A method and system for evaluating the quality of continuous compaction of a roadbed based on spatial heterogeneity features

Abstract

This invention discloses a method and system for evaluating the continuous compaction quality of roadbeds based on spatial heterogeneity characteristics. By time-aligning the vertical acceleration signal information of the vibratory roller and the three-dimensional spatial coordinate information of the compaction roller's center point, a compaction index is generated based on the vertical acceleration signal information of the vibratory roller. Based on the compaction index, the compaction surface is spatially gridded and a digital compaction field is constructed. The constructed digital compaction field is then quantified for multi-dimensional and multi-scale spatial heterogeneity to obtain multi-dimensional spatial heterogeneity indices for the compaction surface. Through multi-source data fusion and gridded reconstruction, discrete compaction measurement points are constructed into a continuous spatial digital field. This transforms the evaluation object from isolated numerical points into field data with spatial location and correlation, achieving a quantitative evaluation of the spatial distribution structure, variation characteristics, and continuity of compaction quality, ensuring that the evaluation results fully correspond to the actual spatial non-uniformity of the engineering entity.

Description

Technical Field

[0001] This invention belongs to the field of intelligent construction and digital geotechnical engineering technology, specifically relating to a method and system for evaluating the continuous compaction quality of roadbeds based on spatial heterogeneity characteristics. Background Technology

[0002] Continuous compaction control technology, which uses sensors installed on road rollers to collect vibration responses in real time and calculate compaction counts (such as CMV), has become a key means of quality process control in modern roadbed construction. However, current mainstream technologies and evaluation methods lack a spatial dimension at the core level, severely restricting their development towards refinement and intelligence. Specifically, this manifests in the following four fundamental limitations: The evaluation dimensions are lacking, and spatial field characteristics are ignored: Existing methods treat continuously collected compaction data as discrete and independent "point values," performing only statistical analyses such as mean, pass rate, and coefficient of variation, completely stripping away their spatial location and correlation. This results in the inability to reveal the true spatial distribution structure, variation characteristics, and continuity of compaction quality, leading to a serious disconnect between the evaluation results and the actual spatial state of the engineering entity.

[0003] Lack of ability to identify and quantify harmful spatial structures: Existing systems struggle to effectively identify and differentiate whether compaction defects (such as under-compaction and over-compaction) are spatially discrete and scattered, or have formed continuous, strip-like or clump-like clusters. There is a lack of effective quantitative methods for key spatial structural information, such as whether the transition between defective and acceptable areas is a clear abrupt change or a gradual, ambiguous one, and whether there are drastic fluctuations within the defective area. This makes it impossible to scientifically assess the structural compatibility and potential instability risks within the fill structure.

[0004] There is a lack of evaluation indicators and standards that consider spatial characteristics: Current evaluations mainly rely on traditional indicators based on numerical statistics (such as mean and coefficient of variation), and a dedicated indicator system has not yet been established to systematically quantify spatial heterogeneity characteristics such as spatial autocorrelation, surface geometric complexity, and local configuration order. Therefore, it is impossible to transform the key quality information of "spatial distribution pattern" into measurable, comparable, and evaluable engineering language.

[0005] The ability to provide dynamic early warning and precise control based on spatial evolution is insufficient: Existing technologies struggle to provide early warnings of areas that may develop into serious defects during construction, based on the evolutionary trends of the spatial distribution patterns of compaction states. When quality problems are discovered, it is also difficult to accurately attribute and define the causes of problems based on the spatial morphology, boundary characteristics, and surrounding environment of abnormal areas, leading to decision-making that relies on experience and lacks specificity and scientific rigor. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for evaluating the continuous compaction quality of roadbeds based on spatial heterogeneity characteristics, so as to overcome the lack of spatial dimension in the existing technology, which makes it impossible to accurately evaluate the continuous compaction quality of roadbeds.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for assessing the continuous compaction quality of roadbeds based on spatial heterogeneity characteristics includes the following steps: S1, real-time acquisition of vertical acceleration signal information of vibrating wheel and three-dimensional spatial coordinate information of the center point of compaction wheel, and time alignment of the acquired vertical acceleration signal information of vibrating wheel and three-dimensional spatial coordinate information of the center point of compaction wheel; S2, based on the vertical acceleration signal information of the vibrating wheel, generates compaction index, and based on the compaction index, spatially grids the compaction surface and constructs a digital compaction field; S3 quantifies the spatial heterogeneity of the constructed digital compaction field in multiple dimensions and at multiple scales, and obtains multi-dimensional spatial heterogeneity indicators of the compaction operation surface.

[0008] Preferably, the vertical acceleration signal of the vibrating wheel is filtered to remove low-frequency vehicle body sway interference and high-frequency noise, resulting in a filtered signal. The filtered signal is then analyzed using consecutive short time windows. Within these short time windows, a Fourier transform is performed on the acceleration signal to extract its fundamental frequency. First harmonic amplitude and vibration frequency and amplitude Calculate the compaction count value for each time window according to the formula. : (1) in, This is a calibration constant; thus, a time stamp corresponding to the timestamp is generated. CMV Time series; Using timestamps CMV The time series is rigorously matched with the 3D spatial coordinates and attitude angle sequence of the compaction wheel center point output by GNSS / IMU; the positioning point coordinates are corrected to the center point of the contact surface between the vibrating wheel and the ground through coordinate transformation, generating a sequence with 3D coordinates (…). X i , Y i , Z i )and CMV i The set of data points of value .

