Traffic Overload Early Warning Method and System Based on Multi-Source Data Fusion

Abstract

This invention discloses a traffic overload warning method and system based on multi-source data fusion, relating to the field of traffic monitoring. The method includes: an interactive multi-source data linkage acquisition system that collects multi-source data from target vehicles to obtain a multi-source data set; using a cross-modal attention mechanism to filter key anchor data in the multi-source data set, determining the key anchor data set and associated multi-source data sets; performing interactive semantic integration analysis on the key anchor data set and associated multi-source data sets respectively to determine multi-source integrated interactive features; pre-constructing a multi-dimensional traffic overload label library, matching and identifying the multi-source integrated interactive features with the multi-dimensional traffic overload label library, and generating a traffic overload warning command based on the identification results. This solves the technical problems of low detection accuracy and high false alarm rate in existing traffic overload warning systems, achieving the technical effect of improving detection accuracy and reducing the false alarm rate.

Description

Technical Field

[0001] This application relates to the field of traffic monitoring, and in particular to a method and system for early warning of traffic overload under multi-source data fusion. Background Technology

[0002] The detection and early warning of traffic overloading plays a crucial role in ensuring road safety, maintaining traffic order, and extending the service life of roads. Currently, the main methods for detecting traffic overloading rely on monitoring from a single data source, such as using weighing equipment to measure vehicle load or relying solely on cameras to identify vehicle type and speed. However, these single-data-source methods have significant limitations. For example, the accuracy of weighing equipment decreases under complex weather conditions, or cameras perform poorly at night or in low light conditions. Furthermore, a single data source cannot comprehensively reflect the extent of vehicle overloading, easily leading to misjudgments or missed detections.

[0003] Currently, traffic overload warning systems suffer from low detection accuracy and high false alarm rate. Summary of the Invention

[0004] This application provides a traffic overload warning method and system based on multi-source data fusion. It employs an interactive multi-source data linkage acquisition system to collect comprehensive data from target vehicles, forming a multi-source data set. A cross-modal attention mechanism is used to filter out the core key anchor point data set and its related associated multi-source data sets. Interactive semantic analysis technology is used to process these two types of data separately, extracting multi-source integrated interactive features with comprehensive representation capabilities. These features are then intelligently matched with a pre-built multi-dimensional traffic overload label library. When risk features are detected, a traffic overload warning command is immediately generated. These technical means solve the technical problems of low detection accuracy and high false alarm rate in existing traffic overload warning systems, achieving the technical effect of improving detection accuracy and reducing false alarm rate.

[0005] This application provides a traffic overload warning method based on multi-source data fusion, comprising: an interactive multi-source data linkage acquisition system for acquiring multi-source data of a target vehicle to obtain a multi-source data set; using a cross-modal attention mechanism to filter key anchor data in the multi-source data set to determine a key anchor data set and a related multi-source data set; performing interactive semantic integration analysis on the key anchor data set and the related multi-source data set respectively to determine multi-source integrated interactive features; pre-constructing a multi-dimensional tag library for traffic overload, matching and identifying the multi-source integrated interactive features with the multi-dimensional tag library for traffic overload, and generating a traffic overload warning instruction based on the identification results.

[0006] In a possible implementation, an interactive multi-source data linkage acquisition system collects multi-source data from a target vehicle to obtain a multi-source data set, and performs the following processing: It uses the axle group-based weighing and toll collection system within the multi-source data linkage acquisition system to collect the target vehicle's axle group weight, total weight, and speed to obtain weighing data; it uses the vehicle identification system within the multi-source data linkage acquisition system to collect the target vehicle's axle type, number of wheels, and axle spacing to obtain vehicle characteristic data; it uses the environmental monitoring system within the multi-source data linkage acquisition system to collect the road conditions and weather characteristics of the target vehicle during measurement to obtain environmental data; and it summarizes the weighing data, vehicle characteristic data, and environmental data to obtain the multi-source data set.

[0007] In a possible implementation, a cross-modal attention mechanism is used to filter key anchor data in the multi-source dataset to determine the key anchor data set and the associated multi-source dataset. The following processing is then performed: the multi-source dataset is embedded and encoded, mapped to a feature space, to generate a multi-source embedding representation set; the attention weight between each multi-source embedding representation and other multi-source embedding representations in the multi-source embedding representation set is calculated using the cross-modal attention mechanism to determine the multi-source embedding representation attention coefficient set; the multi-source data corresponding to the top m multi-source embedding representations in the multi-source embedding representation attention coefficient set are added to the key anchor data set; the key anchor data set is removed from the multi-source dataset to obtain the associated multi-source dataset.

[0008] In a possible implementation, the attention weights between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations are calculated using a cross-modal attention mechanism to determine the set of multi-source embedding representation attention coefficients. The following processing is then performed: The cosine similarity function in the cross-modal attention mechanism is extracted to calculate the representation similarity between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations, obtaining a representation similarity set; the softmax function in the cross-modal attention mechanism is extracted to perform attention weight transformation on the representation similarity set, obtaining an attention weight set; using each multi-source embedding representation in the multi-source embedding representation set as an index, the attention weight set is retrieved to determine the attention weight set in which each multi-source embedding representation participates and its mean is calculated. The calculation result is used as the multi-source embedding representation attention coefficient, obtaining the set of multi-source embedding representation attention coefficients.

[0009] In a possible implementation, interactive semantic integration analysis is performed on the key anchor data set and the associated multi-source data set respectively to determine multi-source integrated interactive features, and the following processing is performed: first key anchor data is randomly extracted from the key anchor data set; interactive attention weights are calculated between the first key anchor data and the key anchor data set respectively to determine a first interactive attention weight set; interactive semantic analysis is performed based on the first interactive attention weight set and the key anchor data set to determine key anchor interaction features; interactive semantic analysis is performed on the associated multi-source data set to determine associated interaction features; and the key anchor interaction features and the associated interaction features are integrated to obtain the multi-source integrated interaction features.

