions, and the clarity of the vehicle's exterior features. The complexity of the vehicle's trajectory refers to dynamic behaviors such as acceleration, deceleration, or curved driving. The deviation between the monitoring viewpoint and the vehicle's direction of travel refers to whether the second-view model's shooting angle is aligned with the vehicle's direction. Ambient lighting conditions include the impact of light intensity (e.g., sunny, cloudy, nighttime) on image quality. The clarity of the vehicle's exterior features includes factors such as the vehicle's color, pattern, and degree of mud, which affect the model's accuracy in feature extraction. The second speed influencing factors include: radar measurement range and accuracy, vehicle surface material, environmental interference, and sudden speed changes. The radar measurement range and accuracy factors include the radar's effective measurement distance and speed measurement accuracy. Vehicle surface material factors include the impact of materials such as metal and glass on radar signal reflection. Environmental interference factors include interference from surrounding obstacles, vehicles in adjacent lanes, or weather. Sudden speed changes refer to the impact of sudden acceleration or deceleration of the vehicle on the continuous measurement of the radar speedometer. Confirm the actual number of the first and second speed influencing factors among the above factors, and perform a fusion calculation of the target vehicle speed information based on the number of these factors, the first vehicle speed information, and the second vehicle speed information, that is, calculate the target vehicle speed information using the following formula:

[0051] ;

[0052] in, The target vehicle speed information, The number of factors affecting the first speed. The number of factors affecting the second speed. The first viewpoint model obtained by the second viewpoint model The first vehicle speed information, The first speed sensor obtained by the dual radar speed detector The second vehicle speed information, The first weight corresponds to the first vehicle speed information, and , The second weight corresponds to the second vehicle speed information, and Among them, weight and This reflects the relative importance of the two sources of influence, ensuring that the weighting is appropriate to their respective levels of impact.

[0053] Step S104: By performing third-view monitoring on the target vehicle, obtain multi-feature information corresponding to the target vehicle, including contour information, position information, axle information, and wheelbase information.

[0054] Specifically, multi-dimensional dynamic monitoring includes third-view monitoring, which acquires vehicle images of the target vehicle from the corresponding viewpoint, thereby obtaining multi-feature information of the target vehicle, including contour information, position information, axle information, and wheelbase information.

[0055] In one possible implementation, step S104 further includes: capturing a vehicle image corresponding to the target vehicle using a preset camera; extracting features from the vehicle image to obtain a feature image corresponding to the target vehicle, the feature image including a first feature image, a second feature image, and a third feature image, wherein the third feature image contains an initial number of channels greater than that of the second feature image, and the second feature image contains an initial number of channels greater than that of the first feature image; and encoding and decoupling the first feature image, the second feature image, and the third feature image respectively to obtain the multi-feature information corresponding to the target vehicle.

[0056] Specifically, vehicle images of the target vehicle are acquired through third-person perspective monitoring at the corresponding viewpoint, and feature extraction is performed on the vehicle images to obtain feature images corresponding to the target vehicle. The feature images include a first feature image and a second feature image. and the third feature image Among them, the first feature image Second feature image Third feature image C represents the initial number of channels in the feature image, and H and W represent the initial width and height of the vehicle image, respectively. Therefore, the third feature image contains a greater number of initial channels than the second feature image, and the second feature image contains a greater number of initial channels than the first feature image. The first feature image mainly contains edge features and texture features of the vehicle image, representing a lower-level feature. The second feature image further extracts more complex shape and local structural features, including vehicle axle features. The highest-level third feature image integrates global semantic information, more accurately reflecting the vehicle's contour and position information. The global semantic information includes: overall vehicle contour information reflecting the vehicle model category, and vehicle position-related information reflecting the vehicle's specific coordinates in the image. Subsequently, the first, second, and third feature images are encoded and decoupled to obtain the multi-feature information corresponding to the target vehicle. The encoding and decoupling of the first, second, and third feature images are performed using the following formula:

[0057] ;

[0058] in, These are the first feature image, the second feature image, and the third feature image, respectively. These represent the transposes of the weight matrices corresponding to the first feature image, the second feature image, and the third feature image, respectively. The first feature image, the second feature image, and the third feature image are respectively represented by the output images after encoding and decoupling. The coordinates of the first pixel in the feature image. The coordinates of the second pixel in the output image are given; the correspondence between the coordinates of the first pixel and the coordinates of the second pixel is determined by the convolution operation in the above formula.