[0009] Preferably, the compaction work surface is divided into rectangular grids, and robust processing is performed on the compaction index data points within each grid cell to generate a weighted composite for each grid cell. CMV value: (2) in, The median of the data within the grid cell. The interquartile range after removing outliers. The average value of the interpolation points of the triangular mesh. The weighting coefficients are dynamically adjusted based on the distribution quality of grid data; the weighting coefficients are determined using a direct assignment method, based on the effective data contained in each grid cell. CMV The proportions of the data points calculated as the median, interquartile range, and interpolation average are used as weighting coefficients.

[0010] Preferably, the constructed digital compaction field is subjected to multi-dimensional and multi-scale spatial heterogeneity quantification, specifically including: quantification of the macro-spatial characteristics of the compaction operation surface and fine identification and quantification of engineering risk areas.

[0011] Preferably, the quantification of the macroscopic spatial characteristics of the compaction surface specifically involves a comprehensive evaluation of the overall spatial heterogeneity of the compaction surface from three dimensions: spatial correlation, geometric complexity, and information order.

[0012] Preferably, the global Moran index is used to determine whether there is significant spatial clustering in the quality distribution, quantitatively assess the overall strength and pattern of spatial dependence of compaction quality, and after confirming the existence of significant spatial structure in compaction quality through the global Moran index, the essential characteristics of spatial heterogeneity are more profoundly quantified at the overall level; spatial configuration entropy is used to characterize the degree of order or disorder of the arrangement and combination of different compaction grade regions in the local spatial range, reflecting the stability of spatial configuration from the perspective of information theory.

[0013] Preferably, the detailed identification and quantification of engineering risk areas specifically includes: using the density-based DBSCAN algorithm to perform cluster analysis on defective mesh cells; after completing the cluster identification and basic feature extraction of defective areas, precise quantitative descriptions are performed from the perspectives of boundary transition characteristics, spatial neighborhood environment, internal structural state, and multi-zone correlation.

[0014] Preferably, early warning is based on obtaining multi-dimensional spatial heterogeneity indicators of the compaction surface.

[0015] Preferably, early warning based on the acquisition of multi-dimensional spatial heterogeneity indicators of the compaction surface specifically includes: establishing a threshold database and early warning level database containing overall compaction surface indicators and local defect clustering indicators, and adopting a two-layer early warning mechanism of overall compaction surface early warning and local defect clustering early warning for early warning.

[0016] A roadbed continuous compaction quality assessment system based on spatial heterogeneity characteristics includes a preprocessing module, a spatial grid module, and an assessment module. The preprocessing module collects the vertical acceleration signal information of the vibrating wheel and the three-dimensional spatial coordinate information of the center point of the compaction wheel in real time, and performs time alignment on the collected vertical acceleration signal information of the vibrating wheel and the three-dimensional spatial coordinate information of the center point of the compaction wheel. The spatial grid module generates compaction indices based on the vertical acceleration signal information of the vibrating wheel, and then uses these indices to spatially grid the compaction surface and construct a digital compaction field. The evaluation module quantifies the spatial heterogeneity of the constructed digital compaction field in multiple dimensions and at multiple scales, and obtains multi-dimensional spatial heterogeneity indicators of the compaction operation surface.

[0017] Compared with the prior art, the present invention has the following beneficial technical effects: This invention provides a method for evaluating the continuous compaction quality of roadbeds based on spatial heterogeneity. It involves real-time acquisition of vertical acceleration signals from a vibratory roller and the three-dimensional spatial coordinates of the roller's center point. The acquired vertical acceleration signals and coordinates are time-aligned. Compaction indices are generated based on the vertical acceleration signals, and the compaction surface is spatially gridded to construct a digital compaction field. The constructed digital compaction field undergoes multi-dimensional and multi-scale spatial heterogeneity quantification to obtain multi-dimensional spatial heterogeneity indices for the compaction surface. Through multi-source data fusion and gridded reconstruction, discrete compaction measurement points are transformed into a continuous spatial digital field. This transforms the evaluation object from isolated numerical points into field data with spatial location and correlation, enabling a quantitative assessment of the spatial distribution structure, variability, and continuity of compaction quality. The evaluation results fully correspond to the actual spatial non-uniformity of the engineering entity.

[0018] With the preferred approach of using a multi-scale spatial analysis framework and local refined clustering analysis, this method can effectively distinguish between sporadic point defects and continuous sheet defects, and quantify the clarity of defect boundaries, internal fluctuations, and neighborhood environment, thereby accurately identifying spatial clustering patterns and anomalous structures with high engineering risks. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the process for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics in an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram illustrating the quantification of the macroscopic spatial characteristics of the compaction surface in an embodiment of the present invention.

