AI-based cargo transportation real-time safety early warning and path optimization system

Abstract

The invention discloses an AI-based cargo transportation real-time safety early warning and path optimization system, and relates to the technical field of cargo transportation intelligent management, and the system comprises the following steps: obtaining real-time image data of a to-be-transported cargo, carrying out the target detection, obtaining a detection result, and obtaining a non-abnormal to-be-transported cargo; obtaining vehicle running state data, driving behavior risk coefficients and road condition state data corresponding to the transport vehicle; carrying out initial comprehensive evaluation on the transport vehicle to obtain an initial comprehensive evaluation score, and carrying out initial task allocation; constructing the remaining safe driving mileage of each vehicle at the generation moment of the vehicle energy consumption prediction model, and generating early warning information; generating a vehicle supply method based on the early warning information and re-planning a path; real-time image data of transported cargos during transportation and real-time image data of unusual cargos to be transported are acquired in real time to generate a cargo early warning signal; and generating a current vehicle transportation analysis report. According to the invention, the cargo transportation safety and the operation reliability are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology for cargo transportation, specifically an AI-based real-time safety early warning and route optimization system for cargo transportation. Background Technology

[0002] With the rapid development of global trade and the popularization of e-commerce, the logistics industry has continued to expand, and the frequency, quantity and distance of cargo transportation have increased significantly. The huge volume of transportation has made the issues of safety and efficiency in the cargo transportation process increasingly prominent.

[0003] Chinese invention application CN119963079A discloses a vehicle transportation management method and system based on digital technology. The method includes the following steps: real-time collection of vehicle energy status data via a vehicle monitoring module; transmission of the collected vehicle data to a big data platform using Internet of Things (IoT) technology, analysis of the data, and generation of an energy status warning signal; determination of the optimal refueling route based on the vehicle's current location and traffic information, and advance arrangement of energy refueling services when the warning signal is triggered; and dynamic adjustment of the vehicle's transportation route and task allocation based on real-time traffic information, vehicle load, and destination distance via an intelligent scheduling module. Through real-time monitoring and analysis, the system can automatically trigger a refueling warning signal when the energy status falls below a safe threshold, thereby avoiding unplanned shutdowns due to energy depletion and significantly improving the accuracy and reliability of energy management.

[0004] Chinese invention application CN118941186A discloses a vehicle transportation management system and method based on digital technology. This system integrates four functional modules: vehicle monitoring, intelligent scheduling, cargo management, and report generation. The vehicle monitoring module monitors energy status in real time and provides intelligent early warnings of replenishment needs; the intelligent scheduling module optimizes task allocation and route planning to reduce empty runs; the cargo management module strictly monitors transportation trajectories and cargo status to ensure safety; and the report generation module periodically assesses vehicle status and provides maintenance suggestions. This invention significantly improves transportation efficiency, reduces operating costs, and ensures the safety and reliability of cargo transportation through precise data analysis and intelligent decision-making.

[0005] Existing technologies suffer from the following problems: Firstly, they only issue warnings when vehicles run out of energy or cargo malfunctions, lacking real-time capability and intelligence. They often only react after an accident has occurred, hindering early prevention and mitigation, potentially leading to significant losses. Secondly, current route optimization relies heavily on fixed-rule route planning, failing to dynamically integrate real-time traffic data, which can result in longer routes and increased travel time. Therefore, an effective real-time safety warning and route optimization system for cargo transportation is urgently needed for real-time safety monitoring and route optimization. Summary of the Invention

