An adaptive filtering method, system, and device for road load response signals.

By using spectrum-adaptive initial cutoff frequency calculation and multi-dimensional physical constraints on the reference signal, the problem of unstable filtering effect under low signal-to-noise ratio conditions was solved, and the consistency and reliability of the road load response signal monitoring effect under different working conditions were achieved.

CN123122500APending Publication Date: 2026-11-17JIANGXI GANYUE EXPRESSWAY +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611448547.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-17
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively distinguish between the true response spectrum and the noise spectrum under low signal-to-noise ratio conditions, resulting in unstable filtering effects, an inability to effectively suppress high-frequency noise, and an impact on the monitoring quality of road load response signals.

Method used

A dual cutoff frequency determination mechanism is adopted, which combines adaptive initial cutoff frequency calculation with minimizing multi-dimensional physical constraints on the reference signal. By acquiring the spectral characteristics of the road load response signal, the upper limit of physical constraints is set using the energy concentration law and noise suppression effect of the reference signal, thereby determining the adaptive cutoff frequency.

Benefits of technology

Under different signal-to-noise ratio conditions, it ensures the consistency and reliability of the filtering effect, effectively suppresses high-frequency noise, maintains the stability of the load response pulse, and adapts to the monitoring needs under different working conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122500A_ABST
    Figure CN123122500A_ABST
Patent Text Reader

Abstract

This invention discloses an adaptive filtering method, system, and device for road load response signals, relating to the field of road engineering structure monitoring and dynamic signal processing technology. The method includes: acquiring the time-domain signal of the road load response and determining the initial cutoff frequency based on its spectral amplitude; selecting a reference signal with a spectral energy concentration higher than a preset standard to determine a candidate cutoff frequency range; selecting a noise-sensitive reference signal, comprehensively evaluating the peak value retention rate and root mean square noise reduction rate at each candidate cutoff frequency, and determining an upper limit constraint on the cutoff frequency; selecting the smaller of the initial cutoff frequency and the upper limit constraint as the adaptive cutoff frequency and performing filtering. This method avoids the problem of the cutoff frequency following the abnormal expansion of the noise floor under low signal-to-noise ratio conditions through a dual constraint mechanism, while employing forward-reverse zero-phase filtering to maintain the timing of the load response pulses, making it suitable for batch processing of multi-channel monitoring data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of road engineering structure monitoring and dynamic signal processing technology, specifically to an adaptive filtering method, system, and device for road surface load response signals. Background Technology

[0002] Under the action of moving vehicle loads, pavement structures will generate dynamic mechanical responses such as strain and pressure. By collecting these dynamic responses through response sensors deployed on or inside the pavement structure, a data foundation can be provided for pavement structure condition evaluation, mechanical model verification, fatigue analysis, and long-term health monitoring. In pavement structure health monitoring, strain sensors and pressure sensors embedded inside or on the surface of the pavement structure are often used to collect dynamic response signals under moving vehicle loads. In order to extract effective load response pulses from the raw signals, low-pass filtering is usually required to suppress high-frequency noise.

[0003] Existing signal processing methods typically set a relative amplitude threshold based on the amplitude of the dominant frequency in the signal spectrum, and determine the highest frequency whose spectral amplitude is not lower than the threshold as the effective signal bandwidth, and then combine it with a safety margin to obtain the filter cutoff frequency.

[0004] However, when the amplitude of the road load response signal is low or the background noise is high, the noise floor in the high-frequency band of the spectrum will be relatively raised, making it difficult for the relative threshold based on the dominant frequency amplitude to effectively distinguish between the true response spectrum and the noise spectrum. As a result, the determined effective signal bandwidth expands abnormally with broadband noise, and the final cutoff frequency is much higher than the actual required filtering frequency. In this case, the low-pass filter cannot effectively suppress high-frequency noise, the number of local extrema of the load response pulse increases, the peak identification stability decreases, and the filtering effect is inconsistent, making it difficult to meet the processing needs of batch and long-term monitoring data. Summary of the Invention

[0005] To address the shortcomings of existing technologies in effectively distinguishing between the true response spectrum and the noise spectrum, which leads to the abnormal expansion of the determined effective signal bandwidth with broadband noise, this invention proposes an adaptive filtering method, system, and device for road load response signals. It proposes a dual cutoff frequency determination mechanism that combines adaptive initial cutoff frequency calculation and minimizing the reference signal based on multi-dimensional physical constraints, thereby solving the problems existing in the prior art.

