A road construction quality detection method based on intelligent sensors

Abstract

The application discloses a road construction quality detection method based on an intelligent sensor, which comprises the following steps: step one, collecting drum vibration acceleration signals and performing sliding time window segmentation to form a vibration signal segment sequence; step two, performing direct current removal on the vibration signal segment sequence and calculating a frequency spectrum amplitude distribution to screen effective vibration signal segments; step three, extracting a main peak coupling frequency, a half-power bandwidth, a harmonic energy ratio and a self-correlation delay parameter to construct a characteristic vector; step four, normalizing the characteristic vector and calculating a vector angle to obtain a characteristic angle distance mapping value; step five, constructing a characteristic spectrum representation matrix by using an improved Gegenbauer polynomial and a fractional order modulation; step six, performing a fractional order spectrum operator mapping calculation to obtain a construction quality calculation value; and step seven, performing section determination according to a quality threshold value. The application realizes stable detection and evaluation of road construction quality.

Description

Technical Field

[0001] This invention relates to the field of intelligent signal processing and data analysis technology, and in particular to a method for detecting road construction quality based on intelligent sensors. Background Technology

[0002] With the continuous expansion of road infrastructure construction, vibratory compaction technology using road rollers is widely applied in asphalt pavement and subgrade construction. The quality of compaction directly affects the load-bearing capacity, durability, and safety of the road structure in later use. To improve the level of construction quality control, engineering practice has gradually adopted the use of vibration sensors installed on road roller equipment to collect the vibration state of the drum, and to conduct real-time detection and evaluation of road compaction quality based on vibration signal analysis. In existing technologies, the degree of compaction in the construction area is typically estimated by collecting the drum vibration acceleration signal and performing simple time-domain or frequency-domain analysis, such as root mean square value analysis, peak frequency analysis, or empirical threshold judgment, thereby achieving indirect detection and control of road construction quality.

[0003] In actual road construction scenarios, due to the significant variability in factors such as the type of construction materials, the condition of the base structure, and the vibration parameters of construction equipment, a single vibration characteristic parameter is often insufficient to comprehensively characterize the complex dynamic changes during road compaction, resulting in poor stability of the detection results. Furthermore, most existing technologies rely solely on a limited number of statistical features to judge vibration signals, lacking the ability to deeply model the correlations between multidimensional features of vibration signals. This leads to a decrease in detection accuracy when facing nonlinear changes in vibration signals and rapid changes in construction conditions. In addition, existing methods typically employ simple feature vectors or traditional spectral representations for vibration signal feature representation, failing to effectively describe the structural relationships of vibration features in higher-order spectral spaces, thus limiting their ability to express the differences in complex construction conditions.

[0004] Therefore, how to provide a road construction quality inspection method based on intelligent sensors is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a road construction quality inspection method based on intelligent sensors. This invention constructs a road construction quality inspection method based on drum vibration acceleration signals, combining vibration signal fragmentation processing, multi-dimensional vibration feature extraction, and a feature angle distance mapping mechanism. It introduces an improved Gegenbauer polynomial recursive sequence and a fractional-order modulation mechanism to construct a feature spectrum representation matrix, and calculates construction quality through fractional-order spectral operator mapping. Simultaneously, it combines section energy consistency calculation and a multi-threshold section determination mechanism to achieve stable assessment and high-precision detection of road construction compaction quality.

[0006] A road construction quality inspection method based on intelligent sensors according to an embodiment of the present invention includes the following steps: Step 1: Collect the vibration acceleration signal of the drum and slide it into segments according to the preset time window length and time step to form a sequence of vibration signal segments; Step 2: Perform DC removal processing on the vibration signal segment sequence and calculate the spectral amplitude distribution. Calculate the excitation effectiveness index based on the root mean square value, spectral peak amplitude, and vibration energy density, and retain vibration signal segments with an excitation effectiveness index greater than a preset excitation threshold as valid vibration signal segments. Step 3: Calculate the main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameter for the effective vibration signal segment, and construct a feature vector; Step 4: Perform normalization processing on the feature vectors and calculate the vector angle between adjacent normalized feature vectors to obtain the feature angle distance mapping value; Step 5: Calculate the improved Gegenbauer polynomial recursive sequence based on the characteristic angular distance mapping value, and introduce a fractional-order modulation mechanism to extend the order of the improved Gegenbauer polynomial recursive sequence to form a characteristic spectrum representation matrix; Step 6: Perform fractional-order spectral operator mapping calculations based on the feature spectral representation matrix to obtain the calculated construction quality value; Step 7: Determine the sections of the calculated construction quality based on the preset quality threshold, and generate road construction quality inspection results.

[0007] Optionally, step one is as follows: A triaxial accelerometer is fixedly installed at the position of the vibrating drum of the road roller. A sensor coordinate system is established and the radial axis direction consistent with the radial direction of the drum is determined. The triaxial acceleration components are collected and the acceleration component corresponding to the radial axis is selected as the drum vibration acceleration signal. Write the sampling frequency parameter into the drum vibration acceleration signal and record the timestamp. Organize the drum vibration acceleration signal into a discrete sampling point sequence according to the sampling frequency. Set the time window length and time step, determine the start and end sampling point indices of each time window by successively moving the time step forward from the starting sampling point, and truncate the discrete sampling point sequence within each time window range to form multiple vibration signal segments; An overlapping interval is set for two adjacent vibration signal segments. The length of the overlapping interval is determined by the difference between the time window length and the time step, forming a sequence of vibration signal segments arranged continuously along the construction time.

[0008] Optionally, step two is as follows: DC removal processing is performed on the discrete sampling point sequence in the vibration signal segment sequence. Specifically, the average value of the amplitude of all sampling points in the vibration signal segment is calculated, and the average value is subtracted from the amplitude of each sampling point to form a zero-mean vibration signal sequence. Spectral analysis is performed on the zero-mean vibration signal sequence to convert the discrete sampling point sequence in the zero-mean vibration signal sequence into a frequency domain amplitude sequence, and then arrange them in frequency order to form a spectral amplitude distribution; The root mean square value is calculated based on the zero-mean vibration signal sequence. The calculation method is to square the amplitude of each sampling point in the vibration signal segment, sum all the square values ​​and divide them by the total number of sampling points, and then take the square root of the result. The amplitude of the spectral peak is determined based on the distribution of the spectral amplitude. The magnitude of the spectral amplitude is compared point by point within the preset vibration frequency range, and the amplitude corresponding to the spectral point with the largest amplitude is selected as the amplitude of the spectral peak. A vibration energy density mechanism is introduced. The amplitude of each frequency point in the spectral amplitude distribution is squared to obtain the spectral energy value. The spectral energy values ​​are accumulated according to the frequency interval. The accumulated spectral energy value is divided by the width of the corresponding frequency interval to obtain the vibration energy density. The excitation effectiveness index is calculated based on the root mean square value, spectral peak amplitude, and vibration energy density. The excitation effectiveness index is then compared with a preset excitation threshold, and vibration signal segments with an excitation effectiveness index greater than the preset excitation threshold are retained as valid vibration signal segments.

