Engineering pile length estimation method and system

Abstract

The invention discloses a method and a system for estimating the pile length of an engineering pile. The method comprises the following steps: acquiring a multi-source detection signal reflecting the striking response of a pile hammer and stratum prior information; on the basis of stratum prior information, carrying out physically-driven adaptive analysis on sound wave components in the multi-source detection signals, and extracting preliminary sound wave reflection features; performing correlation verification on the preliminary sound wave reflection characteristics by using vibration components in the multi-source detection signals to generate corrected reflection characteristics; and the signal quality of the corrected reflection characteristics is evaluated, an intelligent hammering control instruction is generated according to the signal quality so as to execute closed-loop enhanced signal excitation, finally, multi-modal information before and after enhancement is fused, and a final pile length estimated value is obtained through calculation. Through sound and vibration coupling verification and intelligent active excitation, the accuracy, reliability and detection depth of pile length estimation under the conditions of strong noise, complex stratum and large depth are improved.

Description

Technical Field

[0001] The present invention belongs to the field of construction engineering detection, and in particular to a method and system for estimating the length of engineering piles. Background Art

[0002] The length of an engineering pile is one of the core parameters that determines the bearing capacity of a single pile. It is directly related to whether the pile end can be embedded in the predetermined stable stratum with sufficient bearing capacity. In engineering practice, if the pile length is insufficient, the bearing capacity of the building foundation will not meet the design requirements, which may cause uneven settlement or even structural failure. If the pile length exceeds the design by too much, it will not only result in a huge waste of materials and labor hours, but also increase the cost of the project. Especially in areas with complex and changing geological conditions, the actual pile depth often differs from the initial design. Therefore, developing a technical means to accurately and reliably estimate the pile length in real time during the pile foundation construction process has important engineering value and economic significance for ensuring project quality, optimizing construction decisions, and controlling construction costs.

[0003] Currently, methods for measuring pile length in construction projects are primarily divided into two categories: static testing and dynamic testing. Static testing, such as the low-strain reflection wave method, is often performed after pile construction and is one of the most widely used nondestructive testing techniques. Dynamic testing, which provides real-time estimation during construction, is primarily based on stress wave theory. Its basic principle is that the impact of the pile hammer generates a stress wave at the pile top. This stress wave propagates downward along the pile shaft, generating a reflected wave upon encountering the pile bottom or a stratum interface with a significant difference in acoustic impedance. Sensors placed on the pile shaft (such as accelerometers) collect signals containing both the initial and reflected waves. By measuring the time difference Δt between the initial wave and the reflected wave at the pile bottom and combining it with the known wave propagation velocity c in the pile shaft material, the pile length can be estimated using the formula L = 1 / 2cΔt. To extract the weak reflected signal from the intense construction noise, existing solutions typically employ pre-processing techniques such as digital filtering and Fourier transforms. A more advanced solution is to use wavelet transform technology in the field of signal processing to denoise the signal, or use matched filtering technology, that is, presetting a theoretical standard reflection waveform as a template, and finding the segment that best matches it in the collected signal to identify the arrival time of the reflected wave.

[0004] However, when faced with complex field conditions, these existing dynamic detection technologies still face bottlenecks that need to be addressed. These bottlenecks mainly focus on two aspects: first, the inability to identify weak effective signals in strong noise backgrounds; second, the lack of effective means to enhance deep signals and cross-validate analysis results. Summary of the Invention

[0005] The purpose of the invention is to provide a method and system for estimating the length of engineering piles to solve the above-mentioned problems existing in the prior art.

[0006] A technical solution, a method for estimating the length of engineering piles, includes:

[0007] Acquire prior information of the stratum reflecting the pile hammer's impact response, as well as multi-source detection signals including acoustic and vibration components;

[0008] Based on prior information of the formation, the acoustic wave components are adaptively analyzed to extract preliminary acoustic wave reflection characteristics;

[0009] Using vibration components, preliminary acoustic wave reflection features are correlated and verified, false features are suppressed, and corrected reflection features are generated;

[0010] Evaluate the signal quality of the corrected reflection characteristics, generate intelligent hammer control instructions based on this, execute enhanced signal excitation, fuse the information before and after enhancement, and calculate the estimated pile length.

[0011] A system for estimating the length of engineering piles, comprising:

[0012] a sensor assembly configured to acquire multi-source detection signals reflecting a pile hammer striking response;

[0013] an electronically controlled hammering system configured to receive control instructions and execute a hammering action with adjustable parameters;

[0014] A signal processing unit is operatively coupled to the sensor assembly and the electronically controlled hammer system, the signal processing unit being configured to:

[0015] Extracting preliminary acoustic wave reflection characteristics based on acoustic wave components in multi-source detection signals obtained from the sensor assembly and pre-stored formation prior information;

[0016] The vibration components in the multi-source detection signals are used to correlate and verify the preliminary acoustic wave reflection characteristics to generate the revised reflection characteristics;

[0017] Evaluate and modify the reflection characteristics, and generate intelligent hammer control instructions based on them and send them to the electronically controlled hammer system to perform enhanced hammer excitation;

[0018] The information before and after the enhanced hammering excitation is integrated to calculate the estimated final pile length.

[0019] Beneficial effect: The present invention improves the accuracy, reliability and detection depth of pile length estimation under conditions of strong noise, complex strata and great depth. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flowchart of the steps of a method for estimating the length of engineering piles provided in an embodiment of the present application.

[0021] Figure 2A flowchart of the steps for calculating the estimated pile length provided in an embodiment of the present application.

[0022] Figure 3 A flowchart of the steps for generating corrected reflection features provided in an embodiment of the present application.

[0023] Figure 4 A flowchart of the steps for calculating the acoustic vibration delay correlation provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or are inherent to these processes, methods, products or apparatus.

[0026] Research has found that at critical interfaces, such as the junction of soft and hard strata, the acoustic reflection coefficient is extremely low (typically only 0.05-0.15). Combined with on-site construction noise levels of up to 80-120 dB, the signal-to-noise ratio often falls below -20 dB. Under these extreme conditions, traditional fixed-parameter filters, while filtering out noise, can easily obliterate the weak true reflection signal. Even wavelet transforms or matched filters employing fixed basis functions are unable to adapt to the waveform distortion, dispersion, and attenuation that occur when acoustic waves propagate through complex strata. This leads to a significant mismatch between the preset template and the actual, diverse reflections, significantly reducing interface recognition efficiency. Furthermore, acoustic wave energy decays exponentially with propagation depth. For piles exceeding 20 meters deep, the reflected signal from the pile bottom may be completely drowned out by the noise, creating blind spots for existing passive detection methods at depth. On the other hand, existing methods rely entirely on the analysis of a single acoustic wave signal. When the complex signal processing algorithm itself introduces seemingly real pseudo-reflection peaks (for example, pseudo-peaks generated at the boundary by segmented processing), the system lacks information in the second physical dimension for cross-validation and is unable to determine whether a detected wave peak is a real physical reflection or an artifact generated by the algorithm, which directly leads to a significant reduction in the reliability of pile length judgment.