[0006] An adaptive filtering method for road load response signals includes the following steps: The road surface load response time-domain signal is acquired, and the initial cutoff frequency is determined based on the spectral amplitude of the road surface load response time-domain signal. A road load response signal with a spectral energy concentration higher than a preset standard is selected as the first reference response signal. A candidate cutoff frequency range is determined based on the spectral energy distribution of the first reference response signal. A noise-sensitive road load response signal is selected as the second reference response signal. Within the candidate cutoff frequency range, each candidate cutoff frequency is used to perform low-pass filtering on the second reference response signal. The load response peak retention rate and the root mean square noise reduction rate for the no-load signal segment corresponding to each candidate cutoff frequency are calculated. Based on the peak retention rate and the root mean square noise reduction rate, the cutoff frequency that simultaneously satisfies the preset peak retention condition and the preset noise suppression condition is determined as the upper limit of the cutoff frequency constraint. The smaller of the initial cutoff frequency and the upper limit of the cutoff frequency constraint is selected as the adaptive cutoff frequency corresponding to the time-domain signal of the road load response. The time-domain signal of the road load response is filtered according to the adaptive cutoff frequency to obtain the filtered road load response processed signal.

[0007] Furthermore, determining the initial cutoff frequency based on the spectral amplitude of the road load response time-domain signal specifically includes the following steps: The DC bias component is determined based on the unloaded signal segment in the time-domain signal of the road load response. The DC bias component is removed from the road load response time domain signal to obtain the baseline correction signal; The baseline correction signal is subjected to frequency domain transformation to obtain the normalized one-sided amplitude spectrum; Within a preset low-frequency analysis range, the frequency corresponding to the maximum value of the normalized single-sided amplitude spectrum is determined as the dominant load frequency. Using a preset proportion of the spectral amplitude corresponding to the dominant load frequency as a relative amplitude threshold, the highest frequency with a spectral amplitude not less than the relative amplitude threshold is determined as the effective signal bandwidth; The initial cutoff frequency is determined based on the sum of the effective signal bandwidth and the preset safety margin.

[0008] Furthermore, the road load response signal with a spectral energy concentration higher than the preset standard is a road load response signal in which the load response frequency components and high-frequency noise frequency components can be distinguished.

[0009] Furthermore, the noise-sensitive road load response signal is a road load response signal under low amplitude or high noise background.

[0010] Further, determining the candidate cutoff frequency range based on the spectral energy distribution of the first reference response signal specifically includes the following steps: Calculate the cumulative spectral energy ratio of the first reference response signal at different frequencies. ;in, The sequence number of the reference response signal in the first reference response signal set; The discrete frequency point number; For the first The physical frequencies corresponding to each discrete frequency point, and ; The target frequency used to calculate the cumulative spectral energy ratio; For the first The reference response signal at frequency Normalized one-sided spectral amplitude at the location; The sampling frequency; This represents the number of sampling points; The Nyquist frequency; Determine how to make the cumulative spectrum energy ratio reach the preset energy maintenance ratio. Minimum frequency: ; The candidate cutoff frequency range is determined based on the minimum frequency.

[0011] Furthermore, the calculation of the load response peak retention rate and the root mean square noise reduction rate of each candidate cutoff frequency specifically includes the following steps: For each candidate cutoff frequency, a low-pass filter is configured using that candidate cutoff frequency to filter the second reference response signal; Based on the peak load response before filtering and after filtering The peak load response was calculated, and the peak load response retention rate was determined. The load response peak value is the maximum amplitude of the second reference response signal within the main loading time window. Based on the root mean square of the unloaded signal segment before filtering and the root mean square of the filtered unloaded signal segment Calculate the root mean square reduction rate of noise in the unloaded signal segment. The no-load signal segment is the signal interval in the second reference response signal that does not contain a moving load response event.

[0012] Further, the road load response time-domain signal is filtered according to the adaptive cutoff frequency to obtain a filtered road load response processing signal. Specifically, a digital low-pass filter is configured according to the adaptive cutoff frequency, and the baseline correction signal is forward-filtered and reverse-filtered using the configured digital low-pass filter to obtain a road load response processing signal in which high-frequency noise is suppressed and the timing of the load response pulses is preserved. The digital low-pass filter is a second-order Butterworth low-pass filter, and the normalized cutoff frequency of the second-order Butterworth low-pass filter is determined according to the ratio of the adaptive cutoff frequency to the Nyquist frequency.