[0021] Figure 3This is a schematic diagram of the defect classification results based on the DBSCAN clustering algorithm in an embodiment of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] like Figure 1 As shown, this invention provides a method for evaluating the continuous compaction quality of roadbeds based on spatial heterogeneity characteristics. Through a three-step logical progression—data field reconstruction, multi-scale heterogeneity analysis, and evaluation and early warning decision-making—it achieves in-depth mining and engineering application of the spatial structural characteristics of compaction quality. Specifically, it includes the following steps: S1, Multi-source data synchronous acquisition and spatiotemporal alignment: Real-time acquisition of vertical acceleration signal information of vibrating wheel and three-dimensional spatial coordinate information of the center point of compaction wheel, and time alignment of the acquired vertical acceleration signal information of vibrating wheel and three-dimensional spatial coordinate information of the center point of compaction wheel; S2, based on the vertical acceleration signal information of the vibrating wheel, generates compaction index, and based on the compaction index, spatially grids the compaction surface and constructs a digital compaction field; S3 quantifies the spatial heterogeneity of the constructed digital compaction field in multiple dimensions and at multiple scales, and obtains multi-dimensional spatial heterogeneity indicators of the compaction operation surface.

[0025] In a specific embodiment of the invention, the multi-source data synchronous acquisition and spatiotemporal alignment are specifically achieved by equipping a standard construction compaction work surface (50m long × 20m wide) with an intelligent compaction system, which includes an integrated vibration sensing module, a positioning module, and a positioning and control module. The vibration sensing module uses an ICP accelerometer (range ±50g, frequency range 0.5Hz to 1000Hz), rigidly mounted on the vibratory roller axle, to acquire the vertical acceleration signal of the vibratory roller in real time at a sampling frequency of 1000Hz. The positioning module uses a fusion positioning terminal integrating GNSS-RTK and IMU, installed at the center of the roller body, outputting the three-dimensional spatial coordinates (longitude, latitude, elevation, planar accuracy ≤2cm, elevation accuracy ≤5cm) and attitude angle of the compaction roller center point at a frequency of not less than 100 Hz. Synchronization and Control Module: Uses satellite synchronization clock (such as GPS PPS signal) as a unified clock source to stamp all sensor data streams with high-precision timestamps in the same time system (synchronization error <1ms), ensuring that the acceleration signal is strictly aligned with each corresponding spatial position.

[0026] The compaction index is generated based on the vertical acceleration signal information of the vibrating wheel. CMV Specifically, it includes the following steps: Signal filtering: A bandpass filter (10Hz to 150Hz) is used to filter out low-frequency vehicle body sway interference and high-frequency noise from the vertical acceleration signal of the vibrating wheel, resulting in a filtered signal; Compaction count calculation: Analysis is performed using consecutive short time windows in the filtered signal. Within these consecutive short time windows (window length...) L Within an integer multiple of the vibration period, perform a Fourier transform on the acceleration to extract its fundamental frequency. The amplitude of the first harmonic of the vibrating wheel (operating frequency, 25Hz to 35Hz) and vibration frequency and amplitude Calculate the compaction count value for each time window using the formula. : (1) in, This is a calibration constant; thus, a time stamp corresponding to the timestamp is generated. CMV Time series; Trajectory point generation: using timestamps, CMV The time series is strictly matched with the three-dimensional spatial coordinates and attitude angle sequence of the compaction wheel center point output by GNSS / IMU; combined with the mounting arm values ​​of the vibrating wheel and the positioning terminal, the positioning point coordinates are corrected to the center point of the contact surface between the vibrating wheel and the ground through coordinate transformation, generating a three-dimensional coordinate system. X i , Yi , Z i )and CMV i The set of data points of value Remove invalid data points caused by satellite signal loss or abnormal speed.

[0027] Spatial Gridding and Digital Compaction Field Construction: In large-scale compaction operations, compaction quality can vary significantly across different locations. Therefore, dividing the work surface into hierarchical grid cells is fundamental to ensuring assessment accuracy. This involves the following steps: The compaction surface (50m long × 20m wide) was divided into rectangular grids of 0.25m long × 2m wide, totaling 2000 grid units. The unit width (2m) matched the width of the compaction wheel, and the unit length (0.25m) corresponded to the contact length between the compaction wheel and the soil. For each grid unit, the compaction index (…) was determined. CMV For each data point, the following robust processing is performed: Outlier filtering: A statistical method based on interquartile range is used to identify and remove outlier data points within each grid cell, eliminating the influence of sensor transient errors or external interference. For each filtered data point within a grid cell... CMV For each data point, the following statistical characteristics are calculated: median, interquartile range, and triangulation mean. A locally hierarchical triangulation network is constructed using the Delaunay triangulation method, and supplementary data points are generated for regions within the grid lacking original data points using linear interpolation. For each grid cell, a weighted composite is finally generated. CMV The value is calculated using the following formula: (2) in, The median of the data within the grid cell. The interquartile range after removing outliers. The average value of the interpolation points of the triangular mesh. These are weighting coefficients that are dynamically adjusted based on the distribution quality of grid data. The weighting coefficients are determined using a direct assignment method, based on the effective data contained within each grid cell. CMV The proportions of the data points calculated as the median, interquartile range, and interpolation average are used as weighting coefficients.

