Construction and precision evaluation method for virtual pavement model of automobile test field

Abstract

The invention relates to the technical field of digital modeling, in particular to a construction and precision evaluation method for a virtual road surface model of an automobile test field. The method comprises the following steps: evaluating the reliability level of data acquired by each component of a vehicle-mounted laser scanning system to screen data meeting virtual road surface modeling requirements, and converting the data into a road surface model format applied to a virtual test field technology; importing the built tire model, the whole vehicle model and the pavement model in a simulation environment to carry out multi-body dynamics simulation; comparing a load spectrum in a simulation environment with a load spectrum in an actual environment, and evaluating vehicle dynamics response characteristics of the pavement model; and evaluating the reduction degree of the physical pavement by using the virtual pavement model in combination with a false damage ratio range in a simulation environment and an actual environment. According to the method, multi-dimensional evaluation is carried out on the precision of the virtual pavement model, and reliable application of the pavement model in virtual simulation is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of digital modeling, and in particular, to a method for constructing and precision evaluating a virtual road surface model of an automotive test field. Background Art

[0002] With the change of the competition pattern in the automotive industry and the progress of automotive R & D technologies, establishing and improving the virtual verification and development system for automotive products is of great significance for enhancing the core competitiveness of automotive enterprises. As a representative progress in the field of virtual simulation technology, virtual test field technology has, to a certain extent, changed the dependence of traditional automotive reliable and durable performance development on physical vehicle tests. The load spectrum acquisition technology based on a virtual test field digitizes the road surface of a physical test field, combines a tire model and a vehicle multi-body dynamics model, and through simulation, the mechanical response characteristics between various chassis connectors and between the connection points of the chassis and the body can be calculated. Using these mechanical characteristics as input conditions, fatigue analysis and verification of various automotive components can be carried out, thus supporting the development of automotive reliable and durable performance. The application of this technology can significantly shorten the automotive development cycle, reduce development costs, and improve product quality at the same time.

[0003] The construction methods of special test road surfaces in an automotive test field are mainly divided into two categories according to road surface characteristics: modeling of regular test road surfaces (such as vibration roads, twist roads, pothole roads, etc.), which mainly relies on construction design drawings and is modeled by means of manual drawing conversion.

[0004] Random excitation test road surfaces (such as Belgian roads, cobblestone roads, etc.) are mostly modeled by laser scanning. The specific method is as follows: the elevation information of each point on the road surface is scanned by a laser scanning system to obtain point cloud data of the road surface, and through grid processing such as screening and interpolation of the point cloud data, a digital road surface (CRG format) available for multi-body dynamics software simulation is obtained. Compared with the manual modeling method, laser scanning technology can achieve millimeter-level precise measurement, accurately obtain road surface characteristic information, and can greatly improve the road surface modeling accuracy.

[0005] The engineering application of virtual test field technology needs to ensure the consistency between the virtual load output and the real vehicle load spectrum. In actual simulation, the accuracies of the vehicle model, tire model, and road surface model will all introduce load errors to varying degrees, resulting in limited accuracy of the road surface model. Moreover, there is currently no precision evaluation method for virtual road surfaces in the existing technology. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for constructing and precision evaluating a virtual road surface model of an automotive test field, to conduct multi-dimensional evaluation of the precision of the virtual road surface model, and to ensure the precision of the road surface model.