[0059] Step S105: Construct a multi-view image structure corresponding to the target vehicle based on the license plate information, width information, vehicle type information, target vehicle speed information, and the multi-feature information.

[0060] Specifically, based on the acquired license plate information, width information, vehicle model information, target vehicle speed information, and multi-feature information, a multi-view graph structure corresponding to the target vehicle is constructed using multi-view dynamic monitoring data. This multi-view graph structure uses the target vehicle as the central node and establishes topological relationships through information extracted from different perspectives. It comprehensively reflects the global and local detailed features of the target vehicle. The specific steps include: using license plate information, width information, vehicle model information, target vehicle speed information, and multi-feature information as attribute features of the nodes, where license plate information serves as the unique identifier of the target vehicle, ensuring that multi-view data corresponds to the same target vehicle; width and vehicle model information are used to describe the vehicle's external features; target vehicle speed information provides the dynamic state of the vehicle's movement; and multi-feature information includes physical characteristics such as vehicle outline, position, axle, and wheelbase. Centered on the target vehicle, feature information extracted from the first, second, and third perspectives is used as input to different sub-views and projected onto the same graph structure; different feature nodes are connected through feature similarity calculations (such as Euclidean distance or cosine similarity) to form association relationships. For example, vehicle model information and contour information nodes can be associated through the similarity of vehicle structural features. The weights of each edge are dynamically assigned based on the confidence level of the viewpoint information or data quality (such as image clarity and radar speed measurement accuracy), with larger weights indicating greater importance of the association. Using the target vehicle's location information, the vehicle's spatial position in the multi-view graph structure is labeled. Global contour semantic information extracted from the third viewpoint is used as the high-level nodes of the graph structure, while feature information from the first and second viewpoints serves as the intermediate and bottom-level nodes, forming a hierarchical graph structure that ensures a unified representation of global and local features. The feature representation of the multi-view graph structure is further optimized using a Graph Neural Network (GNN). A message passing mechanism is used to share information among nodes, enhancing the ability to describe the overall features of the target vehicle. Noise removal is performed on potentially abnormal data nodes to ensure the integrity and robustness of the multi-view graph structure. The final generated multi-view graph structure reflects both the static physical characteristics of the target vehicle and provides a comprehensive vehicle description by incorporating dynamic state features, providing efficient support for subsequent tasks such as dynamic weighing analysis, vehicle classification, and behavior prediction.

[0061] Step S106: Obtain the strain signal generated when the target vehicle passes by using a fiber optic grating sensor.

[0062] Specifically, the sensor density is adjusted according to the possible weight and axle load distribution of the vehicle to improve the accuracy and coverage of data acquisition. When the target vehicle passes by, the Bragg wavelength of the fiber Bragg sensor shifts according to the mechanical deformation caused by the vehicle's gravity and dynamic motion. The wavelength shift data is collected in real time by a high-precision spectral demodulation device using the spectral response characteristics of the fiber Bragg sensor and converted into a strain signal.

[0063] Step S107: Perform time-series analysis on the strain signal using the multi-view diagram structure to calculate the actual weight of the target vehicle.

[0064] Specifically, the Roadside Unit (RSU) uses fiber Bragg grating sensors to convert the physical changes caused by a vehicle passing through into electrical signals, which are continuously transmitted to the Roadside Node (RNN) network. The RNN network receives this signal data, removes noise and outliers, extracts vehicle information from a multi-view graph structure, and extracts and analyzes the time series of the signals to infer the vehicle's weight. By combining the RSU's fiber Bragg grating sensors with the RNN network, the weight of a vehicle weighed at the roadside unit can be obtained efficiently and accurately.

[0065] This application employs the aforementioned method to perform multi-dimensional dynamic monitoring of the target vehicle during dynamic weighing operations. Specifically, it acquires the vehicle's license plate and width information from a first-view perspective, its vehicle model and target speed information from a second-view perspective, and multiple feature information from a third-view perspective. Based on these information, a multi-view image structure is constructed, and strain signals generated as the target vehicle passes are acquired using a fiber optic grating sensor. The strain signals are then analyzed temporally using the multi-view image structure to calculate the vehicle's actual weight. This significantly improves the accuracy and efficiency of dynamic vehicle weighing, enhances the system's adaptability to complex highway traffic scenarios, and enables real-time linkage and sharing of vehicle information and weighing data. This solves the problem that existing static weighing or low-speed dynamic weighing methods are insufficient to meet the weighing needs of vehicles traveling at high speeds on highways.

