A system and method for computing based on VSP distribution
By establishing a vehicle database and using a neural network model to filter data, the computational pressure caused by real-time calculation of VSP distribution was solved, and fast and accurate VSP distribution calculation was achieved, especially improving the evaluation accuracy in the high speed range of heavy vehicles.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SAVABOON INTELLIGENT TECH(QINGDAO) CO LTD
- Filing Date
- 2026-02-28
- Publication Date
- 2026-06-09
AI Technical Summary
In existing technologies, real-time computing of VSP distribution places a huge burden on the server's computing power, resulting in excessive computing pressure.
A calculation method based on VSP distribution is adopted. By establishing a vehicle database, collecting static and driving information of vehicles, and using neural network models to train and filter data, the complex VSP distribution formula calculation steps are reduced, and rapid calculation is achieved.
It reduces computational burden, improves computational efficiency, and ensures the accuracy and reliability of VSP distributed calculations, making it suitable for high-precision evaluation of heavy vehicles with speeds ranging from 70 to 120 km/h.
Smart Images
Figure CN122173757A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of vehicle emissions management, and in particular to a calculation system and method based on VSP distribution. Background Technology
[0002] Currently, Vehicle Specific Power (VSP) distribution is a core parameter for assessing vehicle energy consumption and emissions, widely used in traffic emission modeling, eco-driving evaluation, and urban air quality management. It reflects the instantaneous power per unit mass of a vehicle under different driving conditions, effectively linking driving behavior with pollutant emission levels. For large-displacement heavy-duty vehicles, high load and low transmission efficiency lead to an overall higher VSP distribution, and within the speed range of 70-120 km / h, over 94% conform to the normal distribution assumption, making it suitable for modeling analysis and enabling relatively accurate assessment of energy consumption and emissions. By collecting data on vehicle speed, acceleration, gravitational acceleration, road gradient, rolling resistance coefficient, drag coefficient, frontal area, air density, and vehicle mass, the vehicle's current VSP can be calculated. This data then reflects the vehicle's emission factor, which represents the amount of pollutants emitted per unit distance traveled.
[0003] The existing technical solutions mentioned above have the following drawbacks: although the emission factor of a vehicle can be calculated through VSP distribution, real-time calculation of VSP distribution would require a large amount of computing power, placing a huge burden on the server. Summary of the Invention
[0004] To reduce the computational pressure caused by real-time computation of VSP distribution, this application provides a computation system and method based on VSP distribution.
[0005] On the one hand, the calculation method based on VSP distribution provided in this application adopts the following technical solution: A calculation method based on VSP distribution includes the following steps: Configure the VSP distribution formula; Collect vehicle static information and establish a vehicle database based on the vehicle static information, which includes vehicle body data and vehicle number. When a vehicle is in motion, vehicle driving information is collected, and the vehicle driving information is associated with the corresponding vehicle static information and stored in the vehicle database. The vehicle driving information includes vehicle driving data and vehicle number. The vehicle body data and vehicle driving data are substituted into the VSP distribution formula to calculate the VSP value, and the VSP value is combined with the vehicle body data to calculate the emission factor. Adjust various vehicle driving data and recalculate VSP values and emission factors, obtain experimental group data and store it in the vehicle database; The neural network model was trained using experimental group data. The neural network model established a processing model for each data item in the vehicle driving data. The processing model used a single vehicle driving data item as an independent variable, other vehicle driving data items as invariants, and VSP value and emission factor as dependent variables. When new vehicle driving information is received, the vehicle driving data is filtered and imported into the neural network model. The neural network model searches for experimental group data whose invariants are closest to the vehicle driving data. The searched experimental group data is used to select the corresponding processing model. The processing model is used to calculate the VSP value and emission factor of the vehicle driving data.