[0013] The present invention also includes an adaptive filtering system for road load response signals, comprising: The acquisition module is used to acquire the road load response time-domain signal and determine the initial cutoff frequency based on the spectral amplitude of the road load response time-domain signal. The cutoff frequency determination module is used to select a road load response signal with a spectral energy concentration higher than a preset standard as a first reference response signal, and determine a candidate cutoff frequency range based on the spectral energy distribution of the first reference response signal; select a noise-sensitive road load response signal as a second reference response signal, and perform low-pass filtering on the second reference response signal using each candidate cutoff frequency within the candidate cutoff frequency range, calculating the load response peak retention rate and the root mean square noise reduction rate of the no-load signal segment corresponding to each candidate cutoff frequency; based on the peak retention rate and the root mean square noise reduction rate, determine the cutoff frequency that simultaneously satisfies the preset peak retention condition and the preset noise suppression condition as the upper limit of the cutoff frequency constraint; and select the smaller of the initial cutoff frequency and the upper limit of the cutoff frequency constraint as the adaptive cutoff frequency corresponding to the time-domain signal of the road load response. The filtering module is used to filter the road load response time-domain signal according to the adaptive cutoff frequency to obtain the filtered road load response processed signal.

[0014] The present invention also includes an adaptive filtering computer device for road load response signals, comprising: a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the adaptive filtering method for road load response signals.

[0015] The present invention also includes a readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, are used to perform the steps of the adaptive filtering method for the road load response signal.

[0016] This invention provides an adaptive filtering method for road load response signals, which has the following advantages: To address the technical shortcomings of existing technologies where a single spectral threshold easily follows the noise floor under low signal-to-noise ratio (SNR) conditions, leading to abnormal cutoff frequency expansion, this invention proposes a dual cutoff frequency determination mechanism: adaptive initial cutoff frequency calculation based on the spectrum and minimizing the value of the reference signal's multi-dimensional physical constraints. By introducing a first reference response signal with a spectral energy concentration higher than a preset standard to define the candidate cutoff frequency range, and then using a noise-sensitive second reference response signal to comprehensively evaluate the peak fidelity and noise suppression effect at each candidate cutoff frequency, the cutoff frequency that simultaneously satisfies the preset peak fidelity and noise suppression conditions is determined as the upper constraint limit. This upper constraint limit reflects the energy concentration pattern of the load response signal itself and the dual requirements of peak fidelity and noise suppression in engineering. Finally, the smaller of the initial cutoff frequency and the upper constraint limit is selected as the adaptive cutoff frequency, effectively avoiding the problem of the initial cutoff frequency abnormally expanding along the broadband noise floor under low SNR conditions, and ensuring the consistency and reliability of the filtering effect under different SNR conditions. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the adaptive cutoff frequency determination method in an embodiment of the present invention. Figure 2 This is a schematic diagram of DC bias removal and baseline correction of the original signal in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the spectral relationship between the dominant load frequency, effective signal bandwidth, and adaptive cutoff frequency in an embodiment of the present invention. Figure 4 This is a flowchart illustrating the calibration process for the upper limit of the cutoff frequency constraint in an embodiment of the present invention. Figure 5 This is a flowchart of the forward-reverse zero-phase filtering process in an embodiment of the present invention; Figure 6 This is a block diagram of the adaptive cutoff frequency determination system in an embodiment of the present invention. Detailed Implementation

[0018] 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.

[0019] Currently, during the monitoring of road load response signals, due to the influence of sensor initial state, installation residual strain, self-weight effect, temperature change, electronic noise, electromagnetic interference, and quantization error of the acquisition system, the original road load response signal usually contains DC bias, slowly changing baseline trend, and high-frequency random fluctuations. DC bias will cause the signal to deviate from the theoretical zero baseline, while high-frequency noise may mask the real load pulse, change the number of local extrema, and reduce the stability of peak identification.

[0020] Existing signal processing methods typically employ fixed cutoff frequencies or determine filtering parameters through manual calculation. However, vehicle speed, road surface structure, measurement point depth, sensor type, and response amplitude all alter the spectral characteristics of the load response signal. A fixed cutoff frequency may retain excessive noise under low-speed or high-amplitude conditions, or it may over-smooth the true load pulse under high-speed or low-amplitude conditions. On the other hand, if the effective bandwidth is determined solely based on a single threshold relative to the dominant frequency amplitude, the determined bandwidth may abnormally expand along with broadband noise when the signal amplitude decreases and the noise floor relatively increases.

[0021] Based on this, the present invention proposes an adaptive cutoff frequency determination method for road load response signals based on spectral feature constraints. This method can automatically determine the initial cutoff frequency for the spectral features of each record, and at the same time set the upper limit of physical constraints by utilizing the energy concentration law of the reference signal and the peak preservation and noise suppression effects, so as to take into account high-frequency noise suppression, load response amplitude preservation and event timing preservation under different working conditions.

[0022] like Figure 1 As shown, the method specifically includes the following steps: S1. Acquire the time-domain signal of the road surface load response collected by the road surface response sensor under moving load. And obtain the sampling frequency of the time-domain signal of the road load response. and number of sampling points .

