A Smart Measurement Method and System for Motor Vehicles Based on AI Technology

Abstract

This invention relates to the field of motor vehicle metering technology, specifically disclosing a method and system for establishing intelligent motor vehicle metering based on AI technology. This invention obtains vehicle dynamic data by calculating mass and volume data; performing dynamic operating condition identification and correction; constructing an emission condition prediction model to predict vehicle emission data; constructing a mileage energy consumption prediction model to predict mileage energy consumption data; recording and organizing data to generate motor vehicle metering data, and supplementing it with meteorological data. Meteorological verification processing is performed before data use. This invention, through AI technology, calculates mass and volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data to generate motor vehicle metering data, and supplements it with meteorological data. Meteorological verification processing is performed before data use, providing an effective data verification mechanism to support the determination of the authenticity and reliability of motor vehicle metering data.

Description

Technical Field

[0001] This invention belongs to the field of motor vehicle measurement technology, and in particular relates to a method and system for establishing intelligent motor vehicle measurement based on AI technology. Background Technology

[0002] Motor vehicle metrology is the process of accurately measuring and evaluating key parameters of a motor vehicle, including its physical attributes, operating status, and emission characteristics, using scientific and systematic methods. It primarily includes the quantitative measurement of mass, volume, speed, acceleration, mileage, fuel consumption, and exhaust emissions. Motor vehicle metrology involves not only the measurement of static indicators, such as curb weight and dimensions, but also the collection and analysis of data during dynamic operation, such as real-time speed changes, acceleration performance, and emission concentrations.

[0003] In the current technology, the application scenarios of motor vehicle measurement are becoming increasingly widespread. However, much of the motor vehicle measurement data currently used is obtained through non-official channels, lacking an effective data verification mechanism, and thus unable to fully determine the authenticity and reliability of the motor vehicle measurement data. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for establishing intelligent metering of motor vehicles based on AI technology, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0006] A smart metering method for motor vehicles based on AI technology is established, and the method specifically includes the following steps:

[0007] Collect multi-source static data, perform data fusion and identification, and calculate mass and volume data;

[0008] Collect multi-source dynamic data, perform dynamic operating condition identification and correction, and obtain vehicle dynamic data;

[0009] Based on AI technology, an emission condition prediction model is constructed to predict vehicle emission data according to the mass and volume data and the vehicle dynamic data.

[0010] Based on AI technology, a mileage energy consumption prediction model is constructed to predict mileage energy consumption data according to the mass and volume data, the vehicle dynamic data, and the vehicle emission data.

[0011] Data is recorded and organized to generate motor vehicle metering data, and meteorological metering data is added. Meteorological verification processing is performed before the data is used.

[0012] As a further limitation of the technical solution of this invention embodiment, the step of collecting multi-source static data, performing data fusion and identification, and calculating mass and volume data specifically includes the following steps:

[0013] Collect multi-source static data;

[0014] The multi-source static data is effectively extracted to obtain multiple valid static data; among which, the valid static data includes the vehicle's length, width, height, actual wheelbase, wheel track parameters, and point cloud data;

[0015] Multiple valid static data sets are fused to obtain fused static data.

[0016] The fused static data is identified to calculate the vehicle's mass and volume data; wherein, the vehicle's mass and volume data includes the vehicle's load distribution characteristics and geometric parameters.

[0017] As a further limitation of the technical solution of this embodiment of the invention, the step of collecting multi-source dynamic data, performing dynamic working condition identification and correction, and obtaining vehicle dynamic data specifically includes the following steps:

[0018] Collect dynamic data from multiple sources;

[0019] The multi-source dynamic data is effectively extracted to obtain multiple valid dynamic data.

[0020] Dynamic operating condition identification is performed on multiple valid dynamic data to obtain direct dynamic data;

[0021] By combining the mass and volume data, the direct dynamic data is corrected to obtain vehicle dynamic data.

[0022] As a further limitation of the technical solution of this embodiment of the invention, the step of constructing an emission condition prediction model based on AI technology and predicting vehicle emission data according to the mass-volume data and the vehicle dynamic data specifically includes the following steps:

[0023] Collect emission source data;

[0024] Based on AI technology, an emission condition prediction model is constructed according to the emission source data, the mass and volume data, and the vehicle dynamic data.

[0025] Based on the mass and volume data and the vehicle dynamic data, the first input feature data is generated.

[0026] The first input feature data is imported into the emission condition prediction model to obtain the output vehicle emission data.

[0027] As a further limitation of the technical solution of this invention embodiment, the step of constructing a mileage energy consumption prediction model based on AI technology, and predicting mileage energy consumption data according to the mass-volume data, the vehicle dynamic data, and the vehicle emission data, specifically includes the following steps:

[0028] Collect multi-source mileage data;

[0029] Based on AI technology, a mileage energy consumption prediction model is constructed according to the multi-source mileage data, the vehicle emission data, the mass and volume data, and the vehicle dynamic data.

[0030] The mass volume data and the vehicle dynamic data are processed to generate second input feature data;

[0031] The second input feature data is imported into the mileage energy consumption prediction model to obtain the output mileage energy consumption data.

[0032] As a further limitation of the technical solution of this embodiment of the invention, the process of recording and organizing data, generating motor vehicle metering data, supplementing it with metering meteorological data, and performing meteorological verification processing when using the data specifically includes the following steps:

[0033] The mass and volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data are recorded and organized to generate motor vehicle metering data;

[0034] Multiple target factors are randomly selected from a set of pre-defined meteorological factors.