[0006] By adopting the above scheme, vehicles can input data into the system, and the system establishes a vehicle database to store all vehicle data. The system performs a VSP distribution calculation based on the data obtained from a single vehicle trip, and then trains a neural network model. Subsequently, when the vehicle is in motion, the system can use the neural network model to filter data from the vehicle database based on the driving data and perform rapid calculations, saving the complex VPS distribution formula calculation steps and effectively reducing computational pressure.
[0007] Preferably, the step of "the neural network model searching for the processing model whose invariants are closest to the vehicle driving data" further includes the following steps: If all items of the vehicle driving data are the same in the experimental group data search, then the VSP value and emission factor of the experimental group data are used as the VSP value and emission factor to be calculated. If an experimental group of data is found that differs from only one item in the vehicle driving data, the corresponding processing model is selected using the searched experimental group of data. If no experimental group data is found that differs from only one item in the vehicle driving data, the VSP value and emission factor are calculated directly through a neural network model.
[0008] By adopting the above scheme, if the current vehicle driving data has an identical record in the vehicle database, the existing data can be directly retrieved, eliminating the need for calculation. If no similar data is found, the system will directly perform the conventional VSP distribution calculation.
[0009] Preferably, the following steps are also included: Set the range of vehicle driving simulation data, which includes maximum vehicle speed, maximum acceleration, maximum gravitational acceleration, and maximum road gradient; Based on the vehicle's static information, search for the range of vehicle static simulation data for the same model. The range of vehicle static simulation data includes the error range of rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass. When adjusting various vehicle driving data and recalculating VSP values and emission factors, speed, acceleration, gravitational acceleration, and maximum road gradient are adjusted within the range of vehicle driving simulation data to 0, while rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass are adjusted within the range of vehicle static simulation data.
[0010] By adopting the above scheme, the system presets a data adjustment range when generating experimental data sets, which reduces the workload of generating experimental data sets, maximizes the accuracy of linear regression, and improves the reliability of neural network models.
[0011] Preferably, the following steps are also included: Set the optimal vehicle speed range and driving evaluation model; When the vehicle speed in the vehicle driving information is within the preferred speed range, the current VSP value and emission factor are cached together with the time. Collect vehicle travel route maps and mark data collection segments on the vehicle travel route maps based on the time the vehicle speed is within the optimal speed range. When the vehicle finishes driving, all cached VSP values and emission factors are plotted over time as an eco-driving curve. The eco-driving curve and the data collection road segment are imported into the driving assessment model. The driving assessment model evaluates the vehicle's driving habits based on the emission factor variation pattern in the eco-driving curve and the vehicle speed variation pattern in the data collection road segment.
[0012] By adopting the above scheme, the accuracy of the calculated VPS value and emission factor is highest when the vehicle speed reaches a certain level. The speed range of heavy vehicles is generally 70-120km / h, at which time the accuracy of assessing vehicle driving habits is also the highest. Vehicle driving habits can reflect the direct impact of driving habits on pollutant emissions through the frequency and magnitude of vehicle acceleration and deceleration.
[0013] Preferably, the following steps are also included: Set the test interval; The timing begins once the neural network model has finished training. When the test interval is reached, the vehicle static information is re-acquired. If the re-acquired vehicle static information is different from the stored vehicle static information, the experimental data set is regenerated and the neural network model is trained.
[0014] By adopting the above scheme, since the vehicle's physical data may change due to aging or accident repairs, the system will retrain the neural network model at certain intervals to ensure the accuracy of data calculation.