[0023] The time-domain signal of pavement load response includes at least one of the strain response signal and pressure response signal generated by the pavement structure under the action of moving wheel load; the time-domain signal of pavement load response is collected by at least one pavement response sensor installed on the surface or inside the pavement structure; when multiple pavement response sensors are installed, a corresponding adaptive cutoff frequency is determined for the time-domain signal of pavement load response collected by each pavement response sensor.

[0024] S2. Determine the DC bias component based on the unloaded signal segment in the road load response time-domain signal, and remove the DC bias component from the road load response time-domain signal to obtain the baseline correction signal. .

[0025] Specifically, the DC bias component is estimated from a no-load signal segment that does not contain moving load response events. and in accordance with Obtain baseline correction signal In one implementation, the signal segment representing 5% of the total number of sampling points at the recording start position is selected as the no-load signal segment; specifically, this includes: Determine the number of sampling points for the no-load signal segment. : .

[0026] according to Determine the DC bias component : .

[0027] according to Obtain the baseline correction signal: ;in, .

[0028] S3. Perform frequency domain transformation on the baseline correction signal to obtain the normalized one-sided amplitude spectrum. Within the preset low-frequency analysis interval, determine the dominant load frequency corresponding to the maximum amplitude of the normalized one-sided amplitude spectrum. The effective signal bandwidth is determined based on the spectral amplitude corresponding to the dominant load frequency and a preset relative amplitude threshold. Specifically, it refers to a preset proportion of the spectral amplitude corresponding to the dominant load frequency. As a relative amplitude threshold, the highest frequency with a spectral amplitude not less than this threshold is determined as the effective signal bandwidth. ,in .

[0029] (1) Frequency domain transformation of the baseline correction signal includes: Perform a Discrete Fourier Transform on the baseline correction signal according to the following formula: The normalized one-sided amplitude spectrum is obtained by normalizing the amplitude and the number of sampling points of the discrete Fourier transform result. Among them, the The physical frequency corresponding to each discrete frequency point Determined according to the following formula: Dominant load frequency To ensure the normalized one-sided amplitude spectrum within the preset low-frequency analysis interval The frequency at which the maximum value is reached.

[0030] (2) Determine the effective signal bandwidth include: Based on the spectral amplitude corresponding to the dominant load frequency and preset relative amplitude coefficient The relative amplitude threshold is determined according to the following formula. : The highest frequency with a spectral amplitude not less than the relative amplitude threshold is defined as the effective signal bandwidth. in, Preferably, .

[0031] S4. Based on the effective signal bandwidth and preset safety margin The initial cutoff frequency is determined according to the following formula. : The safety margin is used to preserve higher-order load response components located near the effective bandwidth boundary; the safety margin can be determined based on the frequency resolution, load pulse width, or spectral attenuation characteristics of the reference record.

[0032] Obtain a first reference response signal set, and determine the candidate cutoff frequency range based on the cumulative spectral energy distribution of the first reference response signal set; specifically, this includes: calculating the first reference response signal set for each of the following: The reference response signal at frequency The following cumulative spectral energy ratios : ;in, The sequence number of the reference response signal in the first reference response signal set; The discrete frequency point number; For the first The physical frequencies corresponding to each discrete frequency point, and ; The target frequency used to calculate the cumulative spectral energy ratio; For the first The reference response signal at frequency Normalized one-sided spectral amplitude at the location; The sampling frequency; This represents the number of sampling points; The Nyquist frequency is used to determine the cumulative spectral energy ratio to reach the preset energy retention ratio. minimum frequency : According to the reference response signals corresponding to each reference response signal in the first reference response signal set Determine the candidate cutoff frequency range. The first reference response signal set includes road load response signals whose load response frequency components and high-frequency noise frequency components can be distinguished, with a preset energy retention ratio. Not less than 95%.

[0033] A second reference response signal with low amplitude or high noise sensitivity is selected, and low-pass filtered using multiple candidate cutoff frequencies within the candidate cutoff frequency range. The load response peak retention rate is then calculated. and the root mean square reduction rate of noise in the no-load signal segment Based on the load response peak retention rate and the root mean square noise reduction rate in the no-load signal segment corresponding to different candidate cutoff frequencies, the upper limit of the cutoff frequency constraint is determined. and in accordance with Obtain the adaptive cutoff frequency corresponding to each signal to be processed.

[0034] Peak load response retention rate Determined according to the following formula: .

[0035] Root mean square reduction rate of noise in no-load signal segment Determined according to the following formula: ;in, and These are the peak load responses before and after filtering, respectively. and These are the root mean squares of the unloaded signal segments before and after filtering, respectively.

[0036] The candidate cutoff frequency that simultaneously satisfies the preset peak hold condition and the preset noise suppression condition is determined as the upper limit of the cutoff frequency constraint. .