[0007] To achieve the above purpose, the present invention adopts the following technical solutions:

[0008] The present invention provides a method for constructing and precision evaluating a virtual road surface model of an automobile test field, including:

[0009] Using a metal reference gauge with standard dimensions to evaluate the measurement precision of the vehicle-mounted laser scanning system and confirm the usability of the vehicle-mounted laser scanning system;

[0010] Scanning the test road surface through the vehicle-mounted laser scanning system and obtaining scanning data, wherein the vehicle-mounted laser scanning system includes a radar, a positioning device, an inertial measurement unit, a camera unit, etc.; the scanning data includes laser data, inertial navigation data, positioning data, etc.;

[0011] Confirming the precision level of the point cloud data formed by it through fusion evaluation of multi-source data in the scanning data;

[0012] Constructing a road surface model according to the scanning data in the ASAM OpenCRG standard open-source program software environment;

[0013] Constructing an Ftire tire model and a vehicle multi-body dynamics model in a simulation environment;

[0014] Evaluating the vehicle dynamics response characteristics of the road surface model according to the comparison of the load spectra in the simulation environment and the actual environment;

[0015] If the vehicle dynamics response characteristics meet the requirements, combining the range of the pseudo-damage ratio of the wheel center load in the simulation environment and the actual environment, evaluating the reduction degree of the road surface model to the real road surface.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] The present invention provides a method for constructing and precision evaluating a virtual road surface model of an automobile test field, aiming to evaluate the reduction degree of the real road surface in multiple dimensions. The present invention is based on a vehicle-mounted three-dimensional laser scanning system to scan a specially paved test road surface of the test field, and evaluates the reliability of the original data for modeling through a multi-source scanning data fusion precision evaluation method; based on virtual test field technology, by comparing the characteristics distribution and damage of the load on the real vehicle wheel center caused by the road surface excitation with the virtual load obtained from virtual simulation, further evaluating the reduction degree level of the virtual road surface model to the real road surface in the final application environment. Description of the Drawings

[0018] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is a schematic flowchart of a method for constructing and evaluating the accuracy of a virtual road surface model of an automotive test field provided by an embodiment of the present invention. Specific embodiments

[0020] The following will describe the exemplary embodiments of the present invention with reference to the accompanying drawings. Various details of the embodiments of the present invention are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted below.

[0021] Embodiment 1

[0022] The embodiments of the present invention are applicable to scenarios where a virtual road surface model in a simulation environment is constructed and the accuracy of the constructed model is evaluated. The virtual road surface is to convert the special test road surface of an automotive test field into a 3D digital model to generate a file format that can be used for multi-body dynamics simulation (mainly including RDF three-dimensional equivalent volume road surface, RGR regular grid road surface, and CRG curve regular grid road surface). For test road surfaces whose external features cannot be accurately measured and there are no accurate and detailed construction design drawing parameters for reference (such as Belgian roads, cobblestone roads, etc.), the present invention needs to use laser scanning to construct and convert the model.

[0023] Currently, there are mainly two types of vehicle-mounted laser scanning systems applied to the digitalization of test road surfaces. One is a mechanically rotating lidar, and the other is a solid-state lidar. Among them, the scanning system based on solid-state lidar has advantages such as simple application, low cost, and short cycle. The basic principle of scanning based on solid-state lidar is the triangulation ranging principle: the lidar emits laser light, and the reflected light passes through the receiving lens and hits the linear array CMOS (an image sensor). According to the distance between the light spot hitting the CMOS and the main optical axis, the distance between the road surface point and the lidar is calculated using the principle of similar triangles. An equally spaced and uniformly distributed laser array can simultaneously obtain point cloud data information such as the distance, angle, and position of each point on the road surface in the field of view angle area. The accuracy level of the scanned data largely determines the accuracy level of the road surface model.

[0024] On-vehicle lidar scanning systems are more suitable for scanning and modeling special paved test road surfaces in test sites due to their advantages such as high resolution, high precision, and high efficiency. This system integrates a radar (such as a lidar), a positioning device (such as a GPS), an INS inertial measurement unit, a wheel speed sensor, a digital camera, and an industrial computer. Among them, the radar is used to obtain original measurement data, including the distance, angle, time, and serial number of the scanning points from the target points on the road to the center point of the lidar, that is, point cloud data; the GPS / INS integrated navigation system is used to obtain the real-time position of the vehicle and the vehicle attitude data of the on-vehicle lidar scanning system in the geodetic coordinate system, and specific algorithms are used to fuse GPS and INS data with the GPS measurement data as the time reference. The wheel speed sensor is used to obtain the real-time speed information of the vehicle and the vehicle driving mileage data. The digital camera is used to synchronously obtain the gray-scale information and texture information of the measured road surface. The industrial computer is used for time synchronization and spatial coordinate transformation of the sensor components of the scanning system, so that the start times of the sensors of the system are consistent, and all data is registered in the same coordinate system.

