Vehicle carbon emission management and control system and method based on digital twinning

Abstract

The invention belongs to the field of intelligent traffic systems, and particularly provides a vehicle carbon emission management and control system and method based on digital twinning. By quantifying the freshness of the generated data, the dynamic update of the digital twin and the timeliness and accuracy of the data and driving decision are ensured to a certain extent, so that the reliability of the system is optimized; the advantages of the edge cloud and the center cloud are fused. The edge cloud is used for providing a quick-response vehicle short-term decision, the real-time requirement of the Internet of Vehicles is met, meanwhile, the central cloud performs deep analysis and complex calculation based on a large amount of historical data and in combination with the vehicle short-term decision, and it is ensured that the long-term optimization effect of regional vehicles is achieved; according to the invention, vehicle short-term decision and regional vehicle long-term decision are combined. Vehicle short-term decision-making can reduce carbon emission in real time, regional vehicle long-term decision-making aims at improving driving habits, and excess carbon emission caused by driving behaviors can be rapidly and fundamentally reduced by combining the two decision-making modes.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation systems, and specifically relates to a vehicle carbon emission control system and method based on digital twin technology. Background Art

[0002] With the increase in the number of vehicles and the increasingly serious problem of carbon emissions, modern transportation places more emphasis on low-carbon green travel methods. How to effectively reduce carbon emissions caused by transportation has become an issue that needs to be urgently addressed.

[0003] Digital twin technology has shown its unique advantages in effectively predicting and managing carbon emissions. As a key digital solution in the field of intelligent transportation, digital twin technology can create accurate digital models for physical systems, deeply analyze and simulate vehicle behavior, thereby improving system performance and achieving more refined and dynamic carbon emissions management. At present, many research teams are focusing on edge computing to meet the needs of vehicle networking systems for real-time and efficient data processing. However, these studies mainly focus on the processing capabilities of edge nodes, ignoring the synergistic role that digital twin technology may play in global carbon emissions management. Summary of the invention

[0004] The present invention proposes a vehicle carbon emission control system and method based on digital twins, which cleverly integrates the concepts of edge cloud and central cloud, and combines with digital twin technology to form an efficient three-layer architecture. At the same time, by providing short-term vehicle decisions and long-term regional vehicle decisions, this system can significantly reduce the carbon emissions generated by vehicles during driving, thereby providing strong technical support for achieving low-carbon travel. In addition, the present invention also introduces an information freshness (Freshness of Information, FoI) evaluation mechanism, which quantifies the freshness of information to fully meet the timeliness requirements of data and suggestions.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A vehicle carbon emission control system based on digital twins, the system comprising a physical entity layer, a two-way communication layer and a digital twin layer.

[0007] The physical entity layer includes a physical vehicle module, an edge cloud side physical entity module and a central cloud side physical entity module. The physical vehicle module includes a data acquisition unit, a vehicle physical communication unit and a decision execution unit; the edge cloud side physical entity module includes an edge physical communication unit; and the central cloud side physical entity module includes a cloud side physical communication unit.

[0008] The data acquisition unit is used to collect the vehicle's instantaneous speed, acceleration, altitude increment, distance traveled per second, and environmental data around the vehicle.

[0009] The vehicle physical communication unit is used to upload vehicle and environmental data to the edge cloud side physical entity module, and receive all decisions issued by the edge cloud side physical entity module.

[0010] The decision execution unit is used to execute vehicle short-term decisions and regional vehicle long-term decisions.

[0011] The edge physical communication unit is used to receive vehicle and environmental data, send the data to the edge cloud side digital twin module through the two-way communication layer, and needs to receive the vehicle short-term decisions generated by the digital twin module. At the same time, it uploads the vehicle and environmental data in the area to the physical entity module on the central cloud side, and is also responsible for sending all decisions to the physical vehicle module.

[0012] The cloud-based physical communication unit is used to receive regional vehicle and environmental data, and send the data to the central cloud-side digital twin module through a double-layer communication layer. It is also responsible for receiving the regional vehicle long-term decisions generated by the digital twin module and issuing the vehicle long-term decisions.

[0013] The bidirectional communication layer includes an edge double-layer communication unit and a cloud double-layer communication unit.

[0014] The edge double-layer communication unit is used for the physical entity module on the edge cloud side to transmit vehicle and environmental data to the digital twin module on the edge cloud side, and is also used for the digital twin module on the edge cloud side to send vehicle short-term decisions to the physical entity module on the edge cloud side.

[0015] The cloud-side double-layer communication unit is used for the central cloud-side physical entity module to transmit regional vehicle and environmental data to the central cloud-side digital twin module, and is also used for the central cloud-side digital twin module to send regional vehicle long-term decisions to the central cloud-side physical entity module.

