Automobile carbon footprint database construction method

Abstract

The invention discloses an automobile carbon footprint database construction method, which comprises the following steps of: acquiring multi-source behavior data when an automobile runs improperly, and converting the multi-source behavior data into direct emission data; identifying the multi-source behavior data based on a deep learning network, and predicting to obtain various vehicle change states; inferring indirect emission data according to the vehicle change state, connecting the direct emission data, the vehicle change state and the indirect emission data to form a carbon footprint chain, and storing the carbon footprint chain in a database; acquiring geographic data and environmental data during vehicle driving, and respectively converting the geographic data and the environmental data into geographic factors and environmental factors; probability prediction is carried out on the carbon footprint chain based on the geographic factors and the environmental factors, and the predicted probability value is marked on the carbon footprint chain; through the constructed carbon footprint chains, prediction of carbon emission can be realized, so that the carbon emission can be conveniently fed back to a driver to adjust a driving mode, and meanwhile, by predicting the probability value of each carbon footprint chain, later-stage responsibility determination and vehicle management can be facilitated.

Description

Technical Field

[0001] This invention relates to the field of automotive testing technology, and in particular to a method for constructing an automotive carbon footprint database. Background Technology

[0002] With the advancement of global carbon neutrality goals, the automotive industry, as a key sector for carbon emissions, has made carbon footprint accounting and management crucial. The automotive carbon footprint encompasses greenhouse gas emissions throughout its entire lifecycle, from raw material extraction and parts manufacturing to vehicle use and eventual recycling. Accurately quantifying direct emissions (such as exhaust emissions) and indirect emissions (such as battery production and fuel refining) is of great significance for corporate energy conservation and emission reduction, policy formulation, and the industry's sustainable development. Furthermore, improper driving behaviors during vehicle operation, such as rapid acceleration, sudden braking, and prolonged high-speed driving, can significantly exacerbate wear and tear on core components such as the engine, tires, and brake pads. According to actual measurement data... Studies show that frequent hard braking can increase the wear rate of brake pads, and extreme driving behaviors can accelerate the decline of battery health, leading to an increase in the frequency of parts replacement and generating additional carbon emissions in production and transportation. However, most current automotive carbon footprint calculations only calculate direct or indirect emissions under normal conditions, without specifically collecting and calculating carbon emissions caused by improper vehicle driving. In the full life cycle monitoring of vehicles, existing methods cannot predict the potential increase in carbon emissions through improper driving behaviors such as rapid acceleration and frequent braking, nor can they trace the actual driving situation of the vehicle based on the carbon footprint to determine the allocation of carbon emission responsibility. Summary of the Invention

[0003] In view of this, the present invention proposes a method for constructing a carbon footprint database for automobiles, which can build a database containing the carbon footprint chain to realize carbon emission prediction and subsequent source tracing, so as to accurately locate carbon emission responsibility.

[0004] The technical solution of this invention is implemented as follows: A method for constructing a vehicle carbon footprint database includes the following steps: Step S1: Obtain multi-source behavior data when the vehicle is driving improperly, and convert the multi-source behavior data into direct emission data; Step S2: Based on deep learning networks, identify multi-source behavioral data and predict various vehicle change states; Step S3: Infer indirect emission data based on vehicle change status, connect direct emission data, vehicle change status and indirect emission data to form a carbon footprint chain, and store it in the database; Step S4: Obtain the geographic data and environmental data of the vehicle during driving, and convert them into geographic factors and environmental factors respectively. Step S5: Based on geographical and environmental factors, perform probability prediction on the carbon footprint chain and mark the predicted probability values ​​on the carbon footprint chain.

[0005] Preferably, step S1 includes the following steps: Step S11: When the vehicle is driven improperly, collect the vehicle engine speed, throttle opening, instantaneous fuel consumption, acceleration and tilt angle as operating data; Step S12: Obtain vehicle driving trajectory, speed change and mileage data as driving data, and collect energy consumption data of air conditioning and headlights; Step S13: Timestamp align the running data, driving data, and energy consumption data, and perform noise reduction, missing value filling, and normalization processing. Step S14: Convert the processed operating data, driving data, and energy consumption data into direct emission data based on the emission calculation formula.

[0006] Preferably, the emission calculation formula is obtained based on international standard formulas / industry recommended models / algorithm models in academic literature.