[0025] Generally speaking, the scanning modeling and accuracy evaluation of the test road surface provided in this embodiment mainly include the following six tasks: 1) Scanning preparation. This includes system sensor calibration configuration (keeping time synchronization and spatial basic registration of each sensor under the current installation conditions) and road surface foreign object cleaning (such as gravel, weeds, etc.); 2) Scanning acquisition. The scanning vehicle travels uniformly along the road center line, maintaining a low speed (such as about 15 km / h). At least two sets of effective point cloud data should be collected for each road. The transition road surfaces at the front and rear of the road should be collected for at least 20 meters respectively to ensure the integrity of the later digital road surface during the simulation process; 3) Scanning data verification. Evaluate the data such as the positioning accuracy of the radar and vehicle-mounted laser scanning system and the vehicle attitude during the acquisition process, and check the point cloud quality with the help of point cloud processing software. There should be no situations such as missing point clouds in blocks or strips, distortion of the road surface feature contours, and clusters of burrs; 4) Point cloud data preprocessing. The results of point cloud data preprocessing will directly affect the later modeling accuracy and quality. The main preprocessing methods of point cloud data include: denoising, interpolation, compression, etc.; 5) Point cloud data reconstruction. Generally, the point clouds obtained by scanning only have position information (three-dimensional coordinates) and feature information (intensity values and texture values), but no topological relationship between data points. Point cloud reconstruction is to restore the geometric model of the original surface from the measured point cloud data. First, it is necessary to add topological structure information to the scattered points, usually using triangular mesh subdivision to establish the topological relationship between the scattered points. On this basis, a certain mathematical form is used to describe the target road surface; 6) CRG road surface model conversion. The processed point cloud data first needs to derive the road reference line with the smallest lateral displacement and smooth curvature from the driving trajectory of the scanning vehicle. Then, the reference line is discretized into an equidistant point sequence, and vertical lines of the reference line are generated at each equidistant point. According to the requirements of the CRG grid resolution, reference points are generated on the vertical lines. The reference line is the u-axis (vehicle driving direction) and v-axis (road surface width direction) of the coordinate system, and the u-axis and v-axis are perpendicular to each other. The u-axis is divided at equal intervals, and the interval of the v-axis can be defined flexibly. Finally, by assigning elevation data to each grid node of the point cloud data, the format conversion of the point cloud data to the CRG model can be completed. 7) Construct a road surface model in the simulation model, and conduct load spectrum comparison and pseudo-damage benchmarking to conduct multi-dimensional evaluation of the accuracy of the road surface model.

[0026] Next, in combination with Figure 1 the method provided in the embodiment of the present invention will be described in detail. The method provided in this embodiment can be executed by an electronic device.

[0027] S110. Use a metal reference gauge with a standard size to evaluate the measurement accuracy of the vehicle-mounted laser scanning system, and confirm the usability of the vehicle-mounted laser scanning system.

[0028] "Elevation" refers to the distance from a certain point along the plumb line direction to the absolute reference plane, called absolute elevation, simply elevation. The elevation data (vertical) at each point on the road surface is the most core data of the virtual road surface model, and it is necessary to verify and evaluate the accuracy level of the elevation data. This step requires evaluating the accuracy of the vehicle-mounted laser scanning system before scanning. Optionally, use the vehicle-mounted laser scanning system to scan a metal reference gauge of standard size to obtain the size scanning data of the metal reference gauge; compare the size scanning data with the true size of the metal reference gauge of standard size.

