Vehicle-mounted laser IRI detection method based on acceleration vibration compensation

Abstract

The application discloses a kind of vehicle-mounted laser IRI detection methods based on acceleration vibration compensation, it is related to road flatness detection technical field, including obtaining the road elevation survey data of laser range finder collected in the process of vehicle driving, the vehicle vibration acceleration data of acceleration sensor collected and the vehicle mileage data of encoder collected, and according to the time sequence of collection road elevation survey data, vehicle vibration acceleration data and vehicle mileage data are time-synchronized and arranged to generate time series detection data.In the application, by uniform time marking and time-synchronized and arranged processing to road elevation survey data, vehicle vibration acceleration data and vehicle mileage data, the three types of data form a one-to-one correspondence under the same time axis, so as to establish a complete time series detection data structure, provide a unified data basis for subsequent vibration compensation calculation, improve the overall consistency and continuity of data processing process.

Description

Technical Field

[0001] This invention relates to the field of road smoothness detection technology, and in particular to a vehicle-mounted laser IRI detection method based on acceleration vibration compensation. Background Technology

[0002] Road surface smoothness is an important technical indicator for evaluating road performance and driving comfort. The International Roughness Index (IRI) is a widely used standard evaluation indicator in the field of road engineering. The IRI value is usually calculated from the longitudinal profile elevation data of the road. The result can reflect the vibration response of vehicles when driving on the road surface. Therefore, it is widely used in road construction quality inspection, road maintenance assessment and road performance monitoring.

[0003] Current IRI (Infrared Reflection) detection typically employs vehicle-mounted laser detection systems. A laser rangefinder mounted on the vehicle continuously scans the road surface to acquire longitudinal profile elevation data. This data, combined with vehicle travel distance information, forms a road elevation sequence, which is then used to calculate the IRI value using a suspension response model. However, during vehicle operation, vibrations caused by the suspension system, tire contact, and road surface excitation alter the relative position between the laser rangefinder and the road surface, resulting in an additive effect of vehicle vibration errors on the measured elevation data. To mitigate this impact, some technologies incorporate accelerometers and compensate through integration or filtering. However, vibration acceleration data contains zero bias, low-frequency drift, and noise interference, easily leading to cumulative integration errors and displacement drift. Furthermore, inconsistencies in spatial alignment and sampling intervals between elevation and vibration displacement data also affect the accuracy of IRI calculations. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a vehicle-mounted laser IRI detection method based on acceleration vibration compensation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a vehicle-mounted laser IRI detection method based on acceleration vibration compensation, comprising the following steps: acquiring road elevation measurement data collected by a laser rangefinder, vehicle vibration acceleration data collected by an acceleration sensor, and vehicle mileage data collected by an encoder during vehicle operation; performing time synchronization processing on the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data according to the acquisition time sequence to generate time-series detection data; performing DC removal processing on the vehicle vibration acceleration data in the time-series detection data to eliminate the DC component in the vehicle vibration acceleration data and generate zero-bias correction acceleration data; performing Fourier transform processing on the zero-bias correction acceleration data, extracting the vehicle bump vibration frequency band according to the frequency domain distribution of the vibration signal and filtering out noise signals to generate vibration separation acceleration data; performing Simpson integral algorithm calculation processing on the vibration separation acceleration data according to the acquisition time sequence, constructing an integral calculation unit for the vibration separation acceleration data according to the three consecutive sampling intervals, and performing calculation within each integral calculation unit according to... The Simpson integral weights are reconstructed using midpoint weight adjustment parameters and curvature compensation parameters. The reconstructed acceleration integral is then subjected to velocity and displacement integral calculations to generate vehicle vibration displacement data. Singular spectrum analysis is applied to this data to construct a trajectory matrix of the vehicle vibration displacement sequence. Singular value decomposition is performed on the trajectory matrix, and the number of singular components participating in trend reconstruction is determined based on the trend energy proportion parameter. A weighted diagonal average reconstruction is then applied to the trend matrix using a reconstruction weighting index parameter to separate the trend components from the vehicle vibration displacement data, generating displacement correction data. Spatial distance alignment is performed between the road elevation measurement data and the displacement correction data based on vehicle mileage data. Resampling is then performed on both data according to a predetermined spatial sampling interval to generate equidistant detection data. Vibration displacement compensation is then applied to the road elevation measurement data in the equidistant detection data using the displacement correction data to generate compensated road longitudinal profile elevation data. Finally, the International Roughness Index (IRI) value is calculated based on the compensated road longitudinal profile elevation data.

[0006] As a further description of the above technical solution: The generation of time-series detection data includes the following steps: receiving road elevation measurement data continuously collected by a laser rangefinder, vehicle vibration acceleration data continuously collected by an accelerometer, and vehicle mileage data continuously collected by an encoder during vehicle operation, and recording corresponding acquisition time markers for each of the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data; performing time alignment processing on the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data according to the acquisition time sequence, and combining the corresponding road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data under the same acquisition time marker to form synchronous detection record data; performing sequential sorting processing on the synchronous detection record data under continuous acquisition time markers, so that each synchronous detection record data forms a continuous data sequence according to the acquisition time sequence; and uniformly organizing the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data in the continuous data sequence to generate time-series detection data.

[0007] As a further description of the above technical solution: The generation of zero-bias correction acceleration data includes the following steps: reading vehicle vibration acceleration data from time-series detection data and continuously reading the vehicle vibration acceleration data according to the acquisition time sequence to form a vibration acceleration sequence; performing zero-reference statistical processing on the vehicle vibration acceleration data within the continuous sampling interval of the vibration acceleration sequence, calculating the average acceleration value of the vibration acceleration sequence within the current sampling interval, and determining the average acceleration value as the DC component of the current sampling interval; subtracting the DC component of the corresponding sampling interval from the vehicle vibration acceleration data of each sampling point in the vibration acceleration sequence according to the acquisition time sequence to generate zero-bias correction acceleration values; and continuously organizing the zero-bias correction acceleration values ​​generated at each sampling point according to the acquisition time sequence to form zero-bias correction acceleration data.

[0008] As a further description of the above technical solution: The generation of vibration separation acceleration data includes the following steps: Zero-bias correction acceleration data is read from time-series detection data, and the zero-bias correction acceleration data is continuously read according to the acquisition time sequence to form a zero-bias correction acceleration sequence; the zero-bias correction acceleration sequence is organized into continuous sampling segments according to a fixed sampling interval, and fast Fourier transform processing is performed on the zero-bias correction acceleration data in each continuous sampling segment to generate corresponding frequency domain amplitude data and frequency domain frequency data; based on the frequency domain frequency data, the frequency domain amplitude data is divided into frequency intervals, and the frequencies corresponding to vehicle bump vibrations are identified. The amplitude data within the interval is determined as the vibration component amplitude, and the amplitude data below the vehicle bump vibration frequency range and the amplitude data above the vehicle bump vibration frequency range are determined as the noise component amplitude. The frequency domain data corresponding to the vibration component amplitude is retained, and amplitude suppression processing is performed on the frequency domain data corresponding to the noise component amplitude to generate vibration frequency domain data. The inverse Fourier transform processing is performed on the vibration frequency domain data to recover the time domain acceleration sequence containing only the vehicle bump vibration component, and the time domain acceleration sequence is organized according to the acquisition time order to generate vibration-separated acceleration data.

[0009] As a further description of the above technical solution: The generation of vehicle vibration displacement data includes the following steps: Vibration separation acceleration data is read from time-series detection data, and the vibration separation acceleration data is continuously read and processed according to the acquisition time sequence to form a vibration separation acceleration sequence; based on the sampling time interval of the vibration separation acceleration sequence, the vibration separation acceleration sequence is divided into multiple consecutive three-point sampling intervals, and an integral calculation unit is constructed for each consecutive three-point sampling interval; within each integral calculation unit, the Simpson integral algorithm is executed for calculation, and a midpoint weight adjustment parameter is introduced to adjust the acceleration contribution of the middle sampling point within the integral calculation unit, while a curvature compensation parameter is introduced to adjust the acceleration contribution of the two ends within the integral calculation unit. The acceleration contribution of the sampling points is reconstructed using symmetric weights to generate acceleration integral increment data. The acceleration integral increment data generated by each integral calculation unit is continuously accumulated and updated according to the acquisition time sequence to form a vehicle vibration velocity sequence. Based on the vehicle vibration velocity sequence, a velocity integral calculation unit is constructed using three consecutive sampling points. Within each velocity integral calculation unit, the Simpson integral algorithm is executed. The weights of each sampling point within the velocity integral calculation unit are reconstructed based on the midpoint weight adjustment parameter and curvature compensation parameter to generate velocity integral increment data. The velocity integral increment data is continuously accumulated and updated according to the acquisition time sequence to form vehicle vibration displacement data.

[0010] As a further description of the above technical solution: The generation of displacement correction data includes the following steps: Reading vehicle vibration displacement data from time-series detection data and continuously processing the data according to the acquisition time sequence to form a vehicle vibration displacement sequence; performing an improved singular spectrum analysis algorithm on the vehicle vibration displacement sequence to construct a trajectory matrix according to the embedding window length, and performing singular value decomposition on the trajectory matrix to generate a singular value sequence arranged in the order of singular components and the corresponding singular component matrix sequence; introducing a trend energy proportion parameter to perform cumulative energy proportion calculation on the singular value sequence to determine the number of trend components that meet the trend energy proportion parameter requirements, and extracting the trend component set from the singular component matrix sequence accordingly to generate a trend matrix; introducing a reconstruction weighting index parameter to perform weighted diagonal average reconstruction on the trend matrix, assigning weighting coefficients to the matrix elements participating in the reconstruction according to the diagonal support during the reconstruction process and completing the weighted average reconstruction to generate a trend component sequence; performing sample-point differential subtraction on the vehicle vibration displacement sequence and the trend component sequence to remove the trend component sequence from the vehicle vibration displacement sequence, generating displacement correction data, and continuously organizing and outputting the displacement correction data according to the acquisition time sequence.