[0037] Under the condition of road surface strain response signal processing at vehicle speeds of 5–40 km / h, candidate cutoff frequencies include 20 Hz, 25 Hz, 30 Hz, 40 Hz, and 50 Hz, with an upper limit constraint on the cutoff frequency. It is 25 Hz.

[0038] S5. Based on the initial cutoff frequency and upper limit of cutoff frequency constraint The adaptive cutoff frequency corresponding to the time-domain signal of the road load response is determined according to the following formula. : ;in, .

[0039] S6. Based on the adaptive cutoff frequency Configure a digital low-pass filter and use it to filter the baseline correction signal to obtain a road load response processing signal in which high-frequency noise is suppressed and the timing of the load response pulse is preserved.

[0040] Preferably, the digital low-pass filter is a second-order Butterworth low-pass filter; the normalized cutoff frequency of the second-order Butterworth low-pass filter is based on the adaptive cutoff frequency. With Nyquist frequency The ratio is determined; the baseline correction signal is then subjected to forward filtering and reverse filtering in sequence to eliminate the phase delay caused by causal filtering, so that the position of the load response peak in the time domain remains consistent before and after filtering.

[0041] S7. When event-level feature extraction is required, local peaks and valleys in the road load response processing signal are identified based on terrain prominence. False extremes caused by residual noise are eliminated based on the degree of prominence of local extremes relative to adjacent reference areas. The peak values ​​of tensile response, compression response, and their occurrence times are determined within the main loading time window corresponding to the moving load. The difference between adjacent peaks and valleys is used as the load response amplitude to obtain road load response features including the occurrence time of load response events, load response amplitude, and peak-valley arrangement order.

[0042] Compared with the prior art, the present invention has at least the following beneficial effects.

[0043] (1) By associating the cutoff frequency with the dominant frequency and effective bandwidth of the load response signal of a single road surface, the filtering bandwidth can be automatically adjusted for different vehicle speeds, measuring points and response amplitudes, reducing manual calculations for each road surface.

[0044] (2) By constructing the upper limit of the cutoff frequency constraint using the cumulative spectral energy, peak retention rate and root mean square noise reduction rate, we can avoid the abnormal bandwidth expansion caused by low amplitude and low signal-to-noise ratio signals following the broadband noise floor when using the relative amplitude threshold.

[0045] (3) Using forward-reverse zero-phase filtering, while attenuating high-frequency noise, the timing of the load response peak is maintained, which is beneficial for judging the vehicle load position and comparing the response timing of different measuring points.

[0046] (4) This invention takes road load response signal as the processing object and does not require a specific sensor installation structure as a necessary condition. It can be applied to batch processing of monitoring data from different sensor locations and multiple channels.

[0047] (5) The processed signal can be further used for peak and valley identification, response amplitude statistics, mechanical model verification and pavement structure status evaluation.

[0048] Based on the above methods, the present invention proposes the following embodiments: Example 1: Determination of Adaptive Cutoff Frequency for Road Surface Strain Response Signal like Figure 1 As shown, in this embodiment, response signal acquisition, baseline correction, spectral feature extraction, initial cutoff frequency calculation, physical constraint upper limit application, and phase-preserving filtering are completed sequentially according to S1 to S6.

[0049] S1. Obtain the time-domain signal of the road surface load response. A road surface strain sensor is used to collect the strain time history under a moving single-axle dual-wheel load. In a specific test, the rear axle load of the vehicle was 100 kN, and the vehicle speeds were 5 km / h, 10 km / h, 20 km / h, and 40 km / h, with 5 effective repeated tests at each speed; the strain measurement points were located at different depths within the asphalt layer. The acquired raw signal is denoted as... The above-described load, speed, and measuring point conditions are for illustrative purposes only and do not constitute a limitation on the scope of application of this invention.

[0050] Sampling frequency The sampling frequency should meet the Nyquist sampling condition and be twice the frequency band of the structure to be retained. In one data acquisition implementation, a sampling frequency of 2000 Hz is used. For long-term monitoring data, other sampling frequencies may be used provided that the target frequency band fidelity requirement is met.

[0051] S2, Remove DC bias. For example... Figure 2 As shown, the original signal is generated by the target's dynamic response. Baseline bias and slow trend of change and high frequency noise Composition, that is .

[0052] After confirming that no vehicle loading occurred at the beginning of the recording, the first pN sampling points are selected as the no-load signal segment. In this embodiment... , .according to Calculate the DC bias and follow the instructions. Obtain the baseline correction signal. If the initial recording interval contains loading events, other unloaded intervals can be selected first through vehicle event detection, and then the DC bias can be estimated using the mean or robust location statistics of the selected unloaded intervals.

[0053] S3. Extract spectral features. conduct Point Discrete Fourier Transform: ,in This represents the discrete frequency index. The physical frequencies corresponding to the discrete frequency points are: .

