A multi-source sensing fusion special transport vehicle real-time monitoring and intelligent scheduling system

Abstract

This invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion. By standardizing and hierarchically fusing multi-source data, fused data is obtained, avoiding monitoring distortion caused by outliers and noisy data. The hierarchical fusion strategy solves the problems of simple data overlay and unresolved conflicts through a progressive integration approach. The output fused data possesses accuracy and consistency, and can be directly used as a reliable basis for monitoring and dispatching. By comparing the fused data with preset thresholds for each dimension in real time, the management level and direction are determined based on the comprehensive comparison results, achieving hierarchical and precise management. According to the management level and direction, combined with the fused data, the demand dispatching information for special transport vehicles is determined sequentially, enabling dynamic adjustment of risk-based dispatching for risky states and efficiency-based dispatching for normal states. Furthermore, the dispatching scheme is formulated based on the fused data, making it more practical.

Description

Technical Field

[0001] This invention relates to the field of transportation vehicle management technology, and in particular to a real-time monitoring and intelligent dispatching system for special transportation vehicles using multi-source sensor fusion. Background Technology

[0002] Special transport vehicles, as key equipment in the logistics industry, undertake the transportation of high-value, high-risk, and highly compliant goods. The real-time monitoring and rational scheduling of their operational status directly affect transportation safety, cargo integrity, and efficiency. Currently, special transport vehicle monitoring and scheduling systems have several shortcomings: Single monitoring dimension and insufficient data reliability: Existing systems mostly rely on a single sensor to collect data, such as obtaining location information only through the positioning module or monitoring the surrounding environment only through vision devices. They lack multi-dimensional data collaborative verification, which easily leads to false detection and missed detection. Especially in complex road conditions, the data from a single sensor cannot accurately reflect the status of vehicles and goods after being interfered with.

[0003] Weak multi-source data fusion capability: Although some systems integrate multiple types of sensors, they have not established a scientific fusion mechanism. Data redundancy and conflicts exist, making it impossible to form a unified decision-making basis, resulting in insufficient monitoring accuracy and difficulty in supporting precise scheduling.

[0004] The scheduling strategy lacks specificity: existing scheduling is mostly based on order time limits and route distance planning, without fully taking into account the operating conditions of special transport vehicles, cargo status, road compliance requirements, and driver operating habits. This can easily lead to unreasonable scheduling and delayed response, increasing transportation risks. Summary of the Invention

[0005] This invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion, in order to solve the problems mentioned in the background art.

[0006] A real-time monitoring and intelligent dispatching system for special transport vehicles, comprising: The multi-source data acquisition module is used to collect multi-source data from special transport vehicles. The multi-source data includes positioning data, cargo environment data, vehicle operating condition data, and surrounding visual data. The data processing module is used to perform data standard processing and hierarchical fusion on multi-source data to obtain fused data; The real-time monitoring module is used to compare the fused data with preset thresholds for each dimension in real time, and determine the management level and direction based on the comprehensive comparison results. The intelligent scheduling module is used to determine the demand scheduling information of special transport vehicles in sequence according to management level and management direction, combined with fused data; The push module is used to determine the execution instructions for the target terminal based on the demand scheduling information and push them to the target terminal.

[0007] Preferably, the multi-source data acquisition module includes: The positioning data acquisition unit is used to collect vehicle latitude and longitude coordinates, driving speed, heading angle, trajectory deviation and altitude data based on GNSS and Beidou dual-mode positioning to obtain positioning data; The environmental data acquisition unit is used to collect gas concentration, temperature and humidity, sealing pressure and impact acceleration based on explosion-proof and vibration-resistant sensors to obtain cargo environmental data. The operating condition data acquisition unit is used to collect engine speed, braking system hydraulic value, transmission oil temperature and tire pressure based on sensor probes to obtain vehicle operating condition data; The visual data acquisition unit is used to acquire multi-source visual data about the surroundings of special transport vehicles based on lidar, cameras and infrared sensors, and obtain surrounding visual data.

[0008] Preferably, the data processing module includes: A standard processing unit is used to process the multi-source data for outlier and noise removal, fill in missing values, and unify the format to obtain standardized data. The hierarchical fusion unit is used to perform hierarchical fusion of standardized data according to coordinate calibration and feature matching to obtain fused data.

[0009] Preferably, the layered fusion unit includes: The first mapping unit is used to establish a vehicle coordinate system with the centroid of the special transport vehicle as the origin. Based on the acquisition parameters of the surrounding visual data, it establishes the relative pose parameters of the internal parameters of the surrounding visual data. The relative pose parameters include rotation and translation matrices. Based on the relative pose parameters, the surrounding visual data is mapped to the vehicle coordinate system to obtain the initial coordinates. The installation offset and attitude angle are introduced, and the initial coordinates are iteratively optimized using the least squares method to obtain the optimized coordinates, the first spatial coordinate system. The second mapping unit is used to establish the geodetic plane coordinate system of the positioning data, and use the Kalman filter smoothing algorithm to eliminate positioning jitter, construct the coordinate mapping relationship between the geodetic plane coordinate system and the first spatial coordinate system, and based on the coordinate mapping relationship, map the positioning data to the first spatial coordinate system to obtain the second spatial coordinate system; The third mapping unit is used to determine the fixed spatial coordinates of the detection instruments in the second spatial coordinate system based on the correspondence between the detection instruments and vehicle parts based on the cargo environmental data and vehicle operating condition data. Based on the detection range of each detection instrument and combined with the fixed spatial coordinates, the detection coordinate range of the cargo environmental data and vehicle operating condition data is determined. The detection coordinate range is then calibrated in the second spatial coordinate system to obtain the target spatial coordinate system. The feature matching unit is used to perform feature matching between data of the same type and between data of different types in the target space coordinate system to obtain feature matching results; The fusion unit is used to perform hierarchical fusion of standard data based on feature matching results to obtain fused data.

[0010] Preferably, the feature matching unit includes: The visual association unit is used to extract key image features from the camera perspective in the target spatial coordinate system and obtain point cloud features from the radar perspective. It matches the key image features and point cloud features, selects matching pairs with an area overlap ratio within a preset range based on the matching results, and establishes association features of surrounding visual data based on the matching pairs. The positioning association unit is used to acquire the running trajectory under different positioning instruments in the target spatial coordinate system, align the running trajectories and determine the trajectory similarity. When the trajectory similarity is greater than a preset threshold, the association feature of the positioning data is established. Cross-association units are used to associate matching pairs and running trajectories to obtain cross-association features of surrounding visual data and positioning data; The cross-association unit is also used to associate vehicle operating condition data and cargo environment data in the target spatial coordinate system based on timestamps, so as to obtain cross-association features of vehicle operating condition data and cargo environment data. The integration unit is used to integrate associated features and cross-associated features to obtain feature matching results.