[0006] To address the aforementioned technical problems, the present invention aims to provide an AI-based real-time safety early warning and route optimization system for cargo transportation. This system includes the following modules: a first data acquisition module, a second data acquisition module, an evaluation and task allocation module, a prediction module, a vehicle early warning module, a supply route planning module, a cargo management module, and a feedback module. The modules described above are connected via wired and / or wireless means to enable data transmission between them. The first data acquisition module is responsible for acquiring real-time image data and cargo attribute data of the goods to be transported, and for performing target detection on the real-time image data to obtain detection results, and for identifying goods without abnormalities to be transported based on the detection results. The second data acquisition module is responsible for acquiring... At any given time, the data includes vehicle operation status, driving behavior risk coefficient, and road condition data for n types of transport vehicles. n is an integer greater than or equal to 1; Evaluation and task allocation module: Performs an initial comprehensive evaluation on n types of transport vehicles to obtain an initial comprehensive evaluation score, and performs initial task allocation on the transport vehicles based on the initial comprehensive evaluation score; Prediction module; Based on n types of transport vehicles after initial task allocation, construct a vehicle energy consumption prediction model to generate... The remaining safe driving range for each vehicle at any given time; Vehicle warning module; based on At any given moment, a warning message is generated regarding the remaining safe driving range for each vehicle. Supply route planning module; generates vehicle supply methods based on early warning information and replans vehicle routes; Cargo Management Module; Real-time Acquisition The system collects real-time image data of goods in transit and real-time image data of goods awaiting transport without any abnormalities from the first data acquisition module, and compares and analyzes the pixel values ​​to generate a cargo warning signal. Feedback module: Obtains vehicle operation status data after the completion of the transportation task and generates a vehicle transportation analysis report.

[0007] In a preferred embodiment, the vehicle operating status data includes: vehicle load, energy status data, vehicle location, tire pressure, component wear data, fault codes, vehicle maintenance data, etc.; the road condition status data includes: traffic flow, road segment traffic status, traffic light status, road surface condition, and weather condition.

[0008] In a preferred embodiment, the process of obtaining the driving behavior risk coefficient includes: Historical driving data for n types of vehicles is acquired, including: driving speed, acceleration, driving time, driving mileage, braking force, accelerator force, and sharp turning maneuvers. Noise and outliers are removed from the historical driving data, and driving behavior risk indicators are constructed, including: speed change rate, frequency of rapid acceleration, frequency of rapid deceleration, frequency of sharp turns, and average idling time. Weights are assigned to the driving behavior risk indicators using the analytic hierarchy process (AHP). The weighted driving behavior risk indicators are then transformed into feature vectors, and a feature coefficient matrix is ​​constructed. A behavioral risk coefficient model is built using a machine learning algorithm. The feature coefficient matrix is ​​input into the behavioral risk coefficient model for training, and the corresponding driving behavior risk coefficients are output. Real-time driving data acquired through the vehicle's onboard system is processed by constructing driving behavior risk indicators, assigning weights to the indicators, transforming them into feature vectors, and constructing a feature coefficient matrix. This data is then input into the behavioral risk coefficient model, and the driving behavior risk coefficients are output.

[0009] In a preferred embodiment, the method for obtaining the initial comprehensive evaluation score is as follows: Obtain vehicle inspection data for n types of transport vehicles, and construct an evaluation index system based on the vehicle inspection data. The evaluation index system includes: basic performance score, health status score, historical performance score, and adaptability score. The evaluation index system is then fitted and summed to obtain an initial comprehensive evaluation score, which can be specifically expressed as: ;in, This indicates the initial overall assessment score. This indicates the basic performance score. Indicates a health status score. Indicates the score for historical performance. This indicates the fit score.

[0010] In a preferred embodiment, the specific process of initial task allocation is as follows: The initial comprehensive evaluation scores of the n types of transport vehicles are sorted, and tasks are assigned to the vehicles with the highest initial comprehensive evaluation scores according to cargo attribute data and task priority. If any of the n types of transport vehicles exceeds the maximum number of tasks for the day, the task is skipped and the vehicle with the second highest initial comprehensive evaluation score is selected for task assignment.

[0011] In a preferred embodiment, generating The specific process for determining the remaining safe driving range of each vehicle at a given time includes: The system acquires and preprocesses historical data (i sets) of vehicle operating status, driving behavior risk coefficients, and road condition data for n types of transport vehicles, transforming them into i sets of corresponding feature vectors. A deep learning model is then constructed to predict vehicle energy consumption. These i sets of feature vectors serve as input to the model, which outputs the remaining safe driving mileage. The training objective is to minimize the loss function value of the vehicle energy consumption prediction model. Training stops when the loss function value is less than or equal to the preset target loss value. Based on the data from the second data acquisition module... Real-time data on vehicle operating status, driving behavior risk coefficients, and road conditions for n types of transport vehicles are acquired and then predicted using a vehicle energy consumption prediction model. The remaining safe driving range for each vehicle at any given time.