[0054] according to Normalized one-sided amplitude spectrum constructed from amplitude values Within the preset low-frequency analysis range excluding zero-frequency components, The frequency corresponding to the maximum value is determined as the dominant load frequency. The preset low-frequency analysis range can be set according to the vehicle speed range, load application time, and expected response frequency band of the road structure; in the above 5-40 km / h test, the dominant load frequency is mainly below 10 Hz.

[0055] by Determine the relative amplitude threshold. In this embodiment, we take... and will satisfy The highest frequency is determined as the effective signal bandwidth. ,Right now .like Figure 3 As shown, Describe the location of the main load excitation. Describe the upper boundary of the structural response that still has a significant spectral amplitude.

[0056] S4. Determine the initial cutoff frequency. To avoid truncating higher-order response components near the effective signal bandwidth boundary, in Increase safety margin on the basis ,get . It can be set to frequency resolution. It can be a positive integer multiple, or it can be determined based on the attenuation law of harmonics near the main frequency in the reference signal.

[0057] When using only the relative amplitude threshold, low amplitude signals This may increase with the increase in broadband noise floor. Therefore, this embodiment further determines the upper limit of the cutoff frequency constraint using a reference signal. .

[0058] S5, calibrate the upper limit of the cutoff frequency constraint. For example... Figure 4 As shown, a first reference response signal set with relatively clear separation between the load response components and the noise frequency band is first selected. For the... Calculate the cumulative spectral energy ratio of the reference signal and determine that it meets the requirements. minimum frequency This embodiment takes .

[0059] In the reference records at vehicle speeds of 5 km / h, 10 km / h, and 20 km / h, more than 95% of the strain signal spectral energy is concentrated below approximately 25 Hz. Based on this, 20–50 Hz is set as the candidate upper limit analysis range, and 20 Hz, 25 Hz, 30 Hz, 40 Hz, and 50 Hz are selected as candidate cutoff frequencies.

[0060] Subsequently, a low-amplitude noise-sensitive strain signal at a depth of 25 cm and a speed of 40 km / h was selected as the second reference response signal, and filtered using the aforementioned candidate cutoff frequencies. Calculate the load response peak retention rate, where and These are the peak values ​​of the main loading window before and after baseline correction and filtering, respectively; according to... Calculate the root mean square reduction rate of noise in the unloaded signal segment.

[0061] When the candidate cutoff frequencies are 20 Hz, 25 Hz, 30 Hz, 40 Hz, and 50 Hz, the average peak retention rates are 76.1%, 86.0%, 92.0%, 97.5%, and 99.0%, respectively, with corresponding root mean square noise reduction rates of 41.8%, 39.2%, 36.8%, 32.0%, and 27.5% for the unloaded signal segment. There is a tendency for excessive smoothing of load pulses at 20 Hz, while 40–50 Hz retains more high-frequency fluctuations; 25–30 Hz forms a stable trade-off between peak retention and noise suppression. To obtain stronger noise suppression capabilities, this embodiment will... Set to 25 Hz.

[0062] For each signal to be processed, the final adaptive cutoff frequency is based on... Therefore, when the initial cutoff frequency obtained from the relative threshold is lower than the upper limit of the constraint, the narrow-band result of the record itself is retained; when the low signal-to-noise ratio causes the initial cutoff frequency to increase abnormally, it is adjusted accordingly. Limit bandwidth expansion. For other road structures, speed ranges, or sensor systems, recalibration should be performed using the corresponding reference signal. 25 Hz should not be interpreted as a fixed value for all road load response signals.

[0063] S6. Implement phase-preserving filtering. For example... Figure 5 As shown, based on the normalized cutoff frequency Design a second-order Butterworth low-pass filter. First, along the positive time direction... Filter the result, then reverse the result and filter it again using the same filter. Finally, restore the time sequence to obtain the final result. Forward-reverse processing eliminates phase delay and gives the synthesized amplitude-frequency response a steeper roll-off characteristic.

[0064] After the above processing is completed, the noise suppression effect can be evaluated based on the root mean square or signal-to-noise ratio of the unloaded signal segment, and the peak value of the main loading window before and after filtering can be compared to verify the fidelity of the response amplitude. In the multi-velocity strain record, the overall signal-to-noise ratio of the processed signal is significantly improved, and the peak and trough occurrence times corresponding to the vehicle load remain stable.

[0065] S7. Optional peak and valley feature extraction. The terrain prominence is calculated using local extrema, and small noise spurious peaks are eliminated based on the height difference between the local extrema and adjacent reference areas. Tensile peaks, compression peaks, and their occurrence times are identified within the main loading window, and the load response amplitude is represented by the difference between adjacent peaks and valleys. Compared to using only a fixed absolute amplitude threshold, terrain prominence is more adaptable to residual baseline shifts and small local fluctuations.