[0010] In a possible implementation, interactive semantic analysis is performed based on the first set of interactive attention weights and the set of key anchor data to determine the interactive features of key anchors, and the following processing is performed: a semantic analyzer is pre-built, wherein the semantic analyzer is obtained after training the framework constructed based on the convolutional neural network based on sample data; the semantic analyzer is used to identify the first set of interactive attention weights and the set of key anchor data to determine the interactive features of key anchors.

[0011] In a possible implementation, the interaction attention weights of the first key anchor data and the key anchor data set are calculated respectively to determine the first interaction attention weight set, and the following processing is performed: the similarity between the first key anchor data and the key anchor data in the key anchor data set is calculated respectively to obtain the first similarity set; each first similarity in the first similarity set is divided by the sum of the first similarity set to obtain the first interaction attention weight set.

[0012] In a possible implementation, a multi-dimensional tag library for traffic overload is pre-built, and the following processes are performed: obtaining a historical traffic overload data set; extracting data from the historical traffic overload data set according to overload type, overload level, and overload behavior characteristics to construct an initial overload tag set; performing a union operation on the initial overload tag set, and adding the processed tags to the initially empty database to obtain the multi-dimensional tag library for traffic overload.

[0013] In a possible implementation, the following process is performed: according to the traffic overload warning instruction, a preset feedback window is obtained; within the preset feedback window, the target vehicle is statically weighed again; if the weighing result meets the requirements, a passage instruction is obtained.

[0014] This application also provides a traffic overload warning system based on multi-source data fusion, comprising: a multi-source data acquisition module for interacting with a multi-source data linkage acquisition system to acquire multi-source data from target vehicles and obtain a multi-source data set; a key anchor point data filtering module for using a cross-modal attention mechanism to filter key anchor point data from the multi-source data set, determining the key anchor point data set and associated multi-source data sets; an interactive semantic analysis module for performing interactive semantic integration analysis on the key anchor point data set and the associated multi-source data set respectively, determining multi-source integrated interactive features; and a traffic overload warning instruction generation module for pre-constructing a multi-dimensional traffic overload label library, matching and recognizing the multi-source integrated interactive features with the multi-dimensional traffic overload label library, and generating a traffic overload warning instruction based on the recognition results.

[0015] The proposed traffic overload warning method and system based on multi-source data fusion, as described in this application, firstly involves an interactive multi-source data acquisition system to collect multi-source data on the target vehicle, obtaining a multi-source data set. Next, a cross-modal attention mechanism is used to filter key anchor data within the multi-source data set, determining the key anchor data set and associated multi-source data sets. Then, interactive semantic integration analysis is performed on the key anchor data set and the associated multi-source data sets respectively to determine the multi-source integrated interactive features. Finally, a pre-constructed multi-dimensional traffic overload label library is used to match and identify the multi-source integrated interactive features against the multi-dimensional traffic overload label library, generating a traffic overload warning command based on the identification results. This achieves the technical effect of improving detection accuracy and reducing false alarm rate. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a flowchart illustrating the traffic overload warning method based on multi-source data fusion provided in this application embodiment.

[0018] Figure 2 This is a schematic diagram of the structure of a traffic overload warning system based on multi-source data fusion provided in an embodiment of this application.

[0019] Figure labeling: Multi-source data acquisition module 10, key anchor point data filtering module 20, interactive semantic analysis module 30, traffic overload warning instruction generation module 40. Detailed Implementation

[0020] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below.

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" can be the same or different subsets of all possible embodiments and can be combined with each other without conflict. The terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0023] This application provides a traffic overload warning method based on multi-source data fusion, such as... Figure 1 As shown, the method includes:

[0024] Step S100: The interactive multi-source data linkage acquisition system collects multi-source data from the target vehicle to obtain a multi-source data set.

[0025] Specifically, the multi-source data linkage acquisition system is a system capable of simultaneously acquiring data from different sources (such as cameras, radar, sensors, etc.) and ensuring data consistency in time and space through a linkage mechanism. Multiple sensors, including but not limited to high-definition cameras, lidar, inductive loops, and millimeter-wave radar, are deployed at key traffic sections (such as highways and bridge entrances). Each sensor collects data from target vehicles according to a preset sampling frequency (e.g., a high-definition camera acquires 30 frames per second, and a millimeter-wave radar acquires speed information 10 times per second). The data acquisition module performs preliminary formatting processing on the acquired data, such as converting image data into a unified pixel format and speed information into a unified unit. The formatted data is then transmitted to a data processing center via a data transmission network (such as a fiber optic network or wireless communication network), with timestamps and other synchronization information appended to the data packets.

[0026] In one possible implementation, an interactive multi-source data linkage acquisition system acquires multi-source data from the target vehicle to obtain a multi-source data set. Step S100 further includes step S110, which utilizes the axle-group weighing toll collection system within the multi-source data linkage acquisition system to acquire the axle weight, total weight, and vehicle speed of the target vehicle, obtaining weighing data. Specifically, the axle-group weighing toll collection system uses pressure sensors or weighing sensor arrays installed on the road surface. When a vehicle passes by, the sensors can accurately measure the weight of each axle and the total weight of the vehicle. Simultaneously, the vehicle speed is calculated based on the time difference between the vehicle's passage through the sensors. Specifically, when a vehicle enters the detection area of ​​the axle-group weighing toll collection system, the pressure sensors or weighing sensor array measure the weight of each axle in real time and accumulates it to obtain the total weight of the vehicle. The vehicle speed is calculated based on the time difference between the vehicle's passage through the sensors. The weight and speed data acquired by the sensors undergo preliminary processing (such as filtering and calibration) through the data acquisition module, and then are transmitted to the data processing center via wired or wireless communication. For example, the axle-based weighing system measures the weight of each axle of a truck as 10 tons, 12 tons, and 8 tons, for a total weight of 30 tons, at a speed of 80 km / h. This weighing data is transmitted to the data processing center via a data acquisition module.