[0028] By mesh generation and robust statistical characteristic calculation, representative compaction quality data can be generated on large-area compaction surfaces. This process ensures comprehensive and accurate monitoring of the compaction quality of the surface, avoiding erroneous judgments caused by missing or biased local data. Through the above method, each mesh cell ultimately obtains an apparent compaction modulus value with high reliability and sufficient characterization of spatial structure features, providing a high-quality field data foundation for subsequent spatial heterogeneity analysis.

[0029] The digital compaction field constructed in the above steps is subjected to multi-dimensional and multi-scale spatial heterogeneity quantification. Through systematic spatial statistical and geometric analysis methods, the quality information implicit in the spatial distribution is transformed into quantifiable and interpretable evaluation indicators, laying a spatial analysis foundation for the intelligent comprehensive evaluation of compaction quality. The specific steps include: 2.1 Overall Level: Quantitative System of Macroscopic Spatial Characteristics of Compaction Surface: The overall hierarchical analysis aims to comprehensively characterize the spatial structural features of the compaction face from a macroscopic perspective. It establishes three independent yet complementary indicator systems to comprehensively evaluate the overall spatial heterogeneity of the compaction face from three dimensions: spatial correlation, geometric complexity, and information order. Specifically, it includes the following steps: 2.1.1 Spatial correlation quantification: global spatial autocorrelation analysis; To determine whether compaction quality is systematically affected by explainable factors such as materials and processes, its spatial autocorrelation must first be quantified. If the compaction state exhibits significant spatial structure, this structure must be analyzed to pinpoint specific problems and optimize the process. The global Moran index is used to determine whether there is significant spatial clustering in the quality distribution, and it can be used to quantitatively assess the overall strength and pattern of the spatial dependence of compaction quality; specifically, the following steps are included: The compaction work surface is divided into N Each level of grid cell ( N = 20000, corresponding to a 0.25m × 2m grid. For each grid cell... i Its engineering compaction index value is All grid cells CMV The average value is Define the spatial weight matrix. , representing a grid cell i and j Spatial proximity between them; Global Moran index I The calculation formula is: (3) Spatial weight matrix This reflects the spatial proximity between grid cells; the closer two grid cells are, the greater their weight. The inverse distance weighting method is used for calculation. First, the grid cells are calculated... i and j Horizontal Euclidean distance between the center points : (4) in, and They are grids i and jThe coordinates of the center point, then the spatial weight It can be represented as: (5) in, p This is the distance attenuation parameter (its value is 1 or 2), which is used when two grids are far apart. Approaching 0.

[0030] Significance test: Calculate Moran's index I Expected value E( I ) and variance Var( I ), and thus obtain Z Score: (6) Under randomization, the formulas for calculating the expected value and variance are: (7) (8) in: (9) (10) (11) in, n The number of grid cells. When At a 95% confidence level, spatial autocorrelation can be considered statistically significant. Significant positive correlation ( I >0, Significant: Indicates spatial clusters of high or low compaction values, requiring focused analysis of local process or material anomalies. No significant spatial autocorrelation ( I Approaching 0 (Not significant): This means that the quality variation lacks structural and spatial dependence. This does not indicate uniformity, but rather reveals that the source of variation is global and random, and the root cause needs to be investigated from the overall control of material source consistency, mixing and transportation process, or compaction process.

[0031] 2.1.2 Quantification of Spatial Complexity: Multi-scale Fractal Dimension Analysis After confirming the significant spatial structure of compaction mass using the global Moran index, this study aims to quantify the essential characteristics of spatial heterogeneity at a more comprehensive level. To this end, spatial fractal dimension is employed to quantify the complexity of the spatial distribution of compaction mass and the degree of surface inhomogeneity.

[0032] The engineering compaction index established above based on a 0.25m × 2m grid CMV The value is the input data field. This will be the value of each grid cell. CMVThe values ​​are normalized to the interval [0, 1], forming a coordinate system based on planar coordinates ( X , Y (Based on) and normalized CMV The value is the height ( Z The three-dimensional data surface. Define a set of scale factor sequences. ,in A positive integer representing the multiple by which the original mesh is aggregated on the plane. For example, Corresponding to the original 0.25m × 2m grid, This corresponds to aggregating 2×2 adjacent original grids into a 0.5m×4m supergrid cell. For each scale s The entire three-dimensional space containing the data field is divided into a series of sides with lengths of 1. s A cube. In X - Y On the plane, each cube projects onto one The original grid region. For X - Y Each on the plane The region, assuming it contains all the normalized original grids. CMV The maximum value is The minimum value is The number of cube layers required to cover the data surface in this region along the Z direction is: (12) Accumulate all area , to obtain at scale s Total number of cubes required to cover the entire data field Fractal dimension calculation: for different scales and its corresponding Linear fitting is performed in a double logarithmic coordinate system: (13) Among them, the slope obtained from the fitting This is the compaction index. CMV The spatial fractal dimension. A value close to 2.0 indicates that the compaction quality is highly uniform in space. This indicates that there are some fluctuations, but the structure is relatively simple. This indicates that the compacted surface is extremely rough, suggesting that the compaction quality varies drastically in space, with inhomogeneities and abnormal fluctuations at multiple scales.