[0066] Example 2: Batch Processing of Multi-Channel Response Signals For sensor channels located at multiple depths or in multiple directions, the sampling frequency, recording length, and no-load interval of each channel are read separately, and calculations are performed independently for each record according to Example 1. , and And the same monitoring structure was calibrated using a reference signal. Therefore, each record can obtain a value that matches its own spectral energy distribution. This eliminates the need to manually set the cutoff frequency for each channel.

[0067] When there are significant differences in sensor type, sampling frequency, or effective response bandwidth between different channels, reference signal groups can be established separately and their corresponding values ​​calibrated. For pressure response signals, the same spectral threshold and upper limit constraint process can be used after dimension preservation or normalization processing; the specific energy ratio, candidate cutoff frequency, and peak holding conditions are re-determined based on the physical frequency band of the pressure response.

[0068] Based on the same inventive concept, this invention also proposes an adaptive filtering system for road load response signals, such as... Figure 6 As shown, it includes: The signal acquisition module 101 is used to acquire the time-domain signal of the road load response collected by the road response sensor under the action of moving load, and to acquire the sampling frequency and the number of sampling points of the time-domain signal of the road load response.

[0069] The baseline correction module 102 is used to determine the DC bias component based on the unloaded signal segment in the road load response time domain signal; and to remove the DC bias component from the road load response time domain signal to obtain the baseline correction signal.

[0070] The spectrum feature extraction module 103 is used to perform frequency domain transformation on the baseline correction signal to obtain a normalized one-sided amplitude spectrum; within a preset low-frequency analysis interval, the frequency corresponding to the maximum value of the normalized one-sided amplitude spectrum is determined as the dominant load frequency; using a preset ratio of the spectrum amplitude corresponding to the dominant load frequency as a relative amplitude threshold, the highest frequency with a spectrum amplitude not less than the relative amplitude threshold is determined as the effective signal bandwidth.

[0071] The cutoff frequency determination module 104 is used to select a road load response signal with a spectral energy concentration higher than a preset standard as a first reference response signal, determine a candidate cutoff frequency range based on the spectral energy distribution of the first reference response signal, select a noise-sensitive road load response signal as a second reference response signal, perform low-pass filtering on the second reference response signal using each candidate cutoff frequency within the candidate cutoff frequency range, calculate the load response peak retention rate and the root mean square noise reduction rate of the no-load signal segment corresponding to each candidate cutoff frequency, determine the cutoff frequency that simultaneously satisfies the preset peak retention condition and the preset noise suppression condition as the upper limit of the cutoff frequency constraint based on the peak retention rate and the root mean square noise reduction rate, and select the smaller of the initial cutoff frequency and the upper limit of the cutoff frequency constraint as the adaptive cutoff frequency corresponding to the time-domain signal of the road load response.

[0072] The filtering module 105 is used to filter the road load response time domain signal according to the adaptive cutoff frequency to obtain the filtered road load response processed signal 106.

[0073] The present invention also proposes an adaptive filtering computer device for road load response signals, comprising: a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the adaptive filtering method for road load response signals.

[0074] The present invention also proposes a readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, are used to perform steps of an adaptive filtering method for road load response signals.

[0075] 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 of adaptive filtering of a road surface load response signal, characterized by, Includes the following steps: The road surface load response time-domain signal is acquired, and the initial cutoff frequency is determined based on the spectral amplitude of the road surface load response time-domain signal. A road load response signal with a spectral energy concentration higher than a preset standard is selected as the first reference response signal, and the candidate cutoff frequency range is determined based on the spectral energy distribution of the first reference response signal. A noise-sensitive road load response signal is selected as the second reference response signal. Within the candidate cutoff frequency range, the second reference response signal is low-pass filtered using each candidate cutoff frequency. The load response peak retention rate and the root mean square noise reduction rate of the no-load signal segment corresponding to each candidate cutoff frequency are calculated. Based on the peak retention rate and the root mean square noise reduction rate, the cutoff frequency that simultaneously satisfies the preset peak retention condition and the preset noise suppression condition is determined as the upper limit of the cutoff frequency constraint. The smaller of the initial cutoff frequency and the upper limit of the cutoff frequency constraint is selected as the adaptive cutoff frequency corresponding to the time-domain signal of the road load response. The time-domain signal of the road load response is filtered according to the adaptive cutoff frequency to obtain the filtered road load response processed signal.