[0012] In a preferred embodiment, the specific process of generating early warning information includes: Get The distance of each vehicle from the mission destination and the distance from the supply station at any given time will be... At any given time, the remaining safe driving range of each vehicle and The distance of each vehicle from the mission destination and the location of the supply station at each time point are compared and analyzed; if At any given time, the remaining safe driving range of each vehicle is less than A warning message is generated when each vehicle is more than 100 kilometers from the mission destination than the distance to the supply station; if... At any given time, the remaining safe driving range of each vehicle is greater than or equal to No warning information is generated for the distance of each vehicle from the mission destination at any given time.

[0013] In a preferred embodiment, the process of generating the vehicle resupply method includes: If the vehicle is When a warning message is generated in real time, vehicle operation status data, remaining safe driving mileage for each vehicle, and supply station status data are acquired. The supply station status data includes: supply station location, utilization rate, operating hours, energy type and price, and supply speed. A supply recommendation and scoring model is constructed based on a decision tree. The vehicle operation status data, remaining safe driving mileage for each vehicle, and supply station status data are input into the supply recommendation and scoring model to obtain supply station score data, which is then sorted. Supply stations with high scores are recommended to transport vehicles, thus generating a vehicle supply method.

[0014] In a preferred embodiment, the specific process of replanning the route for a vehicle includes: Will The system uses the vehicle's location, supply station location, and mission destination as nodes, and vehicle operation status data, driving behavior risk coefficient, road condition data, and supply station status data as node attributes. It connects the nodes to construct a network graph structure, and uses a path planning algorithm to traverse the network graph structure. With constraints of shortest path, shortest time, minimum energy consumption, and minimum detour cost, it obtains the optimal subgraph structure for completing the mission, which is then used as the optimal path after replanning.

[0015] In a preferred embodiment, the specific process for generating a cargo warning signal is as follows: Real-time acquisition The system collects real-time image data of the goods being transported during transit and real-time image data of the goods awaiting transport without any abnormalities, obtained by the first data acquisition module. Pixel values ​​are then extracted from each image. The image pixel values ​​of real-time image data of goods being transported during transportation are compared with the image pixel values ​​of real-time image data of goods to be transported without any abnormalities, which are obtained by the first data acquisition module. If the difference is within the preset allowable error threshold range, no cargo warning signal is generated; if the difference is not within the preset allowable error threshold range, a cargo warning signal is generated.

[0016] Compared with the prior art, the beneficial effects of the present invention are: the present invention acquires real-time image data and cargo attribute data of the goods to be transported, performs target detection on the real-time image data to obtain detection results, and acquires goods to be transported without abnormalities based on the detection results; The system collects real-time data on vehicle operation status, driving behavior risk coefficients, and road conditions for n types of transport vehicles. An initial comprehensive evaluation is conducted on each of the n types of transport vehicles to obtain an initial comprehensive evaluation score. Based on this score, initial task assignments are performed on the transport vehicles. Finally, a vehicle energy consumption prediction model is constructed based on the n types of transport vehicles after the initial task assignments. At any given time, each vehicle has a remaining safe driving range; based on At any given time, the system generates warning information about the remaining safe driving range of each vehicle; based on this warning information, it generates vehicle resupply methods and replans vehicle routes; and it acquires real-time data. The system collects real-time image data of goods during transport and real-time image data of goods awaiting transport without abnormalities from the first data acquisition module, compares and analyzes pixel values ​​to generate a cargo warning signal; it also acquires vehicle operation status data after the transport mission is completed and generates a vehicle transport analysis report; it solves the problem of existing technologies that provide early warnings when vehicles run out of energy or cargo is abnormal, avoiding situations where the remaining safe driving range is insufficient to reach the supply station and mission destination, and has real-time and intelligent prediction capabilities; at the same time, it breaks away from traditional path planning that relies on fixed rules, and replans the path based on real-time conditions, improving vehicle transport efficiency and reducing transport costs. Attached Figure Description

[0017] Figure 1 This is a schematic diagram showing the connection of each module in the AI-based real-time safety warning and route optimization system for cargo transportation, as described in an embodiment of this application. Detailed Implementation