[0015] On the other hand, the computing system based on VSP distribution provided in this application adopts the following technical solution: A VSP-based computing system includes a data acquisition module, a data storage module, an emissions calculation module, a model training module, and a driving calculation module. The data acquisition module collects vehicle static information and vehicle driving information. The vehicle static information includes vehicle body data and vehicle number, and the vehicle driving information includes vehicle driving data and vehicle number. The vehicle static information and vehicle driving information are then sent to the data storage module. The data storage module establishes a vehicle database based on the vehicle static information, stores the vehicle static information in the vehicle database, and associates the vehicle driving information with the corresponding vehicle static information and stores it in the vehicle database. The emission calculation module has a preset VSP distribution formula. It calls the vehicle static information and vehicle driving information stored in the data storage module, substitutes the vehicle body data and vehicle driving data into the VSP distribution formula to calculate the VSP value, and uses the VSP value combined with the vehicle body data to calculate the emission factor. It adjusts various vehicle driving data and recalculates the VSP value and emission factor, obtains experimental group data, sends the experimental group data to the data storage module and stores it in the corresponding vehicle database. The model training module has a preset neural network model. It calls the experimental data set of the data storage module and uses the experimental data set to train the neural network model. The neural network model establishes a processing model for each data item in the vehicle driving data. The processing model takes a single vehicle driving data item as an independent variable, other vehicle driving data items as invariants, and VSP value and emission factor as dependent variables. The neural network model is then sent to the driving calculation module. The driving calculation module calls the latest vehicle driving information stored in the data storage module, filters the vehicle driving data and imports it into the neural network model. The neural network model searches for experimental group data whose invariants are closest to the vehicle driving data, selects the corresponding processing model using the searched experimental group data, and uses the processing model to calculate the VSP value and emission factor of the vehicle driving data.
[0016] By adopting the above scheme, vehicles can input data into the system, and the system establishes a vehicle database to store all vehicle data. The system performs a VSP distribution calculation based on the data obtained from a single vehicle trip, and then trains a neural network model. Subsequently, when the vehicle is in motion, the system can use the neural network model to filter data from the vehicle database based on the driving data and perform rapid calculations, saving the complex VPS distribution formula calculation steps and effectively reducing computational pressure.
[0017] Preferably, when the driving calculation module uses a neural network model to search for experimental group data whose invariants are closest to the vehicle driving data, if an experimental group data in which all items of the vehicle driving data are the same is found, the VSP value and emission factor of the experimental group data are used as the calculated VSP value and emission factor. If an experimental group data in which only one item of the vehicle driving data is different is found, the corresponding processing model is selected using the searched experimental group data. If no experimental group data in which only one item of the vehicle driving data is different is found, the VSP value and emission factor are calculated directly through the neural network model.
[0018] By adopting the above scheme, if the current vehicle driving data has an identical record in the vehicle database, the existing data can be directly retrieved, eliminating the need for calculation. If no similar data is found, the system will directly perform the conventional VSP distribution calculation.
[0019] Preferably, the emission calculation module presets a vehicle driving simulation data range, which includes maximum vehicle speed, maximum acceleration, maximum gravitational acceleration, and maximum road gradient. It searches for the static simulation data range of vehicles of the same model based on the vehicle's static information. This static simulation data range includes the error ranges for rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass. When adjusting various vehicle driving data and recalculating the VSP value and emission factor, speed, acceleration, gravitational acceleration, and maximum road gradient are adjusted within the vehicle driving simulation data range up to 0, while rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass are adjusted within the vehicle static simulation data range.
[0020] By adopting the above scheme, the system presets a data adjustment range when generating experimental data sets, which reduces the workload of generating experimental data sets, maximizes the accuracy of linear regression, and improves the reliability of neural network models.
[0021] Preferably, it also includes a driving optimization module. The driving optimization module has a preset vehicle speed optimization range and a driving evaluation model. It calls the vehicle driving information stored in the data storage module and collects the vehicle driving route map. When the vehicle speed in the vehicle driving information is within the vehicle speed optimization range, it caches the current VSP value and emission factor in combination with time. According to the time the vehicle speed is within the vehicle speed optimization range, it marks the data collection segment on the vehicle driving route map. When the vehicle finishes driving, all cached VSP values and emission factors are plotted according to time to create an ecological driving curve. The ecological driving curve and the data collection segment are imported into the driving evaluation model. The driving evaluation model evaluates the vehicle driving habits based on the emission factor change pattern in the ecological driving curve and the vehicle speed change pattern in the data collection segment.