[0009] Optional, step three is as follows: Spectral analysis is performed on the effective vibration signal segments to obtain the spectral amplitude distribution arranged in frequency order. The spectral amplitude is compared point by point within the preset vibration frequency range, and the frequency corresponding to the spectral point with the largest amplitude is selected as the main peak coupling frequency. Using the amplitude of the spectral peak corresponding to the main peak coupling frequency as the reference amplitude, the spectrum amplitude distribution is searched in the low-frequency and high-frequency directions respectively. When the spectrum amplitude decreases to a preset proportion of the reference amplitude, two corresponding frequency points are determined, and the frequency difference between the two frequency points is taken as the half-power bandwidth. The harmonic frequency position is determined based on the main peak coupling frequency. The corresponding spectral amplitude is extracted at integer multiples of the frequency position corresponding to the main peak coupling frequency. The harmonic energy value is obtained by squaring the spectral amplitude corresponding to each multiple of the frequency position. The sum of the harmonic energy values ​​is then compared with the square of the spectral amplitude corresponding to the main peak coupling frequency to obtain the harmonic energy ratio. Autocorrelation operation is performed on the effective vibration signal segment. The vibration signal segment is multiplied point by point with the signal sequence at different time delay positions and accumulated to obtain the autocorrelation sequence. The time delay corresponding to the correlation peak with the largest amplitude in the autocorrelation sequence, except for the zero delay position, is selected as the autocorrelation delay parameter. The main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameters are combined in a fixed order to form a feature vector.

[0010] Optional, step four is as follows: Normalization is performed on each feature parameter in the feature vector. The maximum and minimum values ​​of each feature parameter within a preset feature range are determined. Then, each feature parameter is subtracted from its corresponding minimum value and divided by the difference between the maximum and minimum values ​​to obtain the normalized feature components. The normalized feature components are recombined in the same order as the feature vector to form the normalized feature vector; Perform a vector dot product operation on adjacent normalized feature vectors, multiply the feature components at corresponding positions in adjacent normalized feature vectors one by one and sum them to obtain the vector dot product value; The vector magnitude of the normalized eigenvector is obtained by squaring each eigencomponent in the normalized eigenvector and summing them, and then taking the square root of the summation result. The cosine similarity value is obtained by dividing the inner product of the vectors by the product of the magnitudes of the adjacent normalized feature vectors. The inverse cosine operation is then performed on the cosine similarity value to obtain the vector angle, which is then used as the feature angle distance mapping value.

[0011] Optional, step five is as follows: Calculate the improved Gegenbauer polynomial recursive sequence based on the characteristic angle distance mapping value; The improved Gegenbauer polynomial recursive sequence includes: Determine the initial term of the Gegenbauer polynomial recursion, set the constant term as the first term of the recursion sequence, and multiply the feature angle distance mapping value with the preset Gegenbauer polynomial parameters to obtain the second term of the recursion sequence; The subsequent polynomials are calculated step by step according to the recursive relationship. Each polynomial is obtained by the operation of the previous polynomial and the first two polynomials. The calculation of the previous polynomial includes multiplying the feature angle distance mapping value with the previous polynomial and performing a linear combination operation with the first two polynomials according to the corresponding order recursive coefficients, thereby generating the improved Gegenbauer polynomial recursive sequence step by step. A fractional-order modulation mechanism is introduced to perform order extension on the improved Gegenbauer polynomial recursive sequence. This fractional-order modulation mechanism includes: Calculate the change range between multiple feature angular distance mapping values, perform absolute value operation on the difference between two adjacent feature angular distance mapping values ​​and calculate the average to obtain the angular distance change rate; The fractional order modulation coefficients are calculated based on the rate of change of angular distance. The fractional order parameters are obtained by superimposing the fundamental polynomial order with the fractional order modulation coefficients. The order extension operation is performed on the improved Gegenbauer polynomial recursive sequence according to the fractional order parameter. The integer polynomial terms are multiplied with the fractional order parameter and accumulated to generate a fractional polynomial sequence. Arrange the fractional polynomial sequence according to the order of polynomial order, and form a characteristic spectrum representation matrix for each order of polynomial terms.

[0012] Optional, step six is ​​as follows: Read the feature spectral representation matrix, divide the feature spectral representation matrix into multiple spectral feature vectors by row, and establish a sequence of spectral feature vectors according to the order of the spectral feature vectors in the feature spectral representation matrix; Determine the parameters of the fractional-order spectral operator, and set the spectral operator scaling parameter and the spectral operator coefficient vector. The spectral operator coefficient vector and the spectral operator scaling parameter are preset parameters. For each spectral feature vector, a fractional spectral operator mapping calculation is performed. The fractional spectral operator mapping calculation includes performing a fractional difference operation on the spectral feature vector. The fractional difference operation determines the difference order by the fractional order parameter, and performs a weighted summation of each order spectral component in the spectral feature vector according to the discrete difference coefficients to obtain the fractional difference result. The fractional difference result is multiplied term by term with the coefficient vector of the spectral operator and then summed to obtain the projection value of the spectral operator; Numerical mapping is performed on the projection values ​​of the spectral operator, and the calculated construction quality values ​​are obtained by multiplying them with preset linear mapping coefficients.