[0007] Preferably, step S2 includes the following specific steps: Step S21: Extract features from multi-source behavioral data to obtain key features; Step S22: Construct a multi-output channel LSTM-CNN prediction model and train it; Step S23: Input the key features into the trained LSTM-CNN prediction model, and the LSTM-CNN prediction model outputs multiple vehicle change states in parallel.

[0008] Preferably, the vehicle's changing state includes powertrain wear, component wear, and battery health degradation.

[0009] Preferably, step S3 includes the following specific steps: Step S31: Extract the worn components from the vehicle's changing state, and calculate the current loss amount based on the multi-source behavior data of the worn components. Step S32: Query the standard component corresponding to the worn component, obtain the test life of the standard component, and convert the test life into the total loss. Step S33: Divide the current loss amount by the total loss amount to obtain the loss ratio; Step S34: Obtain the total carbon emission data for the production and transportation of standard components, and output the product of the total carbon emission data and the loss ratio as indirect emission data. Step S35: Connect the direct emission data, vehicle change status and indirect emission data to form a carbon footprint chain and store it in the database.

[0010] Preferably, before storing the carbon footprint chain in the database in step S35, a hash digest is performed on the carbon footprint chain, and the hash value is uploaded to the blockchain for storage.

[0011] Preferably, step S4 includes the following specific steps: Step S41: Obtain the location information of the vehicle when it was improperly driven using the GPS positioning system; Step S42: Collect the road type and road slope when the vehicle is driving based on the location information as the geographic data for this time; Step S43: Collect ambient temperature and humidity data of the vehicle while it is in motion based on the positioning information as the environmental data for this time. Step S44: Convert the current geographic data and current environmental data into geographic factors and environmental factors, respectively.

[0012] Preferably, the specific steps of step S44 are as follows: using an expert evaluation method, comparing the current geographic data and current environmental data with the geographic data and environmental data during standard driving, and converting them into geographic factors and environmental factors.

[0013] Preferably, step S5 includes the following specific steps: Step S51: Construct a Bayesian probabilistic graphical network, using geographical factors and environmental factors as inputs to the Bayesian probabilistic graphical network; Step S52: Calculate the probability value of each carbon footprint chain by using a Bayesian probabilistic graphical network, and mark the probability value on the carbon footprint chain.

[0014] Compared with the prior art, the beneficial effects of the present invention are: ① When a vehicle is driven improperly, multi-source behavioral data of the vehicle is collected. This multi-source behavioral data can be directly converted into direct emission data. At the same time, various vehicle change states can be predicted based on the multi-source behavioral data. Indirect emission data can be inferred based on the vehicle change states, so as to realize the calculation and accurate prediction of carbon emissions and to assess the carbon emission situation of the vehicle. ② The carbon footprint chain, composed of direct emission data, vehicle change status, and indirect emission data, is stored in the database. Geographic and environmental factors are introduced for probability prediction to obtain the probability value of each carbon footprint chain. This allows for easy traceability when carbon emission responsibility needs to be assigned later, thus enabling the management of vehicle carbon emissions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only preferred embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a method for constructing an automotive carbon footprint database according to the present invention; Figure 2 This is a flowchart of step S1 of a method for constructing an automotive carbon footprint database according to the present invention; Figure 3 This is a flowchart of step S2 of a method for constructing an automotive carbon footprint database according to the present invention; Figure 4 This is a flowchart of step S3 of a method for constructing an automotive carbon footprint database according to the present invention; Figure 5 This is a flowchart of step S4 of a method for constructing an automotive carbon footprint database according to the present invention; Figure 6 This is a flowchart of step S5 of a method for constructing an automotive carbon footprint database according to the present invention; Detailed Implementation To better understand the technical content of this invention, a specific embodiment is provided below, and the invention will be further described in conjunction with the accompanying drawings.

[0017] See Figures 1 to 6 The present invention provides a method for constructing a vehicle carbon footprint database, comprising the following steps: Step S1: Obtain multi-source behavior data when the vehicle is driving improperly, and convert the multi-source behavior data into direct emission data; Step S2: Based on deep learning networks, identify multi-source behavioral data and predict various vehicle change states; Step S3: Infer indirect emission data based on vehicle change status, connect direct emission data, vehicle change status and indirect emission data to form a carbon footprint chain, and store it in the database; Step S4: Obtain the geographic data and environmental data of the vehicle during driving, and convert them into geographic factors and environmental factors respectively. Step S5: Based on geographical and environmental factors, perform probability prediction on the carbon footprint chain and mark the predicted probability values ​​on the carbon footprint chain.

