Commercial vehicle intelligent energy consumption management and energy-saving optimization method and system

Abstract

The invention provides an intelligent energy consumption management and energy-saving optimization method and system for a commercial vehicle, and the method comprises the steps: collecting the multi-source data of vehicles, road conditions, weather and the like, carrying out the real-time reasoning of an edge end and the modeling calibration of a cloud end, constructing a multi-target route optimization algorithm, and generating a personalized ecological driving suggestion and a motorcade cooperation strategy. According to the technology, the fuel consumption of a diesel heavy truck is reduced by 8%-15%, the power consumption of a new energy commercial vehicle is reduced by 6%-12%, adverse driving events are reduced by 30%-45%, the deadhead rate of a motorcade is reduced by 10%-20%, and the energy consumption cost of one hundred vehicles is saved by 8%-14%. The system adopts a modular design and low-intrusive transformation, supports multi-vehicle-type adaptation and OTA updating, meets vehicle specification level requirements, and provides an efficient and energy-saving operation solution for commercial motorcades.

Description

Technical Field

[0001] This invention relates to the field of intelligent traffic management, and specifically to a method and system for intelligent energy consumption management and energy-saving optimization of commercial vehicles. Background Technology

[0002] With the rapid development of the commercial vehicle industry and the continuous advancement of energy conservation and emission reduction policies, the refinement and energy efficiency of fleet management have become key to improving operational efficiency and green development capabilities. Currently, fleet management systems are widely used in various fields such as logistics and transportation, and engineering operations. Their core functions revolve around vehicle monitoring, driving behavior analysis, route optimization, and energy management, providing fundamental support for fleet operations.

[0003] Existing fleet management systems generally employ GPS positioning technology combined with OBD / CAN bus data acquisition to monitor basic operating parameters such as vehicle mileage, fuel consumption, and fault codes. They also provide basic data reports and dispatching functions to meet the basic regulatory needs of daily fleet operations. Although existing fleet management technologies have a certain application foundation, many problems and shortcomings remain to be solved in actual operation, making it difficult to meet the needs of refined energy-saving fleet operation and full lifecycle cost optimization. These shortcomings are specifically reflected in the following aspects: First, the granularity and accuracy of energy consumption monitoring are insufficient. Existing systems mostly record statistical fuel or electricity consumption data for vehicle energy consumption monitoring, without fully integrating external influencing factors such as vehicle load, road gradient, wind resistance during driving, road grade, and micro-weather conditions for comprehensive analysis. This results in significant deviations in energy consumption assessment results and fails to provide accurate data support for fleet energy-saving strategy formulation.

[0004] Secondly, a closed-loop optimization system for driving behavior and energy consumption has not been established. Current driving behavior analysis only focuses on the identification and scoring of violations. These scores are not quantitatively coupled with energy consumption models, failing to accurately reflect the impact of different driving behaviors on energy consumption. Furthermore, there is a lack of eco-driving suggestions tailored to individual drivers, and a real-time human-machine interaction loop has not been constructed, making it difficult to guide drivers to proactively adjust their driving behavior to reduce energy consumption.

[0005] Third, route optimization has a singular objective and fails to consider multidimensional constraints. Most fleet management systems' route optimization strategies focus on shortening travel time or distance, ignoring multiple practical operational constraints such as minimizing energy consumption, controlling empty mileage, restricting access to roads, road gradients, and vehicle weight limits. As a result, while the optimized routes may meet time or distance requirements, they have higher energy consumption and increased operating costs, making them difficult to adapt to complex and diverse actual operating scenarios.

[0006] Fourth, edge-cloud collaboration capabilities are insufficient, and model transferability is poor. In existing technologies, there is a disconnect between the local inference capabilities of the vehicle and the batch processing capabilities of the cloud, failing to fully utilize the advantages of real-time data processing at the edge and the massive data analysis capabilities of the cloud. At the same time, there is a lack of a unified energy consumption baseline construction method across vehicle models, making it difficult to form a transferable energy consumption model and unable to adapt to the current operational situation of multiple vehicle models coexisting in a fleet.

[0007] Fifth, there is a lack of a comprehensive strategic scheduling mechanism at the fleet level. Existing technologies mostly focus on the operation monitoring and optimization of individual vehicles, failing to develop energy-consumption-based cross-vehicle collaborative strategies from the perspective of overall fleet operations, such as vehicle platooning optimization, transport connection coordination, and precise control of empty-running rates. In addition, there is a lack of optimization solutions for the total cost of ownership (TCO) of the fleet, making it impossible to maximize the operational efficiency of the fleet.