[0029] Specifically, in the present invention, a cuboid metal reference gauge with standard size is made, placed on a flat road surface, and the metal reference gauge is scanned by the vehicle-mounted laser scanning system to be measured. Error analysis is carried out on the CSV data and LAS point cloud data obtained by scanning with the true elevation data of the metal reference gauge. For example, the Z coordinate of the point cloud data is the absolute height of the point in the world coordinate system. Select a point on the top surface of the metal reference gauge and a point on the ground, and the difference between the two vertical coordinates is the height of the metal reference gauge. Select points at multiple locations in the LAS (LIDAR Data Exchange Format) point cloud data and calculate the error from the true height of the metal reference gauge, and this error must be controlled within 1 mm. In addition, as the original data scanned by the device, find the elevation data near the metal reference gauge in the CSV table, plot it as a curve graph, and calculate the measured value of the gauge height through the elevation change in the curve graph. For example, the average value of the ordinates (vertical direction) of multiple curves at the high and low positions of the curve, and the error from the true height of the metal reference gauge must be controlled within 1 mm. In addition, for the accuracy of the scanned data in the horizontal and vertical directions, the elevation sampling data points of each point cloud data, each pixel of the PGM (Portable Gray Map) grayscale image, and the grid node positions of the final CRG model correspond exactly. The length and width dimensions of the entire road surface in the PGM grayscale image can be compared with the actual values of the test road surface, and the deviation needs to be controlled within 1%.

[0030] S120. Scan the test road surface through the vehicle-mounted laser scanning system and obtain the scan data. The vehicle-mounted laser scanning system includes a radar, a positioning device, an inertial measurement unit, and a camera unit; the scan data includes laser data, inertial navigation data, and positioning data.

[0031] S130. Confirm the accuracy level of the point cloud data formed by fusing and evaluating the multi-source data in the scan data.

[0032] The vehicle carrying the vehicle-mounted laser scanning device travels at a low speed in a steady state and collects data.

[0033] The raw data collected by the vehicle-mounted laser scanning system is multi-dimensional, such as laser data, inertial navigation data, and positioning data. These data will all affect the accuracy of the final virtual road surface model. Therefore, to evaluate the accuracy of the virtual road surface model, it is necessary to fuse and evaluate the original multi-source data obtained by the vehicle-mounted laser scanning system to obtain an evaluation of the reliability. If the reliability exceeds the set threshold, the reliability requirement is met. The set threshold can be set according to the accuracy requirement.

[0034] S140. In the ASAM OpenCRG standard open-source program environment, construct a road surface model based on the scanning data.

[0035] The process of constructing the road surface model is to convert the special test road surface of the vehicle test site into a 3D digital model to generate a file format that can be used for multi-body dynamics simulation, mainly including: RDF three-dimensional equivalent volume road surface, RGR regular grid road surface, CRG curve regular grid road surface, etc.

[0036] S150. Construct the Ftire tire model and the vehicle's multi-body dynamics model in the simulation environment.

[0037] The force exerted by the road surface on the whole vehicle is transmitted by the tires. However, the tire structure is complex and its various performances are difficult to grasp, and the grounding condition of the tires during vehicle driving is constantly changing in real time. Therefore, the construction of tire models and the extraction of attribute parameters have always been issues concerned and studied by many scholars at home and abroad. The Ftire tire model applied in the present invention can achieve a relatively high frequency and high simulation accuracy for durability simulation analysis (in engineering practice, the accuracy level of the Ftire tire model can reach more than 80%). It has obvious advantages compared with other tire models. The vehicle's multi-body dynamics model is the most core part of the virtual test field technology. Before the vehicle's multi-body dynamics model is applied in the virtual test field simulation, it is necessary to debug and check the model first. The debugging work includes: flexible debugging of some components (such as components that are prone to deformation like the stabilizer bar), input of parameters of elastic components (such as rubber bushings, springs, and shock absorbers), static characteristic debugging and dynamic characteristic debugging of the suspension (for example, performing simulation according to the suspension KC characteristics, respectively establishing front and rear suspension assembly models, and checking the performance of the suspension model).