[0013] Optionally, step seven is as follows: Obtain the calculated construction quality values ​​and establish a construction quality value sequence according to the arrangement order of vibration signal segments in the construction time series; Set a set of quality thresholds, including qualified thresholds, unqualified thresholds, and segment stability thresholds, where qualified thresholds are greater than unqualified thresholds; Each calculated construction quality value in the construction quality value sequence is compared with the qualified threshold and the unqualified threshold respectively. When the calculated construction quality value is greater than or equal to the qualified threshold, a qualified judgment mark is generated. When the calculated construction quality value is less than the unqualified threshold, an unqualified judgment mark is generated. Perform segment energy consistency calculation on the construction quality value sequence, perform squaring operation on multiple consecutive construction quality calculation values ​​and sum them to obtain the segment energy value, and divide the segment energy value by the corresponding segment length to obtain the segment energy density; The section energy density is compared with the section stability threshold. When the section energy density is greater than the section stability threshold, it is determined as a stable construction section. The stable construction section is then divided into sections based on the qualified and unqualified judgment marks, and the road construction quality inspection results are obtained.

[0014] The beneficial effects of this invention are: This invention constructs a road construction quality inspection method based on drum vibration acceleration signals. Combining vibration signal fragmentation processing, vibration feature parameter extraction, and high-order spectral feature expression mechanisms, it addresses the problems of insufficient vibration signal feature expression capabilities, low accuracy in identifying complex construction states, and insufficient stability of detection results in existing road construction quality inspection processes. It proposes a vibration feature analysis strategy based on an improved Gegenbauer polynomial recursive sequence and fractional-order spectral operator mapping calculation. In the vibration signal processing stage, by performing sliding time window segmentation, DC removal processing, and spectral analysis on the drum vibration acceleration signal, a joint excitation effectiveness index is constructed using the root mean square value, spectral peak amplitude, and vibration energy density to achieve stable screening of effective vibration signal segments. In the feature construction stage, a multidimensional vibration feature vector is formed by calculating the main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameters. Furthermore, by utilizing eigenvector normalization and vector angle calculation to construct a eigenangle distance mapping relationship, the ability to distinguish vibration characteristics between different construction states is improved. In the feature expression stage, an improved Gegenbauer polynomial recursive sequence is introduced, and the polynomial order is continuously expanded through a fractional-order modulation mechanism to construct a feature spectrum representation matrix, making the expression of vibration characteristics in higher-order spectral space more stable and more adaptable. In the quality calculation stage, fractional-order spectral operator mapping calculation is used to perform spectral operator projection operation on the feature spectrum representation matrix to achieve a refined quantitative characterization of construction vibration state. In the quality judgment stage, the construction quality calculation value is analyzed continuously through segment energy consistency calculation, and the segment stability is judged by combining qualified threshold, unqualified threshold and segment stability threshold, thereby obtaining the quality detection results of continuous construction segments and realizing stable assessment and high-precision detection of road construction compaction quality. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the overall process of a road construction quality detection method based on intelligent sensors proposed in this invention; Figure 2 This is a flowchart of the processing of the improved Gegenbauer polynomial recursive sequence and fractional order modulation mechanism in this invention. Figure 3 This is a flowchart of the construction quality section determination process in this invention. Detailed Implementation

[0016] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0017] refer to Figures 1-3 A road construction quality inspection method based on intelligent sensors includes the following steps: Step 1: Collect the vibration acceleration signal of the drum and slide it into segments according to the preset time window length and time step to form a sequence of vibration signal segments; Step 2: Perform DC removal processing on the vibration signal segment sequence and calculate the spectral amplitude distribution. Calculate the excitation effectiveness index based on the root mean square value, spectral peak amplitude, and vibration energy density, and retain vibration signal segments with an excitation effectiveness index greater than the preset excitation threshold as valid vibration signal segments. Step 3: Calculate the main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameters for the effective vibration signal segment, and construct the feature vector; Step 4: Perform normalization on the feature vectors and calculate the vector angle between adjacent normalized feature vectors to obtain the feature angle distance mapping value; Step 5: Calculate the improved Gegenbauer polynomial recursive sequence based on the characteristic angle distance mapping value, and introduce a fractional order modulation mechanism to extend the order of the improved Gegenbauer polynomial recursive sequence to form a characteristic spectrum representation matrix; Step 6: Perform fractional-order spectral operator mapping calculations based on the characteristic spectral representation matrix to obtain the calculated construction quality value; Step 7: Determine the sections of the calculated construction quality based on the preset quality thresholds, and generate the road construction quality inspection results.

[0018] In this embodiment, step one specifically includes: A triaxial accelerometer is fixedly installed at the position of the vibrating drum of the road roller. A sensor coordinate system is established and the radial axis direction consistent with the radial direction of the drum is determined. The triaxial acceleration components are collected and the acceleration component corresponding to the radial axis is selected as the drum vibration acceleration signal. Write the sampling frequency parameter into the drum vibration acceleration signal and record the timestamp. Organize the drum vibration acceleration signal into a discrete sampling point sequence according to the sampling frequency. Set the time window length and time step, determine the start and end sampling point indices of each time window by successively moving the time step forward from the starting sampling point, and truncate the discrete sampling point sequence within each time window range to form multiple vibration signal segments; An overlapping interval is set for two adjacent vibration signal segments. The length of the overlapping interval is determined by the difference between the time window length and the time step, forming a sequence of vibration signal segments arranged continuously along the construction time.

[0019] In this implementation, a triaxial accelerometer is installed on the side wall support of the vibrating drum of the road roller. The sampling frequency is set to 1000Hz to ensure sufficient sampling density of the drum vibration signal within the 5Hz to 80Hz operating frequency band. The time window length is set to 1s, corresponding to 1000 consecutive sampling points; the time step is set to 0.25s, corresponding to 250 sampling points, and a sliding time window is formed by successively moving forward according to the time step from the initial sampling point. Each vibration signal segment contains 1000 discrete sampling points, and an overlap interval of 750 sampling points is formed between adjacent vibration signal segments to maintain the continuity of the construction vibration signal in the time dimension. The sensor coordinate system is established with the drum axis direction as a reference, where the radial axis direction is consistent with the normal direction of the drum's outer circle, and the acceleration component corresponding to the radial axis is selected as the drum vibration acceleration signal input.