[0027] This embodiment provides a system for estimating the length of an engineering pile, comprising a sensor assembly, an electronically controlled hammering system, and a signal processing unit. The sensor assembly is configured to acquire multi-source detection signals reflecting the pile hammer's impact response. Specifically, the sensor assembly may include a hammer force sensor for acquiring the hammer's impact force signal, an accelerometer array (e.g., one sensor every two meters along the pile shaft) for acquiring acoustic signals, and a MEMS vibration sensor (e.g., a triaxial MEMS vibration sensor with a ±50g range) mounted at the pile end for acquiring the soil response at the pile end. A hardware triggering mechanism enables microsecond-level synchronous acquisition of each sensor channel, resulting in a time-synchronized raw signal set. Optionally, in addition to the accelerometer, the pile sensor array may also include strain gauges or fiber Bragg grating sensors to directly measure stress waves or strain. The pile end MEMS vibration sensor may also be replaced by a high-sensitivity geophone.

[0028] The electronically controlled hammering system is configured to receive control commands and execute parameter-adjustable hammering actions. The system can precisely adjust parameters such as the hammering frequency, interval, and force based on commands sent by the signal processing unit.

[0029] A signal processing unit is operatively coupled to the sensor assembly and the electronically controlled hammer system. Optionally, the physical implementation of the unit can be an edge computing device deployed at the construction site, an embedded module integrated into the pile driver control system, or a remote server that uploads data to the cloud for computing. The signal processing unit is configured to perform the following operations: Figure 1 A method for estimating the length of engineering piles shown includes the following steps:

[0030] Acquire prior information of the stratum reflecting the pile hammer's impact response, as well as multi-source detection signals including acoustic and vibration components.

[0031] In this embodiment, the multi-source detection signal is a set of time-series signal data reflecting different physical phenomena (such as pile stress wave propagation and pile tip soil vibration) collected synchronously by different physical sensors under a single hammer excitation. The signal processing unit receives the time-synchronized raw signal set from the sensor assembly and reads pre-stored geological survey borehole data, standard penetration test data, and other data to construct a stratigraphic reference model as prior stratigraphic information. Before processing, the signal processing unit also performs quality assessment on the raw signals, filters out noise, and performs baseline correction to produce pre-processed acoustic and vibration signals.

[0032] Based on prior information of the formation, the acoustic wave components are adaptively analyzed to extract preliminary acoustic wave reflection characteristics.

[0033] This embodiment aims to address the difficulty traditional methods face in identifying weak ground interface reflection signals in environments with strong construction noise (e.g., 80-120dB) and extremely low signal-to-noise ratios (e.g., <-20dB). In this embodiment, the signal processing unit utilizes a physics-driven dynamic signal processing architecture, abandoning the traditional fixed basis function approach and achieving adaptive decomposition of acoustic wave signals.

[0034] The vibration components are used to correlate and verify the preliminary acoustic wave reflection features, suppress the pseudo features generated during the analysis process, and generate corrected reflection features.

[0035] Specifically, this step aims to address the problem of spurious reflection peaks introduced at data boundaries due to algorithmic reasons such as segmented processing. In this embodiment, the signal processing unit establishes a physical correlation model between the acoustic reflection signal and the vibration signal of the pile end soil. The vibration signal is used to verify and correct the analysis results of the acoustic signal, effectively identifying and suppressing spurious features.

[0036] Evaluate the signal quality of the corrected reflection characteristics, generate intelligent hammer control instructions based on the signal quality, perform enhanced signal excitation, fuse the information before and after enhancement, and calculate the estimated pile length.

[0037] In this embodiment, this step aims to address the low detection success rate caused by rapid attenuation of deep signals (for example, those exceeding 20 meters). The signal processing unit dynamically optimizes the parameters for the next hammer strike by evaluating the energy decay of the current signal and instructs the electronically controlled hammer system to perform a customized enhanced excitation. This enables targeted enhancement of weak deep signals, improving detection depth and reliability. Finally, by fusing multimodal information before and after enhancement, a highly accurate final pile length estimate is calculated.

[0038] According to one aspect of the present application, before extracting preliminary acoustic wave reflection features, a step of generating a depth-adaptive atomic dictionary is also included. Specifically, this step includes:

[0039] By combining the acoustic wave components in the multi-source detection signals with prior formation information, a real-time physical parameter field that varies with depth is estimated. This parameter field includes acoustic wave velocity and acoustic impedance information.

[0040] The purpose of this step is to enable the subsequent signal analysis model to match the actual physical characteristics of the formation in real time, rather than relying on fixed, unrealistic mathematical models. Specifically, the steps to estimate the real-time physical parameter field that varies with depth include:

[0041] The acoustic wave components in the multi-source detection signals generated by the initial hammering are analyzed, the first wave arrival time difference Δt(z) of sensors at adjacent depths is calculated, and the real-time wave velocity field v(z) that varies with depth is determined.

[0042] For example, the sliding window method (window length 1 meter, step size 0.2 meters) is used to estimate the local wave velocity. The calculation formula is: v(z) = Δz / Δt(z); where v(z) is the real-time wave velocity at depth z, Δz is the distance between adjacent sensors (for example, 2 meters), and Δt(z) is the time difference between the first waves of the acoustic wave reaching the two sensors.

[0043] The dynamic impedance field Z(z) is constructed by multiplying the real-time velocity field v(z) with the density data ρ(z) extracted from the prior formation information. The calculation formula is: Z(z) = ρ(z) × v(z); where ρ(z) is the formation density at depth z, which can be obtained from the geological survey report. The real-time velocity field v(z) and the dynamic impedance field Z(z) together constitute the real-time physical parameter field.

[0044] Based on the values ​​of the real-time physical parameter field at different depths, a depth-adaptive atomic parameter set is calculated; among them, the frequency, phase and attenuation coefficient of the atom are determined by the acoustic impedance mutation characteristics and wave velocity at that depth, respectively; using the depth-adaptive atomic parameter set, atomic functions with specific mathematical forms are generated for each depth, which together constitute a depth-adaptive atomic dictionary.