[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0019] It should be noted that, unless otherwise defined, the technical or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in one or more embodiments of the present invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0020] Please see Figure 1As shown, the AI-based real-time safety early warning and route optimization system for cargo transportation includes the following modules: a first data acquisition module, a second data acquisition module, an assessment and task allocation module, a prediction module, a vehicle early warning module, a supply route planning module, a cargo management module, and a feedback module. The modules described above are connected via wired and / or wireless means to enable data transmission between them. The first data acquisition module is responsible for acquiring real-time image data and cargo attribute data of the goods to be transported, and for performing target detection on the real-time image data to obtain detection results, and for identifying goods without abnormalities to be transported based on the detection results. Based on the above embodiments, the detection results include: abnormal goods to be transported and goods to be transported without abnormalities; The process of obtaining abnormal goods to be transported is as follows: the abnormal goods to be transported in the detection results obtained by the target detection are processed, including repackaging, etc., to obtain new abnormal goods to be transported. The new abnormal goods to be transported and the abnormal goods to be transported in the detection results obtained by the target detection are used as the abnormal goods to be transported. It should be noted that the real-time image data of the goods to be transported is captured from all directions by multiple high-definition cameras installed in the warehouse. This real-time image data is then sent to the first data acquisition module, where it undergoes normalization preprocessing. A deep learning object detection model is built using YOLO series algorithms such as YOLOv8 or Faster R-CNN. The normalized preprocessed real-time image data is input into the deep learning object detection model, which outputs information on abnormal and normal goods to be transported. This facilitates the determination of whether the goods are intact before transport. Abnormal goods, such as damaged packaging, can be promptly addressed, such as repackaging, to ensure that the goods are in a intact state before transport. Goods attribute data includes, but is not limited to, the weight, volume, type, and destination information of the goods, which can be automatically obtained through barcodes or RFID.

[0021] The second data acquisition module is responsible for acquiring... At any given time, the data includes vehicle operation status, driving behavior risk coefficient, and road condition data for n types of transport vehicles. n is an integer greater than or equal to 1; Based on the above embodiments, the vehicle operating status data includes, but is not limited to: vehicle load, energy status data, vehicle location, tire pressure, component wear data, fault codes, vehicle maintenance data, etc.; the road condition status data includes, but is not limited to: traffic flow, road segment traffic status, traffic light status, road surface condition, weather condition, etc. The process of obtaining driving behavior risk coefficients includes: acquiring historical driving data for n types of vehicles, including driving speed, acceleration, driving time, mileage, braking force, accelerator force, and sharp turning maneuvers; removing noise and outliers from the historical driving data and constructing driving behavior risk indicators, including speed change rate, frequency of rapid acceleration, frequency of rapid deceleration, frequency of sharp turning, and average idling time; assigning weights to the driving behavior risk indicators using the Analytic Hierarchy Process (AHP), converting the weighted driving behavior risk indicators into feature vectors and constructing a feature coefficient matrix; constructing a behavior risk coefficient model using a machine learning algorithm; inputting the feature coefficient matrix into the behavior risk coefficient model for training; and outputting the corresponding driving behavior risk coefficients; and finally, taking driving data acquired in real time through the vehicle's onboard system, constructing driving behavior risk indicators, assigning weights to indicators, converting them into feature vectors, constructing a feature coefficient matrix, inputting it into the behavior risk coefficient model, and outputting the driving behavior risk coefficients. It should be noted that the machine learning algorithms mentioned include decision trees, random forests, neural networks, etc.; in one possible embodiment, for ease of understanding, the feature coefficient matrix is ​​for example: [w1*rate of change of speed, w2*frequency of rapid acceleration, w3*frequency of rapid deceleration, w4*frequency of sharp turns, w5*average idling time], where w1, w2, w3, w4, and w5 are the weights of each driving behavior risk indicator; the vehicle operating status data and road condition status data are acquired by various sensors and monitoring devices built into the vehicle system; historical driving data is obtained from the vehicle computer database; The method for training a behavioral risk coefficient model includes: acquiring historical driving data and constructing driving behavior risk indicators and assigning indicator weights; constructing a classifier; converting the weighted driving behavior risk indicators into feature vectors and constructing a feature coefficient matrix, which is then input into the classifier; training the classifier; and outputting a classifier that meets a preset accuracy rate as the risk coefficient model; applying the risk coefficient model to a second data acquisition module to obtain driving behavior risk coefficients. The classifiers include decision trees, random forests, neural networks, and support vector machines.