[0035] According to the multiple target factors mentioned above, acquire and supplement the measurement meteorological data;

[0036] When using the data, the meteorological data and motor vehicle measurement data are verified, and the verification results are obtained.

[0037] A smart vehicle measurement system based on AI technology is established. The system includes a mass and volume measurement unit, a dynamic identification and correction unit, a vehicle emission prediction unit, a mileage and energy consumption prediction unit, and a meteorological verification and processing unit, wherein:

[0038] The mass and volume measurement unit is used to collect multi-source static data, perform data fusion and identification, and measure mass and volume data.

[0039] The dynamic identification and correction unit is used to collect multi-source dynamic data, perform dynamic operating condition identification and correction, and obtain vehicle dynamic data.

[0040] The vehicle emission prediction unit is used to build an emission condition prediction model based on AI technology and predict vehicle emission data based on the mass and volume data and the vehicle dynamic data.

[0041] The mileage energy consumption prediction unit is used to build a mileage energy consumption prediction model based on AI technology, and predict mileage energy consumption data based on the mass volume data, the vehicle dynamic data and the vehicle emission data.

[0042] The meteorological verification and processing unit is used to record and organize data, generate motor vehicle metering data, and supplement the metering meteorological data. When using the data, meteorological verification processing is performed.

[0043] As a further limitation of the technical solution of this embodiment of the invention, the mass volume calculation unit specifically includes:

[0044] Multi-source static data acquisition module, used to acquire multi-source static data;

[0045] The first effective extraction module is used to effectively extract the multi-source static data to obtain multiple effective static data.

[0046] The data fusion module is used to fuse multiple valid static data to obtain fused static data;

[0047] The mass and volume calculation module is used to identify the fused static data and calculate the mass and volume data of the vehicle.

[0048] As a further limitation of the technical solution of this embodiment of the invention, the dynamic identification and correction unit specifically includes:

[0049] Multi-source dynamic acquisition module, used to acquire multi-source dynamic data;

[0050] The second effective extraction module is used to effectively extract the multi-source dynamic data to obtain multiple effective dynamic data.

[0051] The dynamic operating condition identification module is used to identify the dynamic operating conditions of multiple valid dynamic data and obtain direct dynamic data.

[0052] The correction processing module is used to combine the mass and volume data to correct the direct dynamic data and obtain vehicle dynamic data.

[0053] As a further limitation of the technical solution of this embodiment of the invention, the meteorological verification and processing unit specifically includes:

[0054] The recording and processing module is used to record and process the mass and volume data, the vehicle dynamic data, the vehicle emission data, and the mileage and energy consumption data to generate motor vehicle measurement data.

[0055] The factor selection module is used to randomly select multiple target factors from a set of preset meteorological factors.

[0056] The meteorological data supplementation module is used to acquire and supplement metered meteorological data according to multiple target factors.

[0057] The meteorological verification module is used to verify the metering meteorological data and motor vehicle metering data when using the data, and to obtain the verification results.

[0058] Compared with the prior art, the beneficial effects of the present invention are:

[0059] This invention, through the calculation of mass and volume data, dynamic operating condition identification and correction to obtain vehicle dynamic data, constructs an emission condition prediction model to predict vehicle emission data, constructs a mileage energy consumption prediction model to predict mileage energy consumption data, records and organizes the data to generate vehicle metering data, and supplements it with meteorological data. Meteorological verification processing is performed before data use. Based on AI technology, this invention can calculate mass and volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data to generate vehicle metering data, supplement it with meteorological data, and perform meteorological verification processing before data use. This provides an effective data verification mechanism to support the determination of the authenticity and reliability of vehicle metering data. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0061] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.

[0062] Figure 2 A flowchart illustrating the method for acquiring vehicle dynamic data provided in an embodiment of the present invention is shown.

[0063] Figure 3 A flowchart illustrating the method for predicting mileage energy consumption data provided in an embodiment of the present invention is shown.

[0064] Figure 4 An application architecture diagram of the system provided in an embodiment of the present invention is shown.

[0065] Figure 5 A structural block diagram of the mass and volume measurement unit in the system provided by an embodiment of the present invention is shown.

[0066] Figure 6 A structural block diagram of the dynamic identification correction unit in the system provided by an embodiment of the present invention is shown.

[0067] Figure 7 A structural block diagram of the meteorological verification and processing unit in the system provided by an embodiment of the present invention is shown. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0069] Understandably, the application scenarios of motor vehicle measurement are becoming increasingly widespread in existing technologies. However, much of the motor vehicle measurement data currently used is obtained through non-official channels, lacking an effective data verification mechanism, and thus unable to fully determine the authenticity and reliability of the motor vehicle measurement data.

[0070] To address the aforementioned issues, this invention employs the following methods: First, it collects multi-source static data, performs data fusion and identification, and calculates mass and volume data. Second, it collects multi-source dynamic data, performs dynamic operating condition identification and correction, and obtains vehicle dynamic data. Third, based on AI technology, it constructs an emission condition prediction model to predict vehicle emission data based on mass and volume data and vehicle dynamic data. Fourth, based on AI technology, it constructs a mileage energy consumption prediction model to predict mileage energy consumption data based on mass and volume data, vehicle dynamic data, and vehicle emission data. Fifth, it records and organizes data to generate motor vehicle metering data, supplements it with meteorological data, and performs meteorological verification processing before data use. This invention provides an effective data verification mechanism, supporting the assessment of the authenticity and reliability of motor vehicle metering data, by calculating mass and volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data using AI technology, generating motor vehicle metering data, supplementing it with meteorological data, and performing meteorological verification processing before data use.

[0071] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.