[0045] Specifically, the atomic dictionary is an over-complete signal representation method consisting of a set of basic waveforms (i.e., atoms) with a specific mathematical form. By linearly combining these atoms, complex signals can be sparsely represented. The frequency, phase, and attenuation coefficient of the atomic function are determined by the acoustic impedance mutation characteristics and wave velocity at that depth, respectively. For example, the atomic parameters at each depth can be calculated based on the impedance gradient ΞZ / Ξz, the wave velocity v(z), etc., where Ξ is a partial derivative. The dictionary atom family generated in this way can better match the waveform distortion and attenuation caused by stratum changes during actual propagation, and has stronger adaptability and matching accuracy than the fixed wavelet basis or preset dictionary used in the prior art. Exemplarily, the atomic function ψ(t, z) can be expressed as: ψ(t, z)=A(z)×exp(-α(z)×t)×sin(2πf(z)×t+Φ(z)); wherein t is time, z is depth, A(z) is amplitude, α(z) is attenuation coefficient, f(z) is frequency, and Φ(z) is phase, and these parameters are adaptively adjusted with depth z. In some embodiments, in addition to calculating atomic parameters through physical formulas, a machine learning method can also be used to learn a deep neural network model that can generate optimal atoms by training on a large amount of known formation data and corresponding reflection signals.

[0046] On this basis, preliminary sound wave reflection features are extracted, including:

[0047] The acoustic wave components in the multi-source detection signal are divided into segmented signal sequences with overlapping ends along the depth dimension.

[0048] For example, the signal can be divided into multiple intervals by depth, such as [0-5 meters], [4-10 meters], and [9-15 meters], with adjacent intervals overlapping by 1 meter. Segmented processing can decompose a large-scale global optimization problem into multiple, computationally smaller, local sub-problems to meet real-time processing requirements.

[0049] A local projection operator is configured for each segment of the signal, and cascade projection calculations are performed on the segmented signal sequence in sequence to obtain projection results; wherein the projection calculation of the latter segment of the signal includes the processing residual generated by the projection calculation of the previous segment.

[0050] Specifically, for each interval, the atoms most relevant to the depth range are selected from the depth-adaptive atom dictionary (e.g., by principal component analysis dimensionality reduction), and a local, optimized projection operator P is constructed. k Cascade projection is then performed, which is a fast solution algorithm that decomposes a large-scale global optimization problem into a series of local projection subproblems connected end to end, where the residuals of the previous level calculation are passed and incorporated into the calculation of the next level. For example: x k =P k (x k-1 +r k-1 ); where x k is the projection result of level k, x k-1 is the input signal of the previous stage, r k-1 This is the residual signal generated after the previous projection calculation. By transferring the residual from the previous stage to the next stage calculation, the signal energy and information are effectively transferred between different segments, improving computational efficiency.

[0051] The weighted fusion method is used to smooth the overlapping areas of adjacent projection results to generate continuous projection signals.

[0052] Due to the segmented processing, it is necessary to smoothly transition the processing results of the two adjacent segments in the overlapping area to avoid step mutations. Specifically, the step of generating a continuous projection signal includes: for any data point in the overlapping area, according to the relative position of the data point in the overlapping area, a set of position-related weights is calculated, wherein the closer the data point is to a certain endpoint, the greater the weight of the segmented result where the endpoint is located. The corresponding values ​​of the adjacent projection results in the overlapping area are multiplied by the position-related weights and summed to obtain the fusion value of the data point in the continuous projection signal. For example, the fusion value y 融合 Calculated by the following formula: 融合 =w1·y k +w2·y k+1 ; Among them, y k and y k+1 is the projection result of two adjacent segments, w1 and w2 are position-related weights. Preferably, w1=(1-d / L)2 , w2=(d / L) 2 , where L is the total length of the overlapping region (e.g., 1 meter), and d is the distance from the data point to the end of segment k. Alternatively, in addition to quadratic weighted fusion, linear weighted fusion (equivalent to applying a triangular window) or sine / cosine cross-fusion to ensure constant power can be used to achieve a smooth transition.

[0053] Multi-scale reflection information is extracted and assembled from continuous projection signals to ultimately form preliminary acoustic wave reflection features.

[0054] In this embodiment, the formation interface can be identified by finding the mutation point of the reflection coefficient from the generated continuous projection signals, thereby obtaining preliminary reflection peak position and intensity information, forming a preliminary acoustic wave reflection feature.

[0055] Furthermore, after obtaining the preliminary acoustic wave reflection characteristics, the signal processing unit may also analyze the preliminary acoustic wave reflection characteristics to generate a preliminary pile length estimation.

[0056] Specifically, the system calculates the reflection coefficient at each depth point and searches for a reflection coefficient mutation point (for example, a reflection coefficient value greater than 0.15 and a depth gradient exceeding a preset threshold) to preliminarily identify the formation interface. A reflection coefficient mutation point is a point on the depth-reflection intensity mapping function that satisfies both the amplitude and gradient thresholds. Its physical meaning corresponds to a formation interface with a potential significant acoustic impedance difference. The depth corresponding to the deepest significant formation interface identified is recorded as a preliminary pile length estimate, which is then passed on to the subsequent intelligent hammer parameter optimization step as one of the inputs for calculating the target resonant frequency and dynamic hammer force.

[0057] This embodiment improves the ability to identify and extract weak, effective signals in extremely low signal-to-noise ratio environments. By constructing an atomic dictionary that adaptively changes with depth using real-time estimates of the velocity field v(z) and the dynamic impedance field Z(z), this approach eliminates the fixed mathematical basis functions used in existing technologies, which are disconnected from actual physical processes. This means that the templates used to match signals (i.e., dictionary atoms) can simulate in real time the actual physical changes (such as dispersion and attenuation) that occur when sound waves propagate through complex formations. Therefore, when using this dynamic dictionary for sparse decomposition, it is possible to accurately capture extremely weak, real reflection signals that closely match the physical characteristics of the current depth amidst intense construction noise. This solves the problem of insufficient formation interface recognition due to low signal-to-noise ratios (<-20dB) and mismatch between fixed templates and distorted waveforms, thereby improving recognition rates.

[0058] According to one aspect of this application, although cascade projection improves computational efficiency, it will inevitably produce pseudo-reflection peaks at the boundaries of segment processing (for example, at depths of 4-5 meters, 9-10 meters, etc.), which will seriously interfere with the pile length judgment. Therefore, a cross-modal time delay correlation model between acoustic wave reflection and soil vibration is established, and physical constraints are used to suppress the artifacts generated by mathematical processing. Figure 3 As shown, the modified reflection features are generated, including:

[0059] Detect spectrum mutations of vibration components in multi-source detection signals, and extract time-frequency features corresponding to the spectrum mutations to form vibration mutation features.

