Intelligent vehicle management method and system based on multi-dimensional data perception and dynamic analysis

Abstract

The invention discloses a vehicle intelligent management method and system based on multi-dimensional data perception and dynamic analysis, and aims to solve the problems of lag distortion of vehicle management data, stiffness of an assessment mechanism, passive safety management and information isolated island of an existing aluminum processing enterprise. The system comprises a vehicle-mounted intelligent terminal, a cloud management platform and a user interaction terminal, a vehicle management system is constructed through an integrated scheme of hardware sensing, a software platform and an intelligent algorithm, and the management method comprises dynamic profit assessment, intelligent salary commission, predictive maintenance, dynamic fuel consumption standard management, active safety management and cargo loading intelligent check. Real-time data acquisition and automatic analysis are realized, management is upgraded from experience driving to data and algorithm driving, post-processing is converted into beforehand early warning and in-process intervention, multi-system data flow is broken through, accurate transportation cost management and control, transportation safety beforehand prevention and transportation capacity resource optimization configuration are realized, and the transportation efficiency is improved. And the management efficiency and the decision scientificity are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of transportation vehicle management technology, specifically to a vehicle intelligent management method and system based on multi-dimensional data perception and dynamic analysis. Background Technology

[0002] Transport vehicles are a crucial link connecting production, warehousing, and sales in aluminum processing enterprises, and their management efficiency directly impacts the company's transportation costs, cargo safety, and market responsiveness. Currently, existing vehicle management methods in aluminum processing enterprises largely rely on manual recording, static assessment, and post-event analysis, which have several significant drawbacks:

[0003] Data is delayed and distorted: Core data such as vehicle fuel consumption, mileage, and driving behavior mainly rely on manual reporting by drivers. This not only results in data delays but also makes it easy for false or concealed reports to occur, making it difficult to guarantee data accuracy and leading to a lack of reliable data support for management decisions.

[0004] The assessment mechanism is static and rigid: assessment indicators such as fuel consumption and maintenance costs are mostly fixed quotas, without taking into account dynamic factors such as real-time vehicle operating conditions (such as engine load), road conditions (such as slope and congestion), and weather conditions. The assessment results lack scientificity and rationality, and it is difficult to effectively incentivize drivers to optimize their driving behavior.

[0005] Passive safety management: The existing management model mainly relies on post-incident investigation to monitor unsafe behaviors such as fatigued driving and dangerous driving. It cannot achieve real-time early warning and intervention, resulting in weak accident prevention capabilities and easy to cause safety accidents such as cargo damage and personal injury, which will bring economic losses to enterprises.

[0006] Information silos: The vehicle management system, dispatching system, financial system, warehousing system and other business systems are independent of each other and data cannot be shared. This leads to poor coordination in the planning, execution tracking and cost accounting of transportation tasks, resulting in low management efficiency and difficulty in achieving closed-loop management of the entire process.

[0007] Therefore, there is an urgent need for a comprehensive, data-driven, and dynamically optimized intelligent vehicle management method and system to address the aforementioned shortcomings of existing technologies. Summary of the Invention

[0008] The purpose of this invention is to provide a vehicle intelligent management method and system based on multi-dimensional data perception and dynamic analysis, so as to achieve precise control of transportation costs, pre-emptive prevention of transportation safety and optimized allocation of transportation resources, thereby improving the level of intelligent management and operational efficiency of transportation vehicles in aluminum processing enterprises.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis includes:

[0011] The vehicle-mounted intelligent terminal is installed in each transport vehicle and integrates a GPS module, IoT sensors, ADAS driver assistance camera and DMS driver monitoring system to collect real-time data on vehicle location, fuel consumption, fuel tank level, engine condition, driving behavior, driver status and driving environment.

[0012] The cloud management platform communicates with the in-vehicle intelligent terminal to receive, store, and process various types of data uploaded by the in-vehicle intelligent terminal. It is also associated with a fee database, which stores toll standards, subsidy standards, and commission rate rules. The cloud management platform is equipped with intelligent analysis algorithm modules, including a dynamic fuel consumption benchmark model, a maintenance prediction model, a safety risk profiling model, and a profit calculation model.