[0020] In this embodiment, step two specifically involves: DC removal processing is performed on the discrete sampling point sequence in the vibration signal segment sequence. Specifically, the average value of the amplitude of all sampling points in the vibration signal segment is calculated, and the average value is subtracted from the amplitude of each sampling point to form a zero-mean vibration signal sequence. Spectral analysis is performed on the zero-mean vibration signal sequence to convert the discrete sampling point sequence in the zero-mean vibration signal sequence into a frequency domain amplitude sequence, and then arrange them in frequency order to form a spectral amplitude distribution; The root mean square value is calculated based on the zero-mean vibration signal sequence. The calculation method is to square the amplitude of each sampling point in the vibration signal segment, sum all the square values ​​and divide them by the total number of sampling points, and then take the square root of the result. The amplitude of the spectral peak is determined based on the distribution of the spectral amplitude. The magnitude of the spectral amplitude is compared point by point within the preset vibration frequency range, and the amplitude corresponding to the spectral point with the largest amplitude is selected as the amplitude of the spectral peak. A vibration energy density mechanism is introduced. The amplitude of each frequency point in the spectral amplitude distribution is squared to obtain the spectral energy value. The spectral energy values ​​are accumulated according to the frequency interval. The accumulated spectral energy value is divided by the width of the corresponding frequency interval to obtain the vibration energy density. The excitation effectiveness index is calculated based on the root mean square value, spectral peak amplitude, and vibration energy density. The excitation effectiveness index is then compared with a preset excitation threshold, and vibration signal segments with an excitation effectiveness index greater than the preset excitation threshold are retained as valid vibration signal segments.

[0021] In this implementation, the incentive effectiveness index is represented by a normalized value of 0 to 1, and the preset incentive threshold is set to 0.60. When the excitation effectiveness index is greater than 0.60, the corresponding vibration signal segment is retained as a valid vibration signal segment; when the excitation effectiveness index is less than or equal to 0.60, the corresponding vibration signal segment is discarded. The vibration energy density is obtained by squaring the amplitude of each frequency point in the spectrum amplitude distribution to obtain the square value of the spectrum amplitude, and then accumulating it in segments according to the preset frequency interval to form the frequency band energy value. Then, the energy value of each frequency band is divided by the width of the corresponding frequency interval to obtain the energy distribution value within the unit frequency range, thus forming the vibration energy density sequence. The frequency interval is divided according to the working frequency band of the drum vibration and continuously segmented with a fixed frequency step size, so that the change of vibration energy in the frequency dimension under different construction conditions can be quantitatively expressed. The excitation effectiveness index is obtained by weighted combination of the root mean square value, the peak amplitude, and the vibration energy density. The root mean square value reflects the overall amplitude level of vibration, the peak amplitude characterizes the main vibration frequency response intensity, and the vibration energy density describes the distribution characteristics of vibration energy in the frequency space. A stable excitation intensity characterization is formed by joint calculation of multi-dimensional vibration parameters.

[0022] In this embodiment, step three specifically includes: Spectral analysis is performed on the effective vibration signal segments to obtain the spectral amplitude distribution arranged in frequency order. The spectral amplitude is compared point by point within the preset vibration frequency range, and the frequency corresponding to the spectral point with the largest amplitude is selected as the main peak coupling frequency. Using the amplitude of the spectral peak corresponding to the main peak coupling frequency as the reference amplitude, the spectrum amplitude distribution is searched in the low-frequency and high-frequency directions respectively. When the spectrum amplitude decreases to a preset proportion of the reference amplitude, two corresponding frequency points are determined, and the frequency difference between the two frequency points is taken as the half-power bandwidth. The harmonic frequency position is determined based on the main peak coupling frequency. The corresponding spectral amplitude is extracted at integer multiples of the frequency position corresponding to the main peak coupling frequency. The harmonic energy value is obtained by squaring the spectral amplitude corresponding to each multiple of the frequency position. The sum of the harmonic energy values ​​is then compared with the square of the spectral amplitude corresponding to the main peak coupling frequency to obtain the harmonic energy ratio. Autocorrelation operation is performed on the effective vibration signal segment. The vibration signal segment is multiplied point by point with the signal sequence at different time delay positions and accumulated to obtain the autocorrelation sequence. The time delay corresponding to the correlation peak with the largest amplitude in the autocorrelation sequence, except for the zero delay position, is selected as the autocorrelation delay parameter. The main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameters are combined in a fixed order to form a feature vector.

[0023] In this implementation, the spectrum analysis employs Fast Fourier Transform (FFT) processing, with a sampling frequency set to 1000 Hz and a frequency resolution of 1 Hz. The preset vibration frequency range is set to 5 Hz to 80 Hz, covering the main vibration response range of the vibrating drum of the road roller during compaction. When calculating the half-power bandwidth, a half-power ratio of 0.707 is used as the threshold, meaning the corresponding frequency point is determined when the spectral amplitude decreases to 0.707 times the peak amplitude. The harmonic frequency extraction range is set to 2 to 5 times the main peak coupling frequency, and the square of the spectral amplitude at each harmonic position is calculated as the harmonic energy. For autocorrelation delay calculation, a maximum delay length of 0.5 s is set, and the autocorrelation value is calculated point-by-point at sampling intervals. The delay corresponding to the correlation peak with the largest amplitude, excluding the zero-delay position, is selected as the autocorrelation delay parameter to characterize the periodic structure of the vibration signal.

[0024] In this embodiment, step four specifically includes: Normalization is performed on each feature parameter in the feature vector. The maximum and minimum values ​​of each feature parameter within a preset feature range are determined. Then, each feature parameter is subtracted from its corresponding minimum value and divided by the difference between the maximum and minimum values ​​to obtain the normalized feature components. The normalized feature components are recombined in the same order as the feature vector to form the normalized feature vector; Perform a vector dot product operation on adjacent normalized feature vectors, multiply the feature components at corresponding positions in adjacent normalized feature vectors one by one and sum them to obtain the vector dot product value; The vector magnitude of the normalized eigenvector is obtained by squaring each eigencomponent in the normalized eigenvector and summing them, and then taking the square root of the summation result. The cosine similarity value is obtained by dividing the inner product of the vectors by the product of the magnitudes of the adjacent normalized feature vectors. The inverse cosine operation is then performed on the cosine similarity value to obtain the vector angle, which is then used as the feature angle distance mapping value.