[0027] Step S120: The vehicle identification system in the multi-source data linkage acquisition system collects the vehicle axle type, number of wheels, and axle spacing of the target vehicle to obtain vehicle feature data. Specifically, the vehicle identification system includes a high-definition camera and an image recognition algorithm. The high-definition camera captures images of the vehicle, and the image recognition algorithm (such as a deep learning-based object detection model) analyzes the images to identify features such as the vehicle's axle type, number of wheels, and axle spacing. The identified vehicle feature data is formatted by the data acquisition module and transmitted to the data processing center. For example, the vehicle identification system captures an image of a truck and, through the image recognition algorithm, identifies the vehicle as a three-axle vehicle with 12 wheels and axle spacing of 3 meters and 4 meters. The vehicle feature data is also transmitted to the data processing center.

[0028] Step S130: The environmental monitoring system within the multi-source data linkage acquisition system collects road conditions and weather characteristics of the target vehicle at the time of measurement, obtaining environmental data. Specifically, the environmental monitoring system includes road condition sensors (such as road surface humidity sensors, temperature sensors, etc.) and weather monitoring equipment (such as weather stations, rain gauges, etc.). These devices can monitor road humidity, temperature, icing conditions, and weather characteristics (such as rainfall, wind speed, etc.) in real time. The road condition sensors and weather monitoring equipment collect road condition and weather characteristic data in real time. After formatting the collected environmental data, the data acquisition module transmits the environmental data to the data processing center. For example, if the environmental monitoring system detects that the current road condition is dry, the temperature is 25℃, and the weather is sunny, the environmental data is also transmitted to the data processing center.

[0029] Step S140: The weighing data, vehicle characteristic data, and environmental data are aggregated to obtain the multi-source data set. Specifically, in the data processing center, the data aggregation module receives data from the axle-mounted weighing and toll collection system, the vehicle identification system, and the environmental monitoring system. Based on timestamps or other synchronization information, different data are associated with the same target vehicle. The associated data is then integrated into a multi-source data set for subsequent processing.

[0030] This approach integrates weighing data, vehicle characteristic data, and environmental data to comprehensively reflect the operating status and environment of the target vehicle. This includes not only basic information such as vehicle weight and speed, but also characteristics such as axle type and number of wheels, as well as external factors like road conditions and weather conditions. This comprehensiveness makes traffic overload warnings more accurate and reliable, avoiding misjudgments caused by a single data source. For example, in adverse weather conditions (such as rain and snow), the road friction coefficient decreases, and the vehicle's braking distance increases. By collecting environmental data, the system can consider the impact of these factors on vehicle operation, thereby more accurately determining whether a vehicle is overloaded.

[0031] Step S200: Use a cross-modal attention mechanism to filter key anchor data in the multi-source data set to determine the key anchor data set and the associated multi-source data set.

[0032] Specifically, the cross-modal attention mechanism is a deep learning-based mechanism that can simultaneously process data from different modalities (such as images, text, and audio) and automatically learn the most critical information for a specific task (such as traffic overload warning) from different modalities. It filters key information through attention weight allocation. The key anchor data set is the set of data most critical for traffic overload warning selected from the multi-source dataset through the cross-modal attention mechanism. This data contains the most important information for determining whether a vehicle is overloaded, such as vehicle speed, axle weight, road conditions, and axle spacing. The associated multi-source dataset is the dataset other than the key anchor data. While not the most critical, this data can provide auxiliary information during interactive semantic integration analysis, such as vehicle axle type and road icing conditions.

[0033] Specifically, a cross-modal attention model is constructed based on deep learning frameworks (such as TensorFlow and PyTorch). The collected multi-source data is input into the cross-modal attention model, which extracts features from the multi-source data. For example, convolutional neural networks (CNNs) are used to extract visual features from image data, and recurrent neural networks (RNNs) are used to extract temporal features from radar signals. The model calculates the attention weight for each modality of data through an attention mechanism and selects key anchor data based on the weight. For example, if the license plate area in a vehicle image at a certain moment occupies a high proportion of the attention weight, then that image data is identified as key anchor data. The selected key anchor data is stored separately as a key anchor data set, while the remaining data is stored as a set of associated multi-source data.

[0034] In one possible implementation, a cross-modal attention mechanism is used to filter key anchor data in the multi-source data set, determining the key anchor data set and the associated multi-source data set. Step S200 further includes step S210, where the multi-source data set is embedded and encoded, mapped to a feature space, and a multi-source embedded representation set is generated. Specifically, specialized feature extraction models are designed for different modalities of data (such as weighing data, vehicle feature data, environmental data, etc.). For example, for weighing data (such as vehicle axle weight, total weight, vehicle speed, etc.), numerical feature extraction methods (such as standardization, normalization, etc.) are used to extract key features. For vehicle feature data (such as vehicle axle type, number of wheels, axle spacing, etc.), image recognition models (such as convolutional neural networks) are used to extract image features. For environmental data (such as road conditions, weather features, etc.), feature engineering methods are used to extract key features. The extracted features are mapped to a unified feature space through embedded encoding, generating a multi-source embedded representation set, enabling data from different modalities to be compared and fused in the same space.

[0035] Step S220: The attention weights between each multi-source embedding representation and other multi-source embedding representations in the multi-source embedding representation set are calculated using a cross-modal attention mechanism to determine the set of attention coefficients for multi-source embedding representations. Specifically, a cross-modal attention model is constructed, which can use a self-attention mechanism or its variants (such as the multi-head attention mechanism in the Transformer architecture) to calculate the correlation between different modal embedding representations. The multi-source embedding representation set is input into the cross-modal attention model, and the model calculates the similarity (such as dot product similarity, cosine similarity, etc.) between each multi-source embedding representation and other multi-source embedding representations, generating an attention weight matrix. The attention weights reflect the importance of each multi-source embedding representation in the multi-source data. Based on the attention weight matrix, the attention coefficient (i.e., the importance index of each multi-source embedding representation) is determined for each multi-source embedding representation. All attention coefficients are summarized to form the set of attention coefficients for multi-source embedding representations. For example, the model might calculate that the total weight has the highest attention weight, followed by the axle spacing, then the vehicle speed, and the environmental data has the lowest weight.