[0060] For example, a short-time Fourier transform (100ms window length, 10ms step) can be used to calculate the time-varying spectrum of a vibration signal. By calculating the spectrum distance between adjacent time windows, D(t) = |||S(f, t) - S(f, t - Δt) |||2, the points where the spectrum changes significantly are detected, thereby obtaining a sequence of vibration mutation moments. Here, S(f, t) is the frequency distribution obtained by the short-time Fourier transform at time t, and Δt is the time difference.

[0061] For each reflection peak in the preliminary acoustic wave reflection characteristics, the acoustic-vibration delay correlation between each reflection peak and the vibration mutation characteristics is calculated based on the theoretical time delay model of acoustic wave propagation.

[0062] The acoustic-vibration delay correlation is a quantitative statistical indicator used to measure the strength of the linear relationship between an acoustic reflection event and a subsequent pile-end vibration event after taking into account the theoretical propagation delay. Specifically, Figure 4 As shown, the acoustic vibration delay correlation is calculated, including:

[0063] For each reflection peak, the theoretical propagation delay is calculated by integrating the real-time wave velocity that varies with depth along the depth path of the reflection peak.

[0064] In this embodiment, for each detected reflection peak depth L, the theoretical time delay τ is calculated by considering the integral form of the wave velocity changing with depth. 理论 , to improve the accuracy τ 理论 =2∫0 L (dz / v(z))+t 系统 ; where v(z) is the real-time velocity field that varies with depth, t 系统 is the inherent delay of the system (for example, 3.5ms).

[0065] A delay search window is set for each theoretical propagation delay, wherein the size of the window is proportional to the value of the theoretical propagation delay, so as to adaptively compensate for the uncertainty of the deep signal.

[0066] Preferably, the range of the search window can be set to [τ 理论 -Δτ, τ 理论 +Δτ], where the calculation formula of the window half-width Δτ is: Δτ=0.2·τ 理论 +5ms; the greater the depth and the longer the delay, the wider the search window will be, so that it can better adapt to the situation where the deep signal propagation path is more complex and the uncertainty is higher.

[0067] In the delay search window, the narrowband correlation coefficient is calculated in the characteristic frequency band of each reflection peak and vibration mutation pair, and the narrowband correlation coefficient is energy-weighted integrated, and its maximum value is taken as the acoustic vibration delay correlation.

[0068] For example, first calculate the narrowband correlation coefficient spectrum, then calculate the frequency weight w(f) according to the signal energy distribution, and finally get the correlation coefficient ρ 综合 (τ) = ∑ f w(f) ·ρ(f, τ), and find its maximum value as the final acoustic vibration delay correlation ρ max . Used to describe how to quantitatively calculate the correlation strength between acoustic wave reflections and vibration mutations. Where ρ(f, τ) is the narrowband correlation coefficient at frequency f and time delay τ. Optionally, in addition to calculating the energy-weighted correlation coefficient, the phase consistency or mutual information of the two signals in the time-frequency domain can also be calculated as an alternative or supplementary indicator to measure the degree of correlation.

[0069] Based on the calculated acoustic vibration time delay correlation, the intensity of each reflection peak in the preliminary acoustic wave reflection feature is weighted and corrected to suppress false features and enhance true features to obtain a corrected reflection feature.

[0070] Specifically, the intensity of each reflection peak in the preliminary acoustic wave reflection feature is weighted and corrected to obtain a corrected reflection feature, including:

[0071] For each reflection peak in the preliminary acoustic wave reflection characteristics, the pseudo peak discrimination index is calculated by comprehensively considering the acoustic vibration time delay correlation, spectrum bandwidth and phase continuity of each reflection peak to identify the pseudo peaks generated at the cascade processing boundary.

[0072] In this embodiment, a pseudo peak is a peak that appears in the signal processing result and is similar in form to a real reflection, but is essentially caused by mathematical operations (such as boundary effects and algorithmic artifacts) and does not correspond to a real physical event. For example, a reflection peak satisfies the following conditions at the same time: max If two or more of the following three conditions are met, the peak is judged to be a pseudo peak:

[0073] Through the S-type nonlinear weight function, which receives the acoustic vibration time delay correlation as input and applies a penalty factor in combination with the pseudo peak discrimination index, an adaptive correction weight is generated for each reflection peak.

[0074] Preferably, the S-shaped weight function W(ρ) can be designed as: W(ρ) = 1 / (1+exp(-k(ρ-ρ0))); where ρ is the input acoustic vibration time delay correlation, the parameter k controls the transition steepness of the function curve (e.g., k=10), and ρ0 is the turning point (e.g., ρ0=0.5). For reflection peaks identified as pseudo-peaks, an additional penalty factor (e.g., 0.1) can be applied to their weight to further enhance the suppression. Optionally, the S-shaped nonlinear weight function can be replaced by other functions with similar monotonically increasing characteristics, such as the hyperbolic tangent function (tanh) or a piecewise linear function customized based on empirical data.

[0075] The original intensity of each reflection peak in the preliminary acoustic wave reflection feature is multiplied by the adaptive correction weight, and weighted correction is performed to obtain the corrected reflection feature.

[0076] For example, the corrected reflection intensity R 修正 (z) can be calculated as follows: 修正 (z)=R 原始 (z) [1+α (W(ρ)-0.5)]; where, R 原始 (z) is the original reflection intensity, and α is the enhancement factor (e.g., α = 2.5). This formula can enhance true reflection peaks with high correlation (W(ρ) → 1) while effectively suppressing false peaks with low correlation (W(ρ) → 0).

[0077] It should be noted that in the process of weighted correction of each reflection peak, the calculated acoustic vibration delay correlation ρ max After completing the correlation verification of all preliminary reflection peaks, these correlation values ​​are compiled into a data structure corresponding to each reflection peak, namely the correlation weight matrix. This matrix will be passed to the subsequent multimodal feature fusion step as the direct basis for calculating the confidence weight of each pile length estimate.

[0078] This embodiment introduces an independent, physically constrained cross-validation dimension for verifying the authenticity of acoustic detection results. This dimension accurately identifies and suppresses pseudo-reflection peaks generated by the signal processing algorithm itself. This approach exploits the physical coupling effect that true physical interface reflections are inevitably accompanied by a predictable vibration response of the soil at the pile end. By calculating the time-delay correlation coefficient between the acoustic reflection peak and the subsequent vibration mutation event, the system obtains an objective, quantitative credibility indicator: true reflection peaks have a strong correlation (e.g., ρ > 0.7), while algorithmic artifacts have almost no correlation (e.g., ρ < 0.3). Based on this, a weighted correction is applied to the reflection intensity, selectively enhancing true signals and suppressing false signals. This approach not only solves the boundary pseudo-peak problem introduced by cascaded projection processing (pseudo-peak identification accuracy > 95%), but also addresses the inherent flaws of single-source detection, which are difficult to distinguish and susceptible to algorithmic interference, thereby improving the reliability of detection results.