[0008] In summary, existing fleet management technologies have significant shortcomings in terms of energy consumption monitoring accuracy, synergistic optimization of driving behavior and energy consumption, multi-objective route optimization, edge-cloud collaboration, and fleet-level overall planning strategies, making it difficult to meet the current development needs of refined, energy-saving, and low-cost fleet operations. Summary of the Invention

[0009] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for intelligent energy consumption management and energy-saving optimization of commercial vehicles.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: a commercial vehicle intelligent energy consumption management and energy-saving optimization system, comprising an on-board edge terminal, vehicle sensors and ECU / OBD interfaces, an on-board human-machine interface (HMI), a cloud platform, a fleet scheduling and capacity management module, and a third-party data service interface; the on-board edge terminal is connected to the vehicle's CAN / OBD interface and communicates with the cloud platform via 4G / 5G / V2X; the cloud platform includes a data lake and feature warehouse, model services, and an optimization engine; the third-party data interface provides the cloud platform with map road network, real-time traffic, and micro-meteorological data support; the on-board HMI is used to push real-time ecological driving prompts to the driver.

[0011] The vehicle-mounted edge terminal is directly connected to the vehicle's CAN / OBD interface to collect data such as engine speed, fuel injection quantity, torque request, vehicle mass estimation, SOC (for new energy commercial vehicles), brake / pedal opening, and GPS / IMU; it also locally deploys a lightweight energy consumption inference model and a driving behavior scoring model; and it establishes a data interaction link with the cloud platform through 4G / 5G / V2X communication technology. Vehicle sensors and ECU / OBD interfaces provide the system with operating condition signals and fault code data, which are the basic data sources for energy consumption monitoring and behavior analysis. The on-board human-machine interface (HMI) pushes ecological driving prompts to the driver in real time, including operation suggestions such as acceleration, cruise, gear shifting, and braking, to achieve real-time human-machine interaction.

[0012] The data lake and feature warehouse aggregate multi-source information such as vehicle operation data, road condition data, meteorological data, load data, and terrain data to build a standardized data feature system. The model service includes an energy consumption estimation model (designed for different vehicle types such as ICE / HEV / BEV), a driving behavior scoring model, and a platooning energy-saving assessment model, providing core algorithm support for the system. The optimization engine realizes multi-objective route optimization (covering dimensions such as energy consumption, timeliness, toll costs, and risk), fleet scheduling and capacity matching, task connection and empty-run rate control functions.

[0013] The third-party data service interface connects to map road networks (providing information such as road grade, traffic restriction rules, slope, and weight limit), real-time traffic (providing data such as congestion and traffic events), and micro-meteorology (providing parameters such as wind speed, temperature, and precipitation) to provide external data support for system optimization.

[0014] Preferably, the vehicle edge terminal locally deploys a lightweight energy consumption inference model and a driving behavior scoring model to collect engine speed, fuel injection quantity, torque request, vehicle mass estimation, SOC, brake / pedal opening, GPS / IMU data, and performs local inference calculations.

[0015] Preferably, the model service of the cloud platform includes an energy consumption estimation model (ICE / HEV / BEV by vehicle type), a driving behavior scoring model, and a platooning energy-saving assessment model; the optimization engine is used to achieve multi-objective route optimization, fleet scheduling and capacity matching, task connection and empty-run rate control.

[0016] A method for intelligent energy consumption management and energy-saving optimization of commercial vehicles includes the following steps: S1: Data Acquisition and Preprocessing: Collect multi-dimensional data through vehicle edge terminals, vehicle sensors and ECU / OBD interfaces and perform standardized processing; S2: Edge inference outputs energy consumption estimation and behavior score: The lightweight model based on the vehicle edge terminal outputs real-time energy consumption estimation results and driving behavior score; S3: Cloud-based modeling and baseline calibration: The cloud utilizes multi-source data to optimize model parameters and completes energy consumption baseline calibration for different scenarios; S4: Route and Scheduling Optimization: The cloud-based optimization engine combines third-party data to generate the most energy-efficient route planning and scheduling solutions; S5: Issue Eco-driving suggestions and execute strategies: Issue eco-driving suggestions to the vehicle HMI and issue dispatch instructions to the fleet dispatch module; S6: Feedback on execution data and continuous learning: The vehicle sends back execution data and feedback data, and the cloud updates the model and strategy; S7: Generate monthly / quarterly energy saving reports and TCO assessment reports.

[0017] Preferably, in step S2, energy consumption estimation is based on a vehicle dynamics approximation model and environmental correction factors, including road grade, slope, wind resistance, and load, to achieve second-level calculation; driving behavior score is generated by detecting events such as rapid acceleration / sudden braking / high-speed overspeed / long idling, extracting driving rhythm stability and shift timing characteristics, and combining them with energy consumption increment attribution analysis.

[0018] Preferably, in step S4, route optimization constructs a multi-objective cost function, which is: Cost = w1・Energy + w2・Delay + w3・Toll + w4・Risk; under constraints of road weight / traffic restrictions, slope, time window, and charging / gas station coverage, a hierarchical search and local fine-tuning are used to generate energy-priority paths.