[0016] The digital twin layer includes an edge cloud digital twin module and a central cloud digital twin module. The edge cloud digital twin module includes an edge digital communication unit, an edge data storage unit, a FoI evaluation unit, a machine learning unit, and a short-term decision generation unit; the central cloud digital twin module includes a cloud digital communication unit, a cloud data storage unit, a machine learning unit, and a long-term decision generation unit.

[0017] The edge digital communication unit is used to receive vehicle and environmental data, and is also responsible for sending the vehicle's short-term decisions to the edge cloud layer physical entity module through the edge dual-layer communication unit.

[0018] The edge data storage unit is used to store vehicle and environment data and transmit the stored data to the FoI evaluation unit.

[0019] The FoI evaluation unit is used to evaluate the freshness of real-time data and transmit the evaluated data to the machine learning unit.

[0020] The machine learning unit is used to predict carbon emissions based on the evaluated real-time data based on a linear regression model.

[0021] The short-term decision making unit is used to generate short-term decisions to improve the driver's driving behavior habits such as excessive speed, sudden braking and sudden acceleration in real time, improve fuel efficiency, and thus reduce carbon emissions.

[0022] The cloud digital communication unit is used to receive regional vehicle and environmental data, and is also responsible for sending regional vehicle long-term decisions to the central cloud-side physical entity module through the cloud double-layer communication unit.

[0023] The cloud data storage unit is used to store regional vehicle, environment and vehicle short-term decision data, and transmit the stored data to the machine learning unit.

[0024] The machine learning unit is used to predict carbon emissions based on a BP neural network model.

[0025] The long-term decision generating unit is used to generate long-term decisions for regional vehicles, gradually improve drivers' bad driving behavior habits in the long run, improve fuel efficiency, and thus reduce carbon emissions.

[0026] The present invention introduces digital twin technology into the carbon emission control scenario, constructs a vehicle carbon emission control system based on digital twin according to steps S1 to S4, and uses the system to implement a method for controlling carbon emissions under digital twin technology.

[0027] In step S1, the data acquisition unit of the physical vehicle module collects vehicle and environmental data through various sensors deployed on each vehicle. The data collected by the vehicle undergoes preliminary preprocessing, including data cleaning, normalization, and feature extraction to ensure that the data format is consistent. The preprocessed data is uploaded to the physical entity module on the edge cloud side through the vehicle physical communication unit. After the physical entity module on the edge cloud side receives the data, it sends the data to its corresponding digital twin module through the edge double-layer communication unit. In addition, the physical entity module on the edge cloud side also needs to upload the regional vehicle and environmental data to the physical entity module on the central cloud side. After the physical entity module on the central cloud side receives the data, it sends the data to its corresponding digital twin module through the cloud double-layer communication unit.

[0028] In step S2, after receiving the vehicle and environment data, the digital twin module on the edge cloud side stores the data in the edge data storage unit. The FoI evaluation unit evaluates the stored data, and the freshness of the data is measured by the information age indicator to meet the system's real-time requirements.

[0029] Step S3: After the freshness evaluation is completed, the machine learning unit is used to predict the carbon emissions using a linear regression model trained based on the instantaneous vehicle speed, acceleration, altitude increment and driving distance per second. The predicted result is used to generate a short-term vehicle decision through a short-term decision generation unit, and the decision is sent to the edge cloud side physical entity module through the edge double-layer communication unit. The edge cloud side physical entity module sends the received decision to the physical vehicle module for execution, thereby helping the driver reduce carbon emissions.

[0030] Step S4, after the digital twin module on the central cloud side receives the regional vehicle and environmental data, it uses its cloud data storage unit to store it. The stored data is used to further train, verify and test the BP neural network model deployed in the machine learning unit, and the trained model is used to predict carbon emissions. At the same time, combined with the short-term decisions uploaded by the digital twin module on the edge cloud side, the regional vehicle long-term decisions are finally generated. The generated decisions are sent to the physical entity module on the central cloud side through the cloud-side double-layer communication unit, and then sent down by the module to the physical entity module on the edge cloud side, and finally passed to the physical vehicle module to help reduce excessive carbon emissions caused by driving behavior habits.

[0031] Furthermore, in step S1, the carbon emission rate is related to vehicle speed, acceleration and specific power (VSP). In order to accurately monitor the carbon emissions of the vehicle, it is necessary to obtain the instantaneous vehicle speed, acceleration, altitude increment and distance traveled per second in real time. To achieve this goal, the vehicle end of the physical entity layer is deployed with a vehicle speed sensor (for measuring vehicle speed), an acceleration sensor (for measuring acceleration), a barometric altimeter (for measuring altitude increment), an odometer (for measuring distance traveled per second) and sensors related to environmental monitoring.