[0033] 2.1.3 Information Orderliness: Spatial Configuration Entropy Analysis; Spatial configuration entropy is used to characterize the degree of order or disorder in the arrangement and combination of regions with different compaction levels within a local spatial range, reflecting the stability of spatial configuration from the perspective of information theory.

[0034] Discretization of compaction state: Based on engineering design requirements, the engineering compaction index of each grid is discretized. CMV The values ​​are divided into four discrete state levels. Typically, they are divided into three levels: L 1 (Uncompacted): , L 2 (Qualified): , L 3 (over-compaction): . The target compaction value. Pattern extraction: using a size of... w × w The sliding window traverses the entire discretized compaction surface, where w This indicates the number of grid cells contained on each side of the window, and is an odd number (value is 3). For example... Figure 2 As shown, each window covers 3×3 adjacent grid cells. Given that the physical size of the grid cell in this embodiment is 0.25m×2m, therefore when... w When = 3, the actual physical area corresponding to this window is 0.75 × 6 m. Within each window... w 2 The state levels of each grid cell constitute a unique "local spatial configuration pattern". Pattern statistics and entropy calculation: statistically analyze all distinct local patterns occurring throughout the entire compaction surface. Let there be a total of... M A unique local pattern, the first i The frequency of this pattern is Then the spatial configuration entropy of the compaction surface. for: (14) The larger the value, the more diverse the local spatial patterns and the more chaotic the spatial configuration. Low spatial entropy ( ): This indicates that the local patterns are singular or have few types. Combining a high global Moran index and low fractal dimension allows for further analysis, indicating a well-defined, ordered structure in large blocks, with the entire compaction surface approaching global uniformity. High spatial entropy ( This indicates a wide variety of local patterns, with different levels of areas highly intertwined and mixed, resulting in chaotic spatial configurations and internal instability. The compaction quality exhibits typical characteristics of instability and unevenness, suggesting potential risks even if the overall statistical pass rate meets the standard. This indicator is highly sensitive to "spatial arrangement disorder" and can effectively identify hidden quality problems—numerically acceptable but spatially chaotic—that are easily missed by traditional statistical methods.

[0035] 2.2 Local Level: Refined Identification and Quantification of Engineering Risk Areas; Based on macroscopic diagnosis, a refined analysis is conducted on the identified abnormal areas to accurately locate, classify, and quantify engineering risk areas. The analysis includes under-compacted and over-compacted areas, achieving dual-end risk management.

[0036] 2.2.1 Local Anomaly Pattern Recognition: Spatial Cluster Analysis; To effectively distinguish between isolated defect points and continuous defect regions, and to overcome the sensitivity of traditional methods to parameter selection, this invention employs the density-based DBSCAN algorithm to perform cluster analysis on defect mesh cells.

[0037] The first step is to screen for abnormal mesh elements, extracting all compaction quality CMV values ​​that are lower than the compaction degree threshold. The center point of the grid cell and the point above the compaction threshold. The initial defect point set is constructed from the center points of the grid cells. It includes its spatial coordinates and compaction value.

[0038] This invention employs an adaptive parameter determination method, neighborhood radius Adaptive determination: based on the spatial fractal dimension calculated in step 2.1.2. and the geometric characteristics of the compacted surface. Definition ,in The median distance between all defect points and their nearest neighbors. As a scale factor, Determined. When the surface is complex ( When (high), use a smaller one. To identify more refined clusters; when the surface is smooth, a larger [size / size] is used. To identify a wider range of clusters. Minimum cluster point. Determining: Set as ,in m The total number of defects. p = 0.005 is a very small scaling factor to ensure that the algorithm is insensitive to sparse points.

[0039] The DBSCAN algorithm was applied to cluster undercompacted and overcompacted points respectively. The defective points were divided into three categories: core points (belonging to a certain cluster), boundary points (belonging to a certain cluster but with insufficient density), and noise points (isolated points).

[0040] Clustering feature extraction for compaction defect regions: For each defect cluster Calculate a set of morphological and statistical characteristics: cluster area : Number of grid cells within a cluster × Grid area Total cluster area = Number of clusters N × The percentage of total cluster area = / Total area of ​​the compacted surface. Cluster density. ,in Cluster perimeter. Defect shape index. The value measures the compactness of clusters; the higher the value, the less hierarchical they are.

[0041] Compaction strength characteristics (average defect depth) This quantifies the average severity of the defect area deviating from the acceptable standard: (15) in, The absolute difference between the CMV values ​​of a single defective mesh cell that are lower or higher than the target compaction threshold is used to define whether the region is "under-compacted" or "over-compacted." This applies to clustering. The average of all defect points is obtained. This indicator quantifies the average severity of under-compaction or over-compaction in defective areas. The larger the value, the further the average compaction level in the area deviates from the qualified standard, and the higher the risk of structural weakening.