[0022] By adopting the above scheme, the accuracy of the calculated VPS value and emission factor is highest when the vehicle speed reaches a certain level. The speed range of heavy vehicles is generally 70-120km / h, at which time the accuracy of assessing vehicle driving habits is also the highest. Vehicle driving habits can reflect the direct impact of driving habits on pollutant emissions through the frequency and magnitude of vehicle acceleration and deceleration.
[0023] Preferably, it also includes a test calibration module, whereby the data acquisition module sends the collected vehicle static information to the test calibration module; The test calibration module has a preset test interval time. The test calibration module is connected to the emission calculation module and the model training module. When the model training module completes training, it starts timing. When the timing reaches the test interval time, it sends a collection command to the data acquisition module. The data acquisition module re-collects the vehicle static information and sends it to the test calibration module. If the re-collected vehicle static information is different from the stored vehicle static information, it controls the emission calculation module and the model training module to regenerate the experimental data set and train the neural network model.
[0024] By adopting the above scheme, since the vehicle's physical data may change due to aging or accident repairs, the system will retrain the neural network model at certain intervals to ensure the accuracy of data calculation.
[0025] In summary, the present invention has the following beneficial effects: 1. Vehicles can input data into the system, which establishes a vehicle database to store all vehicle data. The system performs a VSP distribution calculation based on the data obtained from a single vehicle trip, and then trains a neural network model. Subsequently, when the vehicle is in motion, the system can use the neural network model to filter data from the vehicle database based on the driving data and perform rapid calculations, saving the complex VPS distribution formula calculation steps and effectively reducing computational pressure. Attached Figure Description
[0026] Figure 1 This is an overall system block diagram of Embodiment 2 of this application.
[0027] Explanation of reference numerals in the attached figures: 1. Data acquisition module; 2. Data storage module; 3. Emissions calculation module; 4. Model training module; 5. Driving calculation module; 6. Driving optimization module; 7. Test calibration module. Detailed Implementation
[0028] Example 1: This application discloses a calculation method based on VSP distribution, with the following specific steps: Configure the VSP distribution formula, vehicle speed optimization range, driving evaluation model, and test interval.
[0029] Set the range of vehicle driving simulation data, which includes maximum vehicle speed, maximum acceleration, maximum gravitational acceleration, and maximum road gradient.
[0030] Collect vehicle static information and establish a vehicle database based on the vehicle static information, which includes vehicle body data and vehicle number.
[0031] Based on the vehicle's static information, search for the range of static simulation data for the same model of vehicle. The range of static simulation data includes the error range of rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass.
[0032] When a vehicle is in motion, its driving information is collected, associated with the corresponding static information of the vehicle, and stored in the vehicle database. The driving information includes the vehicle driving data and the vehicle number.
[0033] The VSP value is calculated by substituting vehicle body data and vehicle driving data into the VSP distribution formula, and the emission factor is calculated by combining the VSP value with the vehicle body data.
[0034] The system adjusts various vehicle driving data and recalculates VSP values and emission factors, acquiring experimental group data and storing it in the vehicle database. During this adjustment, speed, acceleration, gravitational acceleration, and maximum road gradient are adjusted within the range of simulable vehicle driving data to 0. Rolling drag coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass are adjusted within the range of static vehicle simulation data. When generating experimental data sets, the system pre-sets data adjustment ranges, reducing the workload during data set generation and maximizing the accuracy of linear regression, thus improving the reliability of the neural network model.
[0035] The neural network model was trained using experimental group data. The neural network model established a processing model for each data item in the vehicle driving data. The processing model used individual vehicle driving data items as independent variables, other vehicle driving data items as invariants, and VSP value and emission factor as dependent variables.
[0036] When new vehicle driving information is received, the vehicle driving data is filtered and imported into the neural network model. The neural network model searches for experimental group data whose invariants are closest to the vehicle driving data.
[0037] If all items of the vehicle driving data are found to be the same in the experimental group, then the VSP value and emission factor of that experimental group data are used as the VSP value and emission factor for calculation.