[0032] Furthermore, in step S2, after the data is generated, as time flows unidirectionally, the obsolescence of the information also increases unidirectionally, that is, it continues to "age". Due to the need to ensure the real-time and effectiveness of decision-making, a FoI evaluation mechanism is introduced into the system. This mechanism quantifies the freshness of information through the indicator "Age of Information" (AoI). The smaller the AoI, the higher the FoI. Based on the queue model, the system defines the AoI by calculating the difference between the time when the latest data packet is generated and the time when the data packet is received. At the same time, the system sets an AoI threshold. Information below this threshold will no longer be used for model prediction and training, ensuring that the system uses the latest data for carbon emission prediction and decision-making. In addition, a dynamic adjustment strategy is introduced in the mechanism to automatically update the threshold according to different real-time data characteristics, thereby improving the adaptability and robustness of the system.

[0033] Further, in step S3, the input parameters of the linear regression model are first determined to be vehicle speed, acceleration and specific power. Through historical data training, the model learns the relationship between these three input parameters and carbon emissions. At the same time, an incremental update strategy is adopted. When new data arrives, the model will calculate the error of the data point and update the coefficient according to the error. For example, based on the gradient descent method, the coefficient is adjusted according to the new data sample to continuously optimize the prediction ability of the model. After obtaining the vehicle parameters in real time, the model can predict and generate an estimate of the current carbon emissions. In addition, in order to determine the specific factors that cause carbon emissions to exceed the standard, the regression coefficient of the linear regression model can be analyzed or multi-factor regression analysis can be performed to identify the main influencing factors, thereby generating more targeted short-term vehicle decisions. For example, if the regression coefficient of the vehicle speed is large and significant, it means that the vehicle speed has a strong impact on carbon emissions, and it is recommended that the driver adjust the vehicle speed appropriately to reduce carbon emissions. The system's instant reporting and feedback mechanism combines these analyses to help the driver more accurately identify and adjust the specific factors that cause carbon emissions to exceed the standard, and instantly broadcast the results through the vehicle display or audio, such as "the current carbon emissions forecast exceeds the standard, it is recommended to reduce the speed."

[0034] Further, in step S4, the BP neural network model includes an input layer, a hidden layer and an output layer. The input parameters of the model are vehicle speed, acceleration and specific power, and the output parameter is carbon emissions. Based on the results of cross-validation and hyperparameter search, the number of hidden layer nodes is initially selected as 13 to balance the model complexity and training effect. The model is trained using the Levenberg-Marquardt algorithm, which combines the gradient descent method and the Gauss-Newton method and is suitable for the training of medium-sized neural networks. During the training process, historical and real-time data are uniformly preprocessed, including data cleaning, normalization and feature extraction to ensure consistent data formats. The initial training of the model mainly uses historical data in order to learn the overall trends and laws of the data. After the initial training is completed, real-time data will be gradually introduced into the training process. The Batch Training method is used to accumulate real-time data within a certain time window into mini-batches to balance training stability and real-time performance.

[0035] Furthermore, in step S4, the generation of regional vehicle long-term decision is robust. Through short-term decision-making, the system can count the number of times a vehicle emits excess carbon dioxide within a specific time, that is, the number of carbon emission warning broadcasts received by the driver. At the same time, using the digital twin of the environment, the system can evaluate the impact of external factors such as signal light settings, traffic flow patterns (such as evening rush hour), etc. on driving behavior. After identifying the reasons for frequent broadcast reminders, the system can appropriately adjust the carbon emission excess threshold set in the area to improve the flexibility and adaptability of management measures. The regional vehicle long-term decision will generate a detailed report and send it to the driver through a mobile phone application (App). The report includes changes in carbon emission thresholds in each region, personalized driving habit improvement suggestions, and guidance on adjusting driving behavior according to changes in carbon emissions. Such a feedback mechanism helps drivers understand how their driving behavior affects carbon emissions, thereby optimizing driving habits to reduce carbon emissions.

[0036] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0037] 1. The present invention introduces the FoI evaluation mechanism. By quantifying the freshness of data after it is generated, it ensures the dynamic update of the digital twin and the timeliness and accuracy of data and driving decisions to a certain extent, thereby optimizing the reliability of the system;

[0038] 2. The present invention cleverly combines the advantages of edge cloud and central cloud. The edge cloud is used to provide fast-response short-term vehicle decisions to meet the real-time requirements of the Internet of Vehicles. At the same time, the central cloud conducts in-depth analysis and complex calculations based on a large amount of historical data and combined with vehicle short-term decisions to ensure the long-term optimization effect of regional vehicles.

[0039] 3. The present invention combines short-term vehicle decision-making and long-term regional vehicle decision-making. Short-term vehicle decision-making can reduce carbon emissions in real time, while long-term regional vehicle decision-making is committed to improving driving habits. The combination of the two can more quickly and fundamentally reduce excess carbon emissions caused by driving behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is an architecture diagram of a vehicle carbon emission control system based on digital twins in an embodiment of the present invention;

[0041] Figure 2 It is a digital twin layer architecture diagram of a vehicle carbon emission control system based on digital twin in an embodiment of the present invention;

[0042] Figure 3 It is a bidirectional communication layer architecture diagram of a vehicle carbon emission control system based on digital twins in an embodiment of the present invention;

[0043] Figure 4 It is a physical entity layer architecture diagram of the vehicle carbon emission control system based on digital twin in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0045] In order to enable those skilled in the art to better understand the embodiments of the present invention, the techniques and terms involved in the present invention are introduced and explained below.