[0011] As a further description of the above technical solution: The generation of equidistant detection data includes the following steps: Reading road elevation measurement data, displacement correction data, and vehicle mileage data from time-series detection data, and synchronously processing these data according to the acquisition time sequence to form a corresponding sampling point set; associating a spatial distance marker with each sampling point in the corresponding sampling point set based on the vehicle mileage data, and performing monotonicity verification on the spatial distance markers according to the acquisition time sequence; removing sampling points where spatial distance markers have receded, generating a valid sampling point set; calculating the adjacent distance increment sequence based on the spatial distance markers of adjacent sampling points in the valid sampling point set, and then processing the adjacent distance increments... The sequence undergoes abnormal distance increment removal processing to generate a continuous distance increment sequence. Based on the continuous distance increment sequence, spatial distance alignment processing is performed on the road elevation measurement data and displacement correction data to form aligned sampling pairs under the same spatial distance marker, generating alignment detection data. An equidistant spatial sampling point sequence is generated according to a predetermined spatial sampling interval, and resampling processing is performed on the alignment detection data based on the equidistant spatial sampling point sequence to generate equidistant road elevation measurement data and equidistant displacement correction data that correspond one-to-one with the equidistant spatial sampling point sequence. The equidistant road elevation measurement data and equidistant displacement correction data are combined and organized according to the order of the equidistant spatial sampling point sequence to generate equidistant detection data.

[0012] As a further description of the above technical solution: The generation of compensated pavement longitudinal profile elevation data includes the following steps: Reading equidistant pavement elevation measurement data and equidistant displacement correction data from equidistant detection data, and synchronously reading and processing the equidistant pavement elevation measurement data and equidistant displacement correction data according to the sequence of equidistant spatial sampling points to form an equidistant compensation input sequence; performing sign consistency verification processing on the equidistant displacement correction data in the equidistant compensation input sequence, locating the corresponding equidistant spatial sampling point when a sign change occurs in the equidistant displacement correction data, and performing neighborhood smoothing replacement processing on the equidistant displacement correction data corresponding to the equidistant spatial sampling point to generate smooth displacement correction data; performing continuous processing on the equidistant pavement elevation measurement data in the equidistant compensation input sequence. The system performs a verification process. When abrupt changes occur in the equidistant road elevation measurement data, the corresponding equidistant spatial sampling point is located. Neighborhood interpolation replacement is then performed on the equidistant road elevation measurement data corresponding to this equidistant spatial sampling point to generate smooth road elevation measurement data. A per-equidistant spatial sampling point superposition compensation process is then performed on the smooth road elevation measurement data and the smooth displacement correction data. The smooth displacement correction data is superimposed onto the smooth road elevation measurement data in the order of the sampling points to generate compensated road longitudinal profile elevation data. Finally, the compensated road longitudinal profile elevation data is continuously organized and output according to the sequence of equidistant spatial sampling points, ensuring that the compensated road longitudinal profile elevation data maintains a sequence structure that corresponds one-to-one with the sequence of equidistant spatial sampling points.

[0013] As a further description of the above technical solution: The generation of the International Roughness (IRI) value for pavement includes the following steps: First, continuously read the longitudinal profile elevation data of the compensated pavement according to the sequence of equidistant spatial sampling points from the longitudinal profile elevation data. Perform missing sampling point verification on the compensated pavement longitudinal profile elevation data. If missing sampling points exist, perform interpolation completion on the missing sections to generate a complete longitudinal profile elevation sequence. Second, perform zero-datum alignment processing on the complete longitudinal profile elevation sequence. This is achieved by updating the datum elevation of the complete longitudinal profile elevation sequence through datum elevation translation, ensuring that the elevation values ​​of the starting and ending sampling points of the complete longitudinal profile elevation sequence are consistent, thus generating a datum-aligned longitudinal profile elevation sequence. Finally, in the datum-aligned longitudinal profile... The surface roughness program sequence is divided into multiple continuous evaluation segments according to a predetermined evaluation segment length. Each evaluation segment is organized segment by segment according to the sequence of equidistant spatial sampling points to generate an evaluation segment sequence. For each evaluation segment sequence, a recursive calculation of the suspension response is performed. The evaluation segment sequence is used as the road excitation input to generate a corresponding suspension response sequence. The suspension response sequence is then processed by response difference calculation to obtain a response difference sequence. The response difference sequence corresponding to each evaluation segment is then processed by cumulative aggregation to calculate the segment roughness index value of each evaluation segment. Finally, the segment roughness index values ​​are averaged and converged according to the evaluation segment order to generate the International Roughness Index (IRI) value of the road surface.

[0014] The present invention has the following beneficial effects: 1. In this invention, road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data are first processed by unified time stamping and time synchronization, so that the three types of data form a one-to-one correspondence under the same time axis, thereby establishing a complete time-series detection data structure. This provides a unified data foundation for subsequent vibration compensation calculations and improves the overall consistency and continuity of the data processing process. By performing DC removal processing on the vehicle vibration acceleration data and generating zero-bias correction acceleration data, the DC component and sensor zero-bias influence in the vibration acceleration data can be eliminated, thereby improving the stability of the vibration signal data and providing reliable input data for subsequent vibration signal separation and integration calculations. By performing frequency domain analysis processing on the zero-bias correction acceleration data, the vehicle bump vibration frequency band is extracted and other noise frequency bands are suppressed, so that the effective vibration component in the vehicle vibration acceleration data is retained, thereby achieving effective separation of vehicle bump vibration signal and noise signal and improving the vibration signal extraction accuracy.

[0015] 2. In this invention, by introducing midpoint weight adjustment parameters and curvature compensation parameters during vibration integral calculation, the Simpson integral weight is reconstructed, enabling dynamic weight adjustment based on local vibration change characteristics. This improves the accuracy of integral calculation from acceleration to velocity to displacement and reduces numerical drift during integration. By performing improved singular spectrum analysis on vehicle vibration displacement data and introducing trend energy proportion parameters and reconstruction weighting index parameters, trend change components in the vibration displacement sequence are separated and reconstructed, eliminating the influence of trend drift and improving the stability of displacement correction data. By performing spatial distance alignment processing on road elevation measurement data and displacement correction data based on vehicle mileage data and resampling at a unified spatial sampling interval, the elevation data and vibration displacement data are kept consistent in spatial dimension, thereby improving the accuracy of subsequent compensation calculations. By superimposing the displacement correction data onto the road elevation measurement data to generate compensated road longitudinal profile elevation data, and performing IRI calculation on this basis, the influence of vehicle vibration on the road elevation measurement results is effectively corrected, thereby improving the accuracy and reliability of the road surface international roughness IRI value calculation results. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Reference Figure 1 This invention provides an embodiment of a vehicle-mounted laser IRI detection method based on acceleration vibration compensation, comprising the following steps: acquiring road elevation measurement data collected by a laser rangefinder, vehicle vibration acceleration data collected by an accelerometer, and vehicle mileage data collected by an encoder during vehicle operation; performing time synchronization processing on the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data according to the acquisition time sequence to generate time-series detection data; performing DC removal processing on the vehicle vibration acceleration data in the time-series detection data to eliminate the DC component in the vehicle vibration acceleration data to generate zero-bias correction acceleration data; performing Fourier transform processing on the zero-bias correction acceleration data, extracting the vehicle bump vibration frequency band according to the frequency domain distribution of the vibration signal and filtering out noise signals to generate vibration separation acceleration data; performing Simpson integral algorithm calculation processing on the vibration separation acceleration data according to the acquisition time sequence, constructing an integral calculation unit for the vibration separation acceleration data according to three consecutive sampling intervals, and adjusting the weight of the midpoint within each integral calculation unit. The Simpson integral weights are reconstructed using section parameters and curvature compensation parameters. The reconstructed acceleration integral results are then subjected to velocity and displacement integral calculations to generate vehicle vibration displacement data. Singular spectrum analysis is applied to the vehicle vibration displacement data to construct the trajectory matrix of the vehicle vibration displacement sequence. Singular value decomposition is performed on the trajectory matrix, and the number of singular components participating in trend reconstruction is determined based on the trend energy proportion parameter. A weighted diagonal average reconstruction is performed on the trend matrix using the reconstruction weighting index parameter to separate the trend components from the vehicle vibration displacement data, generating displacement correction data. Spatial distance alignment is performed on the road elevation measurement data and displacement correction data based on vehicle mileage data. Resampling is then performed on the road elevation measurement data and displacement correction data according to a predetermined spatial sampling interval to generate equidistant detection data. Vibration displacement compensation is then performed on the road elevation measurement data in the equidistant detection data using the displacement correction data to generate compensated road longitudinal profile elevation data. Finally, the International Roughness Index (IRI) value of the road surface is calculated based on the compensated road longitudinal profile elevation data.

[0019] In this embodiment, as the vehicle travels along the detection section, the laser rangefinder continuously scans the road surface and collects road elevation measurement data; simultaneously, the accelerometer continuously collects vehicle vibration acceleration data during the vehicle's journey; and the encoder synchronously records the vehicle's mileage data during the journey. All three types of data are recorded with corresponding acquisition time stamps according to the device's acquisition time, forming a continuous time series data as the acquisition process progresses.

[0020] After data acquisition is completed, the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data are first time-aligned according to the acquisition time stamps. The acquisition time stamps of each type of data are read sequentially according to the acquisition time order. At the time node corresponding to the same acquisition time stamp, the corresponding road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data are combined so that the three types of data form a synchronous detection record at the same time node.