[0036] Step S230: Add the multi-source data corresponding to the top m multi-source embedding representations in the multi-source embedding representation attention coefficient set to the key anchor data set. Specifically, based on the multi-source embedding representation attention coefficient set, sort the multi-source embedding representations from high to low attention coefficients. Select the top m multi-source embedding representations with the highest attention coefficients, map the selected multi-source embedding representations back to the original multi-source data, find the corresponding multi-source data, and add this original multi-source data to the key anchor data set. For example, if m=2, select the embedding representations corresponding to total weight and axle spacing, and add their corresponding original data (e.g., total weight 30 tons, axle spacing 3 meters and 4 meters) to the key anchor data set.

[0037] Step S240: Remove the key anchor point data set from the multi-source data set to obtain a correlated multi-source data set. Specifically, determine the data in the key anchor point data set, remove these data from the multi-source data set, and the remaining data is the correlated multi-source data set. For example, vehicle speed (e.g., 80 km / h) and environmental data (e.g., dry road, temperature 25℃, sunny day) can be combined to form a correlated multi-source data set.

[0038] This implementation utilizes a cross-modal attention mechanism to automatically identify the most critical information for traffic overload warnings from multi-source data. For example, if the vehicle's total weight and axle spacing are most important for judging overloading, the attention mechanism assigns these data a higher attention coefficient, thus filtering them as key anchor data. This precise filtering improves the efficiency and accuracy of subsequent analysis. Embedded coding maps data from different modalities to the same feature space, enabling direct comparison of cross-modal data and further enhancing the performance of the traffic overload warning system. The separation of the key anchor data set and the associated multi-source data set allows the system to more clearly identify and process core and auxiliary data. Key anchor data can provide direct decision-making basis, while associated multi-source data can serve as auxiliary information to further verify warning results, thereby improving the reliability of the warning.

[0039] In one possible implementation, attention weights between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations are calculated using a cross-modal attention mechanism to determine the set of multi-source embedding representation attention coefficients. Step S220 further includes step S221, extracting the cosine similarity function from the cross-modal attention mechanism to calculate the representation similarity between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations to obtain a representation similarity set. Specifically, the cosine similarity function is used to calculate the similarity between each multi-source embedding representation. Cosine similarity measures the similarity between two vectors by calculating the cosine of the angle between them. Assume the multi-source embedding representation set is {E1, E2, ..., E...} n}, where Ei It is the i-th multi-source embedding representation. For each multi-source embedding representation E i Calculate it with other multi-source embedding representations E j Cosine similarity between (j≠i). All calculated cosine similarity values ​​are stored in the representation similarity set. For example, assuming the multi-source embedding representation set has 4 multi-source embedding representations {E1,E2,E3,E4}, the calculation results are shown in Table 1.

[0040] Table 1: Examples of similarity sets

[0041]

[0042] Step S222: Extract the softmax function from the cross-modal attention mechanism and perform attention weight transformation on the representation similarity set to obtain an attention weight set. Specifically, the softmax function is used to convert representation similarity into attention weights. The softmax function can convert a set of values ​​into a probability distribution, and the softmax function is applied to each similarity value in the representation similarity set. The transformed values ​​are stored in the attention weight set, as shown in Table 2.

[0043] Table 2: Examples of Attention Weight Sets

[0044]

[0045] Step S223: Using each multi-source embedding representation in the multi-source embedding representation set as an index, the attention weight set is retrieved to determine the attention weight set in which each multi-source embedding representation participates and the mean is calculated. The calculation result is used as the attention coefficient of the multi-source embedding representation to obtain the set of attention coefficients of the multi-source embedding representation. Specifically, for each multi-source embedding representation E... i The system retrieves all weight values ​​that a given multi-source embedding representation participates in within the attention weight set, calculates the mean of these weight values, and uses this mean as the attention coefficient for that multi-source embedding representation. The attention coefficients of all multi-source embedding representations are stored in the attention coefficient set, as shown in Table 3.

[0046] Table 3: Examples of attention coefficient sets for multi-source embedding representations

[0047]

[0048] This implementation, through a combination of cosine similarity and the softmax function, accurately measures the correlation between each multi-source embedding representation and other multi-source embedding representations, converting this correlation into attention weights. This enables the system to more accurately identify the data most critical for traffic overload warnings. The softmax function transforms similarity values ​​into a probability distribution, ensuring the normalization of attention weights and avoiding weight bias caused by excessive differences in similarity values. This approach enhances the reliability and stability of data fusion.

[0049] Step S300: Perform interactive semantic integration analysis on the key anchor data set and the associated multi-source data set respectively to determine the multi-source integrated interactive features.

[0050] Specifically, interactive semantic integration analysis refers to the semantic analysis and fusion of data from different sources to extract interactive features that reflect the vehicle's state and behavior. Multi-source integrated interactive features are features obtained through interactive semantic integration analysis, which integrates semantic information from different modalities, comprehensively reflecting the vehicle's state and behavior, and providing a basis for subsequent matching and recognition.

[0051] Specifically, semantic analysis models are constructed for both the key anchor dataset and the associated multi-source dataset. For example, for image data, object detection models (such as YOLO and Faster R-CNN) can be used to identify semantic information such as vehicle type and license plate information; for radar signal data, signal processing algorithms can be used to extract semantic information such as vehicle speed and acceleration. The semantic analysis results of the key anchor dataset and the associated multi-source dataset are then interactively integrated to extract multi-source integrated interactive features. The integration method can be simple concatenation or more complex feature fusion methods, such as weighted summation or feature embedding. To improve the efficiency and accuracy of subsequent matching and recognition, feature optimization algorithms (such as Principal Component Analysis (PCA)) can be used to optimize and reduce the dimensionality of the extracted multi-source integrated interactive features, extracting the most important feature dimensions.