[0046] Digital twin technology: Digital twin is a technology that simulates, analyzes and optimizes the performance of physical entities by creating virtual models of physical entities based on sensor detection and data integration. As a key tool in the field of advanced manufacturing and production, it significantly improves the efficiency of design, production and management by accurately mapping models of digital and physical spaces.

[0047] Linear regression model: Linear regression model is one of the most commonly used predictive models in statistics. It is used to analyze the relationship between two variables: a dependent variable and one or more independent variables. In its simplest form, a linear regression model assumes that there is a linear relationship, that is, every change in the independent variable will lead to a corresponding change in the dependent variable.

[0048] BP neural network model: BP is a multi-layer feedforward neural network trained according to the error back propagation algorithm. It is the most widely used neural network. The principle is to input the learning sample, and the signal is forward propagated to the output layer. If the expected output cannot be obtained in the output layer, the error signal is reversely propagated. The weights and thresholds of the grid are continuously adjusted through the back propagation algorithm to make the final output value as close to the expected value as possible. Through this process of continuous iteration, the signal error is finally within the allowable range. The BP neural network has strong nonlinear mapping capabilities and high self-learning and self-adaptive capabilities, supports multi-parameter input, etc., and is particularly suitable for nonlinear fitting.

[0049] like Figure 1 As shown, a vehicle carbon emission control system architecture based on digital twin in an embodiment of the present invention can be divided into three layers: physical entity layer, two-way communication layer and digital twin layer.

[0050] The physical entity layer includes the physical vehicle module, the edge cloud side physical entity module, and the central cloud side physical entity module. The physical vehicle module consists of the actual vehicles in the system. The edge cloud side physical entity module consists of distributed edge servers that are deployed close to the vehicles to reduce data transmission delays and increase processing speed. The central cloud side physical entity module consists of cloud servers located in the data center.

[0051] The digital twin layer includes the edge cloud digital twin module and the central cloud digital twin module. The edge cloud digital twin module includes the digital twin of the vehicle and the digital twin of the environment, and is responsible for localized data processing and generating short-term decisions to optimize the vehicle's immediate driving behavior. The central cloud digital twin module is deployed on the cloud server, and mainly includes the digital twin of the vehicle, the digital twin of the environment, and the digital twin of the edge server. It is used to integrate the data of each edge server, analyze the long-term trends of regional vehicles and the environment, and generate long-term decisions.

[0052] In the physical vehicle module, sensors collect the vehicle's operating status and surrounding environment data, and upload the data to the physical entity module on the edge cloud side.

[0053] The physical entity module on the edge cloud side sends the data to its corresponding digital twin module. After receiving the data, the module performs real-time analysis and sends the generated short-term decision back to the physical entity module on the edge cloud side, which is then sent to the physical vehicle module to optimize the real-time driving behavior.

[0054] In addition, the physical entity module on the edge cloud side needs to upload regional vehicle and environmental data to the physical entity module on the central cloud side, while the digital twin module on the edge cloud side needs to upload vehicle short-term decisions to the digital twin module on the central cloud side.

[0055] The physical entity module on the central cloud side sends the received data to its corresponding digital twin module, which conducts a comprehensive analysis of the data combined with short-term decisions, identifies long-term trends and patterns, and ultimately generates long-term decisions for regional vehicles. These long-term decisions are first sent to the physical entity module on the central cloud side, which is then sent down to the physical entity module on the edge cloud side, and finally passed to the physical vehicle module to help drivers adjust their driving habits, thereby achieving long-term carbon emission optimization control.

[0056] The present invention introduces digital twin technology into the carbon emission control scenario, constructs a vehicle carbon emission control method based on digital twin according to steps S1 to S4, and uses the system to implement a method for controlling carbon emissions under digital twin technology.