[0021] Subsequently, the synchronous detection record data corresponding to the continuously acquired time markers are processed sequentially. During the processing, the synchronous detection record data are arranged in the order of the acquisition time markers, so that the synchronous detection record data form a continuous data sequence according to the time sequence. Through this processing method, the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data are kept strictly consistent in the time dimension, and a continuous time series data structure that can reflect the entire vehicle driving process is formed.

[0022] After forming a continuous data sequence, the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data in the continuous data sequence are uniformly organized and processed so that each sampling time node contains the corresponding road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data, and are output as a whole according to the collection time order, thereby generating time-series detection data.

[0023] The above processing method establishes a one-to-one correspondence between road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data on the same time axis, providing a unified data input basis for the subsequent generation of zero-bias correction acceleration data, vibration separation acceleration data, and vehicle vibration displacement data.

[0024] In this embodiment, during the actual testing process, the time-series testing data has been organized into a continuous data sequence according to a unified acquisition time order, with each sampling time node containing corresponding vehicle vibration acceleration data. Based on this continuous data sequence, the vehicle vibration acceleration data is first read sequentially from the time-series testing data according to the acquisition time order, so that the vehicle vibration acceleration data at each sampling time node forms a vibration acceleration sequence in chronological order. Through this processing method, the vehicle vibration acceleration data forms a continuously arranged data sequence structure in the time dimension.

[0025] After obtaining the vibration acceleration sequence, zero-reference statistical processing is performed on the vehicle vibration acceleration data within the sequence, using continuous sampling intervals as the statistical range. Within each continuous sampling interval, the vehicle vibration acceleration data at all sampling points within that interval are statistically calculated to obtain the average acceleration value for that interval. This average acceleration value is then determined as the DC component corresponding to that continuous sampling interval. This processing method reflects the overall offset level of the vehicle vibration acceleration data within that sampling interval.

[0026] After determining the DC component of each continuous sampling interval, DC component subtraction processing is performed on the vehicle vibration acceleration data of each sampling point in the vibration acceleration sequence according to the acquisition time sequence. The DC component corresponding to the sampling interval is subtracted from the vehicle vibration acceleration data of each sampling point to obtain the zero-bias corrected acceleration value for that sampling point. This processing method can eliminate the overall offset effect in the vehicle vibration acceleration data caused by sensor zero bias or installation attitude.

[0027] Subsequently, the zero-bias correction acceleration values ​​generated at each sampling point are continuously organized according to the acquisition time sequence, so that all the zero-bias correction acceleration values ​​form a continuous data sequence structure in chronological order, thereby generating zero-bias correction acceleration data.

[0028] By using the above processing method, the vehicle vibration acceleration data is zero-referenced and corrected while maintaining the original sampling time sequence, thus obtaining zero-biased corrected acceleration data that eliminates the influence of DC components, providing a stable data foundation for the subsequent generation of vibration separation acceleration data.

[0029] In this embodiment, during the actual detection process, the time-series detection data has already formed a continuous data sequence according to a unified acquisition time order, and the zero-bias correction acceleration data has already been generated in the aforementioned processing. First, the zero-bias correction acceleration data is read sequentially from the time-series detection data according to the acquisition time order, so that the zero-bias correction acceleration data at each sampling time node forms a zero-bias correction acceleration sequence in chronological order. Through this processing method, the zero-bias correction acceleration data forms a continuously arranged data structure in the time dimension.

[0030] After obtaining the zero-bias correction acceleration sequence, the sequence is divided into continuous sampling segments according to the sampling interval of the accelerometer. The zero-bias correction acceleration sequence is divided into multiple continuous sampling segments according to a fixed sampling interval, so that each continuous sampling segment contains a set of zero-bias correction acceleration data within a continuous time range.

[0031] Subsequently, a Fast Fourier Transform (FFT) is performed on the zero-bias correction acceleration data in each consecutive sampling segment, converting the corresponding time-domain acceleration data into frequency-domain data, and generating corresponding frequency-domain amplitude and frequency data. This processing method allows us to obtain the vibration amplitude distribution of the zero-bias correction acceleration data within different frequency ranges.

[0032] After obtaining the frequency domain amplitude and frequency domain data, the frequency domain amplitude data is divided into frequency intervals based on the frequency domain data. The amplitude data within the frequency interval corresponding to vehicle bump vibration is determined as the vibration component amplitude, while the amplitude data below and above the vehicle bump vibration frequency interval are determined as the noise component amplitude. This processing method enables the vehicle bump vibration component to be distinguished from other interference signals in the frequency domain.

[0033] Subsequently, the frequency domain data corresponding to the amplitude of the vibration component is retained, and amplitude suppression processing is performed on the frequency domain data corresponding to the amplitude of the noise component to weaken the noise component in the frequency domain, thereby generating vibration frequency domain data.

[0034] After obtaining the vibration frequency domain data, an inverse Fourier transform is performed on the vibration frequency domain data to convert the vibration frequency domain data back into time domain data, thus recovering the time domain acceleration sequence containing only the vehicle bump vibration component.

[0035] Finally, the recovered time-domain acceleration sequence is continuously processed according to the acquisition time sequence to ensure that the time-domain acceleration sequence is consistent with the original sampling time sequence, thereby generating vibration separation acceleration data. Through the above processing method, the vehicle bump vibration component in the vehicle vibration acceleration data is effectively extracted, providing input data for the subsequent generation of vehicle vibration displacement data.

[0036] In this embodiment, during the actual detection process, the vibration separation acceleration data has already been generated in the aforementioned processing, and the time-series detection data has been formed into a continuous data sequence according to a unified acquisition time order. First, the vibration separation acceleration data is read sequentially from the time-series detection data according to the acquisition time order, so that the vibration separation acceleration data corresponding to each sampling time node are arranged continuously in chronological order, thereby organizing a vibration separation acceleration sequence. Through this processing method, subsequent integration calculations can be carried out sequentially around the continuous time order.

[0037] After obtaining the vibration separation acceleration sequence, the sequence is divided into three consecutive sampling intervals based on the sampling time interval. Following the acquisition time sequence, three adjacent and consecutive sampling points are defined as a single three-point sampling interval, which are then arranged sequentially along the vibration separation acceleration sequence. Subsequently, a corresponding integration calculation unit is constructed for each three-point sampling interval, ensuring that each unit contains vibration separation acceleration data corresponding to the starting, intermediate, and ending sampling points. This process transforms the continuous vibration separation acceleration sequence into a serialized processing structure capable of performing integration calculations unit by unit.

[0038] After each integration calculation unit is constructed, the midpoint weight adjustment parameter is calculated first. For the current integration calculation unit, the vibration separation acceleration data corresponding to the previous sampling point, the middle sampling point, and the next sampling point are read sequentially, and the change relationship between the vibration separation acceleration data of the middle sampling point and the vibration separation acceleration data of the two sampling points is compared. When the vibration separation acceleration data of the middle sampling point shows more obvious local fluctuations relative to the two sampling points, the weight adjustment amplitude of the middle sampling point is increased accordingly; when the vibration separation acceleration data of the middle sampling point maintains a relatively smooth change with the two sampling points, the weight adjustment amplitude of the middle sampling point is decreased accordingly. By continuously comparing the degree of local change of the three vibration separation acceleration data in each integration calculation unit, the midpoint weight adjustment parameter corresponding to each integration calculation unit is generated. Through this processing method, the contribution of the middle sampling point in the integration calculation can be adjusted according to the local vibration change state. The formula for calculating the midpoint weight adjustment parameter is as follows: ; Midpoint weight adjustment parameter : Vibration separation acceleration data corresponding to the starting sampling point of the current integral calculation unit. : Vibration separation acceleration data corresponding to the intermediate sampling point of the current integral calculation unit. : Vibration separation acceleration data corresponding to the current integral calculation unit's termination sampling point.

[0039] After the midpoint weight adjustment parameter is calculated, the curvature compensation parameter is calculated. For the vibration separation acceleration data corresponding to the previous, middle, and next sampling points within the current integration calculation unit, the variation between the previous and middle sampling points is first determined, then the variation between the middle and next sampling points is determined, and the curvature change level within the current integration calculation unit is determined based on the degree of difference between the two variations. When the variation trend between the three consecutive sampling points is relatively consistent, the curvature compensation parameter remains at a small value; when there is a significant bending change or aggravated change between the three consecutive sampling points, the curvature compensation parameter is increased accordingly. This processing method allows the contribution of the sampling points at both ends in the integration calculation to be symmetrically reconstructed based on the curvature change within the three consecutive point intervals, thereby improving the ability of the integration processing to express local bending changes. Curvature compensation parameter calculation formula: ; Curvature compensation parameters : Vibration separation acceleration data corresponding to the starting sampling point of the current integral calculation unit. : Vibration separation acceleration data corresponding to the intermediate sampling point of the current integral calculation unit. : Vibration separation acceleration data corresponding to the current integral calculation unit's termination sampling point.

[0040] After determining the midpoint weight adjustment parameter and curvature compensation parameter, the Simpson integral algorithm is executed for each integral calculation unit. First, the vibration separation acceleration data corresponding to the starting, intermediate, and ending sampling points within the current integral calculation unit are read. Then, the acceleration contribution corresponding to the intermediate sampling point within the current integral calculation unit is weighted according to the midpoint weight adjustment parameter, ensuring that the contribution of the intermediate sampling point in the integral calculation is calculated according to the midpoint weight adjustment parameter. Simultaneously, the acceleration contributions corresponding to the two sampling points within the current integral calculation unit are symmetrically weighted according to the curvature compensation parameter, ensuring that the contributions of the starting and ending sampling points in the integral calculation are symmetrically adjusted according to the curvature compensation parameter. After completing the weight adjustment of the intermediate sampling point and the symmetrical weight reconstruction of the two sampling points, a comprehensive integral calculation is performed on the acceleration contributions corresponding to the three sampling points within the current integral calculation unit, thereby generating the acceleration integral increment data corresponding to the current integral calculation unit. Simpson weight reconstruction formula: ; ; ; The weight of the initial sampling point in the integral calculation. The weight of intermediate sampling points in the integral calculation. The weight of the final sampling point in the integral calculation. Midpoint weight adjustment parameter Curvature compensation parameter.