[0052] In one possible implementation, interactive semantic integration analysis is performed on the key anchor data set and the associated multi-source data set respectively to determine the multi-source integrated interaction features. Step S300 further includes step S310, randomly extracting first key anchor data from the key anchor data set. Specifically, a random sampling algorithm is used to randomly select one data point from the key anchor data set as the "first key anchor data". This randomness can increase the robustness of the system and prevent deviations caused by fixed selection of a certain data point. Assuming the key anchor data set is {E1, E2}, one of the data points, such as E1, is randomly selected as the first key anchor data.

[0053] Step S320: Calculate the interaction attention weights between the first key anchor data and the set of key anchor data, respectively, to determine the first interaction attention weight set. Specifically, an attention mechanism is used to calculate the interaction attention weights between the first key anchor data and other key anchor data. For example, this can be achieved by calculating their similarity (such as cosine similarity) and applying the softmax function.

[0054] Step S330: Perform interactive semantic analysis based on the first interactive attention weight set and the key anchor point data set to determine the key anchor point interactive features. Specifically, extract fusion features that reflect the core state of the vehicle (such as over-limit risk) by analyzing the interaction relationship between the first key anchor point data and other data in the key anchor point data set. The interactive attention weight quantifies the semantic association strength between different data, while semantic analysis combines these weights with the original data to generate more discriminative features. Specifically, perform a weighted summation of the data in the key anchor point data set according to the first interactive attention weight set. For example, assuming the key anchor point data set is {E1, E2}, the first key anchor point data is E1, and its interactive attention weight with E2 is α. 12 Key anchor point interaction features F key It can be represented as: F key =α 11 ·E1+α 12 ·E2. Wherein, α 11 The weights of E1 and itself (the normalized residual weights) are then used. The weighted result is input into a multilayer perceptron (MLP) or neural network, and higher-order interaction features are further extracted through a non-linear activation function (such as ReLU).

[0055] Step S340: Perform interactive semantic analysis on the associated multi-source data set to determine the associated interactive features. Specifically, using the same analysis method as the key anchor point data set, perform interactive semantic analysis on the associated multi-source data set to generate associated interactive features reflecting auxiliary information such as environment and vehicle speed.

[0056] Step S350: Integrate the key anchor point interaction features and the associated interaction features to obtain the multi-source integrated interaction features. Specifically, the key anchor point interaction features and the associated interaction features are concatenated to generate the multi-source integrated interaction features.

[0057] This implementation dynamically emphasizes the correlation between key anchor data by randomly selecting the first key anchor data and calculating the interaction attention weight, thereby enhancing the representativeness of the features. Concatenating the key anchor interaction features and related interaction features not only preserves all important information but also simplifies feature representation, facilitating subsequent processing and analysis.

[0058] In one possible implementation, the interaction attention weights of the first key anchor data and the key anchor data set are calculated respectively to determine a first interaction attention weight set. Step S320 further includes step S321, calculating the similarity between the first key anchor data and the key anchor data in the key anchor data set respectively to obtain a first similarity set. Specifically, the first key anchor data and the key anchor data set are determined. For other key anchor data in the key anchor data set, the cosine similarity between them and the first key anchor data is calculated, and all calculated similarity values ​​are stored in the first similarity set.

[0059] Step S322: Divide each first similarity value in the first similarity set by the sum of the first similarity sets to obtain a first interactive attention weight set. Specifically, calculate the sum of all similarity values ​​in the first similarity set, divide each similarity value by the sum, and obtain a normalized attention weight. Store all normalized attention weights in the first interactive attention weight set. This implementation, by normalizing similarity values ​​into attention weights, reflects the relative importance of the first key anchor data compared to other key anchor data. In this way, the system can dynamically adjust the degree of attention given to different data, thereby more effectively capturing key information.

[0060] In one possible implementation, interactive semantic analysis is performed based on the first set of interactive attention weights and the set of key anchor data to determine the interactive features of key anchors. Step S330 further includes step S331, pre-constructing a semantic analyzer, wherein the semantic analyzer is obtained after training a framework constructed based on a convolutional neural network on sample data. Specifically, a convolutional neural network (CNN) framework is used to construct the semantic analyzer. CNN, through structures such as convolutional layers, pooling layers, and fully connected layers, can automatically extract features from data and learn semantic relationships between data. The semantic analyzer is trained using labeled sample data. Sample data may include vehicle weighing data, vehicle feature data, environmental data, etc., and the labeling information may be labels such as whether the vehicle is overloaded or speeding. Through training, the semantic analyzer can learn the semantic relationships between these data, such as: the relationship between the total weight of the vehicle and the overload state, the relationship between the vehicle type and the load capacity, and the impact of environmental conditions on vehicle operation.

[0061] Step S332: The semantic analyzer is used to identify the first set of interactive attention weights and the set of key anchor data to determine the key anchor interaction features. Specifically, the first set of interactive attention weights and the set of key anchor data are used as input features and input into a pre-built semantic analyzer. The semantic analyzer processes the input data through structures such as convolutional layers, pooling layers, and fully connected layers to extract interactive features that reflect the semantic relationships between the data. The key anchor interaction features output by the semantic analyzer can be a vector representing the semantic correlation between the data. In this implementation, the attention weights enable the semantic analyzer to dynamically adjust the degree of attention to different data, thereby more effectively capturing key information.

[0062] Step S400: Pre-construct a multi-dimensional tag library for traffic overload, match and identify the multi-source integrated interactive features with the multi-dimensional tag library for traffic overload, and generate a traffic overload warning instruction based on the identification results.

[0063] Specifically, the traffic overload multi-dimensional label library is a pre-built database containing feature labels for various traffic overload situations. These labels describe the standards and characteristics of overload behavior from multiple dimensions (such as load, speed, vehicle type, etc.) and are used to match and identify the features of actually detected vehicles. Matching and identification is a process of determining whether the current vehicle conforms to the overload behavior described in the traffic overload multi-dimensional label library by calculating the similarity or distance between features and labels. The traffic overload warning instruction is a command generated based on the matching and identification results, used to notify traffic management departments or drivers that their vehicles have overloaded behavior. This instruction includes information such as the vehicle's license plate number, overload type, and degree of overload.