[0057] Step S1, the carbon emission rate is related to vehicle speed, acceleration and specific power (VSP), and the VSP calculation formula is as follows:

[0058] P VSP =v(1.1a+9.8Δh / d+0.132)+0.000302v 3

[0059] Where P VSP is the specific power, v is the instantaneous vehicle speed, a is the instantaneous acceleration, Δh is the altitude increment per second, and d is the distance traveled per second. In order to accurately monitor the carbon emissions of the vehicle, this system needs to obtain the instantaneous vehicle speed, acceleration, altitude increment and distance traveled per second in real time. To achieve this goal, the vehicle side of the physical entity layer deploys a speed sensor (for measuring vehicle speed), an acceleration sensor (for measuring acceleration), a barometric altimeter (for measuring altitude increment), and an odometer (for measuring distance traveled per second). The data acquisition unit of the physical vehicle module collects vehicle and environmental data through the above-mentioned sensors deployed in each vehicle. The collected data is preliminarily preprocessed in the vehicle, including data cleaning, normalization and feature extraction to ensure that the data format is consistent. The preprocessed data is uploaded to the physical entity module on the edge cloud side through the vehicle physical communication unit. After receiving the data, the physical entity module on the edge cloud side sends the data to its corresponding digital twin module through the edge double-layer communication unit. In addition, the physical entity module on the edge cloud side also needs to upload the regional vehicle and environmental data to the physical entity module on the central cloud side. After the physical entity module on the central cloud side receives the data, it sends the data to its corresponding digital twin module through the cloud-side double-layer communication unit.

[0060] Step S2: After the digital twin module on the edge cloud side receives the vehicle and environmental data, it stores the data in the edge data storage unit. Due to the need to ensure the real-time and effectiveness of decision-making, the FoI evaluation mechanism is introduced into the system. This mechanism quantifies the freshness of information through the indicator AoI. The smaller the AoI, the higher the FoI. The system is based on a queue model and defines AoI by calculating the difference between the time when the latest data packet is generated and the time when the data packet is received. Specifically, when the system receives the latest data packet at a given time point t, it records its generation time U(t), and uses tU(t) to represent the instantaneous information age Δ(t). Within a time period t∈[0,T], the average information age Δ τ Calculated by the following formula:

[0061]

[0062] At the same time, the system sets an AoI threshold. Information below this threshold will no longer be used for model prediction and training, ensuring that the system uses the latest data for carbon emission prediction and decision-making. In addition, a dynamic adjustment strategy is introduced in the mechanism to automatically update the threshold according to different real-time data characteristics, thereby improving the adaptability and robustness of the system. The stored data is evaluated through the FoI evaluation unit, and the AoI indicator is used to measure the freshness of the data, which greatly improves the real-time performance and prediction accuracy of the system.

[0063] Step S3, after the freshness evaluation is completed, the carbon emissions are predicted using the linear regression model. The predicted result generates a short-term decision for the vehicle through the short-term decision generation unit, and the decision is sent to the physical entity module on the edge cloud side through the edge double-layer communication unit. The physical entity module on the edge cloud side sends the received decision to the physical vehicle module for execution. First, the input parameters of the linear regression model are determined to be vehicle speed, acceleration and specific power. Through historical data training, the model learns the relationship between these three input parameters and carbon emissions. At the same time, an incremental update strategy is adopted. When new data arrives, the model calculates the error of the data point and updates the coefficient according to the error. For example, based on the gradient descent method, the coefficient is adjusted according to the new data sample to continuously optimize the prediction ability of the model. In addition, in order to determine the specific factors that cause excessive carbon emissions, the regression coefficient of the linear regression model can be analyzed or multi-factor regression analysis can be performed to identify the main influencing factors, thereby generating more targeted short-term vehicle decisions. For example, if the regression coefficient of the vehicle speed is large and significant, it means that the vehicle speed has a strong impact on carbon emissions. It is recommended that the driver adjust the vehicle speed appropriately to reduce carbon emissions. The system's instant reporting and feedback mechanism combines these analyses to help drivers more accurately identify and adjust specific factors that lead to excessive carbon emissions, and instantly report the results through the on-board display or audio, such as "The current carbon emissions forecast exceeds the standard, it is recommended to reduce the speed."

[0064] In step S4, after receiving the regional vehicle and environmental data, the digital twin module on the central cloud side uses its cloud data storage unit to store it. The stored data is used to further train, verify and test the BP neural network model deployed in the machine learning unit. The BP neural network model includes an input layer, a hidden layer and an output layer. The input parameters of the model are vehicle speed, acceleration and specific power, and the output parameter is carbon emissions. Based on the results of cross-validation and hyperparameter search, the number of hidden layer nodes is initially selected as 13 to balance the model complexity and training effect. The model is trained using the Levenberg-Marquardt algorithm, which combines the gradient descent method and the Gauss-Newton method and is suitable for the training of medium-sized neural networks. During the training process, historical and real-time data are uniformly preprocessed, including data cleaning, normalization and feature extraction, to ensure consistent data formats. The initial training of the model mainly uses historical data in order to learn the overall trends and laws of the data. After the initial training is completed, real-time data will be gradually introduced into the training process. The Batch Training method is used to accumulate real-time data within a certain time window into mini-batches to balance training stability and real-time performance. The trained BP model is used to predict carbon emissions. Combined with the short-term decisions uploaded by the digital twin module on the edge cloud side, the system can count the number of times the vehicle emits excess carbon dioxide within a specific time, that is, the number of carbon emission warning broadcasts received by the driver. Using the digital twin of the environment, the system can evaluate the impact of external factors such as signal light settings and traffic flow patterns (such as evening rush hour) on driving behavior. After identifying the reasons for frequent broadcast reminders, the system can appropriately adjust the carbon emission excess threshold set in the area to improve the flexibility and adaptability of management measures, and finally generate regional vehicle long-term decisions. The decision is sent to the physical entity module on the central cloud side through the cloud double-layer communication unit, and then sent to the physical entity module on the edge cloud side by the module, and finally passed to the physical vehicle module. The decision will be sent to the driver in the form of a report by the mobile application (App). The report contains changes in carbon emission thresholds in each region, personalized suggestions for improving driving habits, and guidance on adjusting driving behavior according to changes in carbon emissions. Such a feedback mechanism helps drivers understand how their driving behavior affects carbon emissions, thereby optimizing their driving habits to reduce carbon emissions.