[0072] Specifically, a smart metering method for motor vehicles based on AI technology is established, and the method includes the following steps:

[0073] Step S101: Collect multi-source static data, perform data fusion and identification, and calculate mass and volume data.

[0074] In this embodiment of the invention, multi-source static data is collected by acquiring data such as load, vehicle body deformation, and volume change. The multi-source static data is then effectively extracted to eliminate redundancy and noise, resulting in multiple valid static data. These multiple valid static data are then fused to obtain fused static data. A nonlinear relationship model between the vehicle body structure and the actual loaded weight is established, and the fused static data is then identified to calculate the vehicle's mass and volume data.

[0075] Specifically, in the preferred embodiment provided by the present invention, the steps of collecting multi-source static data, performing data fusion and identification, and calculating mass and volume data specifically include the following steps:

[0076] Collect multi-source static data;

[0077] The multi-source static data is effectively extracted to obtain multiple valid static data; among which, the valid static data includes the vehicle's length, width, height, actual wheelbase, wheel track parameters, and point cloud data;

[0078] Multiple valid static data sets are fused to obtain fused static data.

[0079] The fused static data is identified to calculate the vehicle's mass and volume data; wherein, the vehicle's mass and volume data includes the vehicle's load distribution characteristics and geometric parameters.

[0080] Specifically, multiple valid static data sets are fused to obtain fused static data. The specific steps are as follows:

[0081] The region center point is obtained by multi-source static data. The region center point is used as the coordinate origin. The extension line of the coordinate origin in the east direction is used as the X-axis, the extension line of the coordinate origin in the north direction is used as the Y-axis, and the extension line of the coordinate origin in the vertical upward direction is used as the Z-axis to construct a three-dimensional spatial coordinate system.

[0082] A 3D model is constructed based on valid static data, and 3D modeling data is obtained from the 3D model. Based on the 3D spatial coordinate system, the 3D modeling data and timestamps are aligned to obtain spatiotemporal fusion data.

[0083] Using Dempster-Shafer as the main framework, a conflict detection model is constructed. Based on the conflict detection model, the spatiotemporal fusion dataset is verified and corrected to obtain a verified spatiotemporal fusion dataset. Based on the verified spatiotemporal fusion dataset, a set of confidence values ​​is obtained, and the verified spatiotemporal fusion data corresponding to the maximum confidence value is taken as the optimal fusion data.

[0084] The wheelbase Shannon entropy is calculated based on the wheelbase parameter. The wheelbase variation coefficient and mass percentage vector value are extracted from the optimal fused data. A multidimensional feature space is constructed based on the wheelbase variation coefficient, wheelbase Shannon entropy, and mass percentage vector value. Based on the multidimensional feature space, a manifold learning framework is constructed based on PCA, where mass conservation is the physical constraint. The dimensionality of the optimal fused dataset is reduced using the manifold learning framework to obtain the fused static dataset.

[0085] Furthermore, this invention establishes a standardized fusion process for multi-source heterogeneous data by forming a complete processing chain from spatial benchmark establishment, spatiotemporal alignment, confidence adjustment to conflict verification, thereby improving the reliability of the results. By establishing a three-dimensional coordinate system with the geographic center point, it unifies multi-source data into the same spatial reference system, solving the problem of inconsistent spatial benchmarks caused by differences in data sources. By dynamically allocating initial confidence based on data type, it provides differentiated data weighting for subsequent fusion, enhancing robustness to low-quality data.

[0086] Furthermore, the method for establishing intelligent vehicle measurement based on AI technology also includes the following steps:

[0087] Step S102: Collect multi-source dynamic data, perform dynamic operating condition identification and correction, and obtain vehicle dynamic data.

[0088] In this embodiment of the invention, multi-source dynamic data such as speed and acceleration are collected, and the multi-source dynamic data are effectively extracted, abnormal data are removed, and multiple valid dynamic data are obtained. Then, the multiple valid dynamic data are used to identify the dynamic working conditions such as acceleration, deceleration, idling, and constant speed to obtain direct dynamic data. After that, the direct dynamic data is corrected by combining the influence of vehicle load on the speed change trend in the mass and volume data to obtain vehicle dynamic data.

[0089] Specifically, Figure 2 A flowchart illustrating the method for acquiring vehicle dynamic data provided in an embodiment of the present invention is shown.

[0090] In a preferred embodiment of the present invention, the step of collecting multi-source dynamic data, performing dynamic condition identification and correction, and obtaining vehicle dynamic data specifically includes the following steps:

[0091] Step S1021: Collect multi-source dynamic data;

[0092] Step S1022: Effectively extract the multi-source dynamic data to obtain multiple effective dynamic data.

[0093] Step S1023: Perform dynamic condition identification on multiple valid dynamic data to obtain direct dynamic data;

[0094] Step S1024: Combine the mass and volume data to correct the direct dynamic data and obtain vehicle dynamic data.

[0095] Furthermore, the method for establishing intelligent vehicle measurement based on AI technology also includes the following steps:

[0096] Step S103: Based on AI technology, construct an emission condition prediction model to predict vehicle emission data according to the mass and volume data and the vehicle dynamic data.

[0097] In this embodiment of the invention, emission source data such as NOx, CO2, and particulate matter are collected. Based on AI technology, parameters such as mass, speed, and operating conditions are extracted simultaneously from mass and volume data and vehicle dynamic data. These parameters are then combined with the emission source data to construct an emission operating condition prediction model. The mass and volume data and vehicle dynamic data are processed to generate first input feature data including vehicle mass, driving conditions, and acceleration. This first input feature data is then imported into the emission operating condition prediction model to obtain vehicle emission data for NOx, CO2, and particulate matter emissions per unit time.