[0042] Internal compaction variation coefficient Measuring the relative fluctuation of the compaction state within the defective area: (16) in, For clustering Compaction index of all grid cells The standard deviation measures the absolute dispersion of compaction values ​​within the cluster. for Compaction index of all grid cells The arithmetic mean.

[0043] Comprehensive Risk Index: Construct a weighted comprehensive evaluation index to conduct multi-dimensional risk rating of defective areas, supporting accurate disposal decisions.

[0044] (17) in, and These represent the maximum area and maximum defect depth of all clusters within the compacted surface, respectively. Weights Determined using the analytic hierarchy process, it satisfies... . The higher the value, the greater the risk of the defective area, and it should be addressed first.

[0045] Large-area dense clusters represent continuous, weakly compacted areas, posing the highest engineering risk. Small-area dispersed clusters represent localized, point-like defects, with relatively lower risk; for example... Figure 3 As shown. Isolated points represent sporadic defects, possibly caused by chance. Coefficient of variation. Used to measure the relative fluctuation or uniformity of compaction within a defective region. Even the average defect depth of two defective regions. same, Higher compaction levels indicate greater internal variations in compaction values, potentially including extremely low "range points." This internal inhomogeneity translates to a higher risk of localized stress concentration and uneven settlement potential in engineering, and its harmful effects are more insidious than those of overall uniform undercompaction.

[0046] 2.2.2 In-depth Quantification and Analysis of Defect Areas: After completing the clustering and identification of defective regions and the extraction of basic features (area, shape, average defect depth, internal coefficient of variation), in order to further evaluate their engineering risks and formation mechanisms, this invention proposes the following four in-depth quantitative indicators, which provide precise quantitative descriptions from the perspectives of boundary transition characteristics, spatial neighborhood environment, internal structural state, and multi-region correlation.

[0047] Compaction gradient at the boundary of the defect zone: Quantification of defect region edge CMV The value indicates the degree of change towards the surrounding acceptable area. It is used to determine whether the defect boundary is a sharp abrupt change or a gradual, ambiguous transition, providing a direct basis for determining the scope of repair. Calculation steps: Boundary mesh determination: for any identified defect cluster C k Find all cells that have at least one adjacent cell that do not belong to C k The grid cells constitute its boundary grid set. B k .

[0048] Local gradient calculation: for boundary set B k Each grid cell in i ,calculate The average value of all non-cluster neighboring grids The difference: This difference Reflected in the grid i The local compaction difference at the location.

[0049] Cluster boundary gradient exponent: Calculate the arithmetic mean of the local gradients of all boundary grids in the defect cluster, and use it as the boundary compaction transition gradient of the defect region. GDB k : (18) high GDB Value: Indicates a steep compaction boundary between the defect area and the surrounding area. This may stem from a sudden change in material properties (such as the boundary of segregated agglomerates) or a clear demarcation in the compaction process. Such defects have clear boundaries, and the repair area is easy to determine, but stress concentration may exist at the boundary. Low GDB Value: This indicates a smooth transition between the defect area and the surrounding area, classifying it as a "soft boundary" defect. It is typically caused by gradual changes such as insufficient compaction energy or slight variations in moisture content. Defining the extent of this type of defect is difficult, and the risk lies in the fact that its affected area may extend beyond the currently identified range, necessitating an expansion of the monitoring and treatment scope.

[0050] Local compaction quality index of defect area: Assess the overall compaction quality of the local spatial environment in which defect clusters are located. Used to distinguish between "defects isolated in good areas" and "defects appearing in clusters"; the former indicates localized risk, while the latter predicts a larger-scale systematic anomaly. Calculation steps: Define a neighborhood buffer: clustering based on defects C k Using the circumscribed polygon (or minimum circumscribed rectangle) as a reference, extend outwards by a fixed physical distance. D ( D = 1m), forming a buffer zone. .

[0051] Statistical Environment Grid: Statistical Buffer Within, all grid cells that do not belong to any defect cluster (i.e., areas judged as qualified or over-pressured).

[0052] Calculating the local compaction quality index: Calculating these "environmental grids" CMV average value And normalize it: (19) in, This represents the lower acceptable threshold for CMV. High LEQI Value (≥ 0.8): The defective area is surrounded by a high-quality compacted area and can be considered an "isolated cavity in a good matrix." Engineering risks are relatively limited. Low. LEQI Value (<0.8): The local environment where the defect is located also has low compaction quality. This indicates that the defect may be part of a larger weak zone, with high systemic risk.

[0053] Defect region internal structural heterogeneity index : internal coefficient of variation CV kBased on this, the spatial distribution heterogeneity of compaction values ​​within the defect region is further quantified to identify patterns such as core-shell structures, gradient variations, or random dispersion, providing more refined structural information for defect etiology analysis. Specifically, the following steps are included: Internal gradient field calculation: for defect clustering CV k Each grid cell inside i Calculate its compaction index value With all its internal adjacent grids (i.e., belonging to the same grid) CV k Average compaction index of adjacent grids The absolute difference: .