[0065] like Figure 2 As shown, the present invention implements a digital twin layer of a vehicle carbon emission control system based on digital twin.

[0066] The digital twin layer includes the edge cloud side digital twin module and the central cloud side digital twin module.

[0067] The edge cloud digital twin module consists of an edge digital communication unit, an edge data storage unit, a FoI evaluation unit, a machine learning unit, and a short-term decision generation unit. The edge digital communication unit is used to receive vehicle and environmental data, and is responsible for sending the vehicle's short-term decision to the edge cloud physical entity module; the edge data storage unit is used to store vehicle and environmental data, and transmit the stored data to the FoI evaluation unit; the FoI evaluation unit is used to evaluate the freshness of real-time data and transmit the evaluated data to the machine learning unit; the machine learning unit is used to predict carbon emissions based on the evaluated real-time data based on a linear regression model; the short-term decision generation unit is used to generate short-term decisions, improve the driver's driving behavior habits such as excessive speed, sudden braking and sudden acceleration in real time, improve fuel efficiency, and thus reduce carbon emissions.

[0068] The digital twin module on the central cloud side is composed of a cloud digital communication unit, a cloud data storage unit, a machine learning unit, and a long-term decision generation unit. Among them, the cloud digital communication unit is used to receive regional vehicle and environmental data, and is also responsible for sending regional vehicle long-term decisions to the central cloud side physical entity module; the cloud data storage unit is used to store regional vehicle, environmental and vehicle short-term decision data, and transmit the stored data to the machine learning unit; the machine learning unit is used to predict carbon emissions based on the BP neural network model; the long-term decision generation unit is used to generate regional vehicle long-term decisions, gradually improve drivers' bad driving behavior habits in the long run, improve fuel efficiency, and thus reduce carbon emissions.

[0069] like Figure 3 As shown, the present invention implements a two-way communication layer of a vehicle carbon emission control system based on digital twins.

[0070] The two-way communication layer includes an edge double-layer communication unit and a cloud double-layer communication unit. The edge double-layer communication unit is responsible for realizing the communication between the edge cloud side physical entity module and the edge cloud side digital twin module, while the cloud double-layer communication unit is responsible for realizing the communication between the central cloud side physical entity module and the central cloud side digital twin module.

[0071] like Figure 4 As shown, the present invention implements a physical entity layer of a vehicle carbon emission control system based on digital twins.

[0072] The physical entity layer consists of three parts: the physical vehicle module, the edge cloud side physical entity module and the central cloud side physical entity module, which are interconnected through their respective internal physical communication modules.

[0073] The physical vehicle module consists of a data acquisition unit, a vehicle physical communication unit, and a decision execution unit. The data acquisition unit is used to collect the vehicle's instantaneous speed, acceleration, altitude increment, distance traveled per second, and environmental data around the vehicle; the vehicle physical communication unit is used to upload the vehicle and environmental data to the edge cloud physical entity module, and receive all decisions issued by the edge cloud physical entity module; the decision execution unit is used to execute short-term vehicle decisions and long-term regional vehicle decisions.

[0074] The edge cloud side physical entity module includes an edge physical communication unit. The edge physical communication unit is used to receive vehicle and environmental data, send the data to the edge cloud side digital twin module, and needs to receive the short-term vehicle decisions generated by the digital twin module. At the same time, it uploads the vehicle and environmental data in the area to the central cloud side physical entity module, and is also responsible for sending all decisions to the physical vehicle module.

[0075] The physical entity module on the central cloud side is composed of a cloud-based physical communication unit, which is used to receive regional vehicle and environmental data, send the data to the central cloud-based digital twin module, and receive the regional vehicle long-term decision generated by the digital twin module and issue the vehicle long-term decision.

[0076] The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment. Any equivalent modifications or changes made by ordinary technicians in this field based on the contents disclosed by the present invention should be included in the protection scope recorded in the claims.