[0098] Specifically, in the preferred embodiment provided by the present invention, the step of constructing an emission condition prediction model based on AI technology and predicting vehicle emission data according to the mass-volume data and the vehicle dynamic data specifically includes the following steps:

[0099] Collect emission source data;

[0100] Based on AI technology, an emission condition prediction model is constructed according to the emission source data, the mass and volume data, and the vehicle dynamic data.

[0101] Based on the mass and volume data and the vehicle dynamic data, the first input feature data is generated.

[0102] The first input feature data is imported into the emission condition prediction model to obtain the output vehicle emission data.

[0103] Specifically, based on AI technology, an emission condition prediction model is constructed according to the emission source data, the mass and volume data, and the vehicle dynamic data. The specific steps are as follows:

[0104] Engine parameters, exhaust gas composition detection values, and OBD system data are obtained from emission source data. The engine parameters are hierarchically encoded to obtain the device feature code. The dynamic activity coefficient is obtained based on the exhaust gas composition detection value. The OBD system data is segmented to obtain the time-series state identifier. The encoding matrix is ​​obtained based on the device feature code, dynamic activity coefficient, and time-series state identifier.

[0105] Mass volume distribution characteristics are obtained from mass volume data, and vehicle acceleration, steering angle, motion trajectory parameters, and vehicle wheelbase parameters are obtained from vehicle dynamic data. The mass volume distribution characteristics are matched with vehicle acceleration and steering angle respectively to obtain the mass distribution offset index. Based on the mass distribution offset index and motion trajectory parameters, the inertial force coefficient is calculated. Based on the inertial force coefficient and vehicle wheelbase parameters, a composite feature vector is generated.

[0106] Based on the encoding matrix, an emission-quality correlation matrix is ​​constructed through feature cross-validation. A physical characteristic tensor is then formed based on the emission-quality correlation matrix. Based on the composite feature vector, the operating condition evolution pattern is extracted through a sliding window to form a spatiotemporal characteristic tensor. Based on the physical characteristic tensor and the spatiotemporal characteristic tensor, an attention mechanism is used to allocate the output weights of the physical characteristic tensor and the spatiotemporal characteristic tensor and fuse them to obtain a fused feature tensor.

[0107] Redundant features in the fused feature tensor are removed by feature selection method to obtain a simplified feature tensor. Based on the simplified feature tensor, a preliminary prediction model is constructed with carbon conservation as a constraint.

[0108] The simplified feature tensor is input into the preliminary prediction model to obtain the prediction result. The prediction result and the emission source data are sequentially aligned in the time domain and compared in similarity to obtain the comparison result. The preliminary prediction model is iteratively updated based on the comparison result. After reaching the preset number of iterations, the emission condition prediction model is obtained.

[0109] Furthermore, this invention unifies the processing of multi-dimensional features such as engine parameters, exhaust gas composition, OBD time-series data, vehicle motion state, and mass distribution to construct a highly complete input feature system; through derived parameters such as mass distribution offset index and inertial force coefficient, it quantifies the correlation between mechanical characteristics and emissions during vehicle motion, enabling the model to have clear physical interpretability; based on the temporal alignment and iterative updates of prediction results and measured data, it achieves dynamic calibration of model parameters to improve long-term prediction stability.

[0110] Specifically, the first input feature data is generated by processing the mass-volume data and the vehicle dynamic data. The specific steps are as follows:

[0111] The vehicle's mass and volume data and vehicle dynamic data are phase-synchronized using a sliding window, and then spatiotemporal mapping is performed to obtain a spatiotemporally synchronized fused data matrix.

[0112] Acceleration time-series parameters, steering angle change rate time-series parameters, and center of mass offset are obtained through spatiotemporally synchronized fusion data matrices; waveform parameters are extracted by waveform matching; and equivalent inertial force feature vectors are calculated based on waveform parameters and vehicle mass and volume data.

[0113] Wavelet packet decomposition is used to perform multi-scale decomposition of the steering angle change rate time series parameters to obtain characteristic frequency bands. The energy proportion of the characteristic frequency bands is calculated as the frequency band characterization parameter. The actual wheel track is divided by the wheel track parameter to obtain the wheel track utilization rate. Based on the wheel track utilization rate and the center of mass offset, the coupling characteristic parameter is calculated. The three-dimensional feature vector is obtained based on the equivalent inertial force feature vector, the frequency band characterization parameter, and the coupling characteristic parameter.

[0114] Principal component analysis is used to reduce the dimensionality of the static parameter components in the three-dimensional feature vector and orthogonalize the static parameter components to obtain the processed three-dimensional feature vector. Based on the processed three-dimensional feature vector, a local linear embedding algorithm is used to compress the dimensionality of the dynamic features in the three-dimensional feature vector to obtain a low-dimensional feature vector.

[0115] The isolated forest algorithm is used to detect and verify the parameter outliers of the low-dimensional feature vector to obtain the verified low-dimensional feature vector. Based on the verified low-dimensional feature vector, the abnormal data fragments are repaired using linear interpolation to obtain the repaired feature vector.

[0116] The static features in the repaired feature vector are processed by the maximum-minimum normalization method to obtain the standardized static features; the dynamic features in the repaired feature vector are processed by the sliding window Z-score to obtain the standardized dynamic features; the standardized static features and the standardized dynamic features are weighted and concatenated to obtain the first input feature data.