[0025] In this implementation, the maximum and minimum values ​​of each feature parameter are obtained by statistically analyzing all feature vectors during the construction sampling phase, forming a fixed feature range table. In subsequent calculations, the corresponding feature range is directly called for normalization to ensure consistent feature scale across different construction sections. The angular distance calculation between normalized feature vectors constructs an angular measurement relationship in the feature space, allowing the differences in vibration characteristics under different construction conditions to be expressed in geometric angles. When calculating cosine similarity, a minimum numerical constraint is set on the normalized feature vector magnitude. To avoid numerical instability caused by the denominator approaching 0 in the cosine similarity calculation, the minimum numerical constraint threshold for the normalized feature vector magnitude is set to 0.001. When the normalized feature vector magnitude is less than 0.001, it is replaced by 0.001 in the vector angle calculation. The vector angle calculation results are limited to the range of 0 to π and output in a fixed order to form a feature angular distance sequence.

[0026] In this embodiment, step five specifically includes: Calculate the improved Gegenbauer polynomial recursive sequence based on the characteristic angle distance mapping value; The improved Gegenbauer polynomial recurrence sequence includes: Determine the initial term of the Gegenbauer polynomial recursion, set the constant term as the first term of the recursion sequence, and multiply the feature angle distance mapping value with the preset Gegenbauer polynomial parameters to obtain the second term of the recursion sequence; The subsequent polynomials are calculated step by step according to the recursive relationship. Each polynomial is obtained by the operation of the previous polynomial and the first two polynomials. The calculation of the previous polynomial includes multiplying the feature angle distance mapping value with the previous polynomial and performing a linear combination operation with the first two polynomials according to the corresponding order recursive coefficients, thereby generating the improved Gegenbauer polynomial recursive sequence step by step. Fractional order modulation mechanisms are introduced to perform order extension on the improved Gegenbauer polynomial recursive sequence. These fractional order modulation mechanisms include: Calculate the change range between multiple feature angular distance mapping values, perform absolute value operation on the difference between two adjacent feature angular distance mapping values ​​and calculate the average to obtain the angular distance change rate; The fractional order modulation coefficients are calculated based on the rate of change of angular distance. The fractional order parameters are obtained by superimposing the fundamental polynomial order with the fractional order modulation coefficients. The order extension operation is performed on the improved Gegenbauer polynomial recursive sequence according to the fractional order parameter. The integer polynomial terms are multiplied with the fractional order parameter and accumulated to generate a fractional polynomial sequence. Arrange the fractional polynomial sequence according to the order of polynomial order, and form a characteristic spectrum representation matrix for each order of polynomial terms.

[0027] In this implementation, the basic order of the Gegenbauer polynomial recursive sequence is set to 2 to 6, and polynomial terms are generated sequentially through the recursive relationship. The fractional-order modulation mechanism uses the rate of change of the characteristic angular distance mapping value as the modulation basis. First, the difference between two adjacent values ​​is calculated according to the time sequence of the characteristic angular distance mapping value, and the absolute value of the difference is calculated and averaged to obtain the rate of change of angular distance. Then, the fractional-order modulation coefficient is determined according to the rate of change of angular distance, and this modulation coefficient is superimposed with the basic polynomial order to obtain the fractional-order order parameter, which is set between 2.0 and 6.5. The fractional-order order parameter is used to expand the integer-order Gegenbauer polynomial terms obtained by recursion, so that the polynomial spectrum representation remains continuously changing under different construction vibration states, thereby forming a stable characteristic spectrum representation matrix. The improved Gegenbauer polynomial recursive sequence maintains the same basic structure as the traditional Gegenbauer polynomial recursive sequence, both employing a recursive approach to generate polynomial terms sequentially. Specifically, an initial term is set, with the constant term serving as the first term of the polynomial sequence, and the product of the feature angle distance mapping value and preset polynomial parameters serving as the second term. Based on this, higher-order polynomial terms are calculated sequentially according to a predetermined recursive relationship. Each polynomial is calculated by combining the previous polynomial with the polynomials of the preceding and following orders. By multiplying the feature angle distance mapping value with the preceding polynomial and performing a linear combination operation on the preceding two polynomials using the corresponding recursive coefficients, a complete polynomial recursive sequence is gradually generated, thus constructing the basic polynomial structure for feature representation. Based on the above recursive structure, this invention improves the traditional Gegenbauer polynomial recursive sequence by introducing a fractional-order modulation mechanism to achieve continuous expansion of the polynomial order. Specifically, after calculating the characteristic angular distance mapping value, the difference between two adjacent mapping values ​​is calculated according to their order in the sequence. The absolute value of the difference is then calculated, and the average is taken to obtain the angular distance change rate. Subsequently, the fractional-order modulation coefficient is calculated based on the angular distance change rate, and this modulation coefficient is superimposed on the basic polynomial order to obtain the fractional-order order parameter. The fractional-order order parameter is then used to perform order expansion calculations on the original integer-order Gegenbauer polynomial terms, forming a continuously changing fractional-order polynomial sequence between integer orders, thereby constructing the improved Gegenbauer polynomial recursive sequence. Through the aforementioned improvements, the improved Gegenbauer polynomial recursive sequence, while maintaining the stability of the traditional recursive structure, expands the polynomial order from discrete integer to continuous fractional, enabling the characteristic spectral representation to more precisely characterize the changing trends of the characteristic angular distance mapping values. The fractional-order modulation mechanism allows the polynomial sequence to adaptively adjust according to the degree of angular distance change, maintaining a stable order structure when vibration characteristic changes are relatively gentle, and increasing the polynomial expressive power when vibration characteristic changes are more significant. This improves the accuracy of the characteristic spectral representation in characterizing changes in construction vibration state and enhances the stability and discriminative power of the characteristic spectral representation matrix in subsequent spectral operator mapping calculations.

[0028] In this embodiment, step six specifically includes: Read the feature spectral representation matrix, divide the feature spectral representation matrix into multiple spectral feature vectors by row, and establish a sequence of spectral feature vectors according to the order of the spectral feature vectors in the feature spectral representation matrix; Determine the parameters of the fractional-order spectral operator, and set the spectral operator scaling parameter and the spectral operator coefficient vector. The spectral operator coefficient vector and the spectral operator scaling parameter are preset parameters. For each spectral feature vector, a fractional spectral operator mapping calculation is performed. The fractional spectral operator mapping calculation includes performing a fractional difference operation on the spectral feature vector. The fractional difference operation determines the difference order by the fractional order parameter, and performs a weighted summation of each order spectral component in the spectral feature vector according to the discrete difference coefficients to obtain the fractional difference result. The fractional difference result is multiplied term by term with the coefficient vector of the spectral operator and then summed to obtain the projection value of the spectral operator; Numerical mapping is performed on the projection values ​​of the spectral operator, and the calculated construction quality values ​​are obtained by multiplying them with preset linear mapping coefficients.