[0038] If the search finds experimental group data that differs from only one item in the vehicle driving data, the searched experimental group data is used to select the corresponding processing model, and the processing model is used to calculate the VSP value and emission factor of the vehicle driving data.
[0039] If no experimental group data differs from the vehicle driving data in only one aspect, the VSP value and emission factor are calculated directly using a neural network model. If the current vehicle driving data contains an identical record in the vehicle database, the existing data is directly retrieved, eliminating the calculation step. If no similar data is found, the system will also perform a standard VSP distribution calculation.
[0040] When the vehicle speed in the vehicle driving information is within the preferred speed range, the current VSP value and emission factor are cached together with the time.
[0041] Collect vehicle travel route maps and mark data collection sections on the vehicle travel route maps based on the time the vehicle speed is within the optimal speed range.
[0042] When the vehicle finishes driving, all cached VSP values and emission factors are plotted over time as an eco-driving curve.
[0043] Eco-driving curves and data collection road segments are imported into a driving assessment model. The model assesses vehicle driving habits based on the emission factor variation patterns in the eco-driving curves and the vehicle speed variation patterns in the data collection road segments. The accuracy of the calculated VPS value and emission factor is highest at certain vehicle speeds. For heavy vehicles, the speed range is generally 70-120 km / h, at which point the assessment of driving habits is also most accurate. Vehicle driving habits can be reflected through the frequency and magnitude of vehicle acceleration and deceleration, demonstrating the direct impact of driving habits on pollutant emissions.
[0044] Timing begins once the neural network model has finished training.
[0045] When the test interval is reached, the vehicle's static information is re-collected. If the re-collected static information differs from the previously stored information, a new experimental data set is generated, and the neural network model is trained. Because the vehicle's intrinsic data may change due to aging or accident repairs, the system will retrain the neural network model periodically to ensure the accuracy of data calculations.
[0046] The implementation principle of the VSP distribution-based calculation method in this application embodiment is as follows: Vehicles can input data into the system, the system establishes a vehicle database to store all vehicle data, the system performs a VSP distribution calculation based on the data obtained from a single vehicle trip, and then trains a neural network model. Subsequently, when the vehicle is in motion, the system can use the neural network model to filter the driving data in the vehicle database and perform rapid calculations, saving the complex VSP distribution formula calculation steps and effectively reducing the computational pressure.
[0047] Example 2: This application discloses a computing system based on VSP distribution, such as... Figure 1 As shown, it includes a data acquisition module 1, a data storage module 2, an emissions calculation module 3, a model training module 4, a driving calculation module 5, a driving optimization module 6, and a test calibration module 7.
[0048] Data acquisition module 1 collects vehicle static information and vehicle driving information. The vehicle static information includes vehicle body data and vehicle number, while the vehicle driving information includes vehicle driving data and vehicle number. This static and driving information is then sent to data storage module 2. Data acquisition module 1 then sends the collected vehicle static information to test and calibration module 7.
[0049] Data storage module 2 establishes a vehicle database based on vehicle static information, stores the vehicle static information in the vehicle database, and associates vehicle driving information with the corresponding vehicle static information and stores it in the vehicle database.
[0050] The emission calculation module 3 has a preset VSP distribution formula and vehicle driving simulation data range, and calls the vehicle static information and vehicle driving information stored in the data storage module 2. The vehicle driving simulation data range includes maximum speed, maximum acceleration, maximum gravitational acceleration, and maximum road gradient. Based on the vehicle static information, it searches for the vehicle static simulation data range of the same model. The vehicle static simulation data range includes the error range of rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass. When adjusting various vehicle driving data and recalculating VSP values and emission factors, speed, acceleration, gravitational acceleration, and maximum road gradient are adjusted within the vehicle driving simulation data range up to 0, while rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass are adjusted within the vehicle static simulation data range. The emission calculation module 3 substitutes the vehicle body data and vehicle driving data into the VSP distribution formula to calculate the VSP value, and uses the VSP value combined with the vehicle body data to calculate the emission factor. It adjusts various vehicle driving data and recalculates the VSP value and emission factor, obtains the experimental group data, sends the experimental group data to the data storage module 2 and stores it in the corresponding vehicle database.