[0041] After each integral calculation unit generates acceleration integral increment data, continuous cumulative update processing is performed on the acceleration integral increment data corresponding to each integral calculation unit according to the acquisition time sequence. Starting from the integral calculation unit corresponding to the starting position of the vibration separation acceleration sequence, the acceleration integral increment data generated by each subsequent integral calculation unit is read sequentially, and cumulative updates are performed one by one according to the acquisition time sequence, so that the corresponding velocity update result can be obtained at each sampling time node. After continuous cumulative update processing, the velocity update results corresponding to each sampling time node are organized according to the acquisition time sequence, thus forming a vehicle vibration velocity sequence. Through this processing method, the integral result of vibration separation acceleration data in the time dimension is transformed into a continuous vehicle vibration velocity sequence. Vehicle vibration velocity calculation formula: ; : Vehicle vibration velocity data corresponding to the current sampling time point The vehicle vibration velocity data corresponding to the next sampling time point. Vibration separation acceleration data corresponding to three consecutive sampling intervals. : Integral calculation of weights, : Sampling time interval of the vibration separation acceleration sequence.

[0042] After the vehicle vibration velocity sequence is formed, the second-level integration processing is performed based on the vehicle vibration velocity sequence. Following the same three-point processing method as the vibration separation acceleration sequence, velocity integration calculation units are constructed for the vehicle vibration velocity sequence based on three consecutive sampling points. Each velocity integration calculation unit is arranged sequentially along the vehicle vibration velocity sequence according to the acquisition time. Each velocity integration calculation unit contains vehicle vibration velocity data corresponding to the starting sampling point, intermediate sampling point, and ending sampling point.

[0043] Subsequently, the Simpson integral algorithm is applied to each velocity integral calculation unit. First, the vehicle vibration velocity data corresponding to the starting, intermediate, and ending sampling points within the current velocity integral calculation unit are read. Then, the velocity contribution corresponding to the intermediate sampling point within the current velocity integral calculation unit is weighted according to the midpoint weight adjustment parameter. Simultaneously, the velocity contributions corresponding to the sampling points at both ends of the current velocity integral calculation unit are reconstructed using symmetric weights based on the curvature compensation parameter. After completing the weight reconstruction of each sampling point within the current velocity integral calculation unit, a comprehensive integral calculation is performed on the velocity contributions corresponding to the three sampling points within the current velocity integral calculation unit, thereby generating the velocity integral increment data corresponding to the current velocity integral calculation unit. Through this processing method, the vehicle vibration velocity sequence can be further converted into displacement change results.

[0044] After each velocity integral calculation unit generates velocity integral increment data, continuous cumulative update processing is performed on each velocity integral increment data according to the acquisition time sequence. Starting from the velocity integral calculation unit corresponding to the beginning position of the vehicle vibration velocity sequence, the velocity integral increment data generated by each subsequent velocity integral calculation unit is read sequentially and cumulatively updated one by one according to the acquisition time sequence, so that each sampling time node obtains the corresponding displacement update result. Subsequently, the displacement update results corresponding to each sampling time node are continuously organized according to the acquisition time sequence to form vehicle vibration displacement data. Vehicle vibration displacement calculation formula: ; : Vehicle vibration displacement data corresponding to the current sampling time point : Vehicle vibration displacement data corresponding to the next sampling time node Data from three consecutive sampling points in the vehicle vibration velocity sequence. Improved Simpson integral weights Sampling time interval. Vehicle vibration displacement expression: ; Vehicle vibration displacement data sequence : No. Vehicle vibration displacement data corresponding to each sampling time point Number of sampling points corresponding to the vibration separation acceleration sequence Through the above processing method, the vibration separation acceleration data is first processed by the first-level Simpson integral algorithm to generate the vehicle vibration velocity sequence, and then processed by the second-level Simpson integral algorithm to generate the vehicle vibration displacement data. In both integration processes, the integration weights are reconstructed by adjusting the midpoint weight parameter and the curvature compensation parameter, so that the vehicle vibration displacement data can be continuously converted from vibration separation acceleration data to vehicle vibration displacement data while maintaining the continuity of the acquisition time sequence, thus providing an input basis for the generation of subsequent displacement correction data.

[0045] In this embodiment, during the actual detection process, vehicle vibration displacement data has already been generated in the aforementioned processing, and the time-series detection data has been formed into a continuous data sequence according to a unified acquisition time order. First, vehicle vibration displacement data is sequentially read from the time-series detection data according to the acquisition time order, so that the vehicle vibration displacement data corresponding to each sampling time node are arranged continuously in chronological order, thereby organizing a vehicle vibration displacement sequence. This processing method ensures that subsequent trend separation processing unfolds gradually around the continuous time sequence, and guarantees that the vehicle vibration displacement data maintains a continuous temporal structure before entering the improved singular spectrum analysis algorithm.

[0046] After generating the vehicle vibration displacement sequence, an improved singular value decomposition (SVD) algorithm is applied. First, the vehicle vibration displacement sequence is continuously segmented according to the embedding window length, allowing consecutive sampled values ​​from adjacent positions to sequentially enter different window positions, thus constructing a trajectory matrix. Each column in the trajectory matrix corresponds to a segment of continuous displacement data from adjacent sampling positions, with columns arranged sequentially along the sampling time order of the vehicle vibration displacement sequence. This process transforms the one-dimensional vehicle vibration displacement sequence into a trajectory matrix that represents a continuous temporal structure, providing a matrix-based input foundation for subsequent SVD. The trajectory matrix construction formula is as follows: ; Trajectory matrix The vehicle vibration displacement sequence of the first Vehicle vibration displacement data corresponding to each sampling time point : Embedded window length, The total number of sampling points for the vehicle vibration displacement sequence. The number of columns in the trajectory matrix, satisfying: .

[0047] After the trajectory matrix is ​​constructed, singular value decomposition (SVD) is performed. The trajectory matrix is ​​decomposed into multiple component structures arranged in the order of singular components, generating singular value sequences and corresponding singular component matrix sequences. The singular value sequences characterize the contribution of each singular component to the trajectory matrix, while the singular component matrix sequences characterize the matrix structure corresponding to each singular component. This process allows for the separation of primary and secondary variation components in the vehicle vibration displacement sequence, providing a component basis for trend component extraction. Singular value decomposition formula: ; Trajectory matrix Left singular vector matrix, Singular value matrix : Right singular vector matrix. The singular value matrix is ​​expressed as: ;in: : No. A singular value, : The rank of the trajectory matrix.

[0048] After generating the singular value sequence and singular component matrix sequence, the calculation of the trend energy proportion parameter involves first constructing a trajectory matrix based on the vehicle vibration displacement sequence, and then performing singular value decomposition on the trajectory matrix to obtain a singular value sequence arranged in the order of singular components. Subsequently, following the order of the singular value sequence, each singular value is read sequentially, and the component contribution corresponding to each singular value is used as the energy representation of the current singular component. Next, the component contributions corresponding to all singular values ​​are summarized to obtain the overall energy corresponding to all singular components of the trajectory matrix. Then, following the order of the singular value sequence, the energy representations of each singular component are continuously accumulated from front to back to obtain a progressively expanding cumulative energy result. After each accumulation, the current cumulative energy result is compared with the overall energy to obtain the energy proportion result within the current cumulative range. Finally, from the energy proportion results, the proportion result that stably covers the slowly changing part of the vehicle vibration displacement sequence and is distinguished from the high-frequency fluctuation part is selected, and this proportion result is determined as the trend energy proportion parameter. This processing method allows the trend energy proportion parameter to maintain a correspondence with the component distribution of the vehicle vibration displacement sequence itself, and is used to determine the number of trend components that meet the trend reconstruction requirements. The formula for calculating the trend energy proportion parameter is as follows: ; Trend energy percentage parameter The first singular value in a sequence A singular value, Trend component quantity : The total number of singular components.

[0049] After obtaining the singular value sequence and its corresponding singular component matrix sequence, a cumulative energy percentage calculation is performed on the singular value sequence according to the trend energy percentage parameter. Following the order of the singular value sequence, each singular value is read sequentially and cumulatively calculated step by step, while the current cumulative result is compared with the overall energy corresponding to all singular values. When the cumulative energy percentage reaches the requirement corresponding to the trend energy percentage parameter, the number of singular components within the current cumulative range is determined as the number of trend components that meet the trend reconstruction requirements. Through this processing method, the range of singular components used to express trend changes can be selected from the multiple singular components obtained from singular value decomposition.

[0050] After determining the number of trend components, a corresponding number of singular component matrices are extracted from the singular component matrix sequence. These extracted singular component matrices are then combined to generate a trend matrix. Following the order of the singular component matrix sequence, singular component matrices that meet the required number of trend components are read sequentially. These matrices are then superimposed and combined in their corresponding positions, so that the extracted singular component matrices together form a trend matrix reflecting the trend change. Through this processing method, the trend change portion of the vehicle vibration displacement sequence is reconstructed in matrix form. Trend matrix generation formula: ; Trend matrix : No. A singular value, : Left singular vector : Right singular vector : Number of trend components.