[0117] Furthermore, this invention achieves precise spatiotemporal correlation between mass distribution and vehicle motion through phase synchronization technology, breaking through the limitations of traditional static feature extraction and constructing a complete feature system under dynamic working conditions; based on derived parameters with clear engineering semantics such as equivalent inertial force and wheel track utilization rate, a feature space is constructed to ensure that the features have physical interpretability.

[0118] Furthermore, the method for establishing intelligent vehicle measurement based on AI technology also includes the following steps:

[0119] Step S104: Based on AI technology, construct a mileage energy consumption prediction model to predict mileage energy consumption data according to the mass volume data, the vehicle dynamic data, and the vehicle emission data.

[0120] In this embodiment of the invention, multi-source mileage data is collected through GPS trajectory, wheel speed, map matching, etc. Based on AI technology, a mileage energy consumption prediction model is constructed according to the multi-source mileage data, vehicle emission data, mass and volume data, and vehicle dynamic data. The mass and volume data and vehicle dynamic data are processed to generate second input feature data. By importing the second input feature data into the mileage energy consumption prediction model, the output mileage energy consumption data is obtained.

[0121] Specifically, Figure 3 A flowchart illustrating the method for predicting mileage energy consumption data provided in an embodiment of the present invention is shown.

[0122] In a preferred embodiment of the present invention, the step of constructing a mileage energy consumption prediction model based on AI technology, and predicting mileage energy consumption data according to the mass-volume data, the vehicle dynamic data, and the vehicle emission data, specifically includes the following steps:

[0123] Step S1041: Collect multi-source mileage data;

[0124] Step S1042: Based on AI technology, construct a mileage energy consumption prediction model according to the multi-source mileage data, the vehicle emission data, the mass and volume data, and the vehicle dynamic data;

[0125] Step S1043: Process the mass volume data and the vehicle dynamic data to generate second input feature data;

[0126] Step S1044: Import the second input feature data into the mileage energy consumption prediction model to obtain the output mileage energy consumption data.

[0127] Furthermore, the method for establishing intelligent vehicle measurement based on AI technology also includes the following steps:

[0128] Step S105: Record and organize data to generate motor vehicle metering data and supplement it with meteorological data. When using the data, perform meteorological verification processing.

[0129] In this embodiment of the invention, vehicle metering data is generated by recording and organizing mass volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data. Multiple target factors are randomly selected from a set of preset meteorological factors, and metering meteorological data is obtained according to these target factors. This metering meteorological data is then added to the vehicle metering data. Subsequently, when using the data, meteorological verification data related to the metering meteorological data is obtained. By verifying the meteorological verification data with the metering meteorological data against the target factors and corresponding values, the authenticity and reliability of the vehicle metering data are determined. If the meteorological verification fails, the vehicle metering data is determined to be invalid data.

[0130] Specifically, in the preferred embodiment provided by this invention, the steps of recording and organizing data, generating motor vehicle metering data, supplementing it with meteorological data, and performing meteorological verification processing when using the data specifically include the following steps:

[0131] The mass and volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data are recorded and organized to generate motor vehicle metering data;

[0132] Multiple target factors are randomly selected from a set of pre-defined meteorological factors.

[0133] According to the multiple target factors mentioned above, acquire and supplement the measurement meteorological data;

[0134] When using the data, the meteorological data and motor vehicle measurement data are verified, and the verification results are obtained.

[0135] Specifically, when using the data, the meteorological data and motor vehicle measurement data are verified, and the verification results are obtained. The specific steps are as follows:

[0136] Basic parameters are obtained through metrological meteorological data. Based on the basic parameters and motor vehicle metrological data, correlation indicators are calculated. The basic parameters are then screened according to the correlation indicators to obtain an initial meteorological feature set. A time window is constructed by the temporal correlation between metrological meteorological data and motor vehicle metrological data. The correlation weights in the initial meteorological feature set are adjusted using the time window to obtain the meteorological feature set.

[0137] The meteorological feature set is divided into physical measurement parameters and environmental perception parameters. The physical measurement parameters are standardized to obtain a base vector. Spatial features of the environmental perception parameters are extracted using a graph neural network to obtain an enhancement vector. The base vector and the enhancement vector are weighted and fused using an attention mechanism to obtain a fused vector.

[0138] Vehicle speed, vehicle acceleration, and historical verification data are obtained from motor vehicle measurement data. Based on vehicle speed, vehicle acceleration, and fusion vectors, a meteorological-vehicle coupling model is obtained through joint modeling. The meteorological-vehicle coupling model is trained using historical verification data to obtain a trained coupling model. The dynamic verification threshold is obtained through the trained coupling model.

[0139] The dynamic verification threshold is compared with the motor vehicle measurement data to obtain the comparison results, and the comparison results constitute the primary verification. The time series fluctuation characteristics are extracted from the motor vehicle measurement data, and the CUSUM algorithm is used to construct the intermediate verification based on the dynamic verification threshold and the time series fluctuation characteristics. The true label distribution is obtained from the motor vehicle measurement data, and an ideal meteorological model is constructed based on the dynamic verification threshold, motor vehicle measurement data and basic parameters. The ideal meteorological model is used to generate the ideal data distribution, and the deviation between the true label distribution and the ideal data distribution is calculated. The deviation is used to construct the advanced verification.

[0140] A three-level verification mechanism was constructed based on primary verification, intermediate verification, and advanced verification. Meteorological data and motor vehicle measurement data were processed through the three-level verification mechanism to obtain verification results.