[0066] Please refer to Figure 2 This document illustrates a schematic diagram of a multi-view vehicle dynamic weighing device based on vehicle-road-cloud-road-end collaboration provided in an embodiment of this application. The device includes an acquisition module 21 and a processing module 22.

[0067] The acquisition module 21 is used to respond to the dynamic weighing operation of the target vehicle by performing multi-dimensional dynamic monitoring of the target vehicle, including first-view monitoring, second-view monitoring and third-view monitoring; by performing first-view monitoring of the target vehicle, the module acquires the license plate information and width information of the target vehicle; by performing second-view monitoring of the target vehicle, the module acquires the vehicle type information and target speed information of the target vehicle; by performing third-view monitoring of the target vehicle, the module acquires multi-feature information of the target vehicle, including contour information, position information, axle information and wheelbase information.

[0068] The processing module 22 is used to construct a multi-view image structure corresponding to the target vehicle based on license plate information, width information, vehicle type information, target vehicle speed information and multiple feature information; acquire the strain signal generated when the target vehicle passes by through a fiber optic grating sensor; and perform time-series analysis on the strain signal through the multi-view image structure to calculate the actual weight of the target vehicle.

[0069] In one possible implementation, the acquisition module 21 is used to acquire the license plate information and width information corresponding to the target vehicle, specifically including: performing first-view monitoring through optical character recognition technology and a first-view model, wherein the first-view model is a convolutional neural network model; acquiring license plate information through optical character recognition technology; and acquiring width information through the first-view model.

[0070] In one possible implementation, the acquisition module 21 is used to acquire the vehicle model information and target vehicle speed information corresponding to the target vehicle, specifically including: performing second-view monitoring through dual radar speed detectors and a second-view model. The calculation formulas corresponding to the forward propagation and backward propagation of the second-view model are as follows:

[0071] ;

[0072] in, For the second-person perspective model during forward propagation, the first The intermediate layer features corresponding to the convolutional structure For the second-person perspective model during forward propagation, the first The first convolution weights corresponding to the layer convolution structure, To perform convolution operations, Indicates the first The intermediate layer features corresponding to the convolutional structure are smoothed during computation. Indicates the fusion process. For the second-view model during backpropagation, the first The second convolution weights corresponding to the layer convolution structure For the first Each weight update function The gradient accumulated during the backpropagation of the second-view model is used to transmit the vehicle model information and the first vehicle speed information corresponding to the target vehicle through the second-view model, and the second vehicle speed information when the target vehicle passes by is obtained through dual radar speed detectors; the target vehicle speed information is calculated based on the first vehicle speed information and the second vehicle speed information.

[0073] In one possible implementation, the acquisition module 21 is used to calculate the target vehicle speed information based on the first vehicle speed information and the second vehicle speed information, specifically including: acquiring the first speed influencing factor and the second speed influencing factor, wherein the first speed influencing factor is a factor that affects the calculation of the second view model, and the second speed influencing factor is a factor that affects the calculation of the dual radar speed detector; and calculating the target vehicle speed information based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information.

[0074] In one possible implementation, the acquisition module 21 is used to calculate the target vehicle speed information based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information, specifically including: calculating the target vehicle speed information using the following formula:

[0075] ;

[0076] in, For target vehicle speed information, The number of factors influencing the first speed. The number of factors influencing the second speed. The first one calculated using the second-view model First vehicle speed information The first [speed sensor] obtained through dual radar speed detectors The second vehicle speed information, The first weight corresponds to the first vehicle speed information, and , The second weight corresponds to the second vehicle speed information, and .

[0077] In one possible implementation, the acquisition module 21 is used to acquire multi-feature information corresponding to the target vehicle, specifically including: taking a vehicle image corresponding to the target vehicle through a preset camera; extracting features from the vehicle image to acquire a feature image corresponding to the target vehicle, the feature image including a first feature image, a second feature image and a third feature image, the third feature image containing an initial number of channels greater than the second feature image, and the second feature image containing an initial number of channels greater than the first feature image; encoding and decoupling the first feature image, the second feature image and the third feature image respectively to acquire multi-feature information corresponding to the target vehicle.

[0078] In one possible implementation, the acquisition module 21 is used to encode and decouple the first feature image, the second feature image, and the third feature image respectively, specifically including: encoding and decoupling the first feature image, the second feature image, and the third feature image respectively using the following formula:

[0079] ;

[0080] in, These are the first feature image, the second feature image, and the third feature image, respectively. These represent the transposes of the weight matrices corresponding to the first feature image, the second feature image, and the third feature image, respectively. These represent the first feature image, the second feature image, and the third feature image, respectively, and the output images after encoding and decoupling. The coordinates of the first pixel in the feature image. The coordinates of the second pixel in the output image are given; the correspondence between the coordinates of the first pixel and the coordinates of the second pixel is determined by the convolution operation in the above formula.