[0011] Preferably, the fusion unit includes: The local fusion unit is used to perform visual-positioning data fusion using a spatial weighted fusion algorithm based on the cross-association features of surrounding visual data and positioning data, and to perform working condition-environment data fusion using a temporal weighted fusion algorithm based on the cross-association features of vehicle working condition data and cargo environment data, so as to obtain the local fusion result. The deep coupling unit is used to build a feature association graph based on the feature matching results, iteratively optimize the feature association graph based on the graph neural network, determine the dynamic association coefficient between data features, build a global fusion model based on the dynamic association coefficient, and deeply couple the local fusion results to obtain fused data.

[0012] Preferably, the real-time monitoring module includes: Establish a unit to learn from historical data and historical management data, and build an integrated analysis mechanism for threshold comparison, weighted comprehensive judgment and grade direction determination; The analysis unit is used to input the fused data into the integrated analysis mechanism to obtain comprehensive comparison results, and to match management levels and management directions based on the comprehensive comparison results.

[0013] Preferably, the intelligent scheduling module includes: The data partitioning unit is used to divide the fused data according to the vehicle local management scope, regional collaborative management scope, and global scheduling management scope to obtain the corresponding fused data; The local determination unit is used to establish an inference template from data input and status determination to the generation of scheduling information. Based on the vehicle's local management scope, the inference template is used to establish cargo status fusion data, risk level determination and local emergency scheduling information; as well as vehicle mechanical fusion data, wear status determination and driving parameter scheduling information, to obtain the local logic chain. The region determination unit is used to establish regional multi-source fusion data, path adaptability determination and regional detour and task scheduling information based on the regional collaborative management scope and using inference templates; as well as multi-vehicle status fusion data, resource balance determination and regional collaborative operation scheduling information, to obtain the regional logical chain; The global determination unit is used to establish global fusion data, supply and demand balance determination, and cross-regional scheduling information based on the global scheduling management scope and using inference templates to obtain the global logical chain. The scheduling determination unit is used to establish a scheduling determination mechanism based on local logical chains, regional logical chains, and global logical chains. It analyzes the corresponding logical chains in the scheduling determination mechanism based on the input of fused data, management level, and management direction to generate initial scheduling information for special transport vehicles. The verification unit is used to perform reverse verification according to the inference template to determine whether the state after execution based on the initial scheduling information meets the target state. If so, the initial scheduling information is used as the final demand scheduling information. Otherwise, the state after execution based on the initial scheduling information is returned to the previous stage of fused data input for re-determination until the obtained state meets the target state, and the corresponding scheduling information is used as the final demand scheduling information.

[0014] Preferably, the data partitioning unit includes: The local acquisition unit is used to acquire data related to a single special vehicle and its cargo, the mechanical safety of the vehicle, the stability of the cargo, and the immediate risk management from the fused data, as the fused data for the scope of vehicle-mounted local management; The regional acquisition unit is used to acquire data related to multiple special vehicles, management route conflict avoidance, regional resource adaptation, and multi-vehicle collaborative operation within a specified transportation area from the fused data, and to use this data as the fused data for the scope of regional collaborative management. The global acquisition unit is used to acquire data related to all special vehicles and business orders within the entire operation network, manage the balance of supply and demand of transportation capacity, cross-regional resource allocation, and long-term efficiency optimization from the fused data, as the fused data for the global scheduling management scope.

[0015] 10. A real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion according to claim 1, characterized in that the push module comprises: The mapping determination unit is used to establish a multi-dimensional instruction mapping table based on demand scheduling information, combined with management level, management direction and target terminal type; The instruction generation unit generates execution instructions based on a multi-dimensional instruction mapping table and pushes them to the target terminal.

[0016] Compared with the prior art, the present invention has achieved the following beneficial effects: By collecting multi-source data from special transport vehicles, including location data, cargo environment data, vehicle operating condition data, and surrounding visual data, the integrity of the data source is ensured, avoiding misjudgments due to missing data dimensions. Through data standardization and hierarchical fusion of the multi-source data, fused data is obtained, preventing monitoring distortion caused by outliers and noisy data. The hierarchical fusion strategy solves the problems of simple data overlay and unresolved conflicts through a progressive integration approach. The output fused data is accurate and consistent, and can be directly used as a reliable basis for monitoring and scheduling. By comparing the fused data with preset thresholds for each dimension in real time, the management level and direction are determined based on the comprehensive comparison results, achieving hierarchical and precise classification. This quasi-management approach balances risk control and transportation efficiency, adapting to the differentiated needs of special transportation. By determining the demand and scheduling information of special transportation vehicles according to management level and direction, combined with integrated data, the scheduling strategy no longer relies on fixed rules but is closely integrated with management level, direction, and integrated data. This enables dynamic adjustments to risk scheduling corresponding to risk status and efficiency scheduling corresponding to normal status. Furthermore, the scheduling plan, based on integrated data, is more practical. By determining execution instructions for target terminals based on demand and scheduling information and pushing them to those terminals, the timeliness and accuracy of instruction execution are ensured. Shipper terminals can receive information synchronously, improving the transparency of the transportation process. Simultaneously, instructions are traceable, facilitating subsequent responsibility identification and process optimization.

[0017] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of a real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion, as described in an embodiment of the present invention. Figure 2 This is a structural diagram of the data processing module in an embodiment of the present invention. Detailed Implementation

[0021] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0022] Example 1: This embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion, such as... Figure 1 As shown, it includes: The multi-source data acquisition module is used to collect multi-source data from special transport vehicles. The multi-source data includes positioning data, cargo environment data, vehicle operating condition data, and surrounding visual data. The data processing module is used to perform data standard processing and hierarchical fusion on multi-source data to obtain fused data; The real-time monitoring module is used to compare the fused data with preset thresholds for each dimension in real time, and determine the management level and direction based on the comprehensive comparison results. The intelligent scheduling module is used to determine the demand scheduling information of special transport vehicles in sequence according to management level and management direction, combined with fused data; The push module is used to determine the execution instructions for the target terminal based on the demand scheduling information and push them to the target terminal.