[0079] like Figure 2 As shown, according to one aspect of the present application, calculating the estimated pile length includes the following steps:

[0080] The corrected reflection characteristics are analyzed, the attenuation state of the signal energy is evaluated, and the energy attenuation characteristics are obtained.

[0081] For example, by calculating the sound wave energy attenuation curve E 声 (z) and vibration energy curve E 振 (z), evaluate the energy decay rate, and identify the weak signal depth interval with insufficient energy (e.g., E 振 <0.1g 2 area).

[0082] The energy attenuation characteristics are combined with the preliminary pile length estimation to optimize and generate intelligent hammer control parameters.

[0083] In this embodiment, the intelligent hammer control parameters include at least:

[0084] Optimize hammer frequency: Based on the preliminary pile length estimate and the preset geotechnical physics model, the target resonant frequency in the target depth range is estimated. According to the target resonant frequency and energy attenuation characteristics, the optimized hammer frequency is calculated by sweeping around the target resonant frequency.

[0085] Specifically, for the target depth L target Calculate the basic resonance frequency f0=v(L target ) / (4L target ), and taking soil damping into account to correct the target resonance frequency f 共振 Then, design a frequency modulation strategy, for example, to make the hammer frequency f 锤 In f 共振 The sweep is performed within the range of ±10% to effectively stimulate the resonance response in a wide frequency band.

[0086] Adaptive hammering interval: Adaptively adjust and determine the adaptive hammering interval based on the signal energy strength reflected by the energy attenuation characteristics.

[0087] For example, the vibration energy E 振 The hammer interval T is calculated by the ratio of the reference energy E0 间隔 :T 间隔 =T0+ΔT×(1-E 振 / E0); where T0 is the base interval (e.g., 2 seconds) and ΔT is the maximum adjustment (e.g., 0.5 seconds). This strategy results in a longer hammering interval when the signal energy is weaker, thereby accumulating more energy for the next excitation.

[0088] Dynamic hammer force: Taking into account the target depth and energy attenuation characteristics, the dynamic hammer force is dynamically adjusted and determined.

[0089] For example, the hammer force F can be adjusted by a coefficient that combines the depth effect and the energy attenuation effect: F = F 标准 ×[1+β 深度 ×(L / 10)+β 衰减 ×(1-E / E0)]; where F 标准 is the standard hammer force, L is the target depth, β 深度 and β 衰减 are adjustment coefficients (e.g., 0.3 and 0.5, respectively). This strategy effectively enhances deep signals while avoiding signal overload in shallow formations.

[0090] According to the intelligent hammer control parameters, the directional hammer excitation is performed by the electronically controlled hammer system, and the enhanced signal after resonance enhancement is collected synchronously.

[0091] In this embodiment, the intelligent hammer control parameters (including optimized hammer frequency, adaptive hammer interval, and dynamic hammer force) are loaded into the PLC (Programmable Logic Controller) of the electronically controlled hammer system. The electronically controlled hammer system uses a servo motor to precisely control the hammer frequency (for example, with a control accuracy of ±0.1 Hz) and a hydraulic system to precisely adjust the hammer force (for example, with a control accuracy of ±5%), thereby achieving high-fidelity physical execution of the enhanced sexual stimulation command.

[0092] Within a critical time window after each enhanced hammering (for example, before and after the expected response time calculated based on the preliminary estimate of the target depth), the signal processing unit synchronously collects sound and vibration signals at a higher sampling rate (for example, 50kHz). The response data collected in real time is subjected to fast Fourier transform (FFT) analysis, the purpose of which is to monitor in real time whether the target resonance peak appears as expected and its intensity changes. Furthermore, the system calculates two key enhancement effect evaluation indicators: signal energy ratio G: G=E 增强 / E 原始 ; Spectral peak ratio P: P = P 增强 / P 原始 Among them, E 增强 and P 增强 is the energy and key spectrum peak of the signal after this excitation, E 原始 and P 原始 is the corresponding baseline value before enhancement.

[0093] The system makes an immediate decision based on the enhancement effect evaluation metrics obtained in the previous step. If the evaluation results show G < 1.5 or P < 2, the system determines that the actual effect of this enhancement stimulation has failed to meet the preset enhancement target. In this case, the system automatically fine-tunes the hammering control parameters (for example, fine-tuning the hammering frequency by ±2Hz or the hammering force by ±10%) and immediately instructs the electronically controlled hammering system to perform a second stimulation. This evaluation-decision-fine-tuning-re-stimulation process constitutes a fast-response closed-loop feedback control. Closed-loop feedback control is a type of control system in which the system output (signal enhancement effect) is continuously measured and fed back to the input. It is compared with the desired target, and the difference is used to adjust the control action (hammering parameters) so that the system output automatically approaches the desired target. This cycle repeats until the enhancement effect is detected (i.e., both G and P exceed their thresholds) or until a preset maximum number of attempts is reached. Alternatively, in addition to simple threshold determination and fixed-step fine-tuning, more sophisticated automatic control algorithms such as proportional-integral-derivative (PID) controllers can be employed. The PID controller can dynamically calculate the optimal hammer parameter adjustment amount based on the error size, error accumulation and error change rate between the enhancement effect and the target, thereby achieving faster and more stable closed-loop control.

[0094] After closed-loop control confirms that the enhancement effect meets the required standards, the system performs a complete, long-term signal acquisition to obtain the final enhanced acoustic and vibration signals. It is important to note that the preprocessing performed on the enhanced signals is specially tailored. Specifically, the filtering step employs a protective bandpass strategy to prevent valuable newly excited signal components within the target resonant frequency band (e.g., ±20 Hz of the target resonant frequency) from being accidentally attenuated or eliminated during noise filtering. The final output is the enhanced reflection and vibration signatures.

[0095] The final pile length estimate is calculated by combining the modified reflection characteristics, enhanced signals, and the correlation weight matrix inherited from the acoustic-vibration coupling analysis with multimodal fusion using an acoustic-vibration resonance model. Multimodal fusion is an information processing technology that aims to obtain more accurate, robust, and comprehensive conclusions than any single modality can provide by integrating data or analysis results from multiple different sensors or different physical dimensions (modalities).

[0096] This embodiment transforms the traditional passive detection paradigm into an active, intelligent detection system with closed-loop feedback control, effectively resolving the detection blind spot problem caused by severe signal attenuation at depth. Unlike existing technologies that rely on fixed-parameter hammering, this embodiment evaluates energy attenuation based on preliminary detection results, proactively identifying target depth intervals with weak signals. Rather than abandoning the detection process, it instead precisely calculates the optimal hammering parameters (frequency, interval, and force) for energy focusing based on the resonance characteristics of that depth. This customized excitation is then executed by commanding the electronically controlled hammering system, achieving directional, resonant enhancement of the signal at the target depth. This closed-loop control system of detection-evaluation-optimization-re-excitation improves detection success rates and expands the effective operating range of dynamic detection methods.