[0013] User interaction terminals, including PC and mobile apps, communicate with the cloud management platform, allowing managers, drivers, and financial personnel to view data, receive alerts, and process business.

[0014] Furthermore, IoT sensors include fuel consumption sensors, fuel level sensors, engine condition sensors, and acceleration sensors, which are used to monitor real-time fuel consumption, remaining fuel in the tank, engine operating load and duration, and driving behavior data such as rapid acceleration / sudden braking, respectively; ADAS driver assistance cameras are used to monitor lane departure, forward collision risk, and road conditions; and DMS driver monitoring systems are used to monitor driver fatigue states such as yawning and closing eyes, as well as distracted driving behaviors such as making phone calls and smoking.

[0015] Intelligent vehicle management methods based on multi-dimensional data perception and dynamic analysis include:

[0016] The methods include: an automated profit calculation method for the entire transportation task process; an intelligent wage commission method; a maintenance cost management method based on predictive maintenance; a dynamic fuel consumption benchmark generation and reward / penalty method based on big data learning; a proactive safety risk warning and scoring method integrating ADAS / DMS; and an automated verification method for aluminum loading compliance based on image recognition.

[0017] Furthermore, the method for automatically calculating profits throughout the entire transportation task process includes the following steps:

[0018] S11: Collects actual mileage data of the current transportation task through the GPS module of the vehicle-mounted intelligent terminal, collects actual fuel consumption data through IoT sensors, and obtains toll data through the ETC system interface;

[0019] S12: The cloud management platform connects to the expense database and extracts the subsidy standards and cost accounting parameters corresponding to the current transportation task;

[0020] S13: Calculate the dynamic profit of this transportation task using the profit accounting model. Dynamic profit = transportation revenue - fuel cost - toll fees - subsidies - other fixed costs.

[0021] S14: Real-time synchronization of dynamic profit data to the driver's app on the user interaction terminal enables real-time visualization of performance; S45: Financial personnel verify the data through the user interaction terminal and automatically generate periodic profit reports for each vehicle and each driver, providing data support for management decisions.

[0022] Furthermore, the intelligent payroll commission method includes the following steps:

[0023] S21: The cloud management platform automatically matches the corresponding basic commission rate based on the approved transportation task route;

[0024] S22: Collect data on the on-time rate, cargo integrity rate, and fuel efficiency of transportation tasks through the system, and calculate the comprehensive efficiency coefficient. The comprehensive efficiency coefficient = α × on-time rate + β × cargo integrity rate + γ × fuel efficiency, where α, β, and γ are weighting coefficients, and α + β + γ = 1.

[0025] S23: Calculate the actual commission amount based on the basic commission rate and the comprehensive efficiency coefficient. Actual commission amount = basic commission × comprehensive efficiency coefficient;

[0026] S24: Automatically summarizes the commission amount of all transportation tasks for each driver within the cycle, combines it with basic salary and bonus / penalty amount to generate a payslip, and pushes it to the driver and finance personnel for verification through the user interaction terminal.

[0027] Furthermore, the predictive maintenance-based maintenance cost management method includes the following steps:

[0028] S31: Real-time collection of vehicle engine operating data via in-vehicle intelligent terminal, including operating time, operating load, fault codes and historical maintenance records;

[0029] S32: Establish an electronic digital maintenance file for each vehicle, recording maintenance data, fault data, and operational data throughout the vehicle's entire lifecycle;

[0030] S33: Input engine operating data into the maintenance prediction model. The maintenance prediction model is a model trained based on machine learning algorithms, used to predict the potential types of vehicle faults, the probability of fault occurrence, and the optimal maintenance cycle.

[0031] S34: Automatically generate maintenance suggestions based on the prediction results, including maintenance items, maintenance time and maintenance priority, and push them to administrators through the user interaction terminal;

[0032] S35: Compile statistics on actual vehicle repair costs and conduct deviation analysis between the actual repair costs and the predicted costs. Based on the deviation values, establish a reward and penalty mechanism. If the actual repair costs are lower than the predicted range, a reward will be given; if they exceed the predicted range without a reasonable explanation, a penalty will be imposed.