[0018] This invention provides a method for constructing a vehicle carbon footprint database. Instead of traditional full-cycle carbon emission calculations, it focuses on calculating carbon emissions during improper driving conditions, including sudden braking, rapid acceleration, frequent lane changes, and prolonged high-speed driving. During these improper driving conditions, multi-source behavioral data can be collected. This data can directly calculate the current carbon emissions. Furthermore, based on the differences in the multi-source behavioral data, a deep learning network can identify and predict various possible vehicle states. These states refer to wear and tear, aging, or other deterioration conditions of various vehicle components or structures. After obtaining these various vehicle states, indirect emissions can be inferred from them. Data is used to analyze emissions, such as when brake pads wear out and need to be replaced. The production and transportation of brake pads generate carbon emissions. These emissions are considered indirect emissions data. Deep learning networks can predict various vehicle change states, and different vehicle change states can lead to different indirect emissions data. By creating a carbon footprint chain from direct emissions data to vehicle change states and then to indirect emissions data, multiple carbon footprint chains can be formed and stored in a database. The database allows users to view the predicted carbon emissions data and intuitively determine the causal relationship between direct and indirect emissions data. This data can then be used to alert drivers to reduce improper driving behavior based on vehicle change states and can provide feedback to manufacturers or testing companies to adjust production processes or testing methods.

[0019] After storing multiple carbon footprint chains in the database, this invention also introduces geographical and environmental factors. Geographical and environmental data collected during improper vehicle driving can be converted into geographical and environmental factors. Different geographical and environmental data will affect vehicle changes. Therefore, by combining geographical and environmental factors, probability prediction can be performed on each carbon footprint chain, and the predicted probability value can be marked on the carbon footprint chain. At this point, each carbon footprint chain stored in the database contains a probability value. When tracing the carbon emission behavior of vehicles in the future, carbon emission responsibility can be divided based on the different probability values ​​and the actual situation of the vehicle, making it easier for car manufacturers or insurance companies to manage vehicles.

[0020] Preferably, step S1 includes the following steps: Step S11: When the vehicle is driven improperly, collect the vehicle engine speed, throttle opening, instantaneous fuel consumption, acceleration and tilt angle as operating data; Step S12: Obtain vehicle driving trajectory, speed change and mileage data as driving data, and collect energy consumption data of air conditioning and headlights; Step S13: Timestamp align the running data, driving data, and energy consumption data, and perform noise reduction, missing value filling, and normalization processing. Step S14: Convert the processed operating data, driving data, and energy consumption data into direct emission data based on the emission calculation formula, which is obtained according to the international standard formula / industry recommended model / algorithm model in academic literature.

[0021] When a vehicle engages in improper driving behavior, it is necessary to collect multi-source behavioral data at that moment. This multi-source behavioral data includes operational data, driving data, and energy consumption data. The onboard OBD system can collect real-time data on engine speed, throttle opening, and instantaneous fuel consumption. Accelerometers and gyroscopes acquire data on acceleration and tilt angle during vehicle movement. The onboard navigation system can obtain data on vehicle trajectory, speed changes, and mileage. Simultaneously, onboard sensors can collect energy consumption data from auxiliary equipment such as air conditioning and lights. The collected operational data, driving data, and energy consumption data are then time-stamped and aligned. Preprocessing can be performed, including noise reduction, elimination of abnormal fluctuations, filling missing values ​​in multi-source behavioral data using linear interpolation to ensure data integrity, and normalization of multi-source behavioral data to unify data units for easier subsequent calculations. The processed multi-source behavioral data can be directly substituted into the emission calculation formula to calculate direct emissions data. The emission calculation formula can be a commonly used one, such as the emission calculation formula specified in the international standard "Limits and Measurement Methods for Pollutant Emissions from Light-Duty Vehicles", the industry-recommended model published by the Society of Automotive Engineers, or the emission calculation model mentioned in academic literature.

[0022] Preferably, step S2 includes the following specific steps: Step S21: Extract features from multi-source behavioral data to obtain key features; Step S22: Construct a multi-output channel LSTM-CNN prediction model and train it; Step S23: Input the key features into the trained LSTM-CNN prediction model, and the LSTM-CNN prediction model outputs multiple vehicle change states in parallel. The vehicle change states include power system loss, component loss and battery health degradation.