[0019] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention achieves breakthrough improvements in multiple dimensions, including energy consumption control, driving standards, operational efficiency, and economic benefits, through an edge-cloud collaborative architecture, multi-objective optimization algorithms, and a closed-loop mechanism throughout the entire process. It is adaptable to various commercial vehicle scenarios and possesses significant practical value and promotion potential. In terms of energy saving, this technology achieves precise energy reduction across all scenarios. For diesel heavy-duty trucks in long-haul logistics conditions, fuel consumption is reduced by 8%–15% thanks to the synergistic effect of edge-end second-level energy consumption estimation and cloud-based route optimization; in high-speed platooning, fuel consumption is further reduced by 3%–6% by optimizing vehicle spacing to reduce aerodynamic drag. For new energy commercial vehicles in urban delivery scenarios, electricity consumption is reduced by 6%–12%, while range is simultaneously improved by 5%–10%, effectively alleviating range anxiety. Its core advantage lies in breaking through the limitations of traditional statistical energy consumption monitoring, combining multi-dimensional environmental factors such as load, gradient, and wind resistance to make energy consumption assessment more accurate and optimization strategies more targeted.

[0020] Meanwhile, significant improvements have been made in driving behavior optimization. By establishing a quantitative coupling model between driving behavior and energy consumption, the incidence of adverse driving events has been reduced by 30%–45%, and the duration of long idling times has been shortened by 25%–35%. The system generates personalized ecological driving suggestions based on real-time data and enables real-time human-machine interaction through the in-vehicle HMI, promoting drivers to develop energy-saving driving habits, reducing high-energy-consuming operations from the source, and forming a closed-loop optimization of "monitoring-scoring-suggestion-improvement".

[0021] Furthermore, operational efficiency has been comprehensively improved, with fleet empty-running rate reduced by 10%-20% and average arrival time improved by 5%-9%. The multi-objective route optimization algorithm takes into account multiple dimensions such as energy consumption, timeliness, and road constraints, avoiding congested and restricted roads, and achieving efficient task matching in conjunction with fleet collaborative scheduling; platooning and connection optimization strategies further improve transportation turnover efficiency and reduce ineffective driving costs.

[0022] Finally, the economic benefits are particularly outstanding. Based on a fleet of 100 vehicles, annual fuel / electricity costs are reduced by 8%–14%, and total cost of ownership (TCO) decreases by 5%–10%, significantly reducing fleet operational pressure. Simultaneously, the system adopts a modular design and low-intrusion modification scheme, supporting adaptation to multiple vehicle models including ICE / HEV / BEV, and supports OTA updates at the edge. It meets automotive-grade reliability and safety compliance requirements, requires no major modifications to the original vehicle structure, has low deployment costs, and strong adaptability. It can be quickly applied to various commercial vehicle fleets such as heavy trucks, light trucks, and cold chain logistics vehicles, possessing broad market prospects and industrial value. Attached Figure Description

[0023] Figure 1 This is a diagram of the overall system architecture of the present invention; Figure 2 This is a flowchart illustrating the energy consumption optimization process of this invention. Figure 3 Here is a flowchart of the driving behavior analysis and scoring module; Figure 4 Flowchart of the route optimization algorithm; Figure 5 This is a timeline diagram for edge-cloud collaboration. Detailed Implementation

[0024] To provide a further understanding of the purpose, structure, features, and functions of the present invention, detailed descriptions are provided below with reference to specific embodiments.

[0025] A commercial vehicle intelligent energy consumption management and energy-saving optimization system includes an on-board edge terminal, vehicle sensors and ECU / OBD interfaces, an on-board human-machine interface (HMI), a cloud platform, a fleet dispatching and capacity management module, and a third-party data service interface. The on-board edge terminal is connected to the vehicle's CAN / OBD interface and communicates with the cloud platform via 4G / 5G / V2X. The cloud platform includes a data lake and feature warehouse, model services, and an optimization engine. The third-party data interface provides the cloud platform with map road network, real-time traffic, and micro-meteorological data support. The on-board HMI is used to push real-time ecological driving prompts to the driver.

[0026] The vehicle-mounted edge terminal is directly connected to the vehicle's CAN / OBD interface to collect data such as engine speed, fuel injection quantity, torque request, vehicle mass estimation, SOC (for new energy commercial vehicles), brake / pedal opening, and GPS / IMU; it also locally deploys a lightweight energy consumption inference model and a driving behavior scoring model; and it establishes a data interaction link with the cloud platform through 4G / 5G / V2X communication technology. Vehicle sensors and ECU / OBD interfaces provide the system with operating condition signals and fault code data, which are the basic data sources for energy consumption monitoring and behavior analysis. The on-board human-machine interface (HMI) pushes ecological driving prompts to the driver in real time, including operation suggestions such as acceleration, cruise, gear shifting, and braking, to achieve real-time human-machine interaction.

[0027] The data lake and feature warehouse aggregate multi-source information such as vehicle operation data, road condition data, meteorological data, load data, and terrain data to build a standardized data feature system. The model service includes an energy consumption estimation model (designed for different vehicle types such as ICE / HEV / BEV), a driving behavior scoring model, and a platooning energy-saving assessment model, providing core algorithm support for the system. The optimization engine realizes multi-objective route optimization (covering dimensions such as energy consumption, timeliness, toll costs, and risk), fleet scheduling and capacity matching, task connection and empty-run rate control functions.