[0051] For the calculation of the reconstructed weighted index parameters, after determining the number of trend components and generating the trend matrix, all matrix elements in the trend matrix are first diagonally grouped according to the distribution relationship of matrix elements in the diagonal direction, resulting in multiple diagonal groups. Then, the number of matrix elements participating in the reconstruction within each diagonal group is counted, with the number of matrix elements representing the support of the corresponding diagonal group. Next, a support change sequence is established according to the support magnitude pattern of each diagonal group, and the support change sequence is continuously compared to identify the degree of support difference between the middle and edge diagonal groups of the trend matrix. Then, based on the degree of support difference, the adjustment direction for strengthening or weakening the contribution of matrix elements in the edge diagonal groups during reconstruction is determined; when the support of the edge diagonal groups is significantly lower than that of the middle diagonal groups, the adjustment intensity for the support difference is increased; when the support change among the diagonal groups is relatively gradual, the adjustment intensity for the support difference is decreased. Next, the index value characterizing the intensity of the adjustment is determined as the reconstruction weighted index parameter, and this parameter is used to assign corresponding weighting coefficients to the matrix elements in each diagonal group. This process allows the trend matrix to differentiate between lower-support edge positions and higher-support middle positions when performing weighted diagonal average reconstruction, thereby improving the continuity and stability of the trend component sequence reconstruction results. The formula for calculating the reconstruction weighted index parameter is as follows: ; : Reconstruct the weighted exponent parameters, : No. The number of matrix elements involved in the reconstruction in each diagonal group. : The average number of matrix elements in all diagonal groups Total number of diagonal groups in the trend matrix.

[0052] After determining the trend energy proportion parameter and the reconstruction weighting index parameter, a weighted diagonal average reconstruction process is performed on the trend matrix based on the reconstruction weighting index parameter. First, the matrix elements in the trend matrix are grouped diagonally according to their distribution along the diagonal direction. Then, based on the support corresponding to each diagonal line and the reconstruction weighting index parameter, the weighting coefficients of the matrix elements within each diagonal group are determined. Subsequently, within each diagonal group, a weighted average reconstruction process is performed on the matrix elements participating in the reconstruction based on the determined weighting coefficients, so that the trend matrix is ​​gradually restored to a continuous trend component sequence along the sampling time sequence of the vehicle vibration displacement sequence. This processing method can improve the temporal correspondence between the reconstructed trend component sequence and the original vehicle vibration displacement sequence while maintaining the overall trend expression capability of the trend matrix. Weighted diagonal average reconstruction formula: Weighting function: ; The trend component sequence of the first The trend value corresponding to each sampling point : The trend matrix belongs to the first A set of matrix elements grouped along diagonal lines. The trend matrix Line number Column matrix elements, : The weighting coefficients corresponding to the matrix elements : The support of the diagonal group containing the matrix element. : Reconstruct the weighted exponent parameters.

[0053] After the trend component sequence is generated, a sampling-point differential subtraction process is performed on the vehicle vibration displacement sequence and the trend component sequence. The displacement values ​​of each sampling point in the vehicle vibration displacement sequence and the corresponding trend values ​​in the trend component sequence are read sequentially according to the acquisition time. At each sampling time node, the displacement value of the sampling point in the vehicle vibration displacement sequence is subtracted from the corresponding trend value in the trend component sequence, thus obtaining the displacement correction value for that sampling time node. This processing method removes the trend component that accumulates over time from the vehicle vibration displacement data, allowing the remaining displacement changes to more accurately reflect the effective changes in vehicle vibration displacement. Displacement correction calculation formula: ; : No. Displacement correction values ​​corresponding to each sampling time node The vehicle vibration displacement sequence of the first Vehicle vibration displacement data at each sampling point The trend component sequence of the first Trend values ​​of each sampling point.

[0054] After the displacement correction values ​​for each sampling time node are generated, all displacement correction values ​​are continuously organized and output according to the acquisition time sequence, so that the displacement correction values ​​corresponding to each sampling time node are arranged continuously in chronological order, thus forming displacement correction data. Through this processing method, the displacement correction data maintains the same sampling time sequence as the vehicle vibration displacement data in terms of data structure, and can be directly used as input data for subsequent spatial distance alignment and resampling processing. Displacement correction data sequence: ; Displacement correction data : No. Displacement correction values ​​corresponding to each sampling time node Total number of sampling points for vehicle vibration displacement sequence.

[0055] In this embodiment, during the actual testing process, the equidistant testing data has already been generated according to the aforementioned implementation method, and the equidistant testing data already includes equidistant road surface elevation measurement data and equidistant displacement correction data corresponding one-to-one with the equidistant spatial sampling point sequence. First, following the order of the equidistant spatial sampling point sequence, the equidistant road surface elevation measurement data and equidistant displacement correction data corresponding to each equidistant spatial sampling point are sequentially read from the equidistant testing data. Under the same equidistant spatial sampling point, the corresponding equidistant road surface elevation measurement data and equidistant displacement correction data are synchronously organized, so that each equidistant spatial sampling point forms a corresponding set of input data. As all equidistant spatial sampling points are synchronously read and organized, all corresponding input data are continuously arranged along the order of the equidistant spatial sampling point sequence, thereby forming an equidistant compensation input sequence. Through this processing method, subsequent compensation processing is carried out point-by-point around a unified spatial sampling order.

[0056] After the equidistant compensation input sequence is formed, the equidistant displacement correction data is first subjected to sign consistency verification. The equidistant displacement correction data corresponding to each equidistant spatial sampling point is read sequentially according to the order of the equidistant spatial sampling point sequence, and the continuity of the equidistant displacement correction data at the current equidistant spatial sampling point with the equidistant displacement correction data at the preceding and following equidistant spatial sampling points in terms of the direction of change is compared. When the equidistant displacement correction data corresponding to the current equidistant spatial sampling point maintains the same direction of change as the equidistant displacement correction data corresponding to the preceding and following equidistant spatial sampling points, the current equidistant displacement correction data is considered to meet the sign consistency requirement. When the direction of change is reversed between the equidistant displacement correction data corresponding to the current equidistant spatial sampling point and the equidistant displacement correction data corresponding to the preceding and following equidistant spatial sampling points, and this direction reversal disrupts the continuous change relationship within the local spatial segment, a sign abrupt change is identified at the current position, and the current equidistant spatial sampling point is located as the position requiring correction. This processing method can identify abnormal positions in the equidistant displacement correction data that are inconsistent with the local continuous change relationship.

[0057] After completing the sign consistency check and locating the equidistant spatial sampling points where sign abrupt changes occur, neighborhood smoothing replacement processing is performed on the corresponding equidistant displacement correction data. First, the equidistant displacement correction data of the neighboring positions before and after the current equidistant spatial sampling point are read. Then, based on the spatial adjacency relationship between the current equidistant spatial sampling point and its neighboring positions, the equidistant displacement correction data that maintains a continuous change relationship at the neighboring positions is used as a smoothing reference. Subsequently, the equidistant displacement correction data with sign abrupt changes at the current equidistant spatial sampling point is replaced using this smoothing reference, ensuring that the displacement change direction at the replaced current equidistant spatial sampling point remains continuous and consistent with that of its neighboring positions. After neighborhood smoothing replacement processing at all abnormal locations, the displacement data corresponding to each equidistant spatial sampling point is reorganized continuously according to the sequence of equidistant spatial sampling points, thereby generating smoothed displacement correction data. This processing method can reduce the impact of local abnormal changes in the equidistant displacement correction data on subsequent superposition compensation processing.

[0058] After the smooth displacement correction data is generated, continuity verification processing is performed on the equidistant road surface elevation measurement data in the equidistant compensation input sequence. The equidistant road surface elevation measurement data corresponding to each equidistant spatial sampling point is read sequentially according to the equidistant spatial sampling point sequence, and the change relationship between the equidistant road surface elevation measurement data at the current equidistant spatial sampling point and the equidistant road surface elevation measurement data at the preceding and following adjacent equidistant spatial sampling points is compared. When the current equidistant road surface elevation measurement data maintains a continuous transition relationship with the preceding and following adjacent positions, it is considered to meet the continuity requirement; when the current equidistant road surface elevation measurement data shows a significant abrupt change relative to the preceding and following adjacent positions, and this abrupt change disrupts the continuous elevation change relationship within the local spatial segment, the current position is identified as an abnormal position, and the current equidistant spatial sampling point is located as the position requiring correction. This processing method can identify local abrupt change locations in the equidistant road surface elevation measurement data.

[0059] After completing continuity verification and locating equidistant spatial sampling points with abrupt changes, neighborhood interpolation replacement processing is performed on the corresponding equidistant road elevation measurement data. First, equidistant road elevation measurement data at the neighboring positions before and after the current equidistant spatial sampling point are read. Then, based on the spatial order relationship between the current equidistant spatial sampling point and its neighboring positions, the equidistant road elevation measurement data at the neighboring positions, which maintain a continuous change relationship, are used to interpolate and replace the abrupt elevation data at the current equidistant spatial sampling point. After the replacement, the elevation value at the current equidistant spatial sampling point is reinstated within the continuous change range between its neighboring positions. After neighborhood interpolation replacement processing at all abnormal locations, the elevation data corresponding to each equidistant spatial sampling point is reorganized according to the sequence of equidistant spatial sampling points, thereby generating smooth road elevation measurement data. This processing method can reduce the impact of local abrupt elevation values ​​on the continuity of subsequent compensation results.