[0097] According to one aspect of the present application, a final pile length estimate is calculated by performing multimodal fusion using an acoustic-vibration resonance model, including:

[0098] Based on the modified reflection characteristics and enhanced signals, a multi-source pile length estimation value set containing independent physical meanings is calculated and generated through the acoustic wave reflection, vibration mutation and acoustic vibration resonance relationship model. For example, the set may include: the estimated value L based on the acoustic wave reflection delay 声 ; Estimated value L based on vibration mutation delay 振 ; Based on the estimated value of the resonant frequency L 共振 ; Estimated value L based on energy decay characteristics 能量 .

[0099] According to the correlation weight matrix inherited from the acoustic-vibration coupling analysis, the corresponding credibility weight is calculated for each estimated value in the multi-source pile length estimation value set to form a credibility weight set. Preferably, the credibility weight wi According to the acoustic vibration delay correlation coefficient ρ i Calculated: w i =ρ i 2 / ∑ j ρ j 2 ; where estimates with higher correlations are given greater weights.

[0100] The confidence weight set is used to perform weighted least square fusion on the multi-source pile length estimation value set to obtain the final pile length estimate, and the detection confidence is calculated based on the residual after weighted fusion. The final pile length estimate L final The calculation formula is: L final =∑ i w i L i At the same time, a weighted standard deviation σ can be calculated as a measure of detection confidence, providing reliable data support for engineering decisions. Alternatively, in addition to using weighted least squares, a Kalman filter framework can be employed, treating the pile length estimate as a state variable that is updated with each hammer blow. Alternatively, Bayesian inference methods can be employed, treating each estimate as evidence with uncertainty and fusing them probabilistically.

[0101] This embodiment improves the accuracy and confidence of the final pile length estimate by intelligently fusing information from multiple sources and multiple physical dimensions. It is not satisfied with obtaining an estimate from a single physical phenomenon (such as acoustic wave transit time), but simultaneously calculates multiple independent pile length estimates from multiple physical dimensions such as acoustic wave reflection, vibration mutation, and acoustic resonance. Instead of simply averaging these estimates, a correlation weight matrix is ​​used as an objective and quantitative credibility indicator. The more correlated the estimate, the higher the physical authenticity is considered to be, and therefore it is given a greater weight in the final weighted fusion calculation. This allows the final result to be inclined to the most reliable physical evidence to the greatest extent possible, effectively avoiding the systematic bias and random errors of a single detection method, improving the final pile length estimation accuracy, and outputting a clear detection confidence level, providing data support for engineering decision-making.

[0102] In a specific embodiment, the acoustic vibration time delay correlation is used to distinguish between reflection peaks and pseudo reflection peaks generated by algorithm processing, and enhance and suppress them respectively. In this case, the following initial conditions are set: in the preliminary acoustic wave reflection characteristics, two reflection peaks from different depths are focused on: Reflection peak A: a real formation interface reflection at a deeper depth, with a depth of L A =15.2 m, initial reflection intensity R 原始，A =0.8. Reflection peak B: A suspected pseudo-peak located at the boundary of the cascade projection processing (9-10 meters), with a depth of L B =9.5 m, initial reflection intensity R 原始，B =0.5. Available data also include: real-time wave velocity field v(z), and vibration mutation time series extracted from vibration signals. B = 9.5 m reflection peak B, by integrating the real-time wave velocity field v(z) along the depth path, the two-way propagation time of the sound wave is calculated to be 18.0 ms. Considering the inherent delay t 系统 =3.5ms, the final theoretical propagation delay τ 理论，B is: 理论，B =18.0ms+3.5ms=21.5ms.

[0103] According to the theoretical delay τ 理论，B , calculate the half-width Δτ of the adaptive search window B :ΔτB=0.2×τ 理论，B +5ms=0.2×21.5ms+5ms=4.3ms+5ms=9.3ms; therefore, the delay search window used to find the corresponding vibration event is [21.5ms-9.3ms, 21.5ms+9.3ms], that is, [12.2ms, 30.8ms]. Within this delay search window, the characteristics of reflection peak B are correlated with the vibration mutation time sequence. After calculation, it was found that there is no significant vibration event corresponding to this reflection peak in terms of time and physical cause. The maximum energy-weighted comprehensive correlation coefficient ρ is finally obtained. max，B Only 0.25. Evaluate reflection peak B based on pseudo peak discrimination index: Index 1 (correlation): ρ max，B =0.25, which is less than the threshold of 0.3, and meets this condition. Indicator 2 (spectral bandwidth): After spectrum analysis, the bandwidth of the peak is 110Hz, which is greater than the threshold of 100Hz, and meets this condition. Indicator 3 (phase continuity): After phase analysis, its phase discontinuity is large, and meets this condition. Since multiple pseudo-peak discrimination conditions are met, the reflection peak B is highly confident that it is a pseudo-peak generated during the processing. The S-type nonlinear weight function W(ρ) is used to calculate its correction weight, where the parameters k=10 and ρ0=0.5. W(ρ B )=W(0.25)=1 / (1+exp(-10(0.25-0.5)))=1 / (1+exp(2.5))≈0.076; After obtaining the weight, calculate the corrected reflection intensity R according to the correction formula 修正，B , where the enhancement coefficient α=2.5: R 修正，B =R 原始，B ×[1+α×(W(ρ B)-0.5)]=0.5×[1+2.5×(0.076-0.5)]=0.5×[1-1.06]=-0.03. The calculation results show that the original intensity 0.5 of reflection peak B is effectively suppressed to near zero and is thus essentially eliminated in the corrected reflection feature.

[0104] In contrast, for a depth of L A =15.2 m, the same processing flow is performed on the real reflection peak A. Assuming the calculated theoretical delay τ 理论，A =35.0ms. Window half-width Δτ A =0.2·35.0ms+5ms=12.0ms, the search window is [23.0ms, 47.0ms]. A corresponding strong pile end vibration mutation event is found within this window, and the maximum correlation coefficient ρ is calculated. max，A =0.85. The correlation coefficient is much greater than 0.7, which does not meet the pseudo peak discrimination condition and is confirmed as a true reflection peak. Weight calculation and weighted correction are performed: W(ρ A )=W(0.85)=1 / (1+exp(-10(0.85-0.5)))=1 / (1+exp(-3.5))≈0.97; R 修正，A =R 原始，A ×[1+α×(W(ρ A )-0.5)] =0.8×[1+2.5×(0.97-0.5)]=0.8×[1+1.175]=1.74. The calculation results show that the original intensity of the true reflection peak A of 0.8 is enhanced to 1.74, making it more prominent in the characteristic spectrum. It can be seen that this embodiment can distinguish between real physical reflections and algorithm artifacts through quantitative physical correlation analysis, and impose completely opposite correction effects on the two: suppressing the pseudo peak by more than 90% and enhancing the real signal by more than 100%. The final generated corrected reflection feature is cleaner and more reliable, laying a solid foundation for the subsequent high-precision pile length estimation.