[0038] S160. Compare the load spectra in the simulation environment and the actual environment, and evaluate the vehicle dynamics response characteristics of the road surface model.

[0039] Compare the load spectra of the three-channel signals of the lateral force, longitudinal force, and vertical force at the wheel center in the simulation environment and the actual environment. The comparison dimensions include: signal coincidence degree in the time domain, coincidence degree of load amplitude frequency distribution, and coincidence degree of power spectral density curves in the frequency range of 5 - 20 Hz.

[0040] Specifically, the wheel center forces of an automobile refer to the three forces (longitudinal wheel center force, lateral wheel center force, and vertical wheel center force) acting on the wheel center by the road surface during vehicle driving. As an important parameter in vehicle dynamics analysis, it is an important input signal for the actual driving conditions in fatigue life prediction through CAE analysis and virtual iteration in the development of vehicle reliable and durable performance. For example, the vertical wheel center force is mainly related to factors such as the normal force and traction force of the tire, that is, the so-called tire load. In actual situations, the forces on the four wheels are affected by wheel loads, road surface conditions, friction, steering angle, traction force under different driving conditions, etc. This requires simultaneous and separate acquisition of the wheel center forces of each wheel. In engineering practice, it can be collected during actual driving through a wheel six-component force sensor, or calculated and extracted in simulation software through virtual test field technology. Since the tire is the only component that directly contacts the road surface of the automobile, using the wheel center force to evaluate the road surface conditions has less interference compared to other vehicle load signals.

[0041] The on-vehicle load spectrum refers to the graph or table of the relationship between the load and time when the vehicle is driving on the road, mainly including load signals such as force, acceleration, and displacement. Commonly used load acquisition sensors include wheel six-component force sensors, accelerometers, displacement sensors, and GPS. The acquisition conditions of the on-vehicle load spectrum are compiled according to the benchmarking requirements, and it is necessary to ensure that the wheel loads, tire pressures, vehicle speeds, time, etc. of each wheel of the vehicle are consistent with the simulation conditions. Each characteristic road surface condition should be collected at least 3 times. Before starting the acquisition, zero adjustment is required to ensure that the chassis components are hardly affected by external forces other than the vehicle gravity. After completing the acquisition work number, it is necessary to intercept the data separately according to each characteristic road surface to prepare for the comparison of the virtual and real load spectra.

[0042] When conducting virtual test field simulation, it is necessary to import the debugged vehicle multi-body dynamics model and the road surface model of the test field into the MSC.Adams software, set information such as vehicle speed and time that is the same as the driving conditions of the on-vehicle load spectrum acquisition, and perform MSC.Adams software simulation to obtain the virtual load spectrum. The virtual load signals output by the simulation can be compared with the on-vehicle load signals collected by the wheel six-component force sensor during the actual vehicle driving process. Before the comparison, it is necessary to perform preprocessing such as filtering and deburring on the two load signals.

[0043] The present invention preliminarily evaluates the simulation application accuracy level of the road surface model based on the comparison results of the virtual and real load spectra of the three-channel signals of the longitudinal, lateral, and vertical wheel center forces. The main dimensions include: signal coincidence degree in the time domain, coincidence degree of load amplitude frequency distribution, coincidence degree of power spectral density curves in the frequency domain, etc.