[0064] Specifically, a multi-dimensional tag library for traffic overloading is constructed based on relevant regulations and standards. This library contains feature tags for various overloading situations, such as overload weight range, speed limit range, and vehicle type restrictions. Machine learning or deep learning algorithms (such as Support Vector Machines (SVM) and neural networks) are used to match and identify the multi-source integrated interactive features with the multi-dimensional tag library. By calculating the similarity or distance between features and tags, it is determined whether the vehicle is currently overloaded. Based on the matching and identification results, if the vehicle is determined to be overloaded, a corresponding traffic overloading warning instruction is generated. The warning instruction may include information such as the vehicle's license plate number, overloading type, and degree of overloading, and is sent to traffic management departments or relevant equipment (such as electronic displays) via a communication network.

[0065] For example, if a vehicle's multi-source integrated interaction characteristics are "license plate number XXX, load capacity 50 tons, speed 130 km / h", and the traffic overload multi-dimensional label database specifies that the maximum overload weight for this road section is 49 tons and the maximum speed limit is 120 km / h, then the matching degree between the vehicle's characteristics and the overload and speeding labels is calculated to determine that it simultaneously exhibits both overload and speeding behaviors. Based on the matching results, a traffic overload warning instruction is generated, such as "Vehicle with license plate number XXX is overloaded by 1 ton and speeds by 10 km / h," and this instruction is sent via the communication network to the traffic management department's monitoring system or roadside electronic displays to alert law enforcement officers and drivers.

[0066] In one possible implementation, a multi-dimensional tag library for traffic overload is pre-built. Step S400 further includes step S410, acquiring a historical traffic overload data set. Specifically, historical traffic overload records are obtained from traffic management departments, including vehicle license plate numbers, loads, speeds, and overload types. Vehicle image and video data are obtained from monitoring systems for extracting vehicle features. Vehicle weighing and speed data are obtained from toll stations. The collected data is then processed, and duplicate or erroneous data is removed to form a complete historical traffic overload data set.

[0067] Step S420: Extract data from the historical traffic overload data set according to overload type, overload level, and overload behavior characteristics to construct an initial overload tag set. Specifically, classify the historical traffic overload data according to overload type, for example, separating overload data and speeding data. Under each overload type, further classify according to overload level, for example, dividing overload data into Level 1 overload and Level 2 overload. Under each overload level, further classify according to overload behavior characteristics, for example, dividing Level 1 overload data into load exceeding the limit by 5 tons and load exceeding the limit by 7 tons. Extract tag information under each category to form the initial overload tag set.

[0068] Step S430: Perform a union operation on the initial set of oversized vehicle tags, and add the processed tags to an initially empty database to obtain a multi-dimensional traffic oversized vehicle tag library. Specifically, perform a union operation on all tags in the initial set of oversized vehicle tags to remove duplicate tags and ensure tag uniqueness. Create an initially empty database to store the multi-dimensional traffic oversized vehicle tag library. Add the processed tags one by one to the database to form a complete multi-dimensional traffic oversized vehicle tag library.

[0069] This approach collects and organizes historical traffic overload data to construct a multi-dimensional tag library that comprehensively covers various overload types, levels, and behavioral characteristics, providing rich reference information for subsequent traffic overload warnings. By removing duplicate tags through union processing, the uniqueness and accuracy of the tags in the multi-dimensional tag library are ensured. This allows the system to more accurately determine whether a vehicle is overloaded during matching and identification.

[0070] In one possible implementation, the method further includes: obtaining a preset feedback window according to the traffic overload warning instruction; performing a static weight re-weighing of the target vehicle within the preset feedback window; and obtaining a passage instruction if the re-weighing result meets the requirements.

[0071] Specifically, a time window is set according to actual needs, within which the static weight re-weighing of the target vehicle is completed. The preset feedback window can be dynamically adjusted according to factors such as traffic flow and road conditions. After the system generates a traffic overload warning instruction, it immediately notifies traffic management personnel or automatically guides the vehicle to the static weighing area. The static weighing equipment provides high-precision weight measurement results. Within the preset feedback window (e.g., within 10 minutes), the static weight re-weighing of the target vehicle is completed. The re-weighing results are recorded, including the vehicle's total weight and overload status. The re-weighing results are compared with the preset load standard to determine if they meet the requirements. If the re-weighing results meet the requirements, the system generates a passage instruction, allowing the vehicle to continue driving; if they do not meet the requirements, a further processing instruction is generated (e.g., requiring the vehicle to unload the overloaded portion). This implementation method, through static weight re-weighing, can verify the accuracy of the traffic overload warning instructions generated by the system and reduce the possibility of misjudgment.

[0072] This application employs an interactive multi-source data linkage acquisition system to collect comprehensive data on the target vehicle, forming a multi-source data set. A cross-modal attention mechanism is used to filter out the core key anchor point data set and its related associated multi-source data sets. Interactive semantic analysis technology is then used to process these two types of data, extracting multi-source integrated interactive features with comprehensive representation capabilities. These features are then intelligently matched with a pre-built multi-dimensional traffic overload label library. When a risk feature is detected, a traffic overload warning instruction is immediately generated. These technical means solve the technical problems of low detection accuracy and high false alarm rate in existing traffic overload warning systems, achieving the technical effect of improving detection accuracy and reducing the false alarm rate.

[0073] In the above text, refer to Figure 1 This paper describes in detail a traffic overload warning method based on multi-source data fusion according to embodiments of the present invention. Next, we will refer to... Figure 2 A traffic overload warning system based on multi-source data fusion according to an embodiment of the present invention is described.

[0074] The traffic overload warning system based on multi-source data fusion according to embodiments of the present invention addresses the technical problems of low detection accuracy and high false alarm rate in existing traffic overload warning systems, achieving the technical effect of improving detection accuracy and reducing false alarm rate. The traffic overload warning system based on multi-source data fusion includes: a multi-source data acquisition module 10, a key anchor point data filtering module 20, an interactive semantic analysis module 30, and a traffic overload warning instruction generation module 40.