[0022] Evaluation and task allocation module: Performs an initial comprehensive evaluation on n types of transport vehicles to obtain an initial comprehensive evaluation score, and performs initial task allocation on the transport vehicles based on the initial comprehensive evaluation score; Based on the above embodiments, the method for obtaining the initial comprehensive evaluation score is as follows: Vehicle inspection data for n types of transport vehicles are obtained; an evaluation index system is constructed based on the vehicle inspection data, wherein the evaluation index system includes: basic performance score, health status score, historical performance score, and adaptability score; the initial comprehensive evaluation score is obtained by fitting and summing the evaluation index system, which can be specifically expressed as: ;in, This indicates the initial overall assessment score. This indicates the basic performance score. Indicates a health status score. Indicates the score for historical performance. Indicates the fit score; The specific process of initial task allocation is as follows: sort the initial comprehensive evaluation scores of n types of transport vehicles, and allocate tasks to vehicles with higher initial comprehensive evaluation scores according to cargo attribute data and task priority. If any vehicle among the n types of transport vehicles exceeds the maximum number of tasks for the day, skip it and select the vehicle with the second-highest initial comprehensive evaluation score to allocate tasks. It should be noted that the basic performance score includes the sum of the comprehensive scores of secondary indicators such as maximum load capacity, maximum driving range, and maximum driving speed; the health status score includes the sum of the comprehensive scores of secondary indicators such as component wear, failure frequency, and maintenance completion rate; the historical performance score includes the sum of the comprehensive scores of secondary indicators such as mission on-time rate and accident rate; the adaptability score includes the sum of the comprehensive scores of secondary indicators such as energy type matching degree (e.g., the matching degree between vehicle energy type (fuel / electric / hybrid) and mission route refueling stations) and special function adaptability (e.g., the refrigeration capacity of refrigerated trucks, the explosion-proof level of hazardous chemical trucks, and other special attributes); and the vehicle inspection data includes: vehicle historical maintenance and failure data, vehicle basic attribute data, vehicle braking data, and wear data.

[0023] Prediction module; Based on n types of transport vehicles after initial task allocation, construct a vehicle energy consumption prediction model to generate... The remaining safe driving range for each vehicle at any given time; Based on the above embodiments, generate The specific process for determining the remaining safe driving mileage of each vehicle at any given time includes: acquiring and preprocessing historical data (i sets) of vehicle operating status, driving behavior risk coefficients, and road condition data for n types of transport vehicles, converting them into corresponding i sets of feature vectors; constructing a vehicle energy consumption prediction model using deep learning, taking the i sets of feature vectors as input to the model, and using the remaining safe driving mileage as output; minimizing the loss function value of the vehicle energy consumption prediction model as the training objective; stopping training when the loss function value of the vehicle energy consumption prediction model is less than or equal to the preset target loss value; and based on the data acquired by the second data acquisition module... Real-time data on vehicle operating status, driving behavior risk coefficients, and road conditions for n types of transport vehicles are acquired and then predicted using a vehicle energy consumption prediction model. The remaining safe driving range for each vehicle at any given time; It should be noted that the deep learning mentioned includes Long Short-Term Memory Networks, Convolutional Neural Networks, etc. The remaining safe driving range refers to the distance a vehicle can safely travel without an energy warning; the loss function value of the vehicle energy consumption prediction model is the mean squared error; mean squared error is one of the commonly used loss functions, which is obtained by using the loss function... The model is trained with the goal of minimizing the value, enabling the vehicle energy consumption prediction model to better fit the data, thereby improving the model's performance and accuracy; in the loss function... This represents the value of the loss function. This indicates the number of eigenvector groups. The group number representing the eigenvectors. Indicates the first The actual remaining safe driving mileage of each vehicle corresponding to the group feature vectors. Indicates the first The predicted remaining safe driving mileage for each vehicle corresponds to the group of feature vectors; other model parameters of the vehicle energy consumption prediction model, such as target loss value, optimization algorithm, ratio of training set, test set, and validation set, as well as optimization of loss function, are all obtained through actual engineering implementation and continuous experimental tuning.