[0023] In this embodiment, the positioning data refers to parameter data that reflects the real-time geographical location, driving trajectory, speed, and driving direction of the special transport vehicle.

[0024] In this embodiment, cargo environmental data refers to environmental parameter data that is directly related to the storage status of special transport cargo, and needs to be collected specifically according to the type of cargo.

[0025] In this embodiment, vehicle operating condition data refers to the core parameter data that reflects the operating status of the special transport vehicle itself, and is directly related to the vehicle's driving safety.

[0026] In this embodiment, the surrounding visual data refers to the images and video data of the vehicle's surrounding environment collected by the vehicle-mounted vision device, which is used to identify obstacles, road conditions, driver behavior, etc.

[0027] In this embodiment, the preset threshold is a set of data thresholds pre-defined based on special transportation industry standards, cargo characteristics, and vehicle safety requirements. It serves as the benchmark for determining whether the status is normal.

[0028] In this embodiment, the management direction includes two categories: risk prevention and control and efficiency optimization. For example, prompting drivers to check the cargo compartment seals; adjusting the driving route to shorten the transportation time.

[0029] In this embodiment, demand scheduling information refers to the specific scheduling instructions output by the intelligent scheduling module, including requirements for route adjustment, vehicle management, personnel operation, and other dimensions.

[0030] The beneficial effects of the above design scheme are as follows: By collecting multi-source data from special transport vehicles, including positioning data, cargo environment data, vehicle operating condition data, and surrounding visual data, the integrity of the data source is ensured, avoiding risk misjudgments caused by the lack of a single data dimension. Through data standardization processing and hierarchical fusion of the multi-source data, fused data is obtained, avoiding monitoring distortion caused by outliers and noisy data. The hierarchical fusion strategy solves the problems of simple data overlay and unresolved conflicts through a progressive integration method. The output fused data has accuracy and consistency and can be directly used as a reliable basis for monitoring and scheduling. By comparing the fused data with preset thresholds for each dimension in real time, the management level and direction are determined based on the comprehensive comparison results. This system enables precise, tiered management, balancing risk control and transportation efficiency, and adapting to the differentiated needs of special transportation. By determining the demand and scheduling information of special transportation vehicles according to management level and direction, combined with integrated data, the scheduling strategy no longer relies on fixed rules but is closely integrated with management level, direction, and integrated data. This allows for dynamic adjustments, with risk scheduling corresponding to risk conditions and efficiency scheduling corresponding to normal conditions. Furthermore, the scheduling plan, based on integrated data, is more practical. By determining execution instructions for target terminals based on demand and scheduling information and pushing them to those terminals, the system ensures the timeliness and accuracy of instruction execution. Shipper terminals can receive information synchronously, improving the transparency of the transportation process. Additionally, the instructions are traceable, facilitating subsequent responsibility identification and process optimization.

[0031] Example 2: Based on Example 1, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The multi-source data acquisition module includes: The positioning data acquisition unit is used to collect vehicle latitude and longitude coordinates, driving speed, heading angle, trajectory deviation and altitude data based on GNSS and Beidou dual-mode positioning to obtain positioning data; The environmental data acquisition unit is used to collect gas concentration, temperature and humidity, sealing pressure and impact acceleration based on explosion-proof and vibration-resistant sensors to obtain cargo environmental data. The operating condition data acquisition unit is used to collect engine speed, braking system hydraulic value, transmission oil temperature and tire pressure based on sensor probes to obtain vehicle operating condition data; The visual data acquisition unit is used to acquire multi-source visual data about the surroundings of special transport vehicles based on lidar, cameras and infrared sensors, and obtain surrounding visual data.

[0032] In this embodiment, the vehicle operating data also includes fuel tank level, battery voltage, synchronously monitored brake pedal travel, steering angle, and other operational data.

[0033] The beneficial effects of the above design scheme are: by collecting multi-source data from special transport vehicles, including positioning data, cargo environment data, vehicle operating condition data, and surrounding visual data, the integrity of the data source is ensured, risk misjudgment caused by the lack of a single data dimension is avoided, full-dimensional data coverage is achieved, and a foundation for accurate decision-making is laid.

[0034] Example 3: Based on Example 1, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion, such as... Figure 2 As shown, the data processing module includes: A standard processing unit is used to process the multi-source data for outlier and noise removal, fill in missing values, and unify the format to obtain standardized data. The hierarchical fusion unit is used to perform hierarchical fusion of standardized data according to coordinate calibration and feature matching to obtain fused data.

[0035] In this embodiment, outlier and noise processing employs a dual mechanism of 3σ criterion and scenario-based verification to remove abnormal data. For continuous data such as vehicle operating conditions and cargo environment, the 3σ criterion identifies outliers exceeding the mean ± 3 times the standard deviation (such as instantaneous extreme values ​​of tire pressure caused by sensor failure or sudden changes in gas concentration) and marks the source of the abnormal data. For surrounding visual data, image grayscale threshold segmentation and morphological filtering are used to remove noise from rain, snow, and strong light interference. Kalman filtering is used to smooth the positioning data to eliminate positioning jitter in occluded scenarios.

[0036] In this embodiment, the format unification specifically refers to the unified encapsulation into a standardized JSON data format, with fields including data type, collection timestamp, sensor number, numerical value, and precision identifier, thereby achieving consistency at the syntax level of multi-source data.

[0037] In this embodiment, a scenario-based completion strategy is adopted to address data loss caused by network fluctuations and brief sensor dormancy: when positioning data is missing, linear interpolation is performed based on historical data from the inertial measurement unit and vehicle operating conditions (vehicle speed, heading angle); when cargo environment and vehicle operating condition data are missing, the missing value is estimated by using an LSTM time series prediction model combined with historical data of similar operating conditions to ensure data integrity.

[0038] The beneficial effects of the above design scheme are as follows: by handling outlier noise, filling in missing values, and unifying formats, it eliminates differences and interference in multi-source data formats, removes abnormal data caused by sensor failures and operating condition interference, fills in data gaps caused by network fluctuations or sensor dormancy, avoids misjudgments caused by messy or incomplete data, ensures the integrity and consistency of standardized data, lays the foundation for layered fusion, and achieves spatial alignment and feature association of heterogeneous data such as vision and positioning through layered integration by coordinate calibration and feature matching process, solves the problem of data redundancy and conflict, improves the accuracy and correlation of fused data, provides core support for the hierarchical management of the real-time monitoring module and the precise command generation of the intelligent scheduling module, and ensures the accuracy and efficiency of special transportation monitoring and scheduling.