[0051] The model training module 4 has a pre-set neural network model. It calls the experimental data set from the data storage module 2 and uses the experimental data set to train the neural network model. The neural network model establishes a processing model for each data point in the vehicle driving data. The processing model uses a single vehicle driving data point as the independent variable, other vehicle driving data points as invariants, and VSP value and emission factor as dependent variables. The neural network model is then sent to the driving calculation module 5. When generating the experimental data set, the system has a pre-set data adjustment range, which reduces the workload when generating the experimental data set, maximizes the accuracy of linear regression, and improves the reliability of the neural network model.
[0052] The driving calculation module 5 calls the latest vehicle driving information stored in the data storage module 2, filters the vehicle driving data, and imports it into the neural network model. The neural network model searches for experimental group data whose invariants are closest to the vehicle driving data. If an experimental group data set with all items identical is found, the VSP value and emission factor of that experimental group data are used as the calculated VSP value and emission factor. If an experimental group data set differs from the vehicle driving data set in only one item, the corresponding processing model is selected using the searched experimental group data, and the VSP value and emission factor are calculated using the processing model. If no experimental group data set differs from the vehicle driving data set in only one item, the VSP value and emission factor are calculated directly through the neural network model. If the current vehicle driving data set has an identical record in the vehicle database, the existing data is directly called, eliminating the calculation step. If no similar data is found, the system will directly perform the conventional VSP distribution calculation.
[0053] The driving optimization module 6 has a preset optimal speed range and a driving evaluation model. It calls the vehicle driving information stored in the data storage module 2 and collects the vehicle driving route map. When the vehicle speed in the driving information is within the optimal speed range, the current VSP value and emission factor are cached together with time. The data collection segment is marked on the vehicle driving route map according to the time the vehicle speed is within the optimal speed range. When the vehicle finishes driving, all cached VSP values and emission factors are plotted according to time to create an ecological driving curve. The ecological driving curve and the data collection segment are imported into the driving evaluation model. The driving evaluation model evaluates the vehicle's driving habits based on the emission factor change pattern in the ecological driving curve and the speed change pattern in the data collection segment. The accuracy of the calculated VSP value and emission factor is highest when the vehicle speed reaches a certain level. The speed range of heavy vehicles is generally 70-120 km / h, and the accuracy of the evaluation of vehicle driving habits is also highest at this speed. Vehicle driving habits can reflect the direct impact of driving habits on pollutant emissions through the frequency and magnitude of vehicle acceleration and deceleration.
[0054] The test calibration module 7 has a preset test interval. It connects to the emissions calculation module 3 and the model training module 4. Once the model training module 4 completes training, it starts timing. When the timing reaches the test interval, it sends a data acquisition command to the data acquisition module 1. The data acquisition module 1 then re-acquires the vehicle's static information and sends it to the test calibration module 7. If the re-acquired vehicle static information differs from the previously stored information, the system controls the emissions calculation module 3 and the model training module 4 to regenerate the experimental data set and train the neural network model. Because the vehicle's intrinsic data may change due to aging or accident repairs, the system will retrain the neural network model periodically to ensure the accuracy of data calculations.
[0055] The embodiments described herein are preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape, and principle of the present invention should be covered within the scope of protection of the present invention.