[0033] Furthermore, in S33, the machine learning algorithm is either the random forest algorithm or the support vector machine algorithm, and the model training data includes historical fault data, maintenance data and engine operation data of similar vehicles, and the model prediction accuracy is not less than 85%.

[0034] Furthermore, the generation and reward / penalty method for dynamic fuel consumption benchmarks based on big data learning includes the following steps:

[0035] S41: Collects massive amounts of historical transportation data through in-vehicle intelligent terminals. Historical transportation data includes vehicle information, route information, vehicle weight information, weather information, road condition information, and actual fuel consumption data.

[0036] S42: Preprocess historical transportation data to remove outlier data and obtain valid sample data;

[0037] S43: Based on valid sample data, a dynamic fuel consumption benchmark model is constructed through regression analysis to dynamically generate personalized fuel consumption benchmark values ​​for each vehicle, each route, different seasons, and different weather conditions;

[0038] S44: Collects real-time actual fuel consumption data for the current transportation task and compares it with the corresponding personalized fuel consumption benchmark value;

[0039] S45: The reward and penalty amounts are automatically calculated based on the comparison results. If the actual fuel consumption is lower than the benchmark value, a reward is given; if it is higher than the benchmark value, a penalty is imposed. The results are then synchronized to the driver and management personnel through the user interaction terminal.

[0040] Furthermore, in S43, the regression analysis model is a multiple linear regression model or a gradient boosting regression model. The model input parameters include route length, route gradient, vehicle weight, average vehicle speed, temperature, and rainfall. The output parameter is a personalized fuel consumption benchmark value. The fuel consumption benchmark value is updated and optimized every quarter based on newly added historical data.

[0041] Furthermore, the proactive safety risk warning and scoring method integrating ADAS / DMS includes the following steps:

[0042] S51: Real-time collection of driving behavior data and driver status data during driving through ADAS driver assistance camera and DMS driver monitoring system. Dangerous driving behaviors include following too closely, sharp turns, rapid acceleration and sudden braking.

[0043] S52: Set safety risk assessment thresholds, including thresholds for fatigued driving, distracted driving, and dangerous driving behavior.

[0044] S53: The real-time collected data is compared with the corresponding judgment threshold. If the threshold is triggered, a voice warning is issued to the driver through the vehicle intelligent terminal, and the warning information is uploaded to the cloud management platform.

[0045] S54: Based on the safety risk profile model, a dynamic safe driving score is generated for each driver according to their historical warning records, number of dangerous driving behaviors, duration of fatigue driving, and safety training. The score range is 0-100.

[0046] S55: Use safe driving scores as the core basis for awarding safety prizes and evaluating performance, and provide targeted training programs for drivers whose scores are below the preset passing line.

[0047] Furthermore, in S54, the safety risk profiling model uses the hierarchical analysis method to determine the weight of each evaluation indicator, which includes the number of warnings, the frequency of dangerous behaviors, the duration of fatigued driving, and the completion status of training.

[0048] Furthermore, the automatic verification method for aluminum loading compliance based on image recognition includes the following steps:

[0049] S61: Install high-definition surveillance cameras at the loading point. The cameras should cover the entire loading area and have a resolution of no less than 1080P.

[0050] S62: Collect image data after the cargo loading is completed through monitoring cameras, including images of the number of aluminum ingot bundles, images of tarpaulin covering, and images of cargo bundling;

[0051] S63: Input the image data into the trained image recognition model. The image recognition model is a convolutional neural network model based on deep learning algorithms, which is used to automatically identify whether the number of aluminum ingot bundles meets the order requirements, whether the tarpaulin is properly covered, and whether the goods are bundled in a standardized manner.

[0052] S64: If the identification result is non-compliant, an early warning will be issued to the management personnel through the cloud management platform to notify the on-site personnel to make rectifications; if the identification result is compliant, a loading compliance record will be generated and stored in the transportation task file.