[0044] Specifically, the signal matching degree in the time domain is evaluated by the maximum value, minimum value and root mean square ratio of the force values passing through the same channel. Among them, the root mean square ratio needs to meet the range of 0.5 to 2. The matching degree of the load amplitude frequency distribution is evaluated by the amplitude range, extreme value size and load frequency in the load rain flow counting histogram; the matching degree of the power spectral density curve in the frequency domain mainly focuses on the coincidence degree of the power spectral density (PSD) curve in the frequency band below 50 Hz. Through the research on road surface excitation, the road surface excitation is mainly concentrated in the frequency band of 5 - 20 Hz, which is suitable for the research on the frequency domain response of the road surface. Therefore, this embodiment focuses on evaluating the matching degree of the power spectral density curve in the frequency band of 5 - 20 Hz.

[0045] S170. If the vehicle dynamic response characteristics meet the requirements, the reduction degree of the road surface model to the real road surface is evaluated by combining the range of the pseudo - damage ratio of the wheel center load in the simulation environment and the actual environment.

[0046] According to the Miner linear damage accumulation criterion: when a metal component is subjected to an alternating load, it will absorb the energy caused by deformation. The absorbed energy w i is proportional to the number of alternating load cycles n i That is, W is the total energy that causes fatigue damage to the component, N is the total number of cycles that cause fatigue damage to the component under the load, w i is the actually absorbed energy, and n i is the actual number of load cycles. In automotive durability development, the cumulative fatigue damage (i.e., pseudo - damage) of automotive components after cyclic loading at different load levels can be calculated according to the above - mentioned proportional relationship. Based on this, the present invention comprehensively and quantitatively evaluates the reduction degree of the virtual road surface model to the real road surface through the comparison results of the virtual - to - real load pseudo - damage ratios of the signals in the three channels of the wheel center longitudinal force, lateral force and vertical force. For example, the ratio of the pseudo - damage of the real vehicle to the pseudo - damage of the simulation for the same wheel position should be within the specified range. For example, the pseudo - damage ratio of the wheel center vertical force channel should meet 0.5 - 2, the pseudo - damage ratio of the wheel center longitudinal force channel should meet 0.33 - 3, and the pseudo - damage ratio of the wheel center lateral force channel should meet 0.2 - 5. Then it can be considered that the accuracy level of this road surface model meets the requirements of multi - body dynamics simulation applications and can restore the real test field road surface.

[0047] In summary, based on engineering practice research, the present invention proposes a digital modeling method for special test roads in an automotive test field based on three-dimensional laser scanning technology, and further proposes a multi-dimensional comprehensive evaluation system for the accuracy of a virtual road surface model, including evaluating the measurement accuracy of road surface parameters of the scanning device with the aid of a reference gauge block, evaluating the accuracy level of point cloud data through the fusion of original multi-source scanning data, evaluating the vehicle dynamics response characteristics of the road surface model based on the comparison of virtual and real load spectra using virtual test field technology, and finally, combining the pseudo-damage ratio of virtual and real loads to comprehensively evaluate the accuracy of the road surface model and its degree of restoration of the real road surface. This step-by-step method system can provide an effective basis for an enterprise to carry out fatigue load prediction and analysis by applying virtual test field technology during the vehicle structure design stage of a new vehicle model. At the same time, the present invention also provides a practical solution for the digital technology service project of test road surfaces in an automotive test field. This achievement also lays a solid foundation for the application and development of virtual test technology; at the same time, through this method system, a virtual road surface model with higher accuracy can be obtained.

[0048] Embodiment 2

[0049] On the basis of the above embodiment, this embodiment details the calculation process of reliability. Evaluating the accuracy level of the vehicle-mounted laser scanning system according to the scanning data includes the following steps:

[0050] S210. Confirm the type of the currently scanned test road surface.

[0051] The types of the current test road surface include, but are not limited to, Belgian roads, cobblestone roads, potholed roads, and twisted roads, etc.

[0052] S220. On the current test road surface, determine the first accuracy of the laser data, the second accuracy of the inertial navigation data, and the third accuracy of the positioning data accuracy.