Claims
1. A calculation method based on VSP distribution, characterized in that, Includes the following steps: Configure the VSP distribution formula; Collect vehicle static information and establish a vehicle database based on the vehicle static information, which includes vehicle body data and vehicle number. When a vehicle is in motion, vehicle driving information is collected, and the vehicle driving information is associated with the corresponding vehicle static information and stored in the vehicle database. The vehicle driving information includes vehicle driving data and vehicle number. Substitute vehicle body data and vehicle driving data into the VSP distribution formula to calculate the VSP value, and use the VSP value in combination with vehicle body data to calculate the emission factor. Adjust various vehicle driving data and recalculate VSP values and emission factors, obtain experimental group data and store it in the vehicle database; The neural network model was trained using experimental group data. The neural network model established a processing model for each data item in the vehicle driving data. The processing model used a single vehicle driving data item as an independent variable, other vehicle driving data items as invariants, and VSP value and emission factor as dependent variables. When new vehicle driving information is received, the vehicle driving data is filtered and imported into the neural network model. The neural network model searches for experimental group data whose invariants are closest to the vehicle driving data. The searched experimental group data is used to select the corresponding processing model. The processing model is used to calculate the VSP value and emission factor of the vehicle driving data.
2. The calculation method based on VSP distribution according to claim 1, characterized in that, The step "searching for the processing model whose invariants are closest to the vehicle driving data using a neural network model" also includes the following steps: If all items of the vehicle driving data are the same in the experimental group data search, then the VSP value and emission factor of the experimental group data are used as the VSP value and emission factor to be calculated. If the search finds experimental group data that differs from only one item in the vehicle driving data, then the corresponding processing model is selected using the searched experimental group data; If no experimental group data is found that differs from only one item in the vehicle driving data, the VSP value and emission factor are calculated directly through a neural network model.
3. The calculation method based on VSP distribution according to claim 1, characterized in that, It also includes the following steps: Set the range of vehicle driving simulation data, which includes maximum vehicle speed, maximum acceleration, maximum gravitational acceleration, and maximum road gradient; Based on the vehicle's static information, search for the range of vehicle static simulation data for the same model. The range of vehicle static simulation data includes the error range of rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass. When adjusting various vehicle driving data and recalculating VSP values and emission factors, speed, acceleration, gravitational acceleration, and maximum road gradient are adjusted within the range of vehicle driving simulation data to 0, while rolling resistance coefficient, wind resistance coefficient, frontal area, air density, and vehicle mass are adjusted within the range of vehicle static simulation data.
4. The calculation method based on VSP distribution according to claim 1, characterized in that, It also includes the following steps: Set the optimal vehicle speed range and driving evaluation model; When the vehicle speed in the vehicle driving information is within the preferred speed range, the current VSP value and emission factor are cached together with the time. Collect vehicle travel route maps and mark data collection segments on the vehicle travel route maps based on the time the vehicle speed is within the optimal speed range. When the vehicle finishes driving, all cached VSP values and emission factors are plotted over time as an eco-driving curve. The eco-driving curve and the data collection road segment are imported into the driving assessment model. The driving assessment model evaluates the vehicle's driving habits based on the emission factor variation pattern in the eco-driving curve and the vehicle speed variation pattern in the data collection road segment.
5. The calculation method based on VSP distribution according to claim 1, characterized in that, It also includes the following steps: Set the test interval; The timing begins once the neural network model has finished training. When the test interval is reached, the vehicle static information is re-acquired. If the re-acquired vehicle static information is different from the stored vehicle static information, the experimental data set is regenerated and the neural network model is trained.