[0028] The third-party data service interface connects to map road networks (providing information such as road grade, traffic restriction rules, slope, and weight limit), real-time traffic (providing data such as congestion and traffic events), and micro-meteorology (providing parameters such as wind speed, temperature, and precipitation) to provide external data support for system optimization.

[0029] Preferably, the vehicle edge terminal locally deploys a lightweight energy consumption inference model and a driving behavior scoring model to collect engine speed, fuel injection quantity, torque request, vehicle mass estimation, SOC, brake / pedal opening, GPS / IMU data, and performs local inference calculations.

[0030] Preferably, the model service of the cloud platform includes an energy consumption estimation model (ICE / HEV / BEV by vehicle type), a driving behavior scoring model, and a platooning energy-saving assessment model; the optimization engine is used to achieve multi-objective route optimization, fleet scheduling and capacity matching, task connection and empty-run rate control. Specific Implementation

[0031] A method for intelligent energy consumption management and energy-saving optimization of commercial vehicles includes the following steps: S1: Data Acquisition and Preprocessing: Collect multi-dimensional data through the vehicle edge terminal, vehicle sensors and ECU / OBD interface and perform standardized processing.

[0032] S11: Through the vehicle edge terminal, it links with the vehicle's CAN / OBD interface and sensors to collect high-frequency data at the second level, including: Powertrain data: engine speed, fuel injection quantity, torque request, throttle opening, and transmission gear; New energy-specific data: SOC, charge / discharge power, battery temperature; Operating status data: vehicle speed, brake pedal opening, accelerator pedal travel, mileage, idling time; Positioning and attitude data: Latitude and longitude, altitude, direction of travel, and acceleration are collected via GPS / IMU.

[0033] Environmental and road condition data collection: acquired in real time through third-party data interfaces, including: Micrometeorological data: wind speed, wind direction, ambient temperature, precipitation; Road network and traffic data: road grade (expressway / national highway / provincial highway / county road / township road), gradient, road weight limit (tonnage), restricted hours, real-time congestion index (0-10), traffic incidents (accident / construction / control); Load data: Obtain cargo weight and load distribution (uniform / uneven load) through vehicle load sensors or freight system interfaces.

[0034] Historical and Fleet Data Acquisition: The cloud-based data lake backtracks and collects nearly 6 months of historical data from the target fleet, including energy consumption data, driving behavior data, fault records, etc. for different vehicle models, routes, and seasons.

[0035] S12: Data Preprocessing Flow: Data cleaning: An outlier detection algorithm (3σ principle + isolated forest model) is used to remove abnormal data caused by sensor failure and signal interference (such as sudden changes in instantaneous vehicle speed, negative fuel injection, etc.); short-term data gaps (gap duration ≤ 3 seconds) are filled by linear interpolation, and long-term gaps (> 3 seconds) are marked as invalid segments with the reason noted.

[0036] Data standardization: Numerical standardization: converting data of different dimensions into the [0,1] interval (e.g., vehicle speed: 0-120km / h→0-1, SOC: 0-100%→0-1). Standardized format: Unified data timestamp format (UTC+8), geographic coordinate system (WGS-84), road classification code (1-expressway, 2-national highway, 3-provincial highway, 4-county road, 5-township road); Feature engineering: Constructing derived features for subsequent modeling, including: Operating characteristics: average vehicle speed, acceleration frequency, braking frequency, idling speed ratio, and gradient change rate; Environmental characteristics: drag coefficient (calculated based on wind speed, wind direction and the angle between the driving direction), road friction coefficient (based on the mapping between road grade and precipitation conditions); Energy consumption related characteristics: energy consumption per unit distance (current energy consumption / current distance), load factor (actual load / rated load). Finally, output a standardized data matrix (including basic features and derived features) and a data quality report (percentage of abnormal data, missing rate, and duration of valid data).

[0037] S2: Edge inference outputs energy consumption estimation and behavior score: The lightweight model based on the vehicle edge terminal outputs real-time energy consumption estimation results and driving behavior score.

[0038] S21: The vehicle-mounted edge terminal comes pre-installed with two core lightweight models (model size ≤ 50MB, inference latency ≤ 100ms, meeting automotive-grade real-time requirements): Lightweight energy consumption inference model: a deep neural network model trained based on the TensorFlow Lite framework (16-dimensional input layer features: vehicle speed, acceleration, load coefficient, slope, drag coefficient, etc.; 3 hidden layers with ReLU activation function; 1-dimensional output layer: energy consumption value in seconds (L / s or kWh / s)). Driving behavior scoring model: Based on a rule-based + machine learning hybrid model, the rule part identifies typical events such as rapid acceleration and sudden braking, while the machine learning part (LightGBM) outputs a comprehensive score.