Claims

1. A vehicle carbon emission control system based on digital twins, characterized by: The physical entity layer includes a physical vehicle module, an edge cloud side physical entity module and a central cloud side physical entity module; the physical vehicle module includes a data acquisition unit, a vehicle physical communication unit and a decision execution unit connected in sequence; the edge cloud side physical entity module includes an edge physical communication unit; the central cloud side physical entity module includes a cloud side physical communication unit; wherein the edge physical communication unit is bidirectionally connected with the vehicle physical communication unit and the cloud side physical communication unit respectively; The two-way communication layer includes an edge two-layer communication unit and a cloud two-layer communication unit to achieve two-way communication between the physical entity layer and the digital twin layer; The digital twin layer includes an edge cloud-side digital twin module and a central cloud-side digital twin module; the edge cloud-side digital twin module includes an edge digital communication unit, an edge data storage unit, a FoI evaluation unit, a machine learning unit, and a short-term decision generation unit connected in sequence, wherein the short-term decision generation unit is connected to the edge digital communication unit; the central cloud-side digital twin module includes a cloud-side digital communication unit, a cloud-side data storage unit, a machine learning unit, and a long-term decision generation unit connected in sequence, wherein the long-term decision generation unit is connected to the cloud-side digital communication unit.

2. A vehicle carbon emission control system based on digital twin according to claim 1, characterized in that: In the physical vehicle module, the data acquisition unit is used to collect the vehicle's instantaneous speed, acceleration, altitude increment, distance traveled per second and environmental data around the vehicle; the vehicle physical communication unit is used to upload the vehicle and environmental data to the physical entity module on the edge cloud side, and receive all decisions issued by the physical entity module on the edge cloud side; the decision execution unit is used to execute vehicle short-term decisions and regional vehicle long-term decisions.

3. The vehicle carbon emission control system based on digital twin according to claim 1 is characterized in that: In the edge cloud side physical entity module, the edge physical communication unit is used to receive vehicle and environmental data, send the data to the edge cloud side digital twin module through the two-way communication layer, and needs to receive the vehicle short-term decisions generated by the digital twin module. At the same time, it uploads the vehicle and environmental data in the area to the central cloud side physical entity module, and is also responsible for sending all decisions to the physical vehicle module.

4. The vehicle carbon emission control system based on digital twin according to claim 1 is characterized in that: In the physical entity module on the central cloud side, the cloud-side physical communication unit is used to receive regional vehicle and environmental data, and send the data to the digital twin module on the central cloud side through a double-layer communication layer. It is also responsible for receiving the regional vehicle long-term decisions generated by the digital twin module and issuing the vehicle long-term decisions.

5. The vehicle carbon emission control system based on digital twin according to claim 1 is characterized in that: In the two-way communication layer, the edge double-layer communication unit is used for the edge cloud side physical entity module to transmit vehicle and environmental data to the edge cloud side digital twin module, and is also used for the edge cloud side digital twin module to send vehicle short-term decisions to the edge cloud side physical entity module; the cloud side double-layer communication unit is used for the central cloud side physical entity module to transmit regional vehicle and environmental data to the central cloud side digital twin module, and is also used for the central cloud side digital twin module to send regional vehicle long-term decisions to the central cloud side physical entity module.

6. The vehicle carbon emission control system based on digital twin according to claim 1 is characterized in that: In the edge cloud side digital twin module, the edge digital communication unit is used to receive vehicle and environmental data, and is also responsible for sending the vehicle's short-term decision to the edge cloud layer physical entity module through the edge double-layer communication unit; the edge data storage unit is used to store vehicle and environmental data, and transmit the stored data to the FoI evaluation unit; the FoI evaluation unit is used to evaluate the freshness of real-time data and transmit the evaluated data to the machine learning unit; the machine learning unit is used to predict carbon emissions based on the evaluated real-time data based on a linear regression model; The short-term decision making unit is used to generate short-term decisions to improve the driver's driving behavior habits such as excessive speed, sudden braking and sudden acceleration in real time, improve fuel efficiency, and thus reduce carbon emissions.

7. The vehicle carbon emission control system based on digital twin according to claim 1 is characterized by: In the central cloud-side digital twin module, the cloud-side digital communication unit is used to receive regional vehicle and environmental data, and is also responsible for sending regional vehicle long-term decisions to the central cloud-side physical entity module through the cloud-side double-layer communication unit; the cloud-side data storage unit is used to store regional vehicle, environmental and vehicle short-term decision data, and transmit the stored data to the machine learning unit; the machine learning unit is used to predict carbon emissions based on the BP neural network model; the long-term decision generation unit is used to generate regional vehicle long-term decisions, gradually improve drivers' bad driving behavior habits in the long run, improve fuel efficiency, and thus reduce carbon emissions.