[0039] Example 4: Based on Example 3, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The hierarchical fusion unit includes: The first mapping unit is used to establish a vehicle coordinate system with the centroid of the special transport vehicle as the origin. Based on the acquisition parameters of the surrounding visual data, it establishes the relative pose parameters of the internal parameters of the surrounding visual data. The relative pose parameters include rotation and translation matrices. Based on the relative pose parameters, the surrounding visual data is mapped to the vehicle coordinate system to obtain the initial coordinates. The installation offset and attitude angle are introduced, and the initial coordinates are iteratively optimized using the least squares method to obtain the optimized coordinates, the first spatial coordinate system. The second mapping unit is used to establish the geodetic plane coordinate system of the positioning data, and use the Kalman filter smoothing algorithm to eliminate positioning jitter, construct the coordinate mapping relationship between the geodetic plane coordinate system and the first spatial coordinate system, and based on the coordinate mapping relationship, map the positioning data to the first spatial coordinate system to obtain the second spatial coordinate system; The third mapping unit is used to determine the fixed spatial coordinates of the detection instruments in the second spatial coordinate system based on the correspondence between the detection instruments and vehicle parts based on the cargo environmental data and vehicle operating condition data. Based on the detection range of each detection instrument and combined with the fixed spatial coordinates, the detection coordinate range of the cargo environmental data and vehicle operating condition data is determined. The detection coordinate range is then calibrated in the second spatial coordinate system to obtain the target spatial coordinate system. The feature matching unit is used to perform feature matching between data of the same type and between data of different types in the target space coordinate system to obtain feature matching results; The fusion unit is used to perform hierarchical fusion of standard data based on feature matching results to obtain fused data.

[0040] In this embodiment, the relative pose parameters for establishing the internal parameters of the surrounding visual data include, for example, establishing the camera intrinsic parameter matrix, distortion coefficients, and the relative pose parameters between the LiDAR and the camera.

[0041] In this embodiment, the rotation matrix is ​​a 3×3 orthogonal matrix, which represents the attitude deflection relationship of the lidar coordinate system relative to the camera coordinate system, covering the rotation angles of the three degrees of freedom of roll, pitch, and yaw around the X, Y, and Z axes.

[0042] In this embodiment, the translation matrix is ​​a 3×1 column vector, representing the spatial position offset of the origin of the lidar coordinate system relative to the origin of the camera coordinate system, and the precise quantization value of the installation offset (Δx, Δy, Δz) is used to compensate for the difference in physical distance between the installation positions of the two types of equipment.

[0043] In this embodiment, the vehicle coordinate system has the X-axis along the vehicle's forward direction, the Y-axis perpendicular to the vehicle body pointing to the right, and the Z-axis perpendicular to the ground and pointing upwards.

[0044] The beneficial effects of the above design scheme are as follows: By establishing a reference coordinate system with the vehicle's center of mass as the origin, and quantifying the relative pose within the visual data using rotation and translation matrices, accurate mapping of visual data to the reference coordinate system is achieved. Combined with installation offset, attitude angle, and iterative optimization using the least squares method, installation errors of the visual equipment and data acquisition deviations are eliminated, ensuring the optimized coordinate accuracy of the visual data meets standards and constructing a stable first spatial coordinate system. This provides a unified spatial reference for all subsequent data mappings and enhances the reliability of visual data through iterative optimization, laying the foundation for multi-source data spatial linkage. The Kalman filter smoothing algorithm effectively eliminates positioning data jitter interference, improving the continuity of positioning data. Furthermore, by constructing a mapping relationship between the earth plane coordinate system and the first spatial coordinate system, seamless integration of positioning data to the reference coordinate system is achieved, forming a second spatial coordinate system. This approach addresses the spatial heterogeneity between location and visual data, strengthening the consistency and linkage between these two core spatial data types. It ensures precise correspondence between vehicle location and surrounding environmental data, supporting subsequent cross-type data matching. By calibrating fixed coordinates based on the correspondence between detection instruments and vehicle parts, and defining the detection coordinate range for operating conditions and environmental data based on the detection range, attribute data is anchored to a second spatial coordinate system, generating a target spatial coordinate system. This breaks down the barriers between operating condition, environmental, and spatial data, endowing attribute data with spatial semantics. This allows abnormal operating conditions and environmental changes to be associated with specific vehicle parts or cargo areas, adapting to the precise monitoring needs of special transportation and improving the practicality of data association. By conducting same-type and cross-type data matching within a unified target spatial coordinate system, it avoids matching misalignments caused by coordinate system heterogeneity. Precise feature association is achieved based on spatial benchmarks, reducing the false matching rate. Global feature linkage is used to mine inherent data correlations, generating high-quality matching results that provide a reliable basis for subsequent fusion, while improving matching efficiency and adapting to the real-time requirements of special transportation. Layered fusion based on matching results takes into account the characteristics of both spatial and attribute data, preserving the core value of each data source. By eliminating data redundancy and conflicts through targeted fusion, collaborative fusion, and verification optimization, the accuracy and reliability of fused data are improved. The final output structured fused data can directly support the level determination of the real-time monitoring module and the generation of instructions for the intelligent scheduling module, ensuring the accuracy and efficiency of special transportation monitoring and scheduling, and strengthening the decision support capability of the overall system.

[0045] Example 5: Based on Example 4, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The feature matching unit includes: The visual association unit is used to extract key image features from the camera perspective in the target spatial coordinate system and obtain point cloud features from the radar perspective. It matches the key image features and point cloud features, selects matching pairs with an area overlap ratio within a preset range based on the matching results, and establishes association features of surrounding visual data based on the matching pairs. The positioning association unit is used to acquire the running trajectory under different positioning instruments in the target spatial coordinate system, align the running trajectories and determine the trajectory similarity. When the trajectory similarity is greater than a preset threshold, the association feature of the positioning data is established. Cross-association units are used to associate matching pairs and running trajectories to obtain cross-association features of surrounding visual data and positioning data; The cross-association unit is also used to associate vehicle operating condition data and cargo environment data in the target spatial coordinate system based on timestamps, so as to obtain cross-association features of vehicle operating condition data and cargo environment data. The integration unit is used to integrate associated features and cross-associated features to obtain feature matching results.