[0060] After both the smooth displacement correction data and the smooth road surface elevation measurement data are generated, a point-by-point equidistant spatial sampling point superposition compensation process is performed on both. The smooth road surface elevation measurement data and the corresponding smooth displacement correction data for each equidistant spatial sampling point are read sequentially according to the equidistant spatial sampling point sequence. At the current equidistant spatial sampling point, the smooth displacement correction data is superimposed onto the smooth road surface elevation measurement data, ensuring that the elevation result at the current equidistant spatial sampling point simultaneously reflects both the road surface elevation measurement result and the vehicle vibration displacement correction result. Subsequently, the same superposition compensation process is performed on all equidistant spatial sampling points sequentially, thereby generating a compensation result sequence corresponding one-to-one with all equidistant spatial sampling points. Through this processing method, the influence of vehicle vibration on the road surface elevation measurement results is compensated point-by-point into the elevation results, thus obtaining the compensated road surface longitudinal profile elevation data.

[0061] After the compensated pavement longitudinal profile elevation data is generated, continuous output processing is performed on the data according to the sequence of equidistant spatial sampling points. The compensation results corresponding to each equidistant spatial sampling point are read sequentially, and these results are output continuously according to the arrangement order of the equidistant spatial sampling point sequence. This ensures that all compensation results maintain a consistent spatial order and one-to-one correspondence with the equidistant spatial sampling point sequence. Through this processing method, the compensated pavement longitudinal profile elevation data maintains complete spatial continuity in its data structure and can be directly used as input data for the subsequent generation of International Roughness (IRI) values ​​for pavement.

[0062] In this embodiment, during the actual testing process, the longitudinal profile elevation data of the compensated pavement has been generated according to the aforementioned implementation method, and a one-to-one correspondence is maintained between the longitudinal profile elevation data of the compensated pavement and the sequence of equidistant spatial sampling points. First, following the order of the equidistant spatial sampling point sequence, the elevation values ​​corresponding to each equidistant spatial sampling point are sequentially read from the longitudinal profile elevation data of the compensated pavement, ensuring that the elevation values ​​corresponding to each equidistant spatial sampling point are continuously arranged in spatial order, forming an initial elevation sequence for smoothness calculation. During continuous reading, the existence of a corresponding elevation value for each equidistant spatial sampling point is checked point by point to complete the missing sampling point verification process. When a corresponding elevation value exists for a certain equidistant spatial sampling point, that elevation value is retained; when a corresponding elevation value does not exist for a certain equidistant spatial sampling point, that equidistant spatial sampling point is determined to be a missing sampling point, and the search continues to look for a range of continuous and valid elevation values ​​in the preceding and following adjacent positions to determine the missing section. Through this processing method, incomplete locations in the longitudinal profile elevation data of the compensated pavement can be identified before proceeding with smoothness calculation.

[0063] After verifying the missing sampling points and identifying the missing sections, interpolation completion processing is performed on the missing sections. Specifically, the elevation values ​​corresponding to the adjacent valid equidistant spatial sampling points at the front and rear of the missing section are read first. Then, based on the spatial positional relationship between each missing sampling point within the missing section and its adjacent valid equidistant spatial sampling points at the front and rear, complete elevation values ​​are generated sequentially for each missing sampling point within the missing section. As all missing sampling points are completed, each equidistant spatial sampling point obtains its corresponding elevation value, and these values ​​are continuously organized according to the sequence of equidistant spatial sampling points, thus generating a complete longitudinal profile elevation sequence. This processing method ensures that subsequent flatness calculations are based on a complete and continuous elevation sequence.

[0064] After generating the complete longitudinal profile elevation sequence, zero-datum alignment is performed. Specifically, the elevation values ​​of the starting and ending sampling points in the complete longitudinal profile elevation sequence are first read and compared. Then, based on the difference between the starting and ending sampling point elevation values, a datum elevation translation update is performed on the elevation values ​​of each sampling point in the complete longitudinal profile elevation sequence, ensuring a unified elevation datum along the spatial order. After the datum elevation translation update is completed, the updated starting and ending sampling point elevation values ​​are read again. When they are consistent, the zero-datum alignment is completed, thus generating a datum-aligned longitudinal profile elevation sequence. This process reduces the impact of the overall datum offset of the complete longitudinal profile elevation sequence on subsequent recursive calculations of the suspension response.

[0065] After the benchmark-aligned longitudinal profile height sequence is generated, it is divided into segments according to the predetermined evaluation segment length. Specifically, the predetermined evaluation segment length is first read, and starting from the beginning position of the benchmark-aligned longitudinal profile height sequence, the spatial range is divided segment by segment along the spatial advancement direction corresponding to the equidistant spatial sampling point sequence, according to the predetermined evaluation segment length. Each spatial range contains a set of continuous equidistant spatial sampling points and their corresponding elevation values, constituting a continuous evaluation segment. Subsequently, the same segmentation process is performed on subsequent height sequences along the spatial advancement direction, dividing the benchmark-aligned longitudinal profile height sequence into multiple continuous evaluation segments. Each continuous evaluation segment is then organized segment by segment according to the equidistant spatial sampling point sequence, thereby generating an evaluation segment sequence. This processing method allows subsequent flatness calculations to be carried out segment by segment around spatial segments of uniform length.

[0066] After the evaluation segment sequence is generated, a recursive calculation of the suspension response is performed on each evaluation segment sequence. Specifically, for the current evaluation segment sequence, the elevation values ​​corresponding to each sampling point within the current evaluation segment are read sequentially according to the equidistant spatial sampling point sequence, and the elevation change results corresponding to each sampling point are used as the pavement excitation input. Subsequently, along the spatial advancement direction of the current evaluation segment, the pavement excitation input of the current sampling point and the response state already formed by the previous sampling point are continuously updated recursively, so that each sampling point within the current evaluation segment sequentially forms the corresponding suspension response result. After all sampling points within the current evaluation segment have been processed, the suspension response results corresponding to each sampling point are continuously organized according to spatial order, thereby generating the corresponding suspension response sequence. Through this processing method, the elevation changes of the benchmark-aligned longitudinal profile elevation sequence within the current evaluation segment can be converted into corresponding suspension response change results.

[0067] After the suspension response sequence is generated, response difference calculation is performed on it. Specifically, according to the spatial order of the sampling points within the current evaluation segment, the suspension response results corresponding to adjacent sampling points in the suspension response sequence are read sequentially. The suspension response result corresponding to the next sampling point is compared with the suspension response result corresponding to the previous sampling point to obtain the response change result between the current adjacent sampling points. This process is then repeated for all adjacent sampling points within the current evaluation segment, ensuring that the response change results between adjacent sampling points are arranged continuously in spatial order, thus obtaining the response difference sequence. This processing method quantifies the degree of continuous change in the suspension response sequence into point-by-point difference results, which are used for calculating the subsequent segment flatness index value.

[0068] After the response difference sequence is generated, an accumulation and summarization process is performed on the response difference sequences corresponding to each evaluation segment. Specifically, for the current evaluation segment, each response change result in the response difference sequence is read sequentially according to spatial order, and each response change result is accumulated item by item to obtain the overall response change amount within the current evaluation segment. After all response change results for the current evaluation segment have been accumulated, the overall response change amount corresponding to the current evaluation segment is determined as the segment smoothness index value for the current evaluation segment. Subsequently, the same process is performed on all evaluation segments sequentially according to the evaluation segment sequence, so that a corresponding segment smoothness index value is generated for each evaluation segment. Through this processing method, the suspension response change results within each evaluation segment can be converted into segment results that can characterize the road surface smoothness of the current evaluation segment.

[0069] After the section smoothness index values ​​for each assessment segment are generated, the smoothness index values ​​for each segment are averaged and aggregated according to the assessment segment order. Specifically, the smoothness index values ​​for each assessment segment are read sequentially according to the order of the assessment segment sequence, and all section smoothness index values ​​are aggregated. Then, the aggregated results are averaged based on the number of section smoothness index values, thus obtaining the overall smoothness result covering all assessment segments; this overall smoothness result is the International Roughness Index (IRI) value for road surface. This processing method allows for the obtaining of the overall smoothness evaluation result for the corresponding tested road segment while maintaining the smoothness characteristics of each assessment segment.

[0070] In a specific implementation, as the vehicle travels along the detection section, a laser rangefinder continuously scans the road surface and collects road elevation measurement data. Simultaneously, an accelerometer continuously collects vehicle vibration acceleration data, and an encoder synchronously records vehicle mileage data. All three types of data are recorded with corresponding acquisition timestamps, forming a continuous time series. Subsequently, based on the acquisition timestamps, time alignment processing is performed on the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data. At the time node corresponding to the same acquisition timestamp, the three types of data are combined to form synchronous detection record data, and then arranged according to the acquisition time order to form a continuous data sequence, thereby generating time-series detection data.

[0071] After obtaining the time-series detection data, vehicle vibration acceleration data are sequentially read from the time-series detection data according to the acquisition time order to form a vibration acceleration sequence. Zero-reference statistical processing is performed within the continuous sampling intervals of the vibration acceleration sequence to calculate the average acceleration value of each sampling interval and determine the average acceleration value as the DC component of the corresponding sampling interval. Then, the DC component of the corresponding sampling interval is subtracted from the vehicle vibration acceleration data of each sampling point in the vibration acceleration sequence to obtain the zero-bias correction acceleration value, which is then continuously organized according to the acquisition time order to form zero-bias correction acceleration data.

[0072] Subsequently, zero-bias correction acceleration data is read from the time-series detection data to form a zero-bias correction acceleration sequence. This sequence is divided into multiple continuous sampling segments according to a fixed sampling interval. Fast Fourier Transform is performed on each continuous sampling segment to generate frequency domain amplitude data and frequency domain frequency data. Based on the frequency domain frequency data, frequency intervals are divided into frequency ranges. The amplitude data in the frequency range corresponding to vehicle bump vibration is determined as the vibration component amplitude. Amplitude suppression processing is performed on the amplitude data corresponding to other frequency ranges to obtain vibration frequency domain data. Then, inverse Fourier Transform is performed on the vibration frequency domain data to recover the time domain acceleration sequence, thereby generating vibration separation acceleration data.