[0141] Furthermore, this invention constructs a "meteorological-vehicle" coupled verification framework by synchronously integrating meteorological parameters and vehicle dynamic data to avoid the limitations of single-dimensional data verification; it generates dynamic verification thresholds that change with the environment based on historical data to improve adaptability under complex working conditions; and it achieves comprehensive detection from microscopic anomalies to macroscopic deviations through a three-layer architecture of primary real-time comparison, intermediate trend analysis, and advanced distribution verification.

[0142] Furthermore, Figure 4 An application architecture diagram of the system provided in an embodiment of the present invention is shown.

[0143] In another preferred embodiment of the present invention, a smart vehicle metering system based on AI technology is established, comprising:

[0144] The mass and volume measurement unit 101 is used to collect multi-source static data, perform data fusion and identification, and measure mass and volume data.

[0145] In this embodiment of the invention, the mass and volume measurement unit 101 collects data such as load, vehicle body deformation, and volume change to achieve the collection of multi-source static data, effectively extracts the multi-source static data, eliminates redundancy and noise, obtains multiple valid static data, and then fuses the multiple valid static data to obtain fused static data. A nonlinear relationship model between the vehicle body structure and the actual loaded mass is established, and then the fused static data is identified to calculate the mass and volume data of the vehicle.

[0146] Specifically, Figure 5 A structural block diagram of the mass and volume measurement unit 101 in the system provided in an embodiment of the present invention is shown.

[0147] In a preferred embodiment of the present invention, the mass-volume measurement unit 101 specifically includes:

[0148] The multi-source static data acquisition module 1011 is used to acquire multi-source static data.

[0149] The first effective extraction module 1012 is used to effectively extract the multi-source static data to obtain multiple effective static data.

[0150] Data fusion module 1013 is used to fuse multiple valid static data to obtain fused static data;

[0151] The mass and volume calculation module 1014 is used to identify the fused static data and calculate the mass and volume data of the vehicle.

[0152] Furthermore, the intelligent vehicle metering system based on AI technology also includes:

[0153] The dynamic identification and correction unit 102 is used to collect multi-source dynamic data, perform dynamic working condition identification and correction, and obtain vehicle dynamic data.

[0154] In this embodiment of the invention, the dynamic identification and correction unit 102 collects multi-source dynamic data such as speed and acceleration, effectively extracts the multi-source dynamic data, removes abnormal data, and obtains multiple valid dynamic data. Then, it performs dynamic condition identification on the multiple valid dynamic data for conditions such as acceleration, deceleration, idling, and constant speed to obtain direct dynamic data. After that, it combines the influence of vehicle load on the speed change trend in the mass and volume data to correct the direct dynamic data and obtain vehicle dynamic data.

[0155] Specifically, Figure 6 A structural block diagram of the dynamic identification correction unit 102 in the system provided by an embodiment of the present invention is shown.

[0156] In a preferred embodiment of the present invention, the dynamic identification correction unit 102 specifically includes:

[0157] The multi-source dynamic acquisition module 1021 is used to acquire multi-source dynamic data;

[0158] The second effective extraction module 1022 is used to effectively extract the multi-source dynamic data to obtain multiple effective dynamic data.

[0159] The dynamic operating condition identification module 1023 is used to identify the dynamic operating conditions of multiple valid dynamic data and obtain direct dynamic data.

[0160] The correction processing module 1024 is used to combine the mass and volume data to correct the direct dynamic data and obtain vehicle dynamic data.

[0161] Furthermore, the intelligent vehicle metering system based on AI technology also includes:

[0162] The vehicle emission prediction unit 103 is used to construct an emission condition prediction model based on AI technology and predict vehicle emission data based on the mass and volume data and the vehicle dynamic data.

[0163] In this embodiment of the invention, the vehicle emission prediction unit 103 collects emission source data such as NOx, CO2, and particulate matter. Based on AI technology, it simultaneously extracts parameters such as mass, speed, and operating conditions according to mass-volume data and vehicle dynamic data. These parameters are then combined with the emission source data to construct an emission operating condition prediction model. The mass-volume data and vehicle dynamic data are processed to generate first input feature data including vehicle mass, driving conditions, and acceleration. The first input feature data is then imported into the emission operating condition prediction model to obtain vehicle emission data for NOx, CO2, and particulate matter emissions per unit time.

[0164] The mileage energy consumption prediction unit 104 is used to construct a mileage energy consumption prediction model based on AI technology, and predict mileage energy consumption data according to the mass volume data, the vehicle dynamic data and the vehicle emission data.

[0165] In this embodiment of the invention, the mileage energy consumption prediction unit 104 collects multi-source mileage data through GPS trajectory, wheel speed, map matching, etc. Based on AI technology, it constructs a mileage energy consumption prediction model according to the multi-source mileage data, vehicle emission data, mass and volume data, and vehicle dynamic data. It processes the mass and volume data and vehicle dynamic data to generate second input feature data. By importing the second input feature data into the mileage energy consumption prediction model, the output mileage energy consumption data is obtained.

[0166] The meteorological verification and processing unit 105 is used to record and organize data, generate motor vehicle metering data, and supplement the metering meteorological data. When using the data, meteorological verification processing is performed.

[0167] In this embodiment of the invention, the meteorological verification processing unit 105 records and organizes mass volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data to generate motor vehicle metering data. It randomly selects multiple target factors from a set of preset meteorological factors, acquires metering meteorological data according to the multiple target factors, and adds the metering meteorological data to the motor vehicle metering data. Then, when using the data in subsequent applications, it acquires meteorological verification data related to the metering meteorological data. By verifying the meteorological verification data and the metering meteorological data against the target factors and corresponding values, it determines whether the motor vehicle metering data is true and reliable. If the meteorological verification fails, the motor vehicle metering data is determined to be invalid data.