[0053] Furthermore, in S63, the training data for the image recognition model includes compliant and non-compliant images of aluminum loading, with a training sample size of no less than 10,000 images and a model recognition accuracy of no less than 90%.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows: The vehicle intelligent management method and system based on multi-dimensional data perception and dynamic analysis have the following advantages:

[0055] 1. Precision: Real-time and accurate multi-dimensional data is automatically collected through the vehicle's intelligent terminal, eliminating the data distortion problem caused by manual reporting. All assessments and calculations are based on objective data, making management decisions more precise. The personalized setting of dynamic fuel consumption benchmark values ​​ensures the fairness of fuel consumption assessments, and automatic profit calculation enables precise control of transportation costs.

[0056] 2. Intelligentization: The introduction of intelligent algorithm models such as predictive maintenance, dynamic fuel consumption benchmark, and driving behavior recognition upgrades vehicle management from traditional experience-driven to data and algorithm-driven. The maintenance prediction model can provide early warning of potential faults and reduce the incidence of sudden faults by more than 30%; the dynamic fuel consumption benchmark model improves the rationality of fuel consumption assessment by 50%.

[0057] 3. Proactive approach: Moving the nodes of safety management and maintenance management from "post-event handling" to "pre-event warning and in-event intervention". For example, proactive safety warnings can reduce the incidence of dangerous driving behaviors by more than 40%, and predictive maintenance can reduce maintenance costs by 20%-30%, significantly improving transportation safety and management efficiency.

[0058] 4. Integration: It connects the data flow between vehicles, personnel, tasks, and costs, and realizes the coordinated linkage of vehicle management, dispatch management, financial management, and warehouse management. It solves the problem of information silos and forms a closed-loop optimized intelligent management system for the entire transportation process, improving management efficiency by more than 40%. Detailed Implementation

[0059] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] A vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis includes:

[0061] The in-vehicle intelligent terminal, installed in each transport vehicle, serves as a data collection and real-time interaction terminal device, integrating a GPS module, IoT sensors, ADAS (Advanced Driver Assistance System) cameras, and a DMS (Driver Monitoring System). The GPS module accurately collects vehicle location and mileage data, with a positioning accuracy of at least 10 meters. The IoT sensors include fuel consumption sensors, fuel level sensors, engine condition sensors, and acceleration sensors, used to monitor real-time vehicle fuel consumption, fuel level, engine load and operating time, and driving behaviors such as rapid acceleration / braking. The ADAS cameras monitor lane departure, forward collision risk, and road conditions. The DMS monitors driver fatigue and distracted driving behaviors, including but not limited to yawning and closing eyes, and distracted driving behaviors including but not limited to making phone calls, horseplay, and smoking.

[0062] The cloud management platform, serving as the core data processing and storage center of the system, adopts a distributed architecture design and possesses high concurrency processing capabilities and data security features. It communicates with in-vehicle intelligent terminals to receive, store, and process various types of data uploaded by these terminals, including basic data such as vehicle information, driver information, transportation tasks, and fee standards, as well as historical operational data. It provides data support for the intelligent analysis algorithm module, pushes processing results and warning information to user interaction terminals, and is linked to a fee database that stores toll standards, subsidy standards, and commission rate rules.

[0063] The cloud management platform is equipped with intelligent analysis algorithm modules, including a dynamic fuel consumption benchmark model, a maintenance prediction model, a safety risk profiling model, and a profit calculation model. The dynamic fuel consumption benchmark model, built on regression analysis algorithms, generates personalized fuel consumption benchmark values. The maintenance prediction model, trained using machine learning algorithms, predicts potential vehicle malfunctions and maintenance cycles. The safety risk profiling model employs hierarchical analysis and statistical analysis algorithms to generate driver safety scores. The profit calculation model, based on a preset cost accounting formula, automatically calculates the profit for transportation tasks.

[0064] User interaction terminals, including PC and mobile app, communicate with the cloud management platform, allowing managers, drivers, and finance personnel to view data, receive alerts, and process business. This caters to the needs of different user roles. Managers can view vehicle operating status, transportation task progress, alerts, and data reports via PC or mobile app for dispatch management, maintenance plan development, and performance evaluation and rewards / penalties. Drivers can view real-time performance, transportation task information, safety alerts, and wage commission details via the mobile app. Finance personnel can verify transportation cost and profit data via PC and generate payroll and financial statements.