[0024] Vehicle warning module; based on At any given moment, a warning message is generated regarding the remaining safe driving range for each vehicle. Based on the above embodiments, the specific process of generating early warning information includes: obtaining... The distance of each vehicle from the mission destination and the distance from the supply station at any given time will be... At any given time, the remaining safe driving range of each vehicle and The distance of each vehicle from the mission destination and the location of the supply station at each time point are compared and analyzed; if At any given time, the remaining safe driving range of each vehicle is less than A warning message is generated when each vehicle is more than 100 kilometers from the mission destination than the distance to the supply station; if... At any given time, the remaining safe driving range of each vehicle is greater than or equal to No warning information is generated regarding the distance of each vehicle from the mission destination at any given time; It should be noted that, The distance of each vehicle from its destination at any given time is obtained by the vehicle navigation system. The location at any given time and the destination of the mission are obtained by querying the cargo attribute data, from which the onboard system can calculate... This module measures the distance of each vehicle from the mission destination at any given time. It avoids situations where the predicted remaining safe mileage of a vehicle is insufficient to reach the supply station or mission destination. When predicting the remaining safe mileage of a vehicle at any given time, it simultaneously acquires the distance of each vehicle from the mission destination and the distance from the supply station, compares and analyzes these data, and generates early warning information.

[0025] Supply route planning module; generates vehicle supply methods based on early warning information and replans vehicle routes; Based on the above embodiments, the process of generating a vehicle resupply method includes: if the vehicle is in If a warning message is generated in real time, vehicle operation status data, remaining safe driving mileage for each vehicle, and supply station status data are obtained. The supply station status data includes: supply station location, utilization rate, operating hours, energy type and price, supply speed, etc. A supply recommendation and scoring model is constructed based on a decision tree. The vehicle operation status data, remaining safe driving mileage for each vehicle, and supply station status data are input into the supply recommendation and scoring model to obtain supply station scoring data, which is then sorted. Supply stations with high scores are recommended to transport vehicles, thus generating a vehicle supply method. It should be noted that the training process of the supply recommendation scoring model is as follows: Historical vehicle operating status data, remaining safe driving mileage for each vehicle, and supply station status data are acquired, converted into feature vectors, and a sample set is constructed. The sample set is then divided into a training set and a test set. A classifier is constructed by inputting the feature vectors from the training set into the classifier for training. The classifier is then tested using the test set, and the output classifier that meets the preset accuracy rate is used as the supply recommendation scoring model. The classifier is one of Naive Bayes, Temporal Neural Network, or Random Forest. Based on the above embodiments, the specific process of replanning the route for vehicles includes: Will The system uses the vehicle's location, supply station location, and mission destination as nodes, and vehicle operation status data, driving behavior risk coefficient, road condition data, and supply station status data as node attributes. It connects the nodes to construct a network graph structure, and uses a path planning algorithm to traverse the network graph structure. With constraints of shortest path, shortest time, minimum energy consumption, and minimum detour cost, it obtains the optimal subgraph structure for completing the mission, which is then used as the optimal path after replanning.

[0026] It should be noted that the path planning algorithms mentioned include, but are not limited to, Deep Q-Network (DQN) path planning algorithm, Bellman-Ford algorithm, etc.