[0046] In this embodiment, the matching of key image features and point cloud features is achieved through the FLANN matcher.

[0047] In this embodiment, based on the vehicle position coordinates of the positioning data, a region of interest (ROI) is defined in the vehicle coordinate system of the visual data. Visual features within the ROI are extracted and correlated with the steering and speed features of the positioning trajectory. A matching model is constructed using a logistic regression algorithm to output the correlation results between the visual scene and the positioning status.

[0048] In this embodiment, based on timestamps, the operating conditions and environmental data at the same moment are associated to construct feature association vectors (such as engine speed-cargo compartment temperature and humidity, tire pressure-vibration frequency). The correlation between the data is calculated using the Pearson correlation coefficient, a correlation coefficient threshold is set, and strongly correlated data pairs are filtered to achieve the linkage identification of abnormal operating conditions and environmental changes (such as a sudden increase in engine speed corresponding to excessive vibration in the cargo compartment).

[0049] The beneficial effects of the above design scheme are as follows: Key features from camera images and radar point cloud features are extracted under a unified target spatial coordinate system. After precise matching, the overlapping areas of preset intervals are compared to establish visual data association features. This avoids feature misalignment caused by coordinate system heterogeneity, reduces the mismatch rate of visual data, strengthens the intrinsic linkage between camera and radar data, lays the foundation for accurate fusion of surrounding visual data, adapts to the complex visual monitoring needs of special transportation, integrates the trajectories of different positioning instruments under the target spatial coordinate system, calculates similarity after alignment processing, and establishes positioning association features after meeting the standards. This effectively verifies the reliability of multi-positioning instrument data, compensates for the signal interference and occlusion shortcomings of single positioning instruments, improves the continuity and accuracy of positioning trajectories, provides stable positioning data support for cross-source association, associates visual matching pairs with positioning trajectories, and generates cross-association features. It breaks down the spatial barriers between visual and positioning data, realizes spatial linkage between "surrounding environment - vehicle position," deeply binds visual scene features with positioning status, improves the logic of cross-type data association, supports subsequent accurate fusion, and associates working condition and environmental data in the target spatial coordinate system based on timestamps. By combining temporal consistency and spatial coordinate benchmarks, the two types of attribute data are given clear correlation logic, enabling the linkage identification of changes in operating conditions and environmental responses. This solves the problem of isolated attribute data, adapts to the collaborative monitoring needs of special transportation operating conditions and cargo status, and outputs structured feature matching results. This provides a high-quality basis for subsequent layered fusion, ensures the accuracy and efficiency of data fusion, and strengthens the decision support capability of the overall system.

[0050] Example 6: Based on Example 5, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The fusion unit includes: The local fusion unit is used to perform visual-positioning data fusion using a spatial weighted fusion algorithm based on the cross-association features of surrounding visual data and positioning data, and to perform working condition-environment data fusion using a temporal weighted fusion algorithm based on the cross-association features of vehicle working condition data and cargo environment data, so as to obtain the local fusion result. The deep coupling unit is used to build a feature association graph based on the feature matching results, iteratively optimize the feature association graph based on the graph neural network, determine the dynamic association coefficient between data features, build a global fusion model based on the dynamic association coefficient, and deeply couple the local fusion results to obtain fused data.

[0051] In this embodiment, based on the visual-positioning feature matching results, a spatial weighted fusion algorithm is adopted. Using a unified vehicle coordinate system as a reference, the three-dimensional coordinates of obstacles and lane line position features extracted from the visual data are fused with the real-time vehicle coordinates and heading angle data from the positioning data.

[0052] In this embodiment, based on the strongly correlated data pairs of operating conditions and environment, a time-weighted fusion algorithm is used to fuse operating condition data such as engine speed and tire pressure with environmental data such as cargo compartment temperature and humidity and vibration frequency, using timestamps as the link.

[0053] In this embodiment, visual-positioning spatial fusion data and working condition-environment attribute fusion data are used as model inputs. Through node feature propagation and edge weight update, deep coupling of spatial data and attribute data is achieved. For example, abnormal vehicle braking data is associated and fused with positioning trajectory deviation and visually recognized road obstacle data.

[0054] The beneficial effects of the above design scheme are as follows: It employs a differentiated weighted algorithm to process two types of cross-related features. Visual-positioning data, relying on a spatial weighted fusion algorithm, accurately integrates vehicle location and surrounding environmental data, aligning with the spatial correlation characteristics of both. Operating condition-environment data, through a temporal weighted fusion algorithm, adheres to the core principle of temporal consistency, linking changes in operating conditions with environmental responses. This avoids adaptation conflicts in heterogeneous data fusion while strengthening the intrinsic correlation of local data, improving the accuracy and reliability of local fusion results, and providing a high-quality data foundation for deep coupling across the entire domain. Using feature correlation graphs as a carrier, iterative optimization of dynamic correlation coefficients through graph neural networks accurately quantifies the correlation strength between different features, solving the problems of fixed weights and poor adaptability in traditional fusion methods. The global fusion model built based on dynamic coefficients can achieve deep coupling of local fusion results, breaking down barriers between spatial and attribute data to generate four-dimensional fused data encompassing visual, positioning, operating condition, and environmental aspects. This weakens the impact of redundant interference correlations while strengthening the contribution of core correlation features, making the fused data more aligned with the decision-making needs of special transportation, and providing high-value, structured data support for subsequent monitoring and scheduling.

[0055] Example 7: Based on Example 1, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The real-time monitoring module includes: Establish a unit to learn from historical data and historical management data, and build an integrated analysis mechanism for threshold comparison, weighted comprehensive judgment and grade direction determination; The analysis unit is used to input the fused data into the integrated analysis mechanism to obtain comprehensive comparison results, and to match management levels and management directions based on the comprehensive comparison results.

[0056] In this embodiment, for example, the fusion data across the entire domain is aligned and matched with the corresponding dimension thresholds one by one through a real-time comparison component. A weighted summation algorithm is used to calculate the comprehensive deviation value. In the weighted summation, the spatial dimension and the environmental dimension each account for 30% of the weight, the working condition dimension accounts for 25% of the weight, and the visual dimension accounts for 15% of the weight. Based on the comprehensive deviation value, three management levels are defined: when the comprehensive deviation value is ≤5%, it is determined to be at the normal level; when the comprehensive deviation value is in the range of 5%-15%, it is determined to be at the warning level; and when the comprehensive deviation value is >15%, it is determined to be at the emergency level.