[0075] The multi-source data acquisition module 10 is used to interact with the multi-source data linkage acquisition system to acquire multi-source data from the target vehicle and obtain a multi-source data set; the key anchor point data filtering module 20 is used to filter key anchor point data from the multi-source data set using a cross-modal attention mechanism to determine the key anchor point data set and the associated multi-source data set; the interactive semantic analysis module 30 is used to perform interactive semantic integration analysis on the key anchor point data set and the associated multi-source data set respectively to determine the multi-source integrated interactive features; the traffic overload warning instruction generation module 40 is used to pre-build a traffic overload multi-dimensional label library, match and identify the multi-source integrated interactive features with the traffic overload multi-dimensional label library, and generate a traffic overload warning instruction based on the identification results.

[0076] The specific configuration of the multi-source data acquisition module 10 will be described in detail below. As mentioned above, the interactive multi-source data linkage acquisition system acquires multi-source data from the target vehicle to obtain a multi-source data set. The multi-source data acquisition module 10 may further include: a weighing data acquisition unit for acquiring the vehicle axle group weight, total weight, and vehicle speed of the target vehicle using the axle group-type weighing and toll collection system in the multi-source data linkage acquisition system to obtain weighing data; a vehicle feature data acquisition unit for acquiring the vehicle axle type, number of wheels, and axle spacing of the target vehicle using the vehicle identification system in the multi-source data linkage acquisition system to obtain vehicle feature data; an environmental data acquisition unit for acquiring the road conditions and weather characteristics of the target vehicle at the time of measurement using the environmental monitoring system in the multi-source data linkage acquisition system to obtain environmental data; and a data aggregation unit for aggregating the weighing data, vehicle feature data, and environmental data to obtain the multi-source data set.

[0077] The specific configuration of the key anchor data filtering module 20 will be described in detail below. As mentioned above, the key anchor data filtering module 20 uses a cross-modal attention mechanism to filter the multi-source data set, determining the key anchor data set and the associated multi-source data set. The key anchor data filtering module 20 may further include: an embedded encoding unit for embedding the multi-source data set into a feature space to generate a multi-source embedded representation set; an attention weight calculation unit for calculating the attention weight between each multi-source embedded representation and other multi-source embedded representations in the multi-source embedded representation set using a cross-modal attention mechanism to determine the multi-source embedded representation attention coefficient set; a key anchor data acquisition unit for adding the multi-source data corresponding to the top m multi-source embedded representations in the multi-source embedded representation attention coefficient set to the key anchor data set; and an associated multi-source data acquisition unit for removing the key anchor data set from the multi-source data set to obtain the associated multi-source data set.

[0078] Specifically, the attention weights between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations are calculated using a cross-modal attention mechanism to determine the set of attention coefficients for multi-source embedding representations. The attention weight calculation unit may further include: a representation similarity calculation subunit for extracting the cosine similarity function in the cross-modal attention mechanism to calculate the representation similarity between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations to obtain a representation similarity set; an attention weight transformation subunit for extracting the softmax function in the cross-modal attention mechanism to perform attention weight transformation on the representation similarity set to obtain an attention weight set; and a mean calculation subunit for using each multi-source embedding representation in the multi-source embedding representation set as an index to retrieve the attention weight set, determine the attention weight set in which each multi-source embedding representation participates, and perform mean calculation, using the calculation result as the multi-source embedding representation attention coefficient to obtain the set of attention coefficients for multi-source embedding representations.

[0079] The specific configuration of the interaction semantic analysis module 30 will be described in detail below. As mentioned above, the interaction semantic integration analysis is performed on the key anchor data set and the associated multi-source data set to determine the multi-source integrated interaction features. The interaction semantic analysis module 30 may further include: a first key anchor data extraction unit for randomly extracting first key anchor data from the key anchor data set; an interaction attention weight calculation unit for calculating the interaction attention weights of the first key anchor data and the key anchor data set to determine a first interaction attention weight set; a first interaction semantic analysis unit for performing interaction semantic analysis based on the first interaction attention weight set and the key anchor data set to determine key anchor interaction features; a second interaction semantic analysis unit for performing interaction semantic analysis on the associated multi-source data set to determine associated interaction features; and an interaction feature integration unit for integrating the key anchor interaction features and the associated interaction features to obtain the multi-source integrated interaction features.

[0080] The interaction semantic analysis unit further includes: a semantic analyzer construction subunit for pre-constructing a semantic analyzer, wherein the semantic analyzer is obtained by training a framework constructed based on a convolutional neural network on sample data; and a semantic analysis recognition subunit for using the semantic analyzer to recognize the first interaction attention weight set and the key anchor data set to determine the key anchor interaction features.

[0081] Specifically, the interaction attention weights of the first key anchor data and the key anchor data set are calculated respectively to determine the first interaction attention weight set. The interaction attention weight calculation unit may further include: a similarity calculation subunit for calculating the similarity between the first key anchor data and the key anchor data in the key anchor data set respectively to obtain a first similarity set; and a first interaction attention weight set acquisition subunit for dividing each first similarity in the first similarity set by the sum of the first similarity set to obtain the first interaction attention weight set.

[0082] The specific configuration of the traffic overload warning instruction generation module 40 will be described in detail below. As mentioned above, a multi-dimensional traffic overload tag library is pre-built. The traffic overload warning instruction generation module 40 may further include: a historical traffic overload data set acquisition unit for acquiring historical traffic overload data sets; a data extraction unit for extracting data from the historical traffic overload data set according to overload type, overload level, and overload behavior characteristics to construct an initial overload tag set; and a traffic overload multi-dimensional tag library construction unit for performing a union processing on the initial overload tag set and adding the processed tags to an initially empty database to obtain a traffic overload multi-dimensional tag library.

[0083] The system may further include: a preset feedback window acquisition module for acquiring a preset feedback window according to the traffic overload warning instruction; and a static weight re-weighing module for performing a static weight re-weighing of the target vehicle within the preset feedback window, and acquiring a passage instruction if the re-weighing result meets the requirements.