[0065] The vehicle intelligent management method based on multi-dimensional data perception and dynamic analysis includes six methods:

[0066] The methods include: an automated profit calculation method for the entire transportation task process; an intelligent wage commission method; a maintenance cost management method based on predictive maintenance; a dynamic fuel consumption benchmark generation and reward / penalty method based on big data learning; a proactive safety risk warning and scoring method integrating ADAS / DMS; and an automated verification method for aluminum loading compliance based on image recognition.

[0067] The automatic profit calculation method for the entire transportation task process includes the following steps:

[0068] S11: Collects actual mileage data of the current transportation task through the GPS module of the vehicle-mounted intelligent terminal, collects actual fuel consumption data through IoT sensors, and obtains toll data through the ETC system interface;

[0069] S12: The cloud management platform connects to the expense database and extracts the subsidy standards and cost accounting parameters corresponding to the current transportation task;

[0070] S13: Calculate the dynamic profit of this transportation task using the profit accounting model. Dynamic profit = transportation revenue - fuel cost - toll fees - subsidies - other fixed costs.

[0071] S14: Real-time synchronization of dynamic profit data to the driver's app on the user interaction terminal enables real-time visualization of performance; S45: Financial personnel verify the data through the user interaction terminal and automatically generate periodic profit reports for each vehicle and each driver, providing data support for management decisions.

[0072] The intelligent salary commission method includes the following steps:

[0073] S21: The cloud management platform automatically matches the corresponding basic commission rate based on the approved transportation task route;

[0074] S22: Collect data on the on-time rate, cargo integrity rate, and fuel efficiency of transportation tasks through the system, and calculate the comprehensive efficiency coefficient. The comprehensive efficiency coefficient = α × on-time rate + β × cargo integrity rate + γ × fuel efficiency, where α, β, and γ are weighting coefficients, and α + β + γ = 1.

[0075] S23: Calculate the actual commission amount based on the basic commission rate and the comprehensive efficiency coefficient. Actual commission amount = basic commission × comprehensive efficiency coefficient;

[0076] S24: Automatically summarizes the commission amount of all transportation tasks for each driver within the cycle, combines it with basic salary and bonus / penalty amount to generate a payslip, and pushes it to the driver and finance personnel for verification through the user interaction terminal.

[0077] The maintenance cost management method based on predictive maintenance includes the following steps:

[0078] S31: Real-time collection of vehicle engine operating data via in-vehicle intelligent terminal, including operating time, operating load, fault codes and historical maintenance records;

[0079] S32: Establish an electronic digital maintenance file for each vehicle, recording maintenance data, fault data, and operational data throughout the vehicle's entire lifecycle;

[0080] S33: Input engine operating data into the maintenance prediction model. The maintenance prediction model is a model trained based on machine learning algorithms, used to predict potential vehicle fault types, fault occurrence probabilities, and optimal maintenance cycles. The machine learning algorithm is either a random forest algorithm or a support vector machine algorithm. The model training data includes historical fault data, maintenance data, and engine operating data of similar vehicles. The model prediction accuracy is no less than 85%.

[0081] S34: Automatically generate maintenance suggestions based on the prediction results, including maintenance items, maintenance time and maintenance priority, and push them to administrators through the user interaction terminal;

[0082] S35: Compile statistics on actual vehicle repair costs and conduct deviation analysis between the actual repair costs and the predicted costs. Based on the deviation values, establish a reward and penalty mechanism. If the actual repair costs are lower than the predicted range, a reward will be given; if they exceed the predicted range without a reasonable explanation, a penalty will be imposed.

[0083] The method for generating and rewarding / penalizing dynamic fuel consumption benchmarks based on big data learning includes the following steps:

[0084] S41: Collects massive amounts of historical transportation data through in-vehicle intelligent terminals. Historical transportation data includes vehicle information, route information, vehicle weight information, weather information, road condition information, and actual fuel consumption data.