[0023] After acquiring multi-source behavioral data, feature extraction can be performed to obtain key features. These key features serve as input to a deep learning network, which is a multi-output channel prediction model based on LSTM-CNN. After inputting the key features into the LSTM-CNN prediction model, various vehicle change states can be predicted. These vehicle change states mainly reflect the wear and tear on vehicle components, including power system wear, component wear, and battery health degradation. Examples of power system wear include engine wear, and examples of component wear include reduced brake pad thickness.

[0024] Preferably, step S3 includes the following specific steps: Step S31: Extract the worn components from the vehicle's changing state, and calculate the current loss amount based on the multi-source behavior data of the worn components. Step S32: Query the standard component corresponding to the worn component, obtain the test life of the standard component, and convert the test life into the total loss. Step S33: Divide the current loss amount by the total loss amount to obtain the loss ratio; Step S34: Obtain the total carbon emission data for the production and transportation of standard components, and output the product of the total carbon emission data and the loss ratio as indirect emission data. Step S35: Connect direct emission data, vehicle change status and indirect emission data to form a carbon footprint chain, perform a hash digest on the carbon footprint chain, upload the hash value to the blockchain for storage, and store it in the database.

[0025] Vehicle changes are related to subsequent component replacements. Component replacement involves component production and transportation, generating carbon emissions. Therefore, after acquiring vehicle changes, we can extract wear-prone components. Based on these components and corresponding multi-source behavioral data, we can calculate the current wear amount, such as the reduction in brake pad thickness. For wear-prone components, we can obtain standard components of the same type and their test lifespan during life testing. Based on the testing conditions and test lifespan, we can convert this into the total wear amount of the standard components. Then, dividing the current wear amount by the total wear amount yields the final result. The loss ratio represents the proportion of damage caused by improper driving to the total loss of the entire component. Then, the total carbon emissions of the standard component, including production and transportation, are calculated. Multiplying the total carbon emissions by the loss ratio yields the indirect emissions data for vehicle changes. The direct emissions data, vehicle changes, and indirect emissions data are then linked into a carbon footprint chain and uploaded to the blockchain to improve security. Finally, the data is stored in a database to prevent malicious tampering. The database can then determine the vehicle changes and indirect emissions data that improper driving may cause, thus alerting drivers to reduce improper driving behaviors.

[0026] Preferably, step S4 includes the following specific steps: Step S41: Obtain the location information of the vehicle when it was improperly driven using the GPS positioning system; Step S42: Collect the road type and road slope when the vehicle is driving based on the location information as the geographic data for this time; Step S43: Collect ambient temperature and humidity data of the vehicle while it is in motion based on the positioning information as the environmental data for this time. Step S44: Convert the current geographic data and current environmental data into geographic factors and environmental factors, respectively. Specifically, using expert evaluation methods, compare the current geographic data and current environmental data with the geographic data and environmental data from the standard driving period, and convert them into geographic factors and environmental factors. After the database is constructed, this invention introduces environmental and geographical factors. First, the GPS positioning system is used to confirm the vehicle's location when improper driving occurs, thus determining the vehicle's position. Then, geographical data for the vehicle's location is collected, including road type and road slope. Different types and slopes of roads cause different wear and tear on tires, engines, and brake pads. Temperature and humidity data for the vehicle's location can also be collected as environmental data. Then, an expert evaluation method is used to compare the data with geographical and environmental data under standard driving conditions. By comparing the differences between the current data and the standard data, geographical and environmental factors can be obtained. For example, when the road slope is 10 degrees higher than the standard road, it will affect the engine power. In this case, the geographical factor can be reduced accordingly from 1, for example, to 0.98. And so on, geographical and environmental factors can be obtained by converting different geographical and environmental data.

[0027] Preferably, step S5 includes the following specific steps: Step S51: Construct a Bayesian probabilistic graphical network, using geographical factors and environmental factors as inputs to the Bayesian probabilistic graphical network; Step S52: Calculate the probability value of each carbon footprint chain by using a Bayesian probabilistic graphical network, and mark the probability value on the carbon footprint chain.

[0028] After obtaining geographical and environmental factors, they are used as input to a Bayesian probabilistic graphical network. The probability value of each carbon footprint chain is calculated through the Bayesian probabilistic graphical network. The probability value represents the probability of a vehicle change state occurring. Later, when assigning responsibility, responsibility can be assigned based on different probability values, which facilitates vehicle management by car manufacturers or insurance companies.