8. A control method for a vehicle carbon emission control system based on digital twin according to any one of claims 1 to 7, characterized in that: The method comprises the following steps: Step S1, the data acquisition unit of the physical vehicle module collects vehicle and environmental data through various sensors deployed in each vehicle. The collected data undergoes preliminary preprocessing in the vehicle, including data cleaning, normalization, and feature extraction to ensure consistent data format; The preprocessed data is uploaded to the physical entity module on the edge cloud side through the vehicle physical communication unit; After receiving the data, the physical entity module on the edge cloud side sends the data to its corresponding digital twin module through the edge double-layer communication unit; in addition, the physical entity module on the edge cloud side also uploads the regional vehicle and environmental data to the physical entity module on the central cloud side; after receiving the data, the physical entity module on the central cloud side sends the data to its corresponding digital twin module through the cloud double-layer communication unit; Step S2: After receiving the vehicle and environment data, the digital twin module on the edge cloud side stores the data in the edge data storage unit; The FoI evaluation unit evaluates the stored data and measures the freshness of the data by the information age indicator to meet the system's real-time requirements. Step S3: After the freshness evaluation is completed, the machine learning unit is used to predict the carbon emissions using a linear regression model trained based on the instantaneous vehicle speed, acceleration, altitude increment and driving distance per second. The predicted result is used to generate a short-term decision for the vehicle through a short-term decision generation unit, and the decision is sent to the physical entity module on the edge cloud side through the edge double-layer communication unit. The physical entity module on the edge cloud side sends the received decisions to the physical vehicle module for execution, thereby helping drivers reduce carbon emissions; Step S4, after the central cloud-side digital twin module receives the regional vehicle and environmental data, it uses its cloud data storage unit to store it. The stored data is used to further train, verify and test the BP neural network model deployed in the machine learning unit, and the trained model is used to predict carbon emissions. At the same time, combined with the short-term decisions uploaded by the edge cloud-side digital twin module, the regional vehicle long-term decisions are finally generated; the generated decisions are sent to the central cloud-side physical entity module through the cloud-side double-layer communication unit, and then sent down by the module to the edge cloud-side physical entity module, and finally passed to the physical vehicle module to help reduce excessive carbon emissions caused by driving behavior habits.

9. A vehicle carbon emission control method based on digital twin according to claim 8, characterized in that: In step S3, first, the input parameters of the linear regression model are determined to be vehicle speed, acceleration and specific power. Through historical data training, the model learns the relationship between these three input parameters and carbon emissions; at the same time, an incremental update strategy is adopted. When new data arrives, the model will calculate the error of the data point and update the coefficient according to the error. Based on the gradient descent method, the coefficient is adjusted according to the new data sample to continuously optimize the prediction ability of the model; in addition, in order to determine the specific factors that cause excessive carbon emissions, the regression coefficients of the linear regression model are analyzed and multi-factor regression analysis is performed to identify the main influencing factors, thereby generating more targeted short-term vehicle decisions; the system's instant reporting and feedback mechanism combines these analyses to help the driver identify and adjust the specific factors that cause excessive carbon emissions, and instantly broadcast the results through the vehicle display or audio.

10. The vehicle carbon emission control method based on digital twin according to claim 8 is characterized in that: In step S4, the BP neural network model includes an input layer, a hidden layer and an output layer; the input parameters of the model are vehicle speed, acceleration and specific power, and the output parameter is carbon emissions; based on the results of cross-validation and hyperparameter search, the number of hidden layer nodes is selected to be 13 to balance the model complexity and training effect; the model is trained using the Levenberg-Marquardt algorithm that combines the gradient descent method and the Gauss-Newton method; During the training process, historical and real-time data are uniformly preprocessed, including data cleaning, normalization, and feature extraction, to ensure consistent data formats; The initial training of the model uses historical data to learn the overall trends and patterns of the data; After the initial training is completed, real-time data will be gradually introduced into the training process; The Batch Training method is used to accumulate real-time data within the set time window into small batches, balancing training stability and real-time performance; the trained BP model is used to predict carbon emissions, and at the same time, combined with the short-term decisions uploaded by the digital twin module on the edge cloud side, the number of times the vehicle emits excess carbon dioxide within a certain period of time, that is, the number of carbon emission warning broadcasts received by the driver, is counted; the digital twin of the environment is used to evaluate the impact of external traffic environmental factors on driving behavior; after identifying the cause of the frequent broadcast reminders, the carbon emission excess threshold set in the area is adjusted, and finally a long-term decision for regional vehicles is generated; the decision is sent to the physical entity module on the central cloud side through the cloud-based double-layer communication unit, and then sent down by the module to the physical entity module on the edge cloud side, and finally passed to the physical vehicle module. The decision will be sent to the driver by the mobile application App in the form of a report; The report includes changes in carbon emission thresholds for each region, personalized suggestions for improving driving habits, and guidance on adjusting driving behavior based on changes in carbon emissions.

Images