[0085] S42: Preprocess historical transportation data to remove outlier data and obtain valid sample data;

[0086] S43: Based on valid sample data, a dynamic fuel consumption benchmark model is constructed through regression analysis. The regression analysis model is either a multiple linear regression model or a gradient boosting regression model, which dynamically generates personalized fuel consumption benchmark values ​​for each vehicle, each route, different seasons, and different weather conditions. The model input parameters include route length, route gradient, vehicle weight, average vehicle speed, temperature, and rainfall, and the output parameter is the personalized fuel consumption benchmark value. The fuel consumption benchmark value is updated and optimized quarterly based on newly added historical data.

[0087] S44: Collects real-time actual fuel consumption data for the current transportation task and compares it with the corresponding personalized fuel consumption benchmark value;

[0088] S45: The reward and penalty amounts are automatically calculated based on the comparison results. If the actual fuel consumption is lower than the benchmark value, a reward is given; if it is higher than the benchmark value, a penalty is imposed. The results are then synchronized to the driver and management personnel through the user interaction terminal.

[0089] The proactive safety risk warning and scoring method integrating ADAS / DMS includes the following steps:

[0090] S51: Real-time collection of driving behavior data and driver status data during driving through ADAS driver assistance camera and DMS driver monitoring system. Dangerous driving behaviors include following too closely, sharp turns, rapid acceleration and sudden braking.

[0091] S52: Set safety risk assessment thresholds, including thresholds for fatigued driving, distracted driving, and dangerous driving behavior.

[0092] S53: The real-time collected data is compared with the corresponding judgment threshold. If the threshold is triggered, a voice warning is issued to the driver through the vehicle intelligent terminal, and the warning information is uploaded to the cloud management platform.

[0093] S54: Based on the safety risk profile model, a dynamic safe driving score is generated for each driver according to their historical warning records, number of dangerous driving behaviors, duration of fatigued driving, and safety training status. The score range is 0-100 points. The safety risk profile model uses the hierarchical analysis method to determine the weight of each evaluation indicator. The evaluation indicators include the number of warnings, frequency of dangerous behaviors, duration of fatigued driving, and training completion status.

[0094] S55: Use safe driving scores as the core basis for awarding safety prizes and evaluating performance, and provide targeted training programs for drivers whose scores are below the preset passing line.

[0095] The automatic verification method for compliance of aluminum loading based on image recognition includes the following steps:

[0096] S61: Install high-definition surveillance cameras at the loading point. The cameras should cover the entire loading area and have a resolution of no less than 1080P.

[0097] S62: Collect image data after the cargo loading is completed through monitoring cameras, including images of the number of aluminum ingot bundles, images of tarpaulin covering, and images of cargo bundling;

[0098] S63: Input the image data into the trained image recognition model. The image recognition model is a convolutional neural network model based on deep learning algorithms, which is used to automatically identify whether the number of aluminum ingot bundles meets the order requirements, whether the tarpaulin is properly covered, and whether the goods are bundled in a standardized manner.

[0099] S64: If the identification result is non-compliant, an early warning will be issued to the management personnel through the cloud management platform to notify the on-site personnel to make rectifications; if the identification result is compliant, a loading compliance record will be generated and stored in the transportation task file.

[0100] The operation flow of the vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis of this invention includes the following steps:

[0101] 1. Data collection phase: The vehicle-mounted intelligent terminal collects data such as vehicle location, gas consumption, fuel level, engine condition, driving behavior, and driver status in real time at a frequency of 10 times per minute, and uploads it to the cloud management platform via 4G / 5G network;

[0102] 2. Data Processing and Analysis Stage: After receiving the data, the cloud management platform performs data cleaning and standardization, and then inputs the processed data into the intelligent analysis algorithm module.

[0103] The dynamic fuel consumption benchmark model generates personalized fuel consumption benchmark values ​​based on parameters such as the route, vehicle weight, and weather of the current transportation task.

[0104] The maintenance prediction model analyzes engine operating data to predict potential faults and maintenance cycles;

[0105] The safety risk profiling model analyzes driving behavior and driver status data in real time, generates warning information and calculates a safe driving score when a threshold is triggered.