[0039] S22: Real-time energy consumption estimation: Model input: The real-time feature data of the vehicle end after preprocessing in step S1 (1 time / second input); Calculation process: The edge terminal is based on an approximate vehicle dynamics model (F=m a +F f +F w +F i Where F is the driving force, m a For inertial force, F f For rolling resistance, F w For air resistance, F i The energy consumption value output by the neural network is physically constrained and corrected by combining environmental correction factors (load, slope, wind resistance) to ensure that the result conforms to the actual working conditions. Output results: second-level energy consumption estimate, energy consumption level under current operating conditions (1-low energy consumption, 2-medium energy consumption, 3-high energy consumption).

[0040] S23: Driving Behavior Analysis and Scoring S231: Event Detection Based on threshold rules and a sliding window algorithm (window duration of 5 seconds), driving behavior events are detected: Rapid acceleration: acceleration > 0.8 m / s² and duration > 0.5 seconds; Emergency braking: deceleration < -0.6 m / s² and duration > 0.3 seconds; Speeding on highways: The vehicle's speed exceeds the current road speed limit by more than 10%; Long idling: Idle time > 60 seconds (excluding parking and waiting scenarios); Improper shifting: The optimal shifting range is calculated based on the current vehicle speed and engine speed. If the shifting range deviates by ±200 rpm for more than 1 second, it is considered improper shifting.

[0041] S232: Energy Consumption Increment Attribution: The contribution of each adverse event to energy consumption is quantified by analyzing SHAP (SHapley Additive exPlanations) values ​​(e.g., rapid acceleration increases energy consumption by 15%, and long idling increases energy consumption by 8%).

[0042] S233: Overall Rating: The scoring system is based on a 100-point scale, with a base score of 60 points. The rules are as follows: No adverse events: 1 point is added every 10 seconds, up to a maximum of 100 points; Adverse events detected: Points will be deducted based on energy consumption contribution (5 points / time for rapid acceleration, 4 points / time for sudden braking, 6 points / time for speeding at high speed, 3 points / minute for prolonged idling, and 2 points / time for improper gear shifting). Grading level mapping: 85-100 points (Excellent), 70-84 points (Good), 60-69 points (Pass), <60 points (Fail).

[0043] S234: Personalized suggestion generation: Based on the scoring results and the type of adverse event, targeted eco-driving suggestions are generated (such as "smoothly depress the accelerator pedal after rapid acceleration and maintain acceleration of 0.3-0.5 m / s²"; "when idling for a long time, it is recommended to turn off the engine if there is no load"; and "shift to a low gear in advance and maintain engine speed of 1500-1800 rpm" when going uphill).

[0044] Finally, it outputs second-level energy consumption estimates, energy consumption levels, driving behavior scores (updated once every 5 seconds), and personalized eco-driving suggestions (output in real time when adverse events are triggered). S3: Cloud-based modeling and baseline calibration: The cloud utilizes multi-source data to optimize model parameters and completes energy consumption baseline calibration for different scenarios.

[0045] S31: Joint Training and Optimization of Multiple Models Energy consumption estimation model optimization: Based on historical data from the data lake and real-time data returned from the vehicle (cumulative sample size ≥ 1 million), the energy consumption models for different vehicle types (ICE / HEV / BEV) are retrained in the cloud. The Adam optimizer is used with a learning rate of 0.001 and 100 iterations. Cross-validation (8:2 training set / test set split) is used to ensure model accuracy (energy consumption estimation error ≤ 3% for ICE models and ≤ 5% for BEV models). Driving behavior scoring model optimization: By combining energy consumption data of different drivers, the weight of deduction points for adverse events is adjusted (e.g., for novice drivers, the weight of deduction points for rapid acceleration is increased by 20%; for experienced drivers, the weight of deduction points for improper gear shifting is increased by 15%), so that the scoring is more in line with the correlation between individual driving habits and energy consumption. Training of the formation energy-saving assessment model: Based on the historical formation driving data of the fleet (such as the distance between the front and rear vehicles, the consistency of vehicle speed, and the change of fuel consumption), the formation energy-saving model is trained and the predicted value of energy saving rate is output under different formation distances (20m / 30m / 50m) and different vehicle speeds (60-100km / h).

[0046] S32: Building Energy Consumption Baselines for Different Scenarios Baseline classification dimensions: Scenarios are divided into four dimensions: "vehicle type (heavy truck / light truck / new energy commercial vehicle) + route type (trunk highway / urban delivery / mountain road) + load level (empty / light load / heavy load) + season (spring / summer / autumn / winter)", covering ≥36 typical scenarios; Baseline calculation method: A combination of statistics and modeling is used. After removing outliers from the historical energy consumption data for each scenario, the mean of the 95% confidence interval is calculated as the baseline energy consumption value for that scenario. Then, it is fine-tuned by the current environmental factors (temperature, wind speed) to obtain the final explainable energy consumption baseline (e.g., baseline fuel consumption for the "heavy-load truck - main highway - winter" scenario: 38L / 100km). Baseline Dynamic Updates: The baseline is updated quarterly based on the latest fleet operation data to ensure consistency between the baseline and actual operation.