[0029] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a vehicle carbon footprint database, characterized in that, Includes the following steps: Step S1: Obtain multi-source behavior data when the vehicle is driving improperly, and convert the multi-source behavior data into direct emission data; Step S2: Based on deep learning networks, identify multi-source behavioral data and predict various vehicle change states; Step S3: Infer indirect emission data based on vehicle change status, connect direct emission data, vehicle change status and indirect emission data to form a carbon footprint chain, and store it in the database; Step S4: Obtain the geographic data and environmental data of the vehicle during driving, and convert them into geographic factors and environmental factors respectively. Step S5: Based on geographical and environmental factors, perform probability prediction on the carbon footprint chain and mark the predicted probability values ​​on the carbon footprint chain.

2. The method for constructing a vehicle carbon footprint database according to claim 1, characterized in that, The specific steps of step S1 include: Step S11: When the vehicle is driven improperly, collect the vehicle engine speed, throttle opening, instantaneous fuel consumption, acceleration and tilt angle as operating data; Step S12: Obtain vehicle driving trajectory, speed change and mileage data as driving data, and collect energy consumption data of air conditioning and headlights; Step S13: Timestamp align the running data, driving data, and energy consumption data, and perform noise reduction, missing value filling, and normalization processing. Step S14: Convert the processed operating data, driving data, and energy consumption data into direct emission data based on the emission calculation formula.

3. The method for constructing a vehicle carbon footprint database according to claim 2, characterized in that, The emission calculation formula is obtained based on international standard formulas, industry recommended models, and algorithm models in academic literature.

4. The method for constructing a vehicle carbon footprint database according to claim 1, characterized in that, The specific steps of step S2 include: Step S21: Extract features from multi-source behavioral data to obtain key features; Step S22: Construct a multi-output channel LSTM-CNN prediction model and train it; Step S23: Input the key features into the trained LSTM-CNN prediction model, and the LSTM-CNN prediction model outputs multiple vehicle change states in parallel.

5. A method for constructing an automotive carbon footprint database according to claim 1 or 4, characterized in that, The changes in vehicle status include powertrain wear, component wear, and battery health degradation.

6. The method for constructing a vehicle carbon footprint database according to claim 1, characterized in that, The specific steps of step S3 include: Step S31: Extract the worn components from the vehicle's changing state, and calculate the current loss amount based on the multi-source behavior data of the worn components. Step S32: Query the standard component corresponding to the worn component, obtain the test life of the standard component, and convert the test life into the total loss. Step S33: Divide the current loss amount by the total loss amount to obtain the loss ratio; Step S34: Obtain the total carbon emission data for the production and transportation of standard components, and output the product of the total carbon emission data and the loss ratio as indirect emission data. Step S35: Connect the direct emission data, vehicle change status and indirect emission data to form a carbon footprint chain and store it in the database.

7. The method for constructing a vehicle carbon footprint database according to claim 6, characterized in that, Before storing the carbon footprint chain in the database in step S35, a hash digest is generated for the carbon footprint chain, and the hash value is uploaded to the blockchain for storage.

8. The method for constructing a vehicle carbon footprint database according to claim 1, characterized in that, The specific steps of step S4 include: Step S41: Obtain the location information of the vehicle when it was improperly driven using the GPS positioning system; Step S42: Collect the road type and road slope when the vehicle is driving based on the location information as the geographic data for this time; Step S43: Collect ambient temperature and humidity data of the vehicle while it is in motion based on the positioning information as the environmental data for this time. Step S44: Convert the current geographic data and current environmental data into geographic factors and environmental factors, respectively.

9. A method for constructing an automotive carbon footprint database according to claim 8, characterized in that, The specific steps of step S44 are as follows: using an expert evaluation method, the current geographic data and current environmental data are compared with the geographic data and environmental data during standard driving, and then converted into geographic factors and environmental factors.

10. A method for constructing an automotive carbon footprint database according to claim 1, characterized in that, The specific steps of step S5 include: Step S51: Construct a Bayesian probabilistic graphical network, using geographical factors and environmental factors as inputs to the Bayesian probabilistic graphical network; Step S52: Calculate the probability value of each carbon footprint chain by using a Bayesian probabilistic graphical network, and mark the probability value on the carbon footprint chain.

Images