[0057] In this embodiment, if the spatial dimension or visual dimension is the dominant deviation dimension, it is determined as the direction of risk prevention and control; if the working condition dimension or environmental dimension is the dominant deviation dimension and the management level is normal, it is determined as the direction of efficiency optimization.

[0058] The beneficial effects of the above design scheme are: weighted comprehensive judgment avoids misjudgment due to single-dimensional bias, improves the accuracy of management level classification, adapts to complex risk scenarios of special transportation, links level and direction settings, balances risk prevention and control with transportation efficiency, relies on the high reliability of fused data to ensure the real-time and accuracy of threshold comparison, provides clear decision guidance for subsequent intelligent scheduling, and strengthens the closed-loop control capability of the system.

[0059] Example 8: Based on Example 1, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The intelligent dispatching module includes: The data partitioning unit is used to divide the fused data according to the vehicle local management scope, regional collaborative management scope, and global scheduling management scope to obtain the corresponding fused data; The local determination unit is used to establish an inference template from data input and status determination to the generation of scheduling information. Based on the vehicle's local management scope, the inference template is used to establish cargo status fusion data, risk level determination and local emergency scheduling information; as well as vehicle mechanical fusion data, wear status determination and driving parameter scheduling information, to obtain the local logic chain. The region determination unit is used to establish regional multi-source fusion data, path adaptability determination and regional detour and task scheduling information based on the regional collaborative management scope and using inference templates; as well as multi-vehicle status fusion data, resource balance determination and regional collaborative operation scheduling information, to obtain the regional logical chain; The global determination unit is used to establish global fusion data, supply and demand balance determination, and cross-regional scheduling information based on the global scheduling management scope and using inference templates to obtain the global logical chain. The scheduling determination unit is used to establish a scheduling determination mechanism based on local logical chains, regional logical chains, and global logical chains. It analyzes the corresponding logical chains in the scheduling determination mechanism based on the input of fused data, management level, and management direction to generate initial scheduling information for special transport vehicles. The verification unit is used to perform reverse verification according to the inference template to determine whether the state after execution based on the initial scheduling information meets the target state. If so, the initial scheduling information is used as the final demand scheduling information. Otherwise, the state after execution based on the initial scheduling information is returned to the previous stage of fused data input for re-determination until the obtained state meets the target state, and the corresponding scheduling information is used as the final demand scheduling information.

[0060] In this embodiment, the cargo status fusion data, risk level determination, and local emergency dispatch information are, for example, cargo status fusion data is fused using cargo-specific sensing units, risk level determination is based on a graph neural network cross-domain fusion model to determine the cargo risk level and causes, and local emergency dispatch information is, for example, the drive-by-wire execution feedback module automatically adjusts the vehicle speed, links the hydraulic system to stabilize the cargo support posture, and the vehicle terminal simultaneously pushes a request for temporary stopping and investigation to the regional dispatch center.

[0061] In this embodiment, vehicle mechanical fusion data, loss status determination, and driving parameter scheduling information are used. For example, the vehicle mechanical fusion data is the fusion data of the vehicle state sensing unit (fiber optic stress data of the main beam of the frame, composite data of hydraulic system pressure and flow, and mechanical fault characteristic data of the triaxial vibration acceleration sensor). The loss status determination is to compare the stress data after wavelet threshold denoising with the hydraulic loss model to determine that the frame fatigue accumulation reaches 30%, the wear of the hydraulic system seals is slightly excessive, and the vibration characteristics match the road bump dominant type. The driving parameter scheduling information is to limit the vehicle speed to ≤60km / h, prohibit sharp turns and sharp braking (drive-by-wire system locks extreme operations), and activate the active buffer device to suppress vibration in the middle range.

[0062] In this embodiment, the regional multi-source fusion data, path adaptability determination, and regional detour and task scheduling information are as follows: the regional multi-source fusion data is a summary of multi-vehicle edge fusion data; the path adaptability determination indicates that the original planned route passes through bridges with insufficient carrying capacity; the regional detour and task scheduling information is as follows: high-priority order tanker truck (no excessive hydraulic loss): scheduled to alternative detour route 1 (avoiding strong winds and low-capacity bridges, the detour distance increases by 15km, and the estimated time loss is 20 minutes, which can be compensated by optimizing driving parameters); medium-priority order tanker truck (slight hydraulic loss): scheduled to alternative detour route 2 (the detour distance increases by 10km, passes through maintenance stations, and seals can be replaced along the way).

[0063] In this embodiment, the multi-vehicle status fusion data, resource balance determination, and regional collaborative operation scheduling information are, for example, the fusion data of two ultra-large component transport vehicles in the region; determining that the regional transportation resources are unbalanced, vehicle No. 1 has low wear and short distance, and is suitable for priority loading and unloading; vehicle No. 2 has high wear and long distance, and if it goes directly to the work point, it will have to wait 1.5 hours, which will easily aggravate the fatigue of the frame; vehicle No. 1 goes to the work point for loading and unloading first, vehicle No. 2 temporarily stops at the intermediate station in the region, and maintenance personnel are arranged on-site to check the stress state of the frame. After vehicle No. 1 completes the operation, vehicle No. 2 is dispatched to go.

[0064] In this embodiment, the global fusion data, supply and demand balance determination, and cross-regional scheduling information are, for example, the following: global fusion data; global imbalance between supply and demand of hazardous chemical transportation capacity (a shortage of 2 vehicles in the north and a redundancy of 2 vehicles in the south), and the maintenance cycle of vehicles in the north has expired, so forced transportation could easily cause safety accidents; the idle vehicles in the south have low wear and tear; dispatching 2 low-wear tank trucks from the south to support the north, planning the optimal cross-regional route (avoiding high-risk road sections and areas with transportation congestion), coordinating with stations along the way to provide supplies, and having 3 fatigued vehicles in the north enter the maintenance station in sequence for repair according to the degree of wear and tear (prioritizing chassis fatigue ≥50%), and synchronizing the maintenance data to the cloud predictive maintenance module to update the remaining service life.