[0054] Gradient distribution statistics: Calculate all internal mesh cells mean and standard deviation Heterogeneity index definition: (20) Low ISHI Value (close to 0): This indicates that although the overall compaction state within the defect area is low, the change is gradual and the structure is uniform, possibly due to uniformly weak material areas or insufficient overall compaction. High ISHI Value: This indicates severe fluctuations in compaction within the defect area, suggesting the possible presence of insufficiently mixed material clumps, localized moisture content anomalies, or varying degrees of missed compaction. Such defects pose a higher risk of localized stress concentration and uneven settlement potential, requiring close monitoring.

[0055] The above three deepening indicators The study provides complementary quantitative perspectives on the clarity of defect boundaries, environmental quality, and internal uniformity, which together support in-depth analysis of the risk level, formation mechanism, and treatment strategies of defect areas.

[0056] In a specific embodiment of the present invention, early warning is based on acquiring multi-dimensional spatial heterogeneity indicators of the compaction surface. By establishing specific threshold judgment standards and early warning classification levels, intelligent early warning of compaction quality risks is achieved. Furthermore, defective areas and their risk levels are intuitively displayed through visualization technology. The specific steps include: 3.1 Multi-dimensional indicator threshold library and early warning level library: Establish a threshold database and early warning level database that includes overall compaction surface indicators and local defect clustering indicators: (1) Overall compaction surface index threshold (based on historical engineering data statistics): Table 1. Threshold Analysis of Overall Compacted Surface Indicators

[0057] (2) Threshold for local defect clustering index Table 2 Threshold Analysis of Local Defect Clustering Indicators

[0058] 3.2 Graded Early Warning Judgment Process and Specific Grades: A two-layer early warning mechanism is adopted, consisting of overall compaction surface early warning and local defect clustering early warning: (1) Overall compaction surface early warning judgment (first level early warning) High-risk warning trigger conditions: The corresponding warning will be triggered if any one of the high-risk warning conditions in Table 1 is met. Or, if three or more of the risk conditions in Table 1 are met.

[0059] Medium-risk warning trigger conditions: Meeting any one of the medium-risk warning conditions in Table 1 will trigger the corresponding warning. Alternatively, meeting two or more low-risk conditions simultaneously, but not meeting the high-risk condition.

[0060] Low-risk warning trigger condition: All overall indicators are within the low-risk range.

[0061] (2) Local defect clustering early warning judgment (second level early warning) Warning Level Assessment (based on multiple indicators, determined according to the strictest principle): High-risk defect assessment: Meets at least two of the high-risk conditions in Table 2 simultaneously. Medium-risk defect assessment: Meets at least two of the medium-risk conditions in Table 2 simultaneously. Low-risk defect assessment: Does not meet either the high-risk or medium-risk assessment conditions.

[0062] (3) Overall assessment of early warning level: The final warning level is determined according to the following levels: A red warning is issued if the overall compacted surface is at high risk, or if any high-risk defect exists. A yellow warning is issued if the overall compacted surface is at medium risk, and there are medium-risk defects but no high-risk defects. A green normal state is issued if the overall compacted surface is at low risk and local defects are at low risk or no risk.

[0063] This invention achieves a leap from point statistics to field analysis, solving the problem of missing evaluation dimensions. Through multi-source data fusion and gridded reconstruction, discrete compaction measurement points are constructed into a continuous spatial digital field. This transforms the evaluation object from isolated numerical points into field data with spatial location and correlation, enabling a quantitative assessment of the spatial distribution structure, variation characteristics, and continuity of compaction quality. This ensures that the evaluation results fully correspond to the actual spatial non-uniformity of the engineering entity. It can accurately identify and quantify harmful spatial structures, solving the problem of insufficient risk identification capabilities. Through a multi-scale spatial analytical framework and local refined clustering analysis, this method can effectively distinguish between sporadic point defects and continuous, patchy defects, and quantify the clarity of defect boundaries, internal fluctuations, and neighborhood environment, thereby accurately identifying spatial clustering patterns and abnormal structures with high engineering risks.

[0064] This invention establishes a systematic evaluation index system for spatial heterogeneity. It innovatively proposes and defines a complete set of quantitative indicators for spatial heterogeneity, including macroscopic indicators such as the global Moran index, spatial fractal dimension, and configuration entropy, as well as local in-depth indicators such as defect depth, boundary gradient, and internal heterogeneity index. This provides a systematic evaluation standard for "spatial homogeneity" and "structural stability" of compaction quality that transcends traditional statistics. It also enables early warning of compaction quality based on spatial pattern evolution. Based on a constructed multi-dimensional index threshold library and a two-layer early warning level, this method can provide early risk warnings based on the dynamic changes in the spatial distribution pattern of compaction states. Simultaneously, through the quantitative results of the spatial morphology, boundary characteristics, and compaction quality index of defect areas, it can assist in the precise location and causal analysis of problems, supporting scientific decision-making.