[0081] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0082] This application also provides an electronic device. (See reference...) Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: at least one processor 301, at least one communication bus 302, a user interface 303, at least one network interface 304, and a memory 305.

[0083] The communication bus 302 is used to enable communication between these components.

[0084] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0085] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0086] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 305, and by calling data stored in memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0087] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory 305 may include a non-transitory computer-readable storage medium. The memory 305 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. (Refer to...) Figure 3 The memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a multi-view vehicle dynamic weighing application based on vehicle-road-cloud-road-end collaboration.

[0088] exist Figure 3In the illustrated electronic device, the user interface 303 is primarily used to provide an input interface for the user and acquire user input data; while the processor 301 can be used to call the multi-view vehicle dynamic weighing application based on vehicle-road-cloud-road-end collaboration stored in the memory 305. When executed by one or more processors 301, the electronic device performs one or more of the methods described in the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0089] This application also provides a computer-readable storage medium storing instructions. When executed by one or more processors, these instructions cause an electronic device to perform one or more of the methods described in the above embodiments.

[0090] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0091] In the various embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0092] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0093] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0094] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0095] The above description is merely an exemplary embodiment disclosed in this application and should not be construed as limiting the scope of this application. Any equivalent changes and modifications made in accordance with the teachings of this application shall still fall within the scope of this application. Those skilled in the art, upon considering the disclosure of the specification and practical truths, will readily conceive of other embodiments disclosed in this application.

[0096] This application is intended to cover any variations, uses, or adaptations disclosed herein that follow the general principles disclosed herein and include common knowledge or customary technical means in the art that are not described in this application.

Claims

1. A multi-view dynamic vehicle weighing method based on vehicle-road-cloud-road-device collaboration, characterized in that, The method includes: In response to a dynamic weighing operation on a target vehicle, the target vehicle is subjected to multi-dimensional dynamic monitoring, including first-view monitoring, second-view monitoring and third-view monitoring. By performing first-view monitoring on the target vehicle, the license plate information and width information corresponding to the target vehicle are obtained; By performing second-view monitoring on the target vehicle, the vehicle model information and target speed information corresponding to the target vehicle are obtained; By performing third-view monitoring on the target vehicle, multi-feature information corresponding to the target vehicle is obtained, including contour information, position information, axle information, and wheelbase information. A multi-view graph structure corresponding to the target vehicle is constructed based on the license plate information, the width information, the vehicle type information, the target vehicle speed information, and the multi-feature information. The strain signal generated when the target vehicle passes by is acquired using a fiber optic grating sensor; The strain signal is analyzed in time series using the multi-view diagram structure to calculate the actual weight of the target vehicle. The step of obtaining the multi-feature information corresponding to the target vehicle specifically includes: capturing a vehicle image corresponding to the target vehicle using a preset camera; extracting features from the vehicle image to obtain a feature image corresponding to the target vehicle, wherein the feature image includes a first feature image, a second feature image, and a third feature image, the third feature image having an initial number of channels greater than the second feature image, and the second feature image having an initial number of channels greater than the first feature image; and encoding and decoupling the first feature image, the second feature image, and the third feature image respectively to obtain the multi-feature information corresponding to the target vehicle. The step of encoding and decoupling the first feature image, the second feature image, and the third feature image respectively specifically includes: encoding and decoupling the first feature image, the second feature image, and the third feature image respectively using the following formula: ； in, These are the first feature image, the second feature image, and the third feature image, respectively. These represent the transposes of the weight matrices corresponding to the first feature image, the second feature image, and the third feature image, respectively. The first feature image, the second feature image, and the third feature image are respectively represented by the output images after encoding and decoupling. The coordinates of the first pixel in the feature image. The coordinates of the second pixel in the output image are given; the correspondence between the coordinates of the first pixel and the coordinates of the second pixel is determined by the convolution operation in the formula.

2. The method according to claim 1, characterized in that, The acquisition of the license plate information and width information corresponding to the target vehicle specifically includes: The first perspective monitoring is performed using optical character recognition technology and a first perspective model, where the first perspective model is a convolutional neural network model. The license plate information is obtained using the optical character recognition technology; the width information is obtained using the first viewpoint model.