[0047] S323: Model and baseline distribution: The cloud will send the optimized model parameters (lightweight version) and scenario-specific energy consumption baselines to the edge terminals of the corresponding vehicles via 4G / 5G / V2X communication; An incremental update mechanism is adopted, which only sends out the differences in model parameters, reducing the transmission bandwidth consumption (the amount of data updated in a single update is ≤10MB). After receiving the data, the edge terminal automatically verifies its integrity. Once the verification is successful, the old parameters are overwritten, and the effective time is ≤30 seconds.

[0048] Finally, the optimized vehicle model parameters, scenario-specific energy consumption baselines, model accuracy reports, and baseline confidence reports are output.

[0049] S4: Route and Scheduling Optimization: The cloud-based optimization engine combines third-party data to generate energy-efficient route planning and scheduling schemes.

[0050] S41: Definition of multi-objective constraints: Core objective (energy consumption minimization): Based on the energy consumption baseline and route characteristics (slope, length, road grade) in step S3, predict the total energy consumption of each candidate route as the core optimization objective; Secondary objectives (multi-dimensional constraints): Time constraints: Meet the delivery window requirements (early arrival ≤ 30 minutes, late arrival ≤ 0 minutes). Cost constraints: Considering toll fees (highway tolls, bridge tolls) and charging / fueling costs, the total operating cost shall not exceed 105% of the baseline plan; Road constraints: Meet road weight, traffic, and height restrictions, and avoid dangerous road sections (such as continuous long downhill slopes ≥5km with a gradient >8%). Range constraints (new energy commercial vehicles): The route must cover charging stations, ensuring that the remaining battery power is ≥15% upon arrival at the charging point and ≥10% upon arrival at the destination; Formation constraints (vehicle dispatch): Multiple vehicles on the same route need to plan a platooning route (priority given to highway sections). The number of vehicles in a platoon should be ≤5, and the safe distance should be ≥20m.

[0051] S42: Construction of Multi-Objective Cost Function: Construct a weighted summation cost function, with weights determined using the Analytic Hierarchy Process (AHP) in conjunction with fleet operation requirements (which can be manually adjusted by the user): Cost=w1・Energy+w2・Delay+w3・Toll+w4・Risk Energy: Total energy consumption of the route (predicted based on energy consumption model, unit: L or kWh), w1∈[0.4-0.6] (core weight); Delay: The difference between the estimated travel time of the route and the baseline time (the fastest conventional route time) (in minutes), w2∈[0.2-0.3]; Toll: Total toll cost for the route (unit: yuan), w3∈[0.1-0.2]; Risk: Route risk coefficient (calculated based on accident rate, gradient risk, and congestion probability, ranging from 0 to 1), w4∈[0.05-0.1].

[0052] S43: Optimize algorithm execution: Step 1: Candidate Route Generation Based on map road network data, an improved Dijkstra algorithm is used to generate 5-8 candidate routes (including the main route and alternative routes) that meet the basic constraints (weight limit, traffic limit, time window). Step 2: Hierarchical Search Optimization The first layer (global search) uses a heuristic algorithm (genetic algorithm) to globally optimize the candidate routes. The goal is to find the approximate route interval with the optimal cost function. The number of iterations is 50 and the population size is 100. The second layer (local fine-tuning): The simulated annealing algorithm is used to perform local optimization on the global optimization results, focusing on adjusting key nodes in the route (such as intersections, slope change sections, and charging / gas station locations), with a cooling rate of 0.95 and a termination temperature of 1e-5 to ensure optimal route details; Step 3: Fleet dispatch and coordination: Task matching: Based on vehicle load capacity, current location, and remaining range, transportation tasks are assigned to the optimal vehicle to reduce empty running rate; Formation planning: For multiple vehicles on the same route, plan the formation assembly point (e.g., 5km before the highway entrance), formation driving section (highway section ≥80%), and disbandment point (3km after the highway exit), and assign the formation order (arranged according to vehicle load from largest to smallest to reduce wind resistance of the following vehicles). Connection optimization: For multi-segment relay transportation tasks, plan connection stations (prioritizing logistics parks with charging / refueling facilities) and connection time windows to avoid long vehicle waiting times.

[0053] Finally, the system outputs the optimal route for each vehicle (including steering guidance, energy consumption prediction, estimated travel time, and charging / refueling suggestions), fleet scheduling plan (task allocation results, formation planning, and shuttle plan), and a multi-objective optimization result report (the achievement status of each objective).

[0054] S5: Issue Eco-driving suggestions and execute strategies: Issue eco-driving suggestions to the vehicle HMI and issue dispatch instructions to the fleet dispatch module; S51: In-vehicle HMI Real-time Prompts and Interaction: Eco-driving suggestion push: Real-time prompts are pushed to the driver using a combination of text, voice, and icons, including: Acceleration prompt: "Current energy consumption is high. It is recommended to accelerate smoothly and maintain an acceleration of 0.4 m / s²". Cruise control prompt: "Optimal cruising speed is 85km / h, current speed is 95km / h, deceleration can reduce energy consumption by 12%"; Gear shift prompt: "It is recommended to shift to 6th gear, as the current gear consumes 8% more fuel than the optimal gear." Braking warning: "Red light 500 meters ahead. It is recommended to release the accelerator and coast to avoid sudden braking." Charging / Refueling Notice: "30km from the next recommended charging point. Current battery level is 25%. We recommend you proceed to refuel."