[0168] Specifically, Figure 7 A structural block diagram of the meteorological verification and processing unit 105 in the system provided in an embodiment of the present invention is shown.

[0169] In a preferred embodiment provided by the present invention, the meteorological verification and processing unit 105 specifically includes:

[0170] The recording and processing module 1051 is used to record and process the mass and volume data, the vehicle dynamic data, the vehicle emission data and the mileage energy consumption data to generate motor vehicle measurement data.

[0171] The factor selection module 1052 is used to randomly select multiple target factors from a set of preset meteorological factors;

[0172] The meteorological data supplementation module 1053 is used to acquire and supplement metered meteorological data according to multiple target factors.

[0173] The meteorological verification module 1054 is used to verify the metering meteorological data and motor vehicle metering data when using the data, and to obtain the verification results.

[0174] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0175] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0176] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0177] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0178] 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, and improvements 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 establishing intelligent metering of motor vehicles based on AI technology, characterized in that, The method specifically includes the following steps: Collect multi-source static data, perform data fusion and identification, and calculate mass and volume data; Collect multi-source dynamic data, perform dynamic operating condition identification and correction, and obtain vehicle dynamic data; Based on AI technology, an emission condition prediction model is constructed to predict vehicle emission data according to the mass and volume data and the vehicle dynamic data. Based on AI technology, a mileage energy consumption prediction model is constructed to predict mileage energy consumption data according to the mass and volume data, the vehicle dynamic data, and the vehicle emission data. Data is recorded and organized to generate motor vehicle metering data, and meteorological metering data is added. Meteorological verification is performed before the data is used. The AI-based emission condition prediction model, which predicts vehicle emission data based on the mass-volume data and vehicle dynamic data, specifically includes the following steps: Collect emission source data; Based on AI technology, an emission condition prediction model is constructed according to the emission source data, the mass and volume data, and the vehicle dynamic data. Based on the mass and volume data and the vehicle dynamic data, the first input feature data is generated. The first input feature data is imported into the emission condition prediction model to obtain the output vehicle emission data; The emission condition prediction model is constructed through the following steps: Engine parameters, exhaust gas composition detection values, and OBD system data are obtained from emission source data. The engine parameters are hierarchically encoded to obtain the device feature code. The dynamic activity coefficient is obtained based on the exhaust gas composition detection value. The OBD system data is segmented to obtain the time-series state identifier. The encoding matrix is ​​obtained based on the device feature code, dynamic activity coefficient, and time-series state identifier. Mass volume distribution characteristics are obtained from mass volume data, and vehicle acceleration, steering angle, motion trajectory parameters, and vehicle wheelbase parameters are obtained from vehicle dynamic data. The mass volume distribution characteristics are matched with vehicle acceleration and steering angle respectively to obtain the mass distribution offset index. Based on the mass distribution offset index and motion trajectory parameters, the inertial force coefficient is calculated. Based on the inertial force coefficient and vehicle wheelbase parameters, a composite feature vector is generated. Based on the encoding matrix, an emission-quality correlation matrix is ​​constructed through feature cross-validation. A physical characteristic tensor is then constructed based on the emission-quality correlation matrix. Based on the composite feature vector, the operating condition evolution mode is extracted through a sliding window to construct a spatiotemporal characteristic tensor. Based on the physical characteristic tensor and the spatiotemporal characteristic tensor, an attention mechanism is used to allocate the output weights of the physical characteristic tensor and the spatiotemporal characteristic tensor and then fuse them to obtain a fused feature tensor. Redundant features in the fused feature tensor are removed by feature selection method to obtain a simplified feature tensor. Based on the simplified feature tensor, a preliminary prediction model is constructed with carbon conservation as a constraint. The simplified feature tensor is input into the preliminary prediction model to obtain the prediction result. The prediction result and the emission source data are sequentially aligned in the time domain and compared in similarity to obtain the comparison result. The preliminary prediction model is iteratively updated based on the comparison result. After reaching the preset number of iterations, the emission condition prediction model is obtained.

2. The method for establishing intelligent vehicle metering based on AI technology according to claim 1, characterized in that, The process of collecting multi-source static data, performing data fusion and identification, and calculating mass and volume data specifically includes the following steps: Collect multi-source static data; The multi-source static data is effectively extracted to obtain multiple valid static data; among which, the valid static data includes the vehicle's length, width, height, actual wheelbase, wheel track parameters, and point cloud data; Multiple valid static data sets are fused to obtain fused static data. The fused static data is identified, and the vehicle's mass and volume data are calculated.

3. The method for establishing intelligent vehicle metering based on AI technology according to claim 2, characterized in that, The multiple valid static data sets are fused to obtain fused static data. The specific steps are as follows: The region center point is obtained by multi-source static data. The region center point is used as the coordinate origin. The extension line of the coordinate origin in the east direction is used as the X-axis, the extension line of the coordinate origin in the north direction is used as the Y-axis, and the extension line of the coordinate origin in the vertical upward direction is used as the Z-axis to construct a three-dimensional spatial coordinate system. A 3D model is constructed based on valid static data, and 3D modeling data is obtained from the 3D model. Based on a three-dimensional spatial coordinate system, the three-dimensional modeling data and timestamps are aligned to obtain spatiotemporal fusion data; Using Dempster-Shafer as the main framework, a conflict detection model is constructed. Based on the conflict detection model, the spatiotemporal fusion dataset is verified and corrected to obtain the verified spatiotemporal fusion dataset. A set of confidence values ​​is obtained based on the verified spatiotemporal fusion dataset, and the verified spatiotemporal fusion data corresponding to the maximum confidence value in the set of confidence values ​​is taken as the optimal fusion data. The wheelbase Shannon entropy is calculated based on the wheelbase parameter. The wheelbase variation coefficient and mass percentage vector value are extracted from the optimal fused data. A multidimensional feature space is constructed based on the wheelbase variation coefficient, wheelbase Shannon entropy and mass percentage vector value. Based on a multidimensional feature space, a manifold learning framework is constructed using PCA as a foundation. The optimal fusion dataset is reduced in dimensionality using a manifold learning framework to obtain a fusion static dataset.