[0053] For example, multiple groups of point cloud data are collected on the current test road surface, and the minimum change unit of this group of point cloud data is determined as the first accuracy. Similarly, the second accuracy of the attitude data (such as the minimum change unit of the pitch angle) and the third accuracy of the positioning data (such as the minimum change unit of the vertical height) are obtained.

[0054] The first accuracy, the second accuracy, and the third accuracy need to be quantified and then applied to the reliability calculation formula. The higher the accuracy, the higher the quantization value. For example, if the accuracy of 0.01 is higher than that of 0.1, then the accuracy of 0.01 is quantified as 10, and the accuracy of 0.1 is quantified as 1.

[0055] S230. Calculate the reliability f of the vehicle-mounted laser scanning system according to the following formula:

[0056] f = k1×m1 + k2×m2 + k3×m3; Formula (1)

[0057] Among them, m1 is the quantization value of the first precision, m2 is the quantization value of the second precision, m3 is the quantization value of the third precision, and the higher the precision, the higher the reliability. k1, k2, and k3 are weights, which have a corresponding relationship with the type of the current test road surface.

[0058] Considering that the laser data precision, inertial navigation data precision, and positioning data precision contribute differently to the generation of the road surface model and have different degrees of influence on the reliability of the vehicle-mounted laser scanning system, the weights of various data precisions are introduced.

[0059] For example, for test road surfaces with high-frequency random excitation (such as Belgian roads and cobblestone roads), the weight of laser data can be taken as 0.6, the weight of positioning data can be taken as 0.3, and the weight of inertial navigation data can be taken as 0.1. For test road surfaces with medium-frequency random excitation (such as vibrating roads and potholed roads), the weight of laser data can be taken as 0.6, the weight of positioning data can be taken as 0.2, and the weight of inertial navigation data can be taken as 0.2. For test road surfaces with low-frequency random excitation (such as twisted roads), the elevation range of road surface features is large, the weight of laser data can be taken as 0.5, the weight of positioning data can be taken as 0.2, and the weight of inertial navigation data can be taken as 0.3.

[0060] The following introduces a method for determining the weights (i.e., k1, k2, and k3).

[0061] Perform tests under multiple test road surfaces in advance to obtain multiple sets of scanning data (including laser data, inertial navigation data, and positioning data) and the true elevation data of each test road surface (assuming that the true elevation data of each point on the test road surface is known). By analyzing the scanning data, the quantization values of the precision of various scanning data are obtained. Substitute the quantization values of the precision into the calculation formula of the reliability f, and the weights are taken as the unknowns to be identified and need to be solved. The reliability f is obtained from the difference between the elevation data and the true elevation data. For example, the following formula is used to calculate the elevation difference E:

[0062]

[0063] Among them, N is the number of elevation data points, j is a variable, D fused , j is the elevation data obtained from the jth set of scanning data, and D rea1 , j is the true elevation data at the same geographical location as the jth set of scanning data.

[0067] Then, a multiple linear model is constructed according to the formula of the reliability f, and k1, k2, and k3 are iterated to minimize the gap between the elevation data in the scanned data and the true elevation data, thereby obtaining the weights of different types of road surfaces. For example, for a torque road, the elevation range of the road surface features is relatively large. The weight of the point cloud data accuracy can be taken as 0.5, the weight of the positioning data accuracy can be taken as 0.2, and the weight of the attitude data accuracy can be taken as 0.3.

[0068] After the weights and data accuracies are determined, each data is substituted into formula (1) to calculate the reliability f. If the reliability f exceeds the set threshold, the reliability requirement is met; otherwise, the reliability requirement is not met.

[0069] This embodiment provides a scientific and reasonable reliability evaluation method, which assigns different weights and precision influence factors according to the contributions of each data source to the generation of the road surface model and their precision levels, so as to comprehensively evaluate the reliability level of the scanned data.

[0070] It should be understood that various forms of the processes shown above can be used, reordering, adding, or deleting steps. For example, the steps described in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in the present invention can be achieved. There is no limitation herein.