[0055] Driver feedback mechanism: Drivers can provide feedback on the effectiveness of suggestions ("effective", "ineffective", "needs adjustment") via HMI buttons. Feedback data is transmitted back to the cloud in real time for subsequent model optimization.

[0056] S52: Execution of fleet dispatch instructions: Dispatch instructions are issued: The cloud-based fleet management system issues dispatch instructions to the driver's APP and the fleet dispatcher's terminal, including task allocation notifications, convoy driving instructions ("Please arrive at the XX Expressway entrance before 10:30 and assemble at the position of vehicle number 2 in the convoy"), and pick-up instructions ("Please arrive at XX Logistics Park before 14:00 to pick up the goods"). Execution monitoring: The cloud tracks the vehicle's location and status in real time, determines whether it is traveling on the optimized route and whether it is complying with platooning requirements, and sends reminders to the driver and dispatcher if it deviates from the route by ≥5km or fails to travel in accordance with platooning requirements.

[0057] S53: Adaptive adjustment for special scenarios: Traffic emergencies: If a traffic accident occurs on the route and causes congestion (the congestion is expected to last more than 30 minutes), the cloud will recalculate the optimal route in real time and push route adjustment prompts to the driver through the HMI; Vehicle malfunction: If a vehicle displays a fault code (such as engine failure or battery malfunction), the edge terminal automatically reduces the energy consumption optimization weight, prioritizes pushing a "nearby repair" prompt, and notifies the dispatcher to arrange a backup vehicle for pick-up. Extreme weather: In the event of extreme weather such as heavy rain or strong winds, the cloud adjusts the energy consumption model parameters (such as increasing the weight of the drag coefficient), and the HMI pushes a "Slow down and maintain a safe distance" prompt. When driving in platoons, the safe distance is automatically increased to 50m.

[0058] Finally, the system outputs the driver's operation execution results, real-time vehicle operation data (vehicle speed, energy consumption, route deviation), driver feedback data, and dispatch instruction execution status.

[0059] S6: Feedback on execution data and continuous learning: The vehicle sends back execution data and feedback data, and the cloud updates the model and strategy.

[0060] S61: Feedback Data Collection and Analysis Data collection: The edge terminal transmits the following feedback data back to the cloud in real time, with a collection frequency of 1 time / second: Execution data: actual energy consumption, actual driving route, driving behavior event records, driver operation data (accelerator pedal opening, braking frequency, gear shifting timing); Feedback data includes: drivers' evaluation of the effectiveness of HMI recommendations, dispatchers' satisfaction ratings of dispatching plans (1-5 points), and records of abnormal situations (such as route closures or cargo changes).

[0061] S62: Data Comparison and Analysis: Energy consumption comparison: Compare actual energy consumption with predicted energy consumption, calculate the deviation rate (|actual energy consumption - predicted energy consumption| / predicted energy consumption), and analyze the reasons for the deviation (such as environmental factors, driving behavior, model error). Route execution comparison: Analyze the degree and reasons for the deviation between the actual route and the optimized route, and count the increase in energy consumption caused by the deviation; Behavioral optimization comparison: Compare the changes in the incidence of adverse driving events and energy consumption increments before and after implementation to evaluate the effectiveness of ecological driving recommendations.

[0062] S63: Iterative Optimization of Model and Policy: Model parameter update: Based on feedback data, online learning algorithms (incremental SVM) are used to update the parameters of the energy consumption estimation model and driving behavior scoring model to ensure that the model can adapt to changes in driver habits and road conditions; Strategy weight adjustment: If drivers report that "timeliness takes precedence over energy consumption", then adjust the cost function weights (w1 decreases to 0.3, w2 increases to 0.4); if the actual energy consumption of a certain route is consistently higher than the predicted energy consumption, adjust the energy consumption baseline and optimization algorithm parameters for that route. Personalized optimization is recommended: Adjust the HMI's recommended push frequency (e.g., increase the push frequency by 50% for novice drivers and decrease it by 30% for experienced drivers) and content details (e.g., suggest more specific pedal operation ranges for drivers who are used to rapid acceleration) based on driver operating habits and feedback.

[0063] S64: OTA (Over-the-Air) Canary Release and Verification: The optimized model and strategy adopt a canary release mechanism, initially deploying on 20% of the vehicles in the fleet and running for one week; Verification metrics: energy consumption reduction rate, driving behavior improvement rate, and driver feedback satisfaction. If the metrics are met (e.g., energy consumption reduction rate ≥8%, satisfaction score ≥4 points), then full deployment is implemented; if the metrics are not met, the process is restarted and optimized.

[0064] Finally, the updated model parameters, adjusted policy weights, optimized energy consumption baseline, and canary release verification report are output.