[0084] The traffic overload warning system based on multi-source data fusion provided in this embodiment of the invention can execute the traffic overload warning method based on multi-source data fusion provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0085] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0086] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A traffic overload warning method based on multi-source data fusion, characterized in that, The method includes: The interactive multi-source data linkage acquisition system collects multi-source data from the target vehicle to obtain a multi-source data set. A cross-modal attention mechanism is used to filter key anchor data in the multi-source data set to determine the key anchor data set and the associated multi-source data set; The multi-source data set is embedded and encoded, mapped to the feature space, and a multi-source embedded representation set is generated. The attention weights between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations are calculated using a cross-modal attention mechanism to determine the set of attention coefficients for multi-source embedding representations. Add the multi-source data corresponding to the multi-source embedding representations located in the first m positions of the multi-source embedding representation attention coefficient set to the key anchor data set. The key anchor data set is removed from the multi-source data set to obtain the associated multi-source data set; Interactive semantic integration analysis is performed on the key anchor point data set and the associated multi-source data set to determine the multi-source integrated interaction features; Randomly extract the first key anchor point data from the key anchor point data set; Calculate the interaction attention weights of the first key anchor data and the set of key anchor data respectively, and determine the first interaction attention weight set; Based on the first set of interactive attention weights and the set of key anchor points, interactive semantic analysis is performed to determine the interactive features of key anchor points. Perform interactive semantic analysis on the associated multi-source data set to determine the associated interactive features; By integrating the key anchor point interaction features and the associated interaction features, the multi-source integrated interaction features are obtained; A multi-dimensional tag library for traffic overload is pre-constructed. The multi-source integrated interactive features are matched and identified with the multi-dimensional tag library for traffic overload, and a traffic overload warning instruction is generated based on the identification results.

2. The traffic overload early warning method based on multi-source data fusion as described in claim 1, characterized in that, The interactive multi-source data linkage acquisition system collects multi-source data from the target vehicle to obtain a multi-source data set, including: The axle group weight, total weight, and speed of the target vehicle are collected by the axle group weight, total weight, and vehicle speed in the multi-source data linkage acquisition system to obtain weighing data; The vehicle identification system in the multi-source data linkage acquisition system collects the vehicle axle type, number of wheels, and axle spacing of the target vehicle to obtain vehicle feature data; The environmental monitoring system in the multi-source data linkage acquisition system collects road conditions and weather characteristics of the target vehicle at the time of measurement to obtain environmental data; The weighing data, vehicle characteristic data, and environmental data are aggregated to obtain the multi-source data set.

3. The traffic overload early warning method based on multi-source data fusion as described in claim 1, characterized in that, The attention weights between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations are calculated using a cross-modal attention mechanism to determine the set of multi-source embedding representation attention coefficients, including: Extract the cosine similarity function from the cross-modal attention mechanism to calculate the representation similarity between each multi-source embedding representation in the multi-source embedding representation set and other multi-source embedding representations, and obtain the representation similarity set; Extract the softmax function from the cross-modal attention mechanism and apply it to the representation similarity set to perform attention weight transformation, thereby obtaining the attention weight set; Using each multi-source embedding representation in the set of multi-source embedding representations as an index, the attention weight set is retrieved, the attention weight set in which each multi-source embedding representation participates is determined, and the mean is calculated. The calculation result is used as the attention coefficient of the multi-source embedding representation to obtain the set of attention coefficients of the multi-source embedding representation.

4. The traffic overload early warning method based on multi-source data fusion as described in claim 1, characterized in that, Based on the first set of interactive attention weights and the set of key anchor points, interactive semantic analysis is performed to determine the interactive features of key anchor points, including: A pre-built semantic analyzer is obtained by training a framework constructed based on a convolutional neural network on sample data; The semantic analyzer is used to identify the first set of interaction attention weights and the set of key anchor points to determine the interaction features of the key anchor points.

5. The traffic overload early warning method based on multi-source data fusion as described in claim 4, characterized in that, Calculate the interaction attention weights of the first key anchor data and the set of key anchor data respectively, and determine the first interaction attention weight set, including: Calculate the similarity between the first key anchor data and the key anchor data in the key anchor data set respectively to obtain the first similarity set; The first interaction attention weight set is obtained by dividing each first similarity in the first similarity set by the sum of the first similarity set.

6. The traffic overload early warning method based on multi-source data fusion as described in claim 1, characterized in that, A pre-built multi-dimensional tag library for traffic overload and oversize vehicles, including: Obtain a collection of historical traffic overload data; Data is extracted from the historical traffic over-limit data set according to the over-limit type, over-limit level, and over-limit behavior characteristics to construct an initial over-limit tag set; The initial set of over-limit tags is subjected to union processing, and the processed tags are added to the initially empty database to obtain a multi-dimensional tag library for traffic over-limit.

7. The traffic overload early warning method based on multi-source data fusion as described in claim 1, characterized in that, include: Based on the traffic overload warning instruction, a preset feedback window is obtained; The target vehicle is statically weighed again within the preset feedback window. If the weighing result meets the requirements, a passage instruction is obtained.

8. A traffic overload early warning system based on multi-source data fusion, characterized in that, The system is used to implement the traffic overload warning method based on multi-source data fusion as described in any one of claims 1-7, and the system includes: The multi-source data acquisition module is used to interact with the multi-source data linkage acquisition system to collect multi-source data from the target vehicle and obtain a multi-source data set. The key anchor data filtering module is used to filter key anchor data in the multi-source data set using a cross-modal attention mechanism, and to determine the key anchor data set and the associated multi-source data set. The interactive semantic analysis module is used to perform interactive semantic integration analysis on the key anchor point data set and the associated multi-source data set respectively, and determine the multi-source integrated interactive features. The traffic overload warning instruction generation module is used to pre-build a multi-dimensional traffic overload label library, match and identify the multi-source integrated interactive features with the multi-dimensional traffic overload label library, and generate a traffic overload warning instruction based on the identification result.

Images