[0105] According to one aspect of this application, in actual engineering, information such as drill hole histograms and test data directly obtained from geological survey units has diverse formats and discrete data points, and cannot be directly used for the adaptive analysis required by this application that requires continuous physical parameter input. Therefore, it is necessary to obtain prior information on the formation, including:

[0106] The geological survey drilling data, standard penetration test data and wave velocity test data are read, and through interpolation and data fusion processing, a stratum reference model containing expected wave velocity, density and impedance parameters is constructed as the prior information of the stratum.

[0107] In this embodiment, the stratigraphic reference model is a structured, digital, three-dimensional geological volume data model. It stores the expected distribution of geophysical parameters (such as density and wave velocity) within the target area in the form of a continuous data field by spatially interpolating and fusing discrete survey data. The signal processing unit is first configured to read raw geological survey data from multiple sources. This data may include, but is not limited to: geological survey borehole data: typically, borehole histograms or textual descriptions describing the lithology, thickness, color, and state of soil layers at different depths at a specific location; standard penetration test (SPT) data: providing the number of standard penetration hammer blows (N) at different depths, which can be used to indirectly assess soil density and strength; and wave velocity test data: acoustic or shear wave test results from on-site or nearby sites, providing P-wave or S-wave velocities at specific depths. After reading the discrete borehole data, the signal processing unit uses a spatial interpolation algorithm to expand this point- or line-level information into a continuous three-dimensional stratigraphic model. Preferably, geostatistical methods such as kriging interpolation or inverse distance weighted interpolation are employed. This step aims to infer the distribution of geotechnical layers at any location and depth within the entire area where the engineering piles are located based on limited borehole data. Alternatively, in addition to kriging interpolation and inverse distance weighted interpolation, other modeling methods commonly used in geostatistics, such as sequential Gaussian simulation or triangulated irregular network (TIN) models, can be used to accommodate different types and distributions of geological data.

[0108] Furthermore, based on the three-dimensional formation model, the signal processing unit further integrates various test data to calculate key physical parameters for each grid point in the model. Specifically, the unit can utilize empirical formulas or data fusion algorithms in geotechnical engineering. For example, based on the N value and geotechnical type from the standard penetration test, it can estimate the density ρ and initial wave velocity v at that point. If direct wave velocity test data is available, it is preferred and locally corrected. On this basis, the acoustic impedance Z = ρ × v can be further calculated. After the above processing, a structured formation reference model is ultimately output that can be directly queried and called by computer programs. This model stores physical parameters such as expected wave velocity, density, and acoustic impedance that continuously vary with spatial position and depth throughout the target area in the form of a three-dimensional data field. As a standardized prior information about the formation, this model provides the initial, depth-varying physical parameter input for the generation of the depth-adaptive atomic dictionary, forming the basis for implementing physics-driven adaptive analysis.

[0109] According to one aspect of the present application, the parameters in a general theoretical physical model (for example, parameters reflecting the damping characteristics of the soil layer) will vary due to differences in the soil quality of a specific construction site. Directly using theoretical values ​​will introduce systematic errors into the final pile length estimation. Therefore, by using the on-site measured high signal-to-noise ratio data to calibrate the model online, the model's case applicability and final estimation accuracy are improved. Online calibration is a model parameter optimization process that uses a high-reliability data subset collected in real time at the actual operation site to adjust the parameters of a general theoretical model so that it can better reflect the personalized physical characteristics of the current specific object. Specifically, before fusing the information before and after the enhancement, it also includes:

[0110] Combining the modified reflection characteristics and vibration mutation characteristics, the preset acoustic-vibration resonance relationship is calibrated by the least squares method to generate a calibrated acoustic-vibration resonance model.

[0111] In this embodiment, the signal processing unit screens and prepares a high-confidence data set for model calibration from the existing analysis results. Preferably, the high-confidence data set comes from the detection results of the shallow part of the pile body (for example, less than 10 meters deep). The reason for selecting shallow data is that the sound wave and vibration signal at this depth has a high signal-to-noise ratio and low energy attenuation. The position of the corrected reflection feature identified is accurate and can be regarded as the real label or anchor point of this calibration process. The data set specifically includes: the accurate depth L of each real reflection peak in the shallow part, the corresponding vibration mutation feature, and the resonance frequency f resonance analyzed based on the vibration signal. In some large projects, one or more special test piles can be used for model calibration. The length and geological conditions of these test piles are precisely controlled or measured, and their detection data can provide a more reliable and unified calibration model for the subsequent detection of all engineering piles.

[0112] The least squares method is used to calibrate the parameters. The signal processing unit uses a preset acoustic resonance relationship, which establishes the pile length L, wave velocity c, and resonance frequency f. 共振 And the relationship between parameters such as acoustic vibration delay correlation ρ. An exemplary relationship is: L=(c / 4f 共振 )×(1-β×ρ); In this formula, the parameter β is an undetermined coefficient that reflects the influence of soil on the acoustic-vibration coupling effect, and its theoretical value is difficult to accurately estimate. Therefore, β is regarded as an unknown number, and a high-confidence calibration data set is used to fit it through the least squares method to solve the optimal value of the parameter β that best matches the current site conditions. After completing the parameter solution, the system will obtain a calibrated acoustic-vibration resonance model that includes the optimized parameter β that has been calibrated online. The model is then used in the multimodal feature fusion step as a method for calculating one of the pile length estimates (i.e., the estimated value L based on the resonance frequency). 共振This ensures the accuracy of the estimated value of the path, thus providing a strong guarantee for the overall accuracy of the final fusion result.

[0113] According to one aspect of this application, at a pile foundation construction site, the raw signals collected by the sensor will inevitably be affected by strong environmental noise interference and its own hardware limitations. Without effective preprocessing, these defects will seriously affect the accuracy of all subsequent analysis results. Specifically, the raw signal is refined preprocessed, including:

[0114] After receiving the time-synchronized raw signal set acquired by the sensor assembly, the signal processing unit first performs a comprehensive quality assessment of the signal for each channel.