[0071] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub - combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground, characterized in that: include: Using a metal reference gauge block of standard size, the measurement accuracy of the vehicle-mounted laser scanning system is evaluated to confirm the feasibility of the vehicle-mounted laser scanning system; Scanning the test road surface by means of the vehicle-mounted laser scanning system and obtaining scanning data, wherein the vehicle-mounted laser scanning system includes a radar, a positioning device, an inertial measurement unit and a camera unit; the scanning data includes laser data, inertial navigation data and positioning data; By performing fusion evaluation on the multi-source data in the scan data, the accuracy level of the point cloud data constituted thereby is confirmed; In the ASAM OpenCRG standard open source program software environment, constructing a road surface model based on the scan data; Construct Ftire tire model and vehicle multi-body dynamics model in the simulation environment; By comparing the load spectra in the simulation environment and the actual environment, the vehicle dynamics response characteristics of the road surface model are evaluated; If the vehicle dynamics response characteristics meet the requirements, the degree of restoration of the road surface model to the real road surface is evaluated in combination with the range of the wheel center load pseudo-damage ratio in the simulation environment and the actual environment.

2. The method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground according to claim 1, characterized in that: Use standard size metal reference blocks to evaluate the measurement accuracy of the vehicle-mounted laser scanning system, including: A vehicle-mounted laser scanning system is used to scan a metal reference gauge block of standard size to obtain the size scanning data of the metal reference gauge block; The dimensional scan data is compared with the actual dimensions of the metal reference gauge block.

3. The method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground according to claim 1, characterized in that: The accuracy level of the point cloud data formed by the multi-source data in the scan data is confirmed by fusion evaluation, including: Confirm the type of the test road surface currently being scanned; On the current test road surface, determine the first accuracy of the laser data, the second accuracy of the inertial navigation data and the third accuracy of the positioning data; The reliability f of the vehicle-mounted laser scanning system is calculated according to the following formula: f = k1 × m1 + k2 × m2 + k3 × m3; Among them, k1, k2, k3 are weights, which correspond to the type of the current test road surface, m1 is the quantized value of the first precision, m2 is the quantized value of the second precision, and m3 is the quantized value of the third precision. The higher the precision, the higher the reliability.

4. The method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground according to claim 3, characterized in that: Before calculating the reliability f of the vehicle-mounted laser scanning system according to the following formula, a weight determination process is also included: Conduct tests on various test pavements in advance to obtain multiple sets of scanning data and the actual elevation data of each test pavement; A multivariate linear model is constructed according to the formula of reliability f, and the weights of different types of road surfaces are obtained by minimizing the gap between the elevation data in the scanned data and the actual elevation data.

5. The method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground according to claim 4, characterized in that: Before calculating the reliability f of the vehicle-mounted laser scanning system according to the following formula, a quantification process of the first accuracy, the second accuracy and the third accuracy is also included: The higher the first precision, the second precision, and the third precision are, the higher the quantization value is.

6. The method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground according to claim 1, characterized in that: By comparing the load spectra in the simulation environment with the load spectra in the actual environment, the vehicle dynamics response characteristics of the road model are evaluated, including: Compare the load spectra of the three channel signals of wheel center lateral force, wheel center longitudinal force and wheel center vertical force in the simulation environment and the actual environment; The comparison dimensions include: signal consistency in the time domain, load amplitude frequency distribution consistency, and power spectrum density curve consistency in the 5-20 Hz frequency band in the frequency domain.

7. The method for constructing and evaluating the accuracy of a virtual road model of an automobile testing ground according to claim 1, characterized in that: Combined with the range of pseudo damage ratios in the simulation environment and the actual environment, the restoration degree of the pavement model is evaluated, including: The restoration degree of the road surface model is evaluated by the range of the virtual load and real load pseudo-damage ratio of the three channel signals of the wheel center longitudinal force, lateral force and vertical force.

Images