2. The method of claim 1, wherein, The initial cutoff frequency is determined based on the spectral amplitude of the time-domain signal of the road load response, specifically including the following steps: The DC bias component is determined based on the unloaded signal segment in the time-domain signal of the road load response. The DC bias component is removed from the road load response time domain signal to obtain the baseline correction signal; The baseline correction signal is subjected to frequency domain transformation to obtain the normalized one-sided amplitude spectrum; Within the preset low-frequency analysis range, the frequency corresponding to the maximum value of the normalized single-sided amplitude spectrum is determined as the dominant load frequency; Using a preset proportion of the spectral amplitude corresponding to the dominant load frequency as a relative amplitude threshold, the highest frequency with a spectral amplitude not less than the relative amplitude threshold is determined as the effective signal bandwidth; The initial cutoff frequency is determined based on the sum of the effective signal bandwidth and the preset safety margin.

3. The method of claim 1, wherein, The road load response signal with a spectral energy concentration higher than the preset standard is a road load response signal in which the load response frequency components and high-frequency noise frequency components can be distinguished.

4. The method of claim 1, wherein, The noise-sensitive road load response signal is the road load response signal under low amplitude or high noise background.

5. The method of claim 1, wherein, Determining the candidate cutoff frequency range based on the spectral energy distribution of the first reference response signal specifically includes the following steps: respectively calculate a cumulative spectral energy proportion of the first reference response signal under different frequencies ; wherein, is a sequence number of a central reference response signal in the first reference response signal; is a discrete frequency point sequence number; is a physical frequency corresponding to the th discrete frequency point, and ; is a target frequency for calculating the cumulative spectral energy proportion; is a normalized single sideband spectral amplitude of the th reference response signal at the frequency ; is a sampling frequency; is a sampling point number; is a Nyquist frequency; determining that the accumulated spectral energy ratio reaches a preset energy maintenance ratio of the minimum frequency: ; The candidate cutoff frequency range is determined based on the minimum frequency.

6. The method of claim 1, wherein, The calculation of the load response peak retention rate and the root mean square noise reduction rate of the no-load signal segment corresponding to each candidate cutoff frequency specifically includes the following steps: For each candidate cutoff frequency, a low-pass filter is configured using that candidate cutoff frequency to filter the second reference response signal; According to the load response peak value before filtering and the load response peak value after filtering , the load response peak value retention rate is calculated ; the load response peak value is the maximum amplitude of the second reference response signal within a main loading time window; According to the root mean square of the pre-filtering no-load signal segment and the root mean square of the post-filtering no-load signal segment , the noise root mean square reduction rate of the no-load signal segment is calculated ; the no-load signal segment is a signal interval in the second reference response signal that does not contain a moving load response event.

7. The method of claim 2, wherein, The road load response time-domain signal is filtered according to the adaptive cutoff frequency to obtain a filtered road load response processing signal. Specifically, a digital low-pass filter is configured according to the adaptive cutoff frequency, and the baseline correction signal is forward-filtered and reverse-filtered using the configured digital low-pass filter to obtain a road load response processing signal in which high-frequency noise is suppressed and the timing of the load response pulses is preserved. The digital low-pass filter is a second-order Butterworth low-pass filter, and the normalized cutoff frequency of the second-order Butterworth low-pass filter is determined according to the ratio of the adaptive cutoff frequency to the Nyquist frequency.

8. An adaptive filtering system for a road load response signal, characterized by include: The acquisition module is used to acquire the road load response time-domain signal and determine the initial cutoff frequency based on the spectral amplitude of the road load response time-domain signal. The cutoff frequency determination module is used to select a road load response signal with a spectral energy concentration higher than a preset standard as a first reference response signal, and determine the candidate cutoff frequency range based on the spectral energy distribution of the first reference response signal. A noise-sensitive road load response signal is selected as the second reference response signal. Within the candidate cutoff frequency range, the second reference response signal is low-pass filtered using each candidate cutoff frequency. The load response peak retention rate and the root mean square noise reduction rate of the no-load signal segment corresponding to each candidate cutoff frequency are calculated. Based on the peak retention rate and the root mean square noise reduction rate, the cutoff frequency that simultaneously satisfies the preset peak retention condition and the preset noise suppression condition is determined as the upper limit of the cutoff frequency constraint. The smaller of the initial cutoff frequency and the upper limit of the cutoff frequency constraint is selected as the adaptive cutoff frequency corresponding to the time-domain signal of the road load response. The filtering module is used to filter the road load response time-domain signal according to the adaptive cutoff frequency to obtain the filtered road load response processed signal.

9. An adaptive filtering computer device for pavement load response signals, characterized by, include: The memory, the processor, and the computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the adaptive filtering method for road load response signals according to any one of claims 1-7.

10. A readable storage medium, characterized by, The readable storage medium stores a computer program, which includes program instructions that, when executed by a processor, are used to perform the steps of the adaptive filtering method for road load response signals according to any one of claims 1-7.