3. The method according to claim 1, characterized in that, The acquisition of the vehicle model information and target speed information corresponding to the target vehicle specifically includes: The second-view monitoring is performed using dual radar speedometers and a second-view model. The calculation formulas for the forward and backward propagation of the second-view model are as follows: ； in, For the second perspective model during forward propagation, the first The intermediate layer features corresponding to the convolutional structure For the second perspective model during forward propagation, the first The first convolution weights corresponding to the layer convolution structure, To perform convolution operations, Indicates the first The intermediate layer features corresponding to the convolutional structure are smoothed using computation. Indicates the fusion process. For the second perspective model during backpropagation, the first The second convolution weights corresponding to the layer convolution structure For the first Each weight update function This represents the gradient accumulated during the backpropagation process of the second-view model; The vehicle model information and first vehicle speed information corresponding to the target vehicle are obtained through the second perspective model, and the second vehicle speed information when the target vehicle passes by is obtained through the dual radar speed detector. The target vehicle speed information is calculated based on the first vehicle speed information and the second vehicle speed information.

4. The method according to claim 3, characterized in that, The step of calculating the target vehicle speed information based on the first vehicle speed information and the second vehicle speed information specifically includes: Obtain a first speed influencing factor and a second speed influencing factor, wherein the first speed influencing factor is the factor affecting the calculation of the second viewpoint model, and the second speed influencing factor is the factor affecting the calculation of the dual radar speed detector; The target vehicle speed information is calculated based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information.

5. The method according to claim 4, characterized in that, The calculation of the target vehicle speed information based on the first speed influencing factor, the second speed influencing factor, the first vehicle speed information, and the second vehicle speed information specifically includes: The target vehicle speed information is calculated using the following formula: ； in, The target vehicle speed information, The number of factors affecting the first speed. The number of factors affecting the second speed. The first viewpoint model obtained by the second viewpoint model The first vehicle speed information, The first speed sensor obtained by the dual radar speed detector The second vehicle speed information, The first weight corresponds to the first vehicle speed information, and , The second weight corresponds to the second vehicle speed information, and .

6. A multi-view vehicle dynamic weighing device based on vehicle-road-cloud-road-device collaboration, characterized in that, The device includes an acquisition module and a processing module, wherein, The acquisition module is configured to respond to a dynamic weighing operation of a target vehicle by performing multi-dimensional dynamic monitoring of the target vehicle, including first-view monitoring, second-view monitoring, and third-view monitoring. By performing first-view monitoring, the module acquires the license plate information and width information corresponding to the target vehicle. By performing second-view monitoring, the module acquires the vehicle model information and target speed information corresponding to the target vehicle. By performing third-view monitoring, the module acquires multi-feature information corresponding to the target vehicle, including contour information, position information, axle information, and wheelbase information. Acquiring the multi-feature information specifically includes: capturing images of the target vehicle using a preset camera. The system obtains a corresponding vehicle image; performs feature extraction on the vehicle image to obtain a feature image corresponding to the target vehicle, the feature image including a first feature image, a second feature image, and a third feature image, wherein the third feature image contains a greater number of initial channels than the second feature image, and the second feature image contains a greater number of initial channels than the first feature image; the first feature image, the second feature image, and the third feature image are respectively encoded and decoupled to obtain the multi-feature information corresponding to the target vehicle; the encoding and decoupling of the first feature image, the second feature image, and the third feature image specifically includes: encoding and decoupling the first feature image, the second feature image, and the third feature image respectively using the following formula: ； in, These are the first feature image, the second feature image, and the third feature image, respectively. These represent the transposes of the weight matrices corresponding to the first feature image, the second feature image, and the third feature image, respectively. The first feature image, the second feature image, and the third feature image are respectively represented by the output images after encoding and decoupling. The coordinates of the first pixel in the feature image. The coordinates of the second pixel in the output image are given; the correspondence between the coordinates of the first pixel and the coordinates of the second pixel is determined by the convolution operation in the formula. The processing module is used to construct a multi-view image structure corresponding to the target vehicle based on the license plate information, width information, vehicle type information, target vehicle speed information, and the multi-feature information; acquire the strain signal generated when the target vehicle passes by through a fiber optic grating sensor; and perform time-series analysis on the strain signal through the multi-view image structure to calculate the actual weight of the target vehicle.

7. An electronic device, characterized in that, The device includes a processor, a communication bus, a user interface, a network interface, and a memory. The memory is used to store instructions. The user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1 to 5.

Images