[0065] S7: Generate monthly / quarterly energy saving reports and TCO assessment reports.

[0066] The present invention has been described by the above-described embodiments; however, these embodiments are merely examples for implementing the present invention. It must be noted that the disclosed embodiments do not limit the scope of the present invention. Conversely, any modifications and refinements made without departing from the spirit and scope of the present invention are within the scope of patent protection of the present invention.

Claims

1. A smart energy consumption management and energy-saving optimization system for commercial vehicles, characterized in that: The system includes an in-vehicle edge terminal, vehicle sensors and ECU / OBD interfaces, an in-vehicle human-machine interface (HMI), a cloud platform, a fleet scheduling and capacity management module, and third-party data service interfaces. The in-vehicle edge terminal connects to the vehicle's CAN / OBD interface and communicates with the cloud platform via 4G / 5G / V2X. The cloud platform includes a data lake and feature warehouse, model services, and an optimization engine. The third-party data interfaces provide the cloud platform with map road network, real-time traffic, and micro-weather data support. The in-vehicle HMI is used to push real-time ecological driving prompts to the driver.

2. The intelligent energy consumption management and energy-saving optimization system for commercial vehicles as described in claim 1, characterized in that: The vehicle-mounted edge terminal locally deploys a lightweight energy consumption inference model and a driving behavior scoring model to collect engine speed, fuel injection quantity, torque request, vehicle mass estimation, SOC, brake / pedal opening, GPS / IMU data, and performs local inference calculations.

3. The intelligent energy consumption management and energy-saving optimization system for commercial vehicles as described in claim 1, characterized in that: The cloud platform's model services include energy consumption estimation models (ICE / HEV / BEV by vehicle type), driving behavior scoring models, and platooning energy-saving assessment models; the optimization engine is used to achieve multi-objective route optimization, fleet scheduling and capacity matching, task connection and empty-run rate control.

4. A method for intelligent energy consumption management and energy-saving optimization of commercial vehicles based on the system described in any one of claims 1-3, characterized in that: Includes the following steps: S1: Data Acquisition and Preprocessing: Collect multi-dimensional data through vehicle edge terminals, vehicle sensors and ECU / OBD interfaces and perform standardized processing; S2: Edge inference outputs energy consumption estimation and behavior score: The lightweight model based on the vehicle edge terminal outputs real-time energy consumption estimation results and driving behavior score; S3: Cloud-based modeling and baseline calibration: The cloud utilizes multi-source data to optimize model parameters and completes energy consumption baseline calibration for different scenarios; S4: Route and Scheduling Optimization: The cloud-based optimization engine combines third-party data to generate the most energy-efficient route planning and scheduling solutions; S5: Issue Eco-driving suggestions and execute strategies: Issue eco-driving suggestions to the vehicle HMI and issue dispatch instructions to the fleet dispatch module; S6: Feedback on execution data and continuous learning: The vehicle sends back execution data and feedback data, and the cloud updates the model and strategy; S7: Generate monthly / quarterly energy saving reports and TCO assessment reports.

5. The intelligent energy consumption management and energy-saving optimization method for commercial vehicles as described in claim 4, characterized in that: In step S2, energy consumption estimation is based on a vehicle dynamics approximation model and environmental correction factors, including road grade, slope, wind resistance, and load, to achieve second-level calculation; driving behavior score is generated by detecting events such as rapid acceleration / sudden braking / high-speed overspeed / long idling, extracting driving rhythm stability and shift timing characteristics, and combining them with energy consumption increment attribution analysis.

6. The intelligent energy consumption management and energy-saving optimization method for commercial vehicles as described in claim 4, characterized in that: In step S4, route optimization constructs a multi-objective cost function, which is: Cost = w1・Energy + w2・Delay + w3・Toll + w4・Risk. Under constraints of road weight / traffic restrictions, slope, time window, and charging / gas station coverage, a hierarchical search and local fine-tuning are used to generate energy-priority paths.

7. The intelligent energy consumption management and energy-saving optimization method for commercial vehicles as described in claim 4, characterized in that: The feedback execution and continuous learning process in step S6 adopts an online learning algorithm and a canary release mechanism. Specifically, it includes real-time transmission of execution data and feedback data from the vehicle. The execution data includes actual energy consumption values, driving route trajectories, and driving operation behavior records. The feedback data includes the driver's evaluation of the effectiveness of the ecological driving suggestions, the satisfaction score of the scheduling plan, and the explanation of abnormal situations.

8. The intelligent energy consumption management and energy-saving optimization method for commercial vehicles as described in claim 4, characterized in that: Step S7, the report generation process, covers energy-saving effectiveness and full life-cycle cost assessment. Specifically, it includes monthly / quarterly energy-saving reports containing core indicators such as energy consumption data statistics by vehicle type, energy consumption reduction rate, driving behavior score distribution, changes in adverse event incidence, and the proportion of energy-saving contribution from route optimization, along with typical case analysis and optimization suggestions.

Images