[0073] After generating the vibration separation acceleration data, the sequence is divided into three consecutive sampling intervals based on the sampling time interval, and an integration calculation unit is constructed. Within each integration calculation unit, the Simpson integral algorithm is executed. A midpoint weight adjustment parameter is introduced to adjust the acceleration contribution of the intermediate sampling point, while a curvature compensation parameter is introduced to reconstruct the symmetrical weights of the acceleration contributions at both ends of the sampling point, thereby generating incremental acceleration integral data. This incremental data is then continuously accumulated and updated to generate the vehicle vibration velocity sequence. Subsequently, a second-level integration process is performed on the vehicle vibration velocity sequence in the same manner to generate vehicle vibration displacement data.

[0074] After obtaining vehicle vibration displacement data, an improved singular spectrum analysis algorithm is applied. First, a trajectory matrix is ​​constructed based on the embedding window length, and singular value decomposition (SVD) is performed on the trajectory matrix to generate a singular value sequence and a sequence of singular component matrices. Then, a trend energy percentage parameter is introduced to calculate the cumulative energy percentage of the singular value sequence, determining the number of trend components that meet the trend reconstruction requirements, and extracting the corresponding singular component matrices to generate the trend matrix. Next, a reconstruction weighting index parameter is introduced to perform a weighted diagonal average reconstruction of the trend matrix to generate a trend component sequence. Finally, a sampling-point differential subtraction process is performed between the vehicle vibration displacement sequence and the trend component sequence to generate displacement correction data.

[0075] Subsequently, road elevation measurement data, displacement correction data, and vehicle mileage data are read from the time-series detection data. Spatial distance markers are established for each sampling point based on the vehicle mileage data. Monotonicity verification is performed on the spatial distance markers to generate a set of valid sampling points. Then, spatial distance alignment processing is performed on the road elevation measurement data and displacement correction data based on the continuous distance increment sequence. Finally, resampling processing is performed according to the predetermined spatial sampling interval to generate equidistant detection data.

[0076] After generating equidistant detection data, sign consistency verification is performed on the equidistant displacement correction data in the equidistant detection data, and neighborhood smoothing replacement is performed on the locations where sign changes occur to generate smooth displacement correction data; at the same time, continuity verification is performed on the equidistant pavement elevation measurement data, and neighborhood interpolation replacement is performed on the locations where changes occur to generate smooth pavement elevation measurement data; then the smooth displacement correction data is superimposed on the smooth pavement elevation measurement data point by point at equidistant spatial sampling points to generate compensated pavement longitudinal profile elevation data.

[0077] After the longitudinal profile elevation data of the compensated pavement is generated, the elevation data is read according to the equidistant spatial sampling point sequence and missing sampling point verification processing is performed. When missing sampling points exist, interpolation completion processing is performed on the missing sections to generate a complete longitudinal profile elevation sequence. Then, zero-datum alignment processing is performed on the complete longitudinal profile elevation sequence to generate a datum-aligned longitudinal profile elevation sequence. Subsequently, the datum-aligned longitudinal profile elevation sequence is divided into multiple continuous evaluation sections according to the predetermined evaluation section length. For each evaluation section, suspension response recursive calculation processing is performed to generate a suspension response sequence. Then, response difference calculation processing is performed on the suspension response sequence to obtain a response difference sequence. The section smoothness index value of each evaluation section is generated through cumulative aggregation processing. Finally, the International Roughness Index (IRI) value of the pavement is generated through average convergence processing.

[0078] In a specific implementation example, during the continuous driving of the vehicle along the detection road, a laser rangefinder continuously collects road surface elevation measurement data, an accelerometer continuously collects vehicle vibration acceleration data, and an encoder continuously collects vehicle mileage data, recording a time stamp for each sampled data. Subsequently, time alignment processing is performed on the three types of data based on the time stamps, and they are combined under the same time stamp to form synchronous detection record data. This data is then organized according to the order of collection time to form a continuous data sequence, thereby generating time-series detection data.

[0079] Subsequently, vehicle vibration acceleration data is read from the time-series detection data and a vibration acceleration sequence is formed. Zero-reference statistical processing is performed within the continuous sampling interval to calculate the average acceleration value. This average acceleration value is then used as the DC component. The DC component is subtracted from the vibration acceleration sequence point by point to generate zero-bias correction acceleration data.

[0080] Next, the zero-bias correction acceleration data is processed by Fast Fourier Transform to obtain frequency domain amplitude data and frequency domain frequency data. The frequency domain amplitude data is then divided according to the frequency range of vehicle bump vibration, retaining the amplitude of vibration components and suppressing the amplitude of noise components. Finally, the inverse Fourier transform is used to restore the time domain acceleration sequence containing only vehicle bump vibration components, thereby generating vibration-separated acceleration data.

[0081] Subsequently, an integral calculation unit is constructed according to the three consecutive sampling intervals. Within each integral calculation unit, the Simpson integral weight is reconstructed and calculated using the midpoint weight adjustment parameter and the curvature compensation parameter to generate acceleration integral increment data. The vehicle vibration velocity sequence is then generated through continuous cumulative update processing. Finally, the vehicle vibration velocity sequence is subjected to a second-level integral processing to generate vehicle vibration displacement data.

[0082] After the vehicle vibration displacement data is generated, the vehicle vibration displacement sequence is processed by an improved singular spectrum analysis algorithm. By constructing a trajectory matrix and performing singular value decomposition, singular value sequences and singular component matrix sequences are obtained. Then, the number of trend components is determined according to the trend energy proportion parameter and the trend matrix is ​​extracted. Then, the trend matrix is ​​reconstructed by weighted diagonal average according to the reconstruction weighted index parameter to generate the trend component sequence. Finally, displacement correction data is generated by differential subtraction processing at each sampling point.

[0083] Subsequently, based on vehicle mileage data, spatial distance alignment processing was performed on the road elevation measurement data and displacement correction data, and resampling processing was performed according to a predetermined spatial sampling interval to generate equidistant detection data. Then, sign consistency verification processing and neighborhood smoothing replacement processing were performed on the equidistant road elevation measurement data, and continuity verification processing and neighborhood interpolation replacement processing were performed on the equidistant road elevation measurement data. Finally, the smoothed displacement correction data was superimposed on the smoothed road elevation measurement data to generate compensated road longitudinal profile elevation data.

[0084] After obtaining the longitudinal profile elevation data of the compensated pavement, the data undergoes missing sampling point verification and interpolation completion to generate a complete longitudinal profile elevation sequence. This sequence is then subjected to zero-datum alignment to generate a datum-aligned longitudinal profile elevation sequence. Subsequently, the datum-aligned longitudinal profile elevation sequence is divided into segments according to the predetermined evaluation segment length, generating an evaluation segment sequence. For each evaluation segment, recursive suspension response calculation and response difference calculation are performed. The response difference sequence is then cumulatively summarized to generate segment roughness index values. Finally, the International Roughness Index (IRI) value is obtained through average convergence processing.

[0085] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A vehicle-mounted laser IRI detection method based on acceleration vibration compensation, characterized in that: Includes the following steps: The system acquires road elevation measurement data collected by a laser rangefinder, vehicle vibration acceleration data collected by an accelerometer, and vehicle mileage data collected by an encoder during vehicle operation. It then performs time synchronization processing on the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data according to the acquisition time sequence to generate time-series detection data. DC removal processing is performed on the vehicle vibration acceleration data in the time-series detection data to eliminate the DC component in the vehicle vibration acceleration data and generate zero-bias correction acceleration data. Fourier transform processing is performed on the zero bias correction acceleration data. Based on the frequency domain distribution of the vibration signal, the frequency band of vehicle bump vibration is extracted and the noise signal is filtered out to generate vibration separation acceleration data. The vibration separation acceleration data is processed by Simpson integral algorithm according to the acquisition time sequence. The vibration separation acceleration data is constructed into an integral calculation unit according to the three consecutive sampling intervals. In each integral calculation unit, the Simpson integral weight is reconstructed according to the midpoint weight adjustment parameter and the curvature compensation parameter. The velocity integral and displacement integral are calculated sequentially on the reconstructed acceleration integral results to generate vehicle vibration displacement data. The vehicle vibration displacement data is processed by a singular spectrum analysis algorithm to construct the trajectory matrix of the vehicle vibration displacement sequence and perform singular value decomposition on the trajectory matrix. The number of singular components participating in trend reconstruction is determined according to the trend energy proportion parameter. The trend matrix is ​​reconstructed by weighted diagonal average using the reconstruction weighting index parameter. The trend components are separated from the vehicle vibration displacement data to generate displacement correction data. Based on vehicle mileage data, spatial distance alignment processing is performed on road elevation measurement data and displacement correction data, and resampling processing is performed on road elevation measurement data and displacement correction data according to a predetermined spatial sampling interval to generate equidistant detection data; Vibration displacement compensation processing is performed on the pavement elevation measurement data in the equidistant detection data using displacement correction data to generate compensated pavement longitudinal profile elevation data. The International Roughness Index (IRI) value of the pavement is calculated based on the longitudinal profile elevation data of the compensated pavement.

2. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of time series detection data includes the following steps: It receives road elevation measurement data continuously collected by a laser rangefinder, vehicle vibration acceleration data continuously collected by an accelerometer, and vehicle mileage data continuously collected by an encoder during vehicle operation, and records the corresponding collection time markers for the road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data respectively. The road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data are time-aligned according to the acquisition time sequence. Under the same acquisition time mark, the corresponding road elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data are combined to form synchronous detection record data. The synchronous detection record data under continuous acquisition time markers are processed to form a continuous data sequence according to the acquisition time order; Road surface elevation measurement data, vehicle vibration acceleration data, and vehicle mileage data from a continuous data sequence are organized in a unified manner to generate time-series detection data.

3. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of zero-bias correction acceleration data includes the following steps: Vehicle vibration acceleration data is read from time-series detection data, and the vehicle vibration acceleration data is continuously read in the order of acquisition time to form a vibration acceleration sequence; Within the continuous sampling interval of the vibration acceleration sequence, zero-reference statistical processing is performed on the vehicle vibration acceleration data to calculate the average acceleration value of the vibration acceleration sequence within the current sampling interval, and the average acceleration value is determined as the DC component of the current sampling interval. According to the acquisition time sequence, the DC component of the corresponding sampling interval is subtracted from the vehicle vibration acceleration data of each sampling point in the vibration acceleration sequence to generate the zero bias correction acceleration value. The zero-bias correction acceleration values ​​generated at each sampling point are continuously organized according to the acquisition time sequence to form zero-bias correction acceleration data.

4. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of vibration separation acceleration data includes the following steps: Zero bias correction acceleration data is read from the time-series detection data and continuously read in the order of acquisition time to form a zero bias correction acceleration sequence. The zero-bias correction acceleration sequence is organized into continuous sampling segments according to a fixed sampling interval, and the zero-bias correction acceleration data in each continuous sampling segment is processed by fast Fourier transform to generate corresponding frequency domain amplitude data and frequency domain frequency data. The frequency domain amplitude data is divided into frequency intervals based on the frequency domain frequency data. The amplitude data in the frequency interval corresponding to vehicle bump vibration is determined as the vibration component amplitude, and the amplitude data below the vehicle bump vibration frequency interval and the amplitude data above the vehicle bump vibration frequency interval are determined as the noise component amplitude. The frequency domain data corresponding to the vibration component amplitude is retained, and the frequency domain data corresponding to the noise component amplitude is subjected to amplitude suppression processing to generate vibration frequency domain data. The vibration frequency domain data is processed by inverse Fourier transform to recover the time domain acceleration sequence containing only the vehicle bump vibration component. The time domain acceleration sequence is then organized according to the acquisition time order to generate vibration-separated acceleration data.

5. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of vehicle vibration displacement data includes the following steps: Vibration separation acceleration data is read from time-series detection data, and the vibration separation acceleration data is continuously read and processed according to the acquisition time sequence to form a vibration separation acceleration sequence; Based on the sampling time interval of the vibration separation acceleration sequence, the vibration separation acceleration sequence is divided into multiple consecutive three-point sampling intervals according to three consecutive sampling points, and an integral calculation unit is constructed for each consecutive three-point sampling interval. Simpson's integral algorithm is executed in each integral calculation unit. A midpoint weight adjustment parameter is introduced to adjust the acceleration contribution of the middle sampling point in the integral calculation unit. At the same time, a curvature compensation parameter is introduced to reconstruct the acceleration contribution of the sampling points at both ends in the integral calculation unit with symmetrical weights, so as to generate incremental acceleration integral data. The acceleration integral increment data generated by each integral calculation unit is continuously accumulated and updated according to the acquisition time sequence, and organized to form a vehicle vibration velocity sequence. Based on the vehicle vibration velocity sequence, a velocity integral calculation unit is constructed according to three consecutive sampling points. The Simpson integral algorithm is executed in each velocity integral calculation unit for calculation. The weights of each sampling point in the velocity integral calculation unit are reconstructed based on the midpoint weight adjustment parameter and the curvature compensation parameter to generate velocity integral incremental data. The incremental data of each velocity integral is continuously accumulated and updated according to the acquisition time sequence to form vehicle vibration displacement data.

6. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of displacement correction data includes the following steps: Vehicle vibration displacement data is read from time-series detection data, and the vehicle vibration displacement data is continuously read and processed according to the acquisition time sequence to form a vehicle vibration displacement sequence. An improved singular spectrum analysis algorithm is applied to the vehicle vibration displacement sequence. The vehicle vibration displacement sequence is constructed into a trajectory matrix according to the embedding window length, and singular value decomposition is performed on the trajectory matrix to generate a singular value sequence and a corresponding singular component matrix sequence arranged in the order of singular components. A trend energy percentage parameter is introduced to perform cumulative energy percentage calculation on the singular value sequence, determine the number of trend components that meet the requirements of the trend energy percentage parameter, and extract the trend component set from the singular component matrix sequence accordingly to generate a trend matrix. A weighted index parameter is introduced to perform a weighted diagonal average reconstruction on the trend matrix. During the reconstruction process, a weighted coefficient is assigned to the matrix elements involved in the reconstruction according to the diagonal support, and the weighted average reconstruction is completed to generate a trend component sequence. The vehicle vibration displacement sequence and trend component sequence are subjected to sampling point differential subtraction processing to remove the trend component sequence from the vehicle vibration displacement sequence, generate displacement correction data, and continuously organize and output the displacement correction data according to the acquisition time sequence.

7. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of isometric detection data includes the following steps: Road surface elevation measurement data, displacement correction data, and vehicle mileage data are read from the time-series detection data. The road surface elevation measurement data, displacement correction data, and vehicle mileage data are read and processed synchronously according to the collection time order, and the corresponding sampling point set is organized. In the corresponding set of sampling points, a spatial distance marker is associated with each sampling point based on the vehicle mileage data, and the spatial distance marker is subjected to monotonicity verification in the order of collection time. When there is backtracking in the spatial distance marker, the sampling point that has backtracked is removed to generate a set of valid sampling points. The adjacent distance increment sequence is calculated based on the spatial distance markers of adjacent sampling points in the effective sampling point set, and abnormal distance increments are removed from the adjacent distance increment sequence to generate a continuous distance increment sequence. Based on the continuous distance increment sequence, spatial distance alignment processing is performed on the road surface elevation measurement data and displacement correction data to form an aligned sampling pair under the same spatial distance marker, thereby generating alignment detection data; Based on the predetermined spatial sampling interval, an equidistant spatial sampling point sequence is generated, and resampling processing is performed in the alignment detection data with the equidistant spatial sampling point sequence as the reference to generate equidistant road surface elevation measurement data and equidistant displacement correction data that correspond one-to-one with the equidistant spatial sampling point sequence. The equidistant road surface elevation measurement data and equidistant displacement correction data are combined and processed in the order of the equidistant spatial sampling point sequence to generate equidistant detection data.

8. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of compensated road longitudinal profile elevation data includes the following steps: Equidistant road surface elevation measurement data and equidistant displacement correction data are read from the equidistant detection data, and the equidistant road surface elevation measurement data and equidistant displacement correction data are synchronously read and processed according to the sequence of equidistant spatial sampling points to form an equidistant compensation input sequence. The sign consistency check is performed on the equidistant displacement correction data in the equidistant compensation input sequence. When a sign change occurs in the equidistant displacement correction data, the corresponding equidistant spatial sampling point is located, and the neighborhood smoothing replacement processing is performed on the equidistant displacement correction data corresponding to the equidistant spatial sampling point to generate smooth displacement correction data. The continuity verification process is performed on the equidistant road surface elevation measurement data in the equidistant compensation input sequence. When there is a sudden change in the equidistant road surface elevation measurement data, the corresponding equidistant spatial sampling point is located, and the neighborhood interpolation replacement process is performed on the equidistant road surface elevation measurement data corresponding to the equidistant spatial sampling point to generate smooth road surface elevation measurement data. The smooth road surface elevation measurement data and smooth displacement correction data are subjected to equidistant spatial sampling point superposition compensation processing. The smooth displacement correction data is superimposed onto the smooth road surface elevation measurement data in the order of sampling points to generate compensated road surface longitudinal profile elevation data. The longitudinal profile elevation data of the compensated road surface is continuously organized and output according to the sequence of equidistant spatial sampling points, so that the longitudinal profile elevation data of the compensated road surface maintains a sequence structure that corresponds one-to-one with the sequence of equidistant spatial sampling points.

9. The vehicle-mounted laser IRI detection method based on acceleration vibration compensation according to claim 1, characterized in that: The generation of the International Roughness Index (IRI) value for road surface includes the following steps: The longitudinal profile elevation data of the compensated road surface is continuously read from the longitudinal profile elevation data of the compensated road surface in the order of the equidistant spatial sampling point sequence, and the missing sampling point verification processing is performed on the longitudinal profile elevation data of the compensated road surface. When there are missing sampling points, the missing section is filled by interpolation to generate a complete longitudinal profile elevation sequence. Zero datum alignment processing is performed on the complete longitudinal profile high program sequence. By updating the datum elevation through translation of the complete longitudinal profile high program sequence, the elevation values ​​of the starting sampling point and the ending sampling point of the complete longitudinal profile high program sequence are kept consistent, thus generating a datum-aligned longitudinal profile high program sequence. In the benchmark aligned longitudinal profile high program sequence, multiple continuous evaluation segments are divided according to the predetermined evaluation segment length, and each evaluation segment is organized segment by segment according to the sequence of equidistant spatial sampling points to generate an evaluation segment sequence. For each evaluation segment sequence, a recursive calculation of the suspension response is performed. The evaluation segment sequence is used as the road excitation input to generate the corresponding suspension response sequence. The suspension response sequence is then processed by response difference calculation to obtain the response difference sequence. The response difference sequence corresponding to each assessment section is cumulatively summarized to calculate the section smoothness index value of each assessment section. The smoothness index values ​​of each section are then averaged and converged in the order of the assessment sections to generate the International Roughness Index (IRI) value of the road surface.

Images