[0029] In this implementation, the feature spectrum representation matrix is ​​divided into multiple spectral feature vectors according to row index order, with each spectral feature vector containing 6 to 12 spectral components. The order parameter of the fractional-order spectral operator is set to a continuous value between 2.0 and 6.5, and modulated in steps of 0.1. The scale parameter of the spectral operator is set to 0.5 to 2.5 to adjust the scale weight of each order spectral component during the spectral operator mapping process. Discrete difference coefficients are generated according to the fractional-order difference coefficient sequence, and a weighted summation operation is performed between adjacent spectral components to obtain the fractional-order difference result. The length of the spectral operator coefficient vector is set to 6 to 12, corresponding to the weight parameters of each order spectral component. The preset linear mapping coefficients are set to real numbers between 0.8 and 1.2 to perform linear mapping calculations on the spectral operator projection values ​​to obtain the construction quality calculation value.

[0030] In this embodiment, step seven specifically includes: Obtain the calculated construction quality values ​​and establish a construction quality value sequence according to the arrangement order of vibration signal segments in the construction time series; Set a set of quality thresholds, including qualified thresholds, unqualified thresholds, and segment stability thresholds, where qualified thresholds are greater than unqualified thresholds; Each calculated construction quality value in the construction quality value sequence is compared with the qualified threshold and the unqualified threshold respectively. When the calculated construction quality value is greater than or equal to the qualified threshold, a qualified judgment mark is generated. When the calculated construction quality value is less than the unqualified threshold, an unqualified judgment mark is generated. Perform segment energy consistency calculation on the construction quality value sequence, perform squaring operation on multiple consecutive construction quality calculation values ​​and sum them to obtain the segment energy value, and divide the segment energy value by the corresponding segment length to obtain the segment energy density; The section energy density is compared with the section stability threshold. When the section energy density is greater than the section stability threshold, it is determined as a stable construction section. The stable construction section is then divided into sections based on the qualified and unqualified judgment marks, and the road construction quality inspection results are obtained.

[0031] In this implementation, the calculated construction quality values ​​are represented by normalized values ​​from 0 to 1. The pass / fail threshold is set to 0.80, the fail / unqualified threshold is set to 0.65, and the section stability threshold is set to 0.75, with the pass / fail threshold being greater than the fail / unqualified threshold. A pass / fail judgment mark is generated when the calculated construction quality value is greater than or equal to 0.80, and a fail / unqualified judgment mark is generated when the calculated construction quality value is less than 0.65. A section energy density greater than 0.75 is considered a stable construction section. The section energy consistency calculation uses the construction quality value sequence as input and performs continuous segmentation processing according to a fixed section length. Each section contains 5 to 10 consecutive calculated construction quality values. The calculated construction quality values ​​within each section are squared sequentially, and the squared results are accumulated to obtain the section energy value. Then, the section energy value is divided by the section length to obtain the section energy density. The section stability threshold is determined based on the statistical results of the calculated construction quality values ​​under the stable vibration state of the construction equipment and is set to a value between 0.65 and 0.85. When the energy density of a section is greater than the section stability threshold, the section is determined to be a stable construction section. Within the stable construction section, the section continuity is divided by combining the qualified and unqualified judgment marks to form a continuous construction quality section.

[0032] Example 1: To verify the feasibility and detection accuracy of this invention in a real engineering environment, it was applied to a highway subgrade construction compaction quality detection scenario. This construction project is located on a section of a provincial highway expansion project, with a construction area length of approximately 1200m. The pavement structure consists of a graded crushed stone base course and an asphalt surface course. A certain model of double-drum vibratory roller was used for vibration compaction during construction. The foundation materials at the construction site have a complex composition, with variations in moisture content and material gradation in local areas. This can easily lead to uneven compaction during the compaction process. Therefore, continuous detection and evaluation of the compaction quality are necessary to ensure stable construction quality.

[0033] A triaxial accelerometer is installed on the construction equipment, fixedly mounted on the side wall of the vibrating drum of the road roller. A sensor coordinate system is established, with one axis aligned with the radial direction of the drum. During construction, the road roller continuously collects the drum's vibration acceleration signals and transmits the sampled signals in real time to a monitoring terminal for analysis. The sampling frequency is set to 1000Hz, and the vibration signal is segmented into sliding windows every 1 second, with a 0.25-second time step between adjacent windows, thus forming multiple vibration signal segment sequences. This method allows for the acquisition of continuous vibration signal data during road construction, enabling real-time analysis of the compaction process.

[0034] After obtaining the vibration signal segment sequence, DC removal processing is performed on the vibration signals, followed by spectral analysis. By calculating the root mean square value, peak amplitude, and vibration energy density of the vibration signals, an excitation effectiveness index is constructed to screen out vibration signal segments with effective excitation characteristics. Signal segments with an excitation effectiveness index below a preset threshold are considered invalid vibration data and are discarded, thus avoiding interference from vibration signals during roller turning, stopping, or low-amplitude states. The resulting effective vibration signal segments are used for subsequent vibration characteristic analysis.

[0035] For each valid vibration signal segment, the peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameter are calculated, and these parameters are combined to form a vibration feature vector. The feature vector is then normalized, and the angle between the normalized feature vectors is calculated using the vector dot product and vector magnitude, thus obtaining the feature angle distance mapping value. This feature angle distance mapping value reflects the degree of difference in vibration characteristics under different construction conditions.

[0036] After obtaining the characteristic angular distance mapping value, an improved Gegenbauer polynomial recursive sequence is constructed, and a fractional-order modulation mechanism is introduced to extend the polynomial order, thereby forming a characteristic spectral representation matrix. This characteristic spectral representation matrix can describe the structural relationship of vibration characteristics in higher-order spectral space. Subsequently, fractional-order spectral operator mapping calculations are performed on the characteristic spectral representation matrix to obtain the construction quality calculation value. The construction quality calculation value reflects the compaction quality state of the road roller in the current construction section.

[0037] During the construction quality assessment phase, the calculated construction quality values ​​are segmented based on preset quality thresholds, and segment energy consistency calculations are performed on continuous construction segments. By calculating the segment energy density and comparing it with the segment stability threshold, it can be determined whether the construction segment is in a stable compaction state. When the segment energy density is higher than the stability threshold, it is determined to be a stable construction segment. The quality grade of the construction segment is then classified based on the pass / fail threshold, thus obtaining the road construction quality inspection results.