Claims

1. A method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics, characterized in that, Includes the following steps: S1, real-time acquisition of vertical acceleration signal information of vibrating wheel and three-dimensional spatial coordinate information of the center point of compaction wheel, and time alignment of the acquired vertical acceleration signal information of vibrating wheel and three-dimensional spatial coordinate information of the center point of compaction wheel; S2, based on the vertical acceleration signal information of the vibrating wheel, generates compaction index, and based on the compaction index, spatially grids the compaction surface and constructs a digital compaction field; S3 quantifies the spatial heterogeneity of the constructed digital compaction field in multiple dimensions and at multiple scales, and obtains multi-dimensional spatial heterogeneity indicators of the compaction operation surface.

2. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 1, characterized in that, The vertical acceleration signal of the vibrating wheel is filtered to remove low-frequency vehicle body sway interference and high-frequency noise, resulting in a filtered signal. The filtered signal is then analyzed using consecutive short time windows. Within these short time windows, a Fourier transform is performed on the acceleration signal to extract its fundamental frequency. First harmonic amplitude and vibration frequency and amplitude Calculate the compaction count value for each time window according to the formula. : （1） in, This is a calibration constant; thus, a time stamp corresponding to the timestamp is generated. CMV Time series; Using timestamps CMV The time series is rigorously matched with the 3D spatial coordinates and attitude angle sequence of the compaction wheel center point output by GNSS / IMU; the positioning point coordinates are corrected to the center point of the contact surface between the vibrating wheel and the ground through coordinate transformation, generating a sequence with 3D coordinates (…). X i , Y i , Z i )and CMV i The set of data points of value .

3. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 1, characterized in that, The compaction work surface is divided into rectangular grids. Robust processing is applied to the compaction index data points within each grid cell to generate a weighted composite for each grid cell. CMV value: （2） in, The median of the data within the grid cell. The interquartile range after removing outliers. The average value of the interpolation points of the triangular mesh. The weighting coefficients are dynamically adjusted based on the distribution quality of grid data; the weighting coefficients are determined using a direct assignment method, based on the effective data contained in each grid cell. CMV The proportions of the data points calculated as the median, interquartile range, and interpolation average are used as weighting coefficients.

4. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 1, characterized in that, The constructed digital compaction field is subjected to multi-dimensional and multi-scale spatial heterogeneity quantification, specifically including: quantification of macro-spatial characteristics of the compaction operation surface and fine identification and quantification of engineering risk areas.

5. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 4, characterized in that, The quantification of the macroscopic spatial characteristics of the compaction surface specifically evaluates the overall spatial heterogeneity of the compaction surface from three dimensions: spatial correlation, geometric complexity, and information order.

6. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 5, characterized in that, The global Moran index is used to determine whether there is significant spatial clustering in the quality distribution, and to quantitatively assess the overall strength and pattern of spatial dependence of compaction quality. After confirming the existence of significant spatial structure in compaction quality through the global Moran index, the essential characteristics of spatial heterogeneity are quantified more profoundly at the overall level. The spatial configuration entropy is used to characterize the degree of order or disorder in the arrangement and combination of different compaction grade regions within a local spatial range, reflecting the stability of spatial configuration from the perspective of information theory.

7. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 5, characterized in that, The detailed identification and quantification of engineering risk areas specifically includes: using the density-based DBSCAN algorithm to perform cluster analysis on defective mesh cells; after completing the cluster identification and basic feature extraction of defective areas, precise quantitative descriptions are performed from the perspectives of boundary transition characteristics, spatial neighborhood environment, internal structural state, and multi-zone correlation.

8. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 1, characterized in that, Early warning is based on obtaining multi-dimensional spatial heterogeneity indicators of the compaction surface.

9. The method for evaluating the continuous compaction quality of roadbed based on spatial heterogeneity characteristics according to claim 8, characterized in that, The early warning system based on the acquisition of multi-dimensional spatial heterogeneity indicators of the compaction surface specifically includes: establishing a threshold database and an early warning level database containing overall compaction surface indicators and local defect clustering indicators, and adopting a two-layer early warning mechanism of overall compaction surface early warning and local defect clustering early warning.

10. A roadbed continuous compaction quality assessment system based on spatial heterogeneity characteristics, characterized in that, It includes a preprocessing module, a spatial mesh module, and an evaluation module: The preprocessing module collects the vertical acceleration signal information of the vibrating wheel and the three-dimensional spatial coordinate information of the center point of the compaction wheel in real time, and performs time alignment on the collected vertical acceleration signal information of the vibrating wheel and the three-dimensional spatial coordinate information of the center point of the compaction wheel. The spatial grid module generates compaction indices based on the vertical acceleration signal information of the vibrating wheel, and then uses these indices to spatially grid the compaction surface and construct a digital compaction field. The evaluation module quantifies the spatial heterogeneity of the constructed digital compaction field in multiple dimensions and at multiple scales, and obtains multi-dimensional spatial heterogeneity indicators of the compaction operation surface.

Images