[0106] The profit calculation model combines collected mileage, gas consumption, toll data and parameters from the cost database to calculate dynamic profit;

[0107] Image recognition models analyze image data collected by cameras at loading points to determine loading compliance;

[0108] 3. Results Output and Feedback Stage: The cloud management platform pushes the analysis results to the corresponding user interaction terminals. The analysis results include dynamic profits, commission amounts, maintenance suggestions, safety warnings, and loading compliance results. Drivers receive warning prompts and performance data, managers receive dispatch information, maintenance suggestions, and report data, and finance personnel receive payroll and financial data, forming a closed-loop management system.

[0109] The above are merely embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis, characterized in that, include: The vehicle-mounted intelligent terminal is installed in each transport vehicle and integrates a GPS module, IoT sensors, ADAS driver assistance camera and DMS driver monitoring system to collect real-time data on vehicle location, fuel consumption, fuel tank level, engine condition, driving behavior, driver status and driving environment. The cloud management platform communicates with the in-vehicle intelligent terminal to receive, store, and process various types of data uploaded by the in-vehicle intelligent terminal. It is also associated with a fee database, which stores toll standards, subsidy standards, and commission rate rules. The cloud management platform is equipped with intelligent analysis algorithm modules, including a dynamic fuel consumption benchmark model, a maintenance prediction model, a safety risk profiling model, and a profit calculation model. User interaction terminals, including PC and mobile apps, communicate with the cloud management platform, allowing managers, drivers, and financial personnel to view data, receive alerts, and process business.

2. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 1, characterized in that, The Internet of Things (IoT) sensors include fuel consumption sensors, fuel level sensors, engine condition sensors, and acceleration sensors, which are used to monitor real-time fuel consumption, remaining fuel in the tank, engine operating load and duration, and rapid acceleration / braking data, respectively; ADAS (Advanced Driver Assistance Systems) cameras are used to monitor lane departure, forward collision risk, and road conditions; and the DMS (Driver Monitoring System) is used to monitor driver fatigue and distracted driving behavior.

3. A vehicle intelligent management method based on multi-dimensional data perception and dynamic analysis, characterized in that: include: The methods include: an automated profit calculation method for the entire transportation task process; an intelligent wage commission method; a maintenance cost management method based on predictive maintenance; a dynamic fuel consumption benchmark generation and reward / penalty method based on big data learning; a proactive safety risk warning and scoring method integrating ADAS / DMS; and an automated verification method for aluminum loading compliance based on image recognition.

4. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 3, characterized in that, The method for automatically calculating profits throughout the entire transportation task process includes the following steps: S11: Collects actual mileage data of the current transportation task through the GPS module of the vehicle-mounted intelligent terminal, collects actual fuel consumption data through IoT sensors, and obtains toll data through the ETC system interface; S12: The cloud management platform connects to the expense database and extracts the subsidy standards and cost accounting parameters corresponding to the current transportation task; S13: Calculate the dynamic profit of this transportation task using the profit accounting model. Dynamic profit = transportation revenue - fuel cost - toll fees - subsidies - other fixed costs. S14: Real-time synchronization of dynamic profit data to the driver's app on the user interaction terminal enables real-time visualization of performance; S45: Financial personnel verify the data through the user interaction terminal and automatically generate periodic profit reports for each vehicle and each driver, providing data support for management decisions.

5. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 3, characterized in that, The intelligent payroll commission method includes the following steps: S21: The cloud management platform automatically matches the corresponding basic commission rate based on the approved transportation task route; S22: Collect data on the on-time rate, cargo integrity rate, and fuel efficiency of transportation tasks through the system, and calculate the comprehensive efficiency coefficient. The comprehensive efficiency coefficient = α × on-time rate + β × cargo integrity rate + γ × fuel efficiency, where α, β, and γ are weighting coefficients, and α + β + γ = 1. S23: Calculate the actual commission amount based on the basic commission rate and the comprehensive efficiency coefficient. Actual commission amount = basic commission × comprehensive efficiency coefficient; S24: Automatically summarizes the commission amount of all transportation tasks for each driver within the cycle, combines it with basic salary and bonus / penalty amount to generate a payslip, and pushes it to the driver and finance personnel for verification through the user interaction terminal.

6. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 3, characterized in that, The maintenance cost management approach based on predictive maintenance includes the following steps: S31: Real-time collection of vehicle engine operating data via in-vehicle intelligent terminal, including operating time, operating load, fault codes and historical maintenance records; S32: Establish an electronic digital maintenance file for each vehicle, recording maintenance data, fault data, and operational data throughout the vehicle's entire lifecycle; S33: Input engine operating data into the maintenance prediction model. The maintenance prediction model is a model trained based on machine learning algorithms, used to predict the potential types of vehicle faults, the probability of fault occurrence, and the optimal maintenance cycle. S34: Automatically generate maintenance suggestions based on the prediction results, including maintenance items, maintenance time and maintenance priority, and push them to administrators through the user interaction terminal; S35: Compile statistics on actual vehicle repair costs and conduct deviation analysis between the actual repair costs and the predicted costs. Based on the deviation values, establish a reward and penalty mechanism. If the actual repair costs are lower than the predicted range, a reward will be given; if they exceed the predicted range without a reasonable explanation, a penalty will be imposed.

7. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 6, characterized in that, In S33, the machine learning algorithm is either the random forest algorithm or the support vector machine algorithm. The model training data includes historical fault data, maintenance data and engine operation data of similar vehicles. The model prediction accuracy is no less than 85%.

8. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 3, characterized in that, The method for generating and rewarding / penalizing dynamic fuel consumption benchmarks based on big data learning includes the following steps: S41: Collects massive amounts of historical transportation data through in-vehicle intelligent terminals. Historical transportation data includes vehicle information, route information, vehicle weight information, weather information, road condition information, and actual fuel consumption data. S42: Preprocess historical transportation data to remove outlier data and obtain valid sample data; S43: Based on valid sample data, a dynamic fuel consumption benchmark model is constructed through regression analysis to dynamically generate personalized fuel consumption benchmark values ​​for each vehicle, each route, different seasons, and different weather conditions; S44: Collects real-time actual fuel consumption data for the current transportation task and compares it with the corresponding personalized fuel consumption benchmark value; S45: The reward and penalty amounts are automatically calculated based on the comparison results. If the actual fuel consumption is lower than the benchmark value, a reward is given; if it is higher than the benchmark value, a penalty is imposed. The results are then synchronized to the driver and management personnel through the user interaction terminal.

9. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 3, characterized in that, The proactive safety risk warning and scoring method integrating ADAS / DMS includes the following steps: S51: Real-time collection of driving behavior data and driver status data during driving through ADAS driver assistance camera and DMS driver monitoring system. Dangerous driving behaviors include following too closely, sharp turns, rapid acceleration and sudden braking. S52: Set safety risk assessment thresholds, including thresholds for fatigued driving, distracted driving, and dangerous driving behavior. S53: The real-time collected data is compared with the corresponding judgment threshold. If the threshold is triggered, a voice warning is issued to the driver through the vehicle intelligent terminal, and the warning information is uploaded to the cloud management platform. S54: Based on the safety risk profile model, a dynamic safe driving score is generated for each driver according to their historical warning records, number of dangerous driving behaviors, duration of fatigue driving, and safety training. The score range is 0-100. S55: Use safe driving scores as the core basis for awarding safety prizes and evaluating performance, and provide targeted training programs for drivers whose scores are below the preset passing line.

10. The vehicle intelligent management system based on multi-dimensional data perception and dynamic analysis according to claim 3, characterized in that, An automated verification method for compliance of aluminum loading based on image recognition includes the following steps: S61: Install high-definition surveillance cameras at the loading point. The cameras should cover the entire loading area and have a resolution of no less than 1080P. S62: Collect image data after the cargo loading is completed through monitoring cameras, including images of the number of aluminum ingot bundles, images of tarpaulin covering, and images of cargo bundling; S63: Input the image data into the trained image recognition model. The image recognition model is a convolutional neural network model based on deep learning algorithms, which is used to automatically identify whether the number of aluminum ingot bundles meets the order requirements, whether the tarpaulin is properly covered, and whether the goods are bundled in a standardized manner. S64: If the identification result is non-compliant, an early warning will be issued to the management personnel through the cloud management platform to notify the on-site personnel to make rectifications; if the identification result is compliant, a loading compliance record will be generated and stored in the transportation task file.