[0065] The beneficial effects of the above design scheme are as follows: It accurately divides and integrates data according to a three-tiered management scope, avoiding data redundancy across different levels, ensuring that data from each management dimension adapts to corresponding scheduling needs, providing precise data support for the construction of a hierarchical logical chain, improving the targeting and efficiency of scheduling analysis, solving the problem of delayed scheduling decisions caused by the chaos of multi-source integrated data, and solidifying the data-judgment-scheduling logic through standardized inference templates. It focuses on the core risks of individual vehicles, quickly generating local emergency and driving parameter scheduling information, achieving millisecond-level risk response, ensuring the safety of individual vehicle cargo and machinery, reducing the probability of the spread of local emergencies, and building regional logical chains based on inference templates. This accurately adapts to multi-vehicle collaboration and detour requirements, effectively avoiding path conflicts, optimizing regional resource allocation, balancing transportation timeliness and safety, and solving the problems of chaotic multi-vehicle scheduling and poor path adaptability within the region. By establishing a global logical chain, dynamic balance between supply and demand of transportation capacity across regions is achieved. This links transportation capacity allocation and vehicle maintenance, maximizing the utilization of idle resources, reducing operating costs, ensuring the timeliness of high-priority orders, and improving overall operational efficiency. The three-layer logical chain is integrated to form a unified scheduling mechanism, achieving precise matching of integrated data, management levels, and directions. It outputs initial scheduling information with layered linkage, ensuring that scheduling instructions are tailored to different management scenarios, balancing local emergencies and global optimization. Reverse verification forms a closed-loop control, promptly correcting unreasonable scheduling instructions to avoid safety accidents or efficiency losses caused by misscheduling, ensuring that scheduling results match the target state, and improving the reliability and fault tolerance of system scheduling. Ultimately, through layered design and closed-loop verification, precise linkage from single vehicle to global special transportation scheduling is achieved, balancing safety, efficiency, and reliability, and adapting to the needs of multiple special transportation scenarios.

[0066] Example 9: Based on Example 8, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion, comprising a data partitioning unit including: The local acquisition unit is used to acquire data related to a single special vehicle and its cargo, the mechanical safety of the vehicle, the stability of the cargo, and the immediate risk management from the fused data, as the fused data for the scope of vehicle-mounted local management; The regional acquisition unit is used to acquire data related to multiple special vehicles, management route conflict avoidance, regional resource adaptation, and multi-vehicle collaborative operation within a specified transportation area from the fused data, and to use this data as the fused data for the scope of regional collaborative management. The global acquisition unit is used to acquire data related to all special vehicles and business orders within the entire operation network, manage the balance of supply and demand of transportation capacity, cross-regional resource allocation, and long-term efficiency optimization from the fused data, as the fused data for the global scheduling management scope.

[0067] The beneficial effects of the above design scheme are: to accurately divide the fused data according to the three-level management scope, avoid the cross-redundancy of data at different levels, make the data of each management dimension adapt to the corresponding scheduling requirements, provide accurate data support for the construction of the hierarchical logic chain, improve the pertinence and efficiency of scheduling analysis, and solve the problem of scheduling decision lag caused by the mess of multi-source fused data.

[0068] Example 10: Based on Example 1, this embodiment of the invention provides a real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion. The push module includes: The mapping determination unit is used to establish a multi-dimensional instruction mapping table based on demand scheduling information, combined with management level, management direction and target terminal type; The instruction generation unit generates execution instructions based on a multi-dimensional instruction mapping table and pushes them to the target terminal.

[0069] In this embodiment, the normal level corresponds to efficiency optimization instructions (such as route fine-tuning and speed suggestions), the warning level corresponds to risk investigation instructions (such as equipment inspection and parameter correction), and the emergency level corresponds to risk avoidance and disposal instructions (such as emergency stop and dispatch of backup vehicles).

[0070] The beneficial effects of the above design scheme are as follows: A mapping table is established by linking demand scheduling information, management level, direction, and target terminal type to achieve multi-dimensional accurate adaptation, avoiding mismatch between instructions and terminals / scenes, solving the problem of insufficient targeting in traditional instruction push, providing a clear basis for standardized instruction generation, ensuring instruction adaptability, and quickly generating and pushing execution instructions based on the mapping table, simplifying the instruction generation process and improving push timeliness; combined with the aforementioned dual-mode communication mechanism, it ensures that instructions accurately reach the corresponding terminals, strengthens the connection between scheduling, instructions, and execution, adapts to the real-time control needs of special transportation, and improves the efficiency of instruction implementation.

[0071] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this application and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A real-time monitoring and intelligent dispatching system for special transport vehicles using multi-source sensor fusion, characterized in that, include: The multi-source data acquisition module is used to collect multi-source data from special transport vehicles. The multi-source data includes positioning data, cargo environment data, vehicle operating condition data, and surrounding visual data. The data processing module is used to perform data standard processing and hierarchical fusion on multi-source data to obtain fused data; The real-time monitoring module is used to compare the fused data with preset thresholds for each dimension in real time, and determine the management level and direction based on the comprehensive comparison results. The intelligent scheduling module is used to determine the demand scheduling information of special transport vehicles in sequence according to management level and management direction, combined with fused data; The push module is used to determine the execution instructions for the target terminal based on the demand scheduling information and push them to the target terminal.

2. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion as described in claim 1, characterized in that, The multi-source data acquisition module includes: The positioning data acquisition unit is used to collect vehicle latitude and longitude coordinates, driving speed, heading angle, trajectory deviation and altitude data based on GNSS and Beidou dual-mode positioning to obtain positioning data; The environmental data acquisition unit is used to collect gas concentration, temperature and humidity, sealing pressure and impact acceleration based on explosion-proof and vibration-resistant sensors to obtain cargo environmental data. The operating condition data acquisition unit is used to collect engine speed, braking system hydraulic value, transmission oil temperature and tire pressure based on sensor probes to obtain vehicle operating condition data; The visual data acquisition unit is used to acquire multi-source visual data about the surroundings of special transport vehicles based on lidar, cameras and infrared sensors, and obtain surrounding visual data.

3. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion as described in claim 1, characterized in that, The data processing module includes: A standard processing unit is used to process the multi-source data for outlier and noise removal, fill in missing values, and unify the format to obtain standardized data. The hierarchical fusion unit is used to perform hierarchical fusion of standardized data according to coordinate calibration and feature matching to obtain fused data.

4. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion according to claim 3, characterized in that, The layered fusion unit includes: The first mapping unit is used to establish a vehicle coordinate system with the centroid of the special transport vehicle as the origin. Based on the acquisition parameters of the surrounding visual data, it establishes the relative pose parameters of the internal parameters of the surrounding visual data. The relative pose parameters include rotation and translation matrices. Based on the relative pose parameters, the surrounding visual data is mapped to the vehicle coordinate system to obtain the initial coordinates. The installation offset and attitude angle are introduced, and the initial coordinates are iteratively optimized using the least squares method to obtain the optimized coordinates, the first spatial coordinate system. The second mapping unit is used to establish the geodetic plane coordinate system of the positioning data, and use the Kalman filter smoothing algorithm to eliminate positioning jitter, construct the coordinate mapping relationship between the geodetic plane coordinate system and the first spatial coordinate system, and based on the coordinate mapping relationship, map the positioning data to the first spatial coordinate system to obtain the second spatial coordinate system; The third mapping unit is used to determine the fixed spatial coordinates of the detection instruments in the second spatial coordinate system based on the correspondence between the detection instruments and vehicle parts based on the cargo environmental data and vehicle operating condition data. Based on the detection range of each detection instrument and combined with the fixed spatial coordinates, the detection coordinate range of the cargo environmental data and vehicle operating condition data is determined. The detection coordinate range is then calibrated in the second spatial coordinate system to obtain the target spatial coordinate system. The feature matching unit is used to perform feature matching between data of the same type and between data of different types in the target space coordinate system to obtain feature matching results; The fusion unit is used to perform hierarchical fusion of standard data based on feature matching results to obtain fused data.

5. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion according to claim 4, characterized in that, The feature matching unit includes: The visual association unit is used to extract key image features from the camera perspective in the target spatial coordinate system and obtain point cloud features from the radar perspective. It matches the key image features and point cloud features, selects matching pairs with an area overlap ratio within a preset range based on the matching results, and establishes association features of surrounding visual data based on the matching pairs. The positioning association unit is used to acquire the running trajectory under different positioning instruments in the target spatial coordinate system, align the running trajectories and determine the trajectory similarity. When the trajectory similarity is greater than a preset threshold, the association feature of the positioning data is established. Cross-association units are used to associate matching pairs and running trajectories to obtain cross-association features of surrounding visual data and positioning data; The cross-association unit is also used to associate vehicle operating condition data and cargo environment data in the target spatial coordinate system based on timestamps, so as to obtain cross-association features of vehicle operating condition data and cargo environment data. The integration unit is used to integrate associated features and cross-associated features to obtain feature matching results.

6. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion according to claim 5, characterized in that, The fusion unit includes: The local fusion unit is used to perform visual-positioning data fusion using a spatial weighted fusion algorithm based on the cross-association features of surrounding visual data and positioning data, and to perform working condition-environment data fusion using a temporal weighted fusion algorithm based on the cross-association features of vehicle working condition data and cargo environment data, so as to obtain the local fusion result. The deep coupling unit is used to build a feature association graph based on the feature matching results, iteratively optimize the feature association graph based on the graph neural network, determine the dynamic association coefficient between data features, build a global fusion model based on the dynamic association coefficient, and deeply couple the local fusion results to obtain fused data.

7. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion according to claim 1, characterized in that, The real-time monitoring module includes: Establish a unit to learn from historical data and historical management data, and build an integrated analysis mechanism for threshold comparison, weighted comprehensive judgment and grade direction determination; The analysis unit is used to input the fused data into the integrated analysis mechanism to obtain comprehensive comparison results, and to match management levels and management directions based on the comprehensive comparison results.

8. The real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion according to claim 1, characterized in that, The intelligent scheduling module includes: The data partitioning unit is used to divide the fused data according to the vehicle local management scope, regional collaborative management scope, and global scheduling management scope to obtain the corresponding fused data; The local determination unit is used to establish an inference template from data input and status determination to the generation of scheduling information. Based on the vehicle's local management scope, the inference template is used to establish cargo status fusion data, risk level determination and local emergency scheduling information; as well as vehicle mechanical fusion data, wear status determination and driving parameter scheduling information, to obtain the local logic chain. The region determination unit is used to establish regional multi-source fusion data, path adaptability determination and regional detour and task scheduling information based on the regional collaborative management scope and using inference templates; as well as multi-vehicle status fusion data, resource balance determination and regional collaborative operation scheduling information, to obtain the regional logical chain; The global determination unit is used to establish global fusion data, supply and demand balance determination, and cross-regional scheduling information based on the global scheduling management scope and using inference templates to obtain the global logical chain. The scheduling determination unit is used to establish a scheduling determination mechanism based on local logical chains, regional logical chains, and global logical chains. It analyzes the corresponding logical chains in the scheduling determination mechanism based on the input of fused data, management level, and management direction to generate initial scheduling information for special transport vehicles. The verification unit is used to perform reverse verification according to the inference template to determine whether the state after execution based on the initial scheduling information meets the target state. If so, the initial scheduling information is used as the final demand scheduling information. Otherwise, the state after execution based on the initial scheduling information is returned to the previous stage of fused data input for re-determination until the obtained state meets the target state, and the corresponding scheduling information is used as the final demand scheduling information.

9. A real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion as described in claim 8, characterized in that, Data partitioning units include: The local acquisition unit is used to acquire data related to a single special vehicle and its cargo, the mechanical safety of the vehicle, the stability of the cargo, and the immediate risk management from the fused data, as the fused data for the scope of vehicle-mounted local management; The regional acquisition unit is used to acquire data related to multiple special vehicles, management route conflict avoidance, regional resource adaptation, and multi-vehicle collaborative operation within a specified transportation area from the fused data, and to use this data as the fused data for the scope of regional collaborative management. The global acquisition unit is used to acquire data related to all special vehicles and business orders within the entire operation network, manage the balance of supply and demand of transportation capacity, cross-regional resource allocation, and long-term efficiency optimization from the fused data, as the fused data for the global scheduling management scope.

10. A real-time monitoring and intelligent dispatching system for special transport vehicles based on multi-source sensor fusion as described in claim 1, characterized in that, The push module includes: The mapping determination unit is used to establish a multi-dimensional instruction mapping table based on demand scheduling information, combined with management level, management direction and target terminal type; The instruction generation unit generates execution instructions based on a multi-dimensional instruction mapping table and pushes them to the target terminal.

Images