[0027] Cargo Management Module; Real-time Acquisition The system collects real-time image data of goods in transit and real-time image data of goods awaiting transport without any abnormalities from the first data acquisition module, and compares and analyzes the pixel values ​​to generate a cargo warning signal. Based on the above embodiments, the specific process for generating cargo warning signals is as follows: real-time acquisition The system collects real-time image data of the goods being transported during transit and real-time image data of the goods awaiting transport without any abnormalities, obtained by the first data acquisition module. Pixel values ​​are then extracted from each image. The image pixel values ​​of real-time image data of goods being transported during transportation are compared with the image pixel values ​​of real-time image data of goods to be transported without any abnormalities, which are obtained by the first data acquisition module. If the difference is within the preset allowable error threshold range, no cargo warning signal is generated; if the difference is not within the preset allowable error threshold range, a cargo warning signal is generated. It should be noted that detecting cargo compression anomalies by comparing differences in image pixel values ​​is a non-contact monitoring method based on computer vision. Its core advantage lies in capturing subtle changes in the appearance of cargo using pixel-level quantitative analysis. Compared with traditional methods such as manual inspection and contact sensors, it has advantages such as high detection accuracy, capturing minute details, automation and efficiency, and reduced labor costs. The driver receives cargo warning signals and adjusts the driving mode or stops to adjust the cargo. The anomaly allowable error threshold is obtained by those skilled in the art through continuous experimentation based on historical transportation cases.

[0028] Feedback module: Obtains vehicle operation status data after the completion of the transportation task and generates a vehicle transportation analysis report.

[0029] It should be noted that the generated vehicle transportation analysis report includes a three-in-one analysis report of "vehicle operation performance + problem diagnosis + optimization suggestions". This analysis report can help drivers to clarify the direction of driving optimization for the next transportation task, maintenance personnel to predict faults in advance, dispatchers to optimize vehicle allocation, and management to grasp the overall health and economy of the fleet, ultimately reducing operating costs (maintenance costs, energy costs) and improving transportation safety.

[0030] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0031] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0032] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AI-based real-time safety early warning and route optimization system for cargo transportation, characterized in that: Includes the following steps: First data acquisition module; Responsible for acquiring real-time image data and cargo attribute data of goods to be transported, and performing target detection on the real-time image data to obtain detection results, and based on the detection results, identifying goods to be transported without any abnormalities; The second data acquisition module is responsible for acquiring... At any given time, the data includes vehicle operation status, driving behavior risk coefficient, and road condition data for n types of transport vehicles. n is an integer greater than or equal to 1; Assessment and task assignment module; An initial comprehensive evaluation is conducted on n types of transport vehicles to obtain an initial comprehensive evaluation score, and the transport vehicles are initially assigned tasks based on the initial comprehensive evaluation score; Prediction module; Based on n types of transport vehicles after initial task allocation, construct a vehicle energy consumption prediction model to generate... The remaining safe driving range for each vehicle at any given time; Vehicle warning module; based on At any given moment, a warning message is generated regarding the remaining safe driving range for each vehicle. Supply route planning module; generates vehicle supply methods based on early warning information and replans vehicle routes; Goods Management Module; Real-time acquisition The system collects real-time image data of goods in transit and real-time image data of goods awaiting transport without any abnormalities from the first data acquisition module, and compares and analyzes the pixel values ​​to generate a cargo warning signal. Feedback module: Obtains vehicle operation status data after the completion of the transportation task and generates a vehicle transportation analysis report.

2. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The vehicle operating status data includes: vehicle load, energy status data, vehicle location, tire pressure, component wear data, fault codes, vehicle maintenance data, etc.; the road condition status data includes: traffic flow, road segment traffic status, traffic light status, road surface condition, and weather condition.

3. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The process of obtaining the driving behavior risk factor includes: Historical driving data for n types of vehicles is acquired, including: driving speed, acceleration, driving time, driving mileage, braking force, accelerator force, and sharp turning maneuvers. Noise and outliers are removed from the historical driving data, and driving behavior risk indicators are constructed, including: speed change rate, frequency of rapid acceleration, frequency of rapid deceleration, frequency of sharp turns, and average idling time. Weights are assigned to the driving behavior risk indicators using the analytic hierarchy process (AHP). The weighted driving behavior risk indicators are then transformed into feature vectors, and a feature coefficient matrix is ​​constructed. A behavioral risk coefficient model is built using a machine learning algorithm. The feature coefficient matrix is ​​input into the behavioral risk coefficient model for training, and the corresponding driving behavior risk coefficients are output. Real-time driving data acquired through the vehicle's onboard system is processed by constructing driving behavior risk indicators, assigning weights to the indicators, transforming them into feature vectors, and constructing a feature coefficient matrix. This data is then input into the behavioral risk coefficient model, and the driving behavior risk coefficients are output.

4. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The method for obtaining the initial comprehensive evaluation score is as follows: Obtain vehicle inspection data for n types of transport vehicles, and construct an evaluation index system based on the vehicle inspection data. The evaluation index system includes: basic performance score, health status score, historical performance score, and adaptability score. The evaluation index system is then fitted and summed to obtain an initial comprehensive evaluation score, which can be specifically expressed as: ;in, This indicates the initial overall assessment score. This indicates the basic performance score. Indicates a health status score. Indicates the score for historical performance. This indicates the fit score.

5. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The specific process for initial task allocation is as follows: The initial comprehensive evaluation scores of the n types of transport vehicles are sorted, and tasks are assigned to the vehicles with the highest initial comprehensive evaluation scores according to cargo attribute data and task priority. If any of the n types of transport vehicles exceeds the maximum number of tasks for the day, the task is skipped and the vehicle with the second highest initial comprehensive evaluation score is selected for task assignment.

6. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, generate The specific process for determining the remaining safe driving range of each vehicle at a given time includes: The system acquires and preprocesses historical data (i sets) of vehicle operating status, driving behavior risk coefficients, and road condition data for n types of transport vehicles, transforming them into i sets of corresponding feature vectors. A deep learning model is then constructed to predict vehicle energy consumption. These i sets of feature vectors serve as input to the model, which outputs the remaining safe driving mileage. The training objective is to minimize the loss function value of the vehicle energy consumption prediction model. Training stops when the loss function value is less than or equal to the preset target loss value. Based on the data from the second data acquisition module... Real-time data on vehicle operation status, driving behavior risk coefficients, and road conditions for n types of transport vehicles are acquired at any given time and predicted using a vehicle energy consumption prediction model. The remaining safe driving range for each vehicle at any given time.

7. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The specific process of generating early warning information includes: Get The distance of each vehicle from the mission destination and the distance from the supply station at any given time will be... At any given time, the remaining safe driving range of each vehicle and The distance of each vehicle from the mission destination and the location of the supply station at each time point are compared and analyzed; if At any given time, the remaining safe driving range of each vehicle is less than A warning message is generated when each vehicle is more than 100 kilometers from the mission destination than the distance to the supply station; if... At any given time, the remaining safe driving range of each vehicle is greater than or equal to No warning information is generated for the distance of each vehicle from the mission destination at any given time.

8. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The process of generating a vehicle resupply method includes: If the vehicle is When a warning message is generated in real time, vehicle operation status data, remaining safe driving mileage for each vehicle, and supply station status data are acquired. The supply station status data includes: supply station location, utilization rate, operating hours, energy type and price, and supply speed. A supply recommendation and scoring model is constructed based on a decision tree. The vehicle operation status data, remaining safe driving mileage for each vehicle, and supply station status data are input into the supply recommendation and scoring model to obtain supply station score data, which is then sorted. Supply stations with high scores are recommended to transport vehicles, thus generating a vehicle supply method.

9. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The specific process of rerouting vehicles includes: Will The system uses the vehicle's location, supply station location, and mission destination as nodes, and vehicle operation status data, driving behavior risk coefficient, road condition data, and supply station status data as node attributes. It connects the nodes to construct a network graph structure, and uses a path planning algorithm to traverse the network graph structure. With constraints of shortest path, shortest time, minimum energy consumption, and minimum detour cost, it obtains the optimal subgraph structure for completing the mission, which is then used as the optimal path after replanning.

10. The AI-based real-time safety early warning and route optimization system for cargo transportation according to claim 1, characterized in that, The specific process for generating cargo early warning signals is as follows: Real-time acquisition The system collects real-time image data of the goods being transported during transit and real-time image data of the goods awaiting transport without any abnormalities, obtained by the first data acquisition module. Pixel values ​​are then extracted from each image. The image pixel values ​​of real-time image data of goods being transported during transportation are compared with the image pixel values ​​of real-time image data of goods to be transported without any abnormalities, which are obtained by the first data acquisition module. If the difference is within the preset allowable error threshold range, no cargo warning signal is generated; if the difference is not within the preset allowable error threshold range, a cargo warning signal is generated.

Images