6. A computing system based on VSP distribution, characterized in that: It includes a data acquisition module (1), a data storage module (2), an emissions calculation module (3), a model training module (4), and a driving calculation module (5); The data acquisition module (1) collects vehicle static information and vehicle driving information. The vehicle static information includes vehicle body data and vehicle number, and the vehicle driving information includes vehicle driving data and vehicle number. The vehicle static information and vehicle driving information are sent to the data storage module (2). The data storage module (2) establishes a vehicle database based on the vehicle static information, stores the vehicle static information in the vehicle database, and associates the vehicle driving information with the corresponding vehicle static information and stores it in the vehicle database. The emission calculation module (3) has a preset VSP distribution formula. It calls the vehicle static information and vehicle driving information stored in the data storage module (2), substitutes the vehicle body data and vehicle driving data into the VSP distribution formula to calculate the VSP value, and uses the VSP value combined with the vehicle body data to calculate the emission factor. It adjusts various vehicle driving data and recalculates the VSP value and emission factor, obtains the experimental group data, sends the experimental group data to the data storage module (2) and stores it in the corresponding vehicle database. The model training module (4) has a preset neural network model. It calls the experimental data group of the data storage module (2) and uses the experimental data group to train the neural network model. The neural network model establishes a processing model for each item of vehicle driving data. The processing model takes a single item of vehicle driving data as an independent variable, other items of vehicle driving data as invariants, and VSP value and emission factor as dependent variables. The neural network model is sent to the driving calculation module (5). The driving calculation module (5) calls the latest vehicle driving information stored in the data storage module (2), filters the vehicle driving data and imports it into the neural network model. The neural network model searches for experimental group data whose invariants are closest to the vehicle driving data. The searched experimental group data is used to select the corresponding processing model. The processing model is used to calculate the vehicle driving data to obtain the VSP value and emission factor.
7. A computing system based on VSP distribution according to claim 6, characterized in that: When the driving calculation module (5) uses a neural network model to search for experimental group data whose invariants are closest to the vehicle driving data, if it finds experimental group data in which all items of the vehicle driving data are the same, it calls the VSP value and emission factor of the experimental group data as the VSP value and emission factor obtained by calculation. If it finds experimental group data that is different from only one item in the vehicle driving data, it uses the searched experimental group data to select the corresponding processing model. If it does not find experimental group data that is different from only one item in the vehicle driving data, it selects to directly calculate the VSP value and emission factor through the neural network model.
8. A computing system based on VSP distribution according to claim 6, characterized in that: The emission calculation module (3) presets the vehicle driving simulation data range, which includes the maximum vehicle speed, maximum acceleration, maximum gravitational acceleration, and maximum road slope. It searches for the vehicle static simulation data range of the same model of vehicle based on the vehicle static information. The vehicle static simulation data range includes the rolling resistance coefficient error range, wind resistance coefficient error range, windward area error range, air density error range, and vehicle mass error range. When adjusting various vehicle driving data and recalculating the VSP value and emission factor, the speed, acceleration, gravitational acceleration, and maximum road slope are adjusted within the vehicle driving simulation data range to 0, while the rolling resistance coefficient, wind resistance coefficient, windward area, air density, and vehicle mass are adjusted within the vehicle static simulation data range.
9. A computing system based on VSP distribution according to claim 6, characterized in that: It also includes a driving optimization module (6), which has a preset vehicle speed optimization range and a driving evaluation model. It calls the vehicle driving information stored in the data storage module (2) and collects the vehicle driving route map. When the vehicle speed in the vehicle driving information is within the vehicle speed optimization range, it caches the current VSP value and emission factor in combination with time. According to the time the vehicle speed is within the vehicle speed optimization range, it marks the data collection section on the vehicle driving route map. When the vehicle ends, all cached VSP values and emission factors are plotted according to time to create an ecological driving curve. The ecological driving curve and the data collection section are imported into the driving evaluation model. The driving evaluation model evaluates the vehicle driving habits according to the emission factor change pattern in the ecological driving curve and the vehicle speed change pattern in the data collection section.
10. A computing system based on VSP distribution according to claim 6, characterized in that: It also includes a test calibration module (7), and the data acquisition module (1) sends the collected vehicle static information to the test calibration module (7); The test calibration module (7) is preset with a test interval time. The test calibration module (7) is connected to the emission calculation module (3) and the model training module (4). When the model training module (4) completes training, it starts timing. When the timing reaches the test interval time, it sends a collection command to the data acquisition module (1). The data acquisition module (1) re-collects the vehicle static information and sends it to the test calibration module (7). If the re-collected vehicle static information is different from the stored vehicle static information, it controls the emission calculation module (3) and the model training module (4) to regenerate the experimental data set and train the neural network model.