[0038] To verify the beneficial effects of this invention, multiple construction sections were selected at the construction site for comparative testing. A traditional root mean square vibration detection method was also used as a comparison method to assess the construction quality. Statistical analysis of the data from multiple construction sections yielded the experimental results, as shown in Table 1.

[0039] Table 1. Comparative Statistical Table of Road Compaction Quality Test Results in Different Construction Sections

[0040] As shown in Table 1, in the testing experiments of 12 construction sections, the method of this invention can stably output the calculated construction quality value in all construction sections and can reasonably judge the construction section based on the calculated construction quality value. Statistical calculation of the detection accuracy in Table 1 shows that the average detection accuracy of the method of this invention in the 12 construction sections reaches 96.0%, while the average detection accuracy of the traditional root mean square vibration detection method is 86.0%, representing an overall improvement in detection accuracy of approximately 10 percentage points. The method of this invention constructs a multidimensional vibration feature vector and introduces an improved Gegenbauer polynomial recursive sequence for feature spectrum expression. Simultaneously, it combines fractional-order spectral operator mapping calculations to perform high-order spectral space analysis of vibration features, enabling the calculated construction quality value to more accurately reflect changes in vibration state. This allows for the correct identification of unqualified construction areas in sections S04, S08, and S10.

[0041] In construction sections with a compaction degree higher than 96% (such as S01, S05, and S07), the quality values ​​detected by this invention remained above 0.90, and the quality judgment results were all qualified. Furthermore, the detection accuracy rate was higher than 97%, indicating that the method of this invention has high stability and repeatability in areas with good construction quality. However, in construction sections with a compaction degree between 92% and 95% (such as S03, S06, and S12), the construction vibration signal usually exhibits significant fluctuations. In this case, traditional methods are prone to unstable detection results. The method of this invention, through section energy consistency calculation and section analysis of continuous construction quality values, can effectively suppress the influence of local vibration fluctuations on the detection results, making the detection results more stable and reliable.

[0042] The experimental results presented in this invention demonstrate that the road construction quality detection method based on intelligent sensors significantly improves the accuracy and reliability of road construction quality detection by constructing a multidimensional feature representation mechanism for vibration signals and combining an improved Gegenbauer polynomial recursive sequence with fractional-order spectral operator mapping calculation. This allows for high-order spectral space modeling of road construction vibration characteristics. Furthermore, the method utilizes a segment energy consistency determination mechanism to enhance the stability of detection results.

[0043] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for road construction quality inspection based on intelligent sensors, characterized in that, Includes the following steps: Step 1: Collect the vibration acceleration signal of the drum and slide it into segments according to the preset time window length and time step to form a sequence of vibration signal segments; Step 2: Perform DC removal processing on the vibration signal segment sequence and calculate the spectral amplitude distribution. Calculate the excitation effectiveness index based on the root mean square value, spectral peak amplitude, and vibration energy density, and retain vibration signal segments with an excitation effectiveness index greater than a preset excitation threshold as valid vibration signal segments. Step 3: Calculate the main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameter for the effective vibration signal segment, and construct a feature vector; Step 4: Perform normalization processing on the feature vectors and calculate the vector angle between adjacent normalized feature vectors to obtain the feature angle distance mapping value; Step 5: Calculate the improved Gegenbauer polynomial recursive sequence based on the characteristic angular distance mapping value, and introduce a fractional-order modulation mechanism to extend the order of the improved Gegenbauer polynomial recursive sequence to form a characteristic spectrum representation matrix; Step 6: Perform fractional-order spectral operator mapping calculations based on the feature spectral representation matrix to obtain the calculated construction quality value; Step 7: Determine the sections of the calculated construction quality based on the preset quality threshold, and generate road construction quality inspection results.

2. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step one specifically involves: A triaxial accelerometer is fixedly installed at the position of the vibrating drum of the road roller. A sensor coordinate system is established and the radial axis direction consistent with the radial direction of the drum is determined. The triaxial acceleration components are collected and the acceleration component corresponding to the radial axis is selected as the drum vibration acceleration signal. Write the sampling frequency parameter into the drum vibration acceleration signal and record the timestamp. Organize the drum vibration acceleration signal into a discrete sampling point sequence according to the sampling frequency. Set the time window length and time step, determine the start and end sampling point indices of each time window by successively moving the time step forward from the starting sampling point, and truncate the discrete sampling point sequence within each time window range to form multiple vibration signal segments; An overlapping interval is set for two adjacent vibration signal segments. The length of the overlapping interval is determined by the difference between the time window length and the time step, forming a sequence of vibration signal segments arranged continuously along the construction time.

3. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step two specifically involves: DC removal processing is performed on the discrete sampling point sequence in the vibration signal segment sequence. Specifically, the average value of the amplitude of all sampling points in the vibration signal segment is calculated, and the average value is subtracted from the amplitude of each sampling point to form a zero-mean vibration signal sequence. Spectral analysis is performed on the zero-mean vibration signal sequence to convert the discrete sampling point sequence in the zero-mean vibration signal sequence into a frequency domain amplitude sequence, and then arrange them in frequency order to form a spectral amplitude distribution; The root mean square value is calculated based on the zero-mean vibration signal sequence. The calculation method is to square the amplitude of each sampling point in the vibration signal segment, sum all the square values ​​and divide them by the total number of sampling points, and then take the square root of the result. The amplitude of the spectral peak is determined based on the distribution of the spectral amplitude. The magnitude of the spectral amplitude is compared point by point within the preset vibration frequency range, and the amplitude corresponding to the spectral point with the largest amplitude is selected as the amplitude of the spectral peak. A vibration energy density mechanism is introduced. The amplitude of each frequency point in the spectral amplitude distribution is squared to obtain the spectral energy value. The spectral energy values ​​are accumulated according to the frequency interval. The accumulated spectral energy value is divided by the width of the corresponding frequency interval to obtain the vibration energy density. The excitation effectiveness index is calculated based on the root mean square value, spectral peak amplitude, and vibration energy density. The excitation effectiveness index is then compared with a preset excitation threshold, and vibration signal segments with an excitation effectiveness index greater than the preset excitation threshold are retained as valid vibration signal segments.

4. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step three specifically involves: Spectral analysis is performed on the effective vibration signal segments to obtain the spectral amplitude distribution arranged in frequency order. The spectral amplitude is compared point by point within the preset vibration frequency range, and the frequency corresponding to the spectral point with the largest amplitude is selected as the main peak coupling frequency. Using the amplitude of the spectral peak corresponding to the main peak coupling frequency as the reference amplitude, the spectrum amplitude distribution is searched in the low-frequency and high-frequency directions respectively. When the spectrum amplitude decreases to a preset proportion of the reference amplitude, two corresponding frequency points are determined, and the frequency difference between the two frequency points is taken as the half-power bandwidth. The harmonic frequency position is determined based on the main peak coupling frequency. The corresponding spectral amplitude is extracted at integer multiples of the frequency position corresponding to the main peak coupling frequency. The harmonic energy value is obtained by squaring the spectral amplitude corresponding to each multiple of the frequency position. The sum of the harmonic energy values ​​is then compared with the square of the spectral amplitude corresponding to the main peak coupling frequency to obtain the harmonic energy ratio. Autocorrelation operation is performed on the effective vibration signal segment. The vibration signal segment is multiplied point by point with the signal sequence at different time delay positions and accumulated to obtain the autocorrelation sequence. The time delay corresponding to the correlation peak with the largest amplitude in the autocorrelation sequence, except for the zero delay position, is selected as the autocorrelation delay parameter. The main peak coupling frequency, half-power bandwidth, harmonic energy ratio, and autocorrelation delay parameters are combined in a fixed order to form a feature vector.

5. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step four specifically involves: Normalization is performed on each feature parameter in the feature vector. The maximum and minimum values ​​of each feature parameter within the preset feature range are determined. The corresponding minimum value is subtracted from each feature parameter and then divided by the difference between the maximum and minimum values ​​to obtain the normalized feature components. The normalized feature components are recombined in the same order as the feature vector to form the normalized feature vector; Perform a vector dot product operation on adjacent normalized feature vectors, multiply the feature components at corresponding positions in adjacent normalized feature vectors one by one and sum them to obtain the vector dot product value; The vector magnitude of the normalized eigenvector is obtained by squaring each eigencomponent in the normalized eigenvector and summing them, and then taking the square root of the summation result. The cosine similarity value is obtained by dividing the inner product of the vectors by the product of the magnitudes of the adjacent normalized feature vectors. The inverse cosine operation is then performed on the cosine similarity value to obtain the vector angle, which is then used as the feature angle distance mapping value.

6. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step five specifically involves: Calculate the improved Gegenbauer polynomial recursive sequence based on the characteristic angle distance mapping value; The improved Gegenbauer polynomial recursive sequence includes: Determine the initial term of the Gegenbauer polynomial recursion, set the constant term as the first term of the recursion sequence, and multiply the feature angle distance mapping value with the preset Gegenbauer polynomial parameters to obtain the second term of the recursion sequence; The subsequent polynomials are calculated step by step according to the recursive relationship. Each polynomial is obtained by the operation of the previous polynomial and the first two polynomials. The calculation of the previous polynomial includes multiplying the feature angle distance mapping value with the previous polynomial and performing a linear combination operation with the first two polynomials according to the corresponding order recursive coefficients, thereby generating the improved Gegenbauer polynomial recursive sequence step by step. A fractional-order modulation mechanism is introduced to perform order extension on the improved Gegenbauer polynomial recursive sequence. This fractional-order modulation mechanism includes: Calculate the change range between multiple feature angular distance mapping values, perform absolute value operation on the difference between two adjacent feature angular distance mapping values ​​and calculate the average to obtain the angular distance change rate; The fractional order modulation coefficients are calculated based on the rate of change of angular distance. The fractional order parameters are obtained by superimposing the fundamental polynomial order with the fractional order modulation coefficients. The order extension operation is performed on the improved Gegenbauer polynomial recursive sequence according to the fractional order parameter. The integer polynomial terms are multiplied with the fractional order parameter and accumulated to generate a fractional polynomial sequence. Arrange the fractional polynomial sequence according to the order of polynomial order, and form a characteristic spectrum representation matrix for each order of polynomial terms.

7. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step six specifically involves: Read the feature spectral representation matrix, divide the feature spectral representation matrix into multiple spectral feature vectors by row, and establish a sequence of spectral feature vectors according to the order of the spectral feature vectors in the feature spectral representation matrix; Determine the parameters of the fractional-order spectral operator, and set the spectral operator scaling parameter and the spectral operator coefficient vector. The spectral operator coefficient vector and the spectral operator scaling parameter are preset parameters. For each spectral feature vector, a fractional spectral operator mapping calculation is performed. The fractional spectral operator mapping calculation includes performing a fractional difference operation on the spectral feature vector. The fractional difference operation determines the difference order by the fractional order parameter, and performs a weighted summation of each order spectral component in the spectral feature vector according to the discrete difference coefficients to obtain the fractional difference result. The fractional difference result is multiplied term by term with the coefficient vector of the spectral operator and then summed to obtain the projection value of the spectral operator; Numerical mapping is performed on the projection values ​​of the spectral operator, and the calculated construction quality values ​​are obtained by multiplying them with preset linear mapping coefficients.

8. The road construction quality inspection method based on intelligent sensors according to claim 1, characterized in that, Step seven specifically involves: Obtain the calculated construction quality values ​​and establish a construction quality value sequence according to the arrangement order of vibration signal segments in the construction time series; Set a set of quality thresholds, including qualified thresholds, unqualified thresholds, and segment stability thresholds, where qualified thresholds are greater than unqualified thresholds; Each calculated construction quality value in the construction quality value sequence is compared with the qualified threshold and the unqualified threshold respectively. When the calculated construction quality value is greater than or equal to the qualified threshold, a qualified judgment mark is generated. When the calculated construction quality value is less than the unqualified threshold, an unqualified judgment mark is generated. Perform segment energy consistency calculation on the construction quality value sequence, perform squaring operation on multiple consecutive construction quality calculation values ​​and sum them to obtain the segment energy value, and divide the segment energy value by the corresponding segment length to obtain the segment energy density; The section energy density is compared with the section stability threshold. When the section energy density is greater than the section stability threshold, it is determined as a stable construction section. The stable construction section is then divided into sections based on the qualified and unqualified judgment marks, and the road construction quality inspection results are obtained.

Images