4. The method for establishing intelligent vehicle metering based on AI technology according to claim 3, characterized in that, The process of collecting multi-source dynamic data, performing dynamic condition identification and correction, and obtaining vehicle dynamic data specifically includes the following steps: Collect dynamic data from multiple sources; The multi-source dynamic data is effectively extracted to obtain multiple valid dynamic data. Dynamic operating condition identification is performed on multiple valid dynamic data to obtain direct dynamic data; By combining the mass and volume data, the direct dynamic data is corrected to obtain vehicle dynamic data.

5. The method for establishing intelligent vehicle metering based on AI technology according to claim 4, characterized in that, Based on the mass and volume data and the vehicle dynamic data, the first input feature data is generated through processing. The specific steps are as follows: The vehicle's mass and volume data and vehicle dynamic data are phase-synchronized using a sliding window, and then spatiotemporal mapping is performed to obtain a spatiotemporally synchronized fused data matrix. Acceleration timing parameters, steering angle change rate timing parameters, and mass center offset are obtained through spatiotemporally synchronized fusion data matrices. Waveform parameters are obtained by extracting acceleration timing parameters using waveform matching. Based on waveform parameters and vehicle mass and volume data, the equivalent inertial force eigenvector is calculated. Wavelet packet decomposition is used to perform multi-scale decomposition of the steering angle change rate time series parameters to obtain characteristic frequency bands. The energy proportion of the characteristic frequency bands is calculated as the frequency band characterization parameter. The actual wheel track is divided by the wheel track parameter to obtain the wheel track utilization rate. Based on the wheel track utilization rate and the center of mass offset, the coupling characteristic parameter is calculated. The three-dimensional feature vector is obtained based on the equivalent inertial force feature vector, the frequency band characterization parameter, and the coupling characteristic parameter. Principal component analysis is used to reduce the dimensionality of the static parameter components in the three-dimensional feature vector and orthogonalize the static parameter components to obtain the processed three-dimensional feature vector. Based on the processed three-dimensional feature vector, a local linear embedding algorithm is used to compress the dimensionality of the dynamic features in the three-dimensional feature vector to obtain a low-dimensional feature vector. The isolated forest algorithm is used to detect and verify the parameter outliers of the low-dimensional feature vector to obtain the verified low-dimensional feature vector. Based on the verified low-dimensional feature vector, the abnormal data fragments are repaired using linear interpolation to obtain the repaired feature vector. The static features in the repaired feature vector are processed by the maximum-minimum normalization method to obtain the standardized static features. The dynamic features in the repaired feature vector are standardized by using a sliding window Z-score to obtain the standardized dynamic features. The standardized static features and standardized dynamic features are weighted and concatenated to obtain the first input feature data.

6. The method for establishing intelligent vehicle metering based on AI technology according to claim 5, characterized in that, The AI-based mileage energy consumption prediction model, which predicts mileage energy consumption data based on the mass-volume data, vehicle dynamic data, and vehicle emission data, specifically includes the following steps: Collect multi-source mileage data; Based on AI technology, a mileage energy consumption prediction model is constructed according to the multi-source mileage data, the vehicle emission data, the mass and volume data, and the vehicle dynamic data. The mass volume data and the vehicle dynamic data are processed to generate second input feature data; The second input feature data is imported into the mileage energy consumption prediction model to obtain the output mileage energy consumption data.

7. The method for establishing intelligent vehicle metering based on AI technology according to claim 6, characterized in that, The process of recording and organizing data to generate motor vehicle metering data, supplementing it with meteorological data, and performing meteorological verification processing before data use specifically includes the following steps: The mass and volume data, vehicle dynamic data, vehicle emission data, and mileage energy consumption data are recorded and organized to generate motor vehicle metering data; Multiple target factors are randomly selected from a set of pre-defined meteorological factors. According to the multiple target factors mentioned above, acquire and supplement the measurement meteorological data; When using the data, the meteorological data and motor vehicle measurement data are verified, and the verification results are obtained.

8. A smart metering system for motor vehicles based on AI technology, characterized in that: The system applies the AI-based intelligent metering method for motor vehicles as described in any one of claims 1 to 7, and the system includes: The mass and volume measurement unit is used to collect multi-source static data, perform data fusion and identification, and measure mass and volume data. The dynamic identification and correction unit is used to collect multi-source dynamic data, perform dynamic operating condition identification and correction, and obtain vehicle dynamic data. The vehicle emission prediction unit is used to build an emission condition prediction model based on AI technology and predict vehicle emission data based on the mass and volume data and the vehicle dynamic data. The mileage energy consumption prediction unit is used to build a mileage energy consumption prediction model based on AI technology, and predict mileage energy consumption data based on the mass volume data, the vehicle dynamic data and the vehicle emission data. The meteorological verification and processing unit is used to record and organize data, generate motor vehicle metering data, and supplement the metering meteorological data. When using the data, meteorological verification processing is performed.

Images