[0115] Specifically, the evaluation process includes calculating key quality indicators for each signal channel, such as: signal-to-noise ratio (SNR): evaluating the relative strength of the effective component and noise in the signal; zero-drift: detecting whether the signal baseline undergoes an unexpected slow shift over time; saturation: checking whether the signal amplitude exceeds the upper limit of the sensor's range, resulting in the top or bottom being flattened. Based on the above evaluation indicators, the system automatically identifies and marks various abnormal conditions, including but not limited to sensor hardware failures, signal saturation caused by a single hammer impact, and abnormal spikes caused by sudden electromagnetic interference. Two key intermediate data are output: a signal quality assessment report and a valid signal period marker. These two data will be used to guide subsequent filtering and analysis steps, so that the algorithm only operates on high-quality, valid signal segments, avoiding interference of contaminated data on the final results.

[0116] After completing the quality assessment, the signal processing unit performs adaptive noise filtering and baseline correction on the valid signal segment according to the valid signal period mark.

[0117] Unlike traditional filters that use fixed parameters, this embodiment recognizes that construction site noise (such as power frequency interference and other mechanical vibration noise) is not constant in the time domain. Therefore, an adaptive filtering strategy is crucial. This embodiment employs an adaptive filter based on the local statistical characteristics of the signal. This filter analyzes the statistical characteristics of the signal (such as variance and kurtosis) in real time within different time windows and automatically adjusts its filtering parameters (such as filter order and cutoff frequency) based on these statistical characteristics. This allows the filter to effectively filter out various types of noise while maximally preserving useful weak signal details generated by reflections from strata interfaces. An adaptive filter is a digital filter whose transfer function (filter parameters) can be self-adjusted based on the statistical characteristics of the input signal through an optimization algorithm, enabling it to achieve optimal filtering in non-stationary noise environments. After filtering, the signal undergoes zero drift compensation and baseline correction to accurately calibrate the signal baseline to zero. After all the above preprocessing steps, the system ultimately outputs two high-quality data streams: a preprocessed acoustic signal and a preprocessed vibration signal. These two signals serve as direct input for all subsequent core analysis. In complex scenarios with multiple independent noise sources, blind source separation (BSS) techniques, such as independent component analysis (ICA), can be used to separate and remove the noise. Alternatively, a deep learning denoising autoencoder model can be pre-trained for nonlinear signal denoising.

[0118] According to one aspect of the present application, after extracting preliminary sound wave reflection features, the method further includes:

[0119] The preliminary acoustic wave reflection characteristics are analyzed to identify the reflection coefficient mutation point, and the depth of the deepest significant mutation point is used as the preliminary pile length estimation.

[0120] In this embodiment, the continuous projection signal is converted into an intuitive reflection intensity function R(z) that varies with depth. Specifically, the signal processing unit extracts the signal amplitude corresponding to each depth point z from the continuous projection signal y(z), and this amplitude is regarded as the reflection wave amplitude A at that depth. 反射 (z). At the same time, the initial hammer signal amplitude measured at the pile top is obtained as the reference incident wave amplitude A 入射 The reflection coefficient of each depth point is calculated by the following formula, thereby constructing a continuous depth-reflection intensity mapping function R(z): R(z)=|A 入射 ∣∣A 反射(z)|; This function R(z) intuitively reflects the reflection of acoustic wave energy at different depths. In order to accurately identify the reflection caused by significant stratum interfaces from the continuous R(z) function, a screening method based on the dual criteria of amplitude and gradient is adopted. Using only the amplitude criterion may misjudge some wider, non-mutated reflection areas as interfaces; using only the gradient criterion may be sensitive to noise. The combination of the dual criteria enables the identified mutation points to have both sufficient reflection energy and clear interface characteristics. The specific implementation is as follows: the signal processing unit traverses all points on the R(z) curve. A point is only judged as a significant mutation point if it meets the following two conditions at the same time: the reflection intensity value of the point must be greater than the preset minimum reflection threshold, that is, R(z)>0.15. It aims to filter out invalid reflections caused by minor stratum heterogeneity or background noise. The depth gradient of the reflection intensity of the point must be greater than the preset gradient threshold, that is, ΞzΞR>threshold G. The depth gradient ΞR / Ξz can be numerically calculated by applying the first-order finite difference method to the R(z) function. The goal is to ensure that the selected point is a sharp, abrupt peak rather than a gentle energy bump. Alternatively, more complex peak detection algorithms from the field of signal processing can be employed. For example, continuous wavelet transforms can be used to find ridges in scale maps at different scales, or the top-hat transform from mathematical morphology can be applied to highlight sharp peaks, achieving more robust abrupt point detection.

[0121] After filtering out all significant mutation points through the above dual criteria, the following algorithm is executed to generate the final preliminary pile length estimate. Set a variable L preliminary And set its initial value to 0. Traverse all the data points that have been identified as significant mutation points. During the traversal process, the depth z corresponding to each significant mutation point is point With variable L preliminary If z point >L preliminary , then update L preliminary =z point After the traversal is completed, the variable L preliminary The value of is the maximum depth of all significant mutation points. This value is formally output as a preliminary pile length estimate. It is passed to the subsequent energy adaptive mutual excitation enhancement module to calculate and optimize the intelligent hammering parameters.

[0122] According to one aspect of the present application, while generating the modified reflection feature, the method further includes:

[0123] The magnitude of the acoustic vibration time delay correlation calculated for each preliminary acoustic wave reflection feature is compiled into a correlation weight matrix.

[0124] In this embodiment, the correlation weight matrix is ​​a structured data array that stores a corresponding acoustic vibration time delay correlation coefficient value that quantifies the physical authenticity of each initially identified reflection peak. It is functionally equivalent to a credibility lookup table that can be queried by subsequent fusion algorithms. Specifically, the correlation weight matrix is ​​a two-dimensional data structure with a size of N×2, where N is the total number of reflection peaks identified from the preliminary acoustic wave reflection characteristics. Matrix dimensions: N rows, 2 columns. Element meaning: The first column (index / key): stores the unique identifier of each reflection peak; preferably, the depth value z corresponding to each reflection peak is used. i (where i = 1, 2, ..., N) as its unique identifier; the second column (value): stores the reflection peak corresponding to the identifier in the first column, and the final acoustic vibration delay correlation coefficient ρ max，i Alternatively, in addition to using an N×2 two-dimensional matrix, a hash table (HashMap) or dictionary object can be used. This key-value pair structure, with reflection peak depth as the key and correlation coefficient as the value, generally offers higher computational efficiency when querying data.

[0125] The signal processing unit executes the following algorithm to construct the matrix: Get the depth list {z1, z2, ..., z N On this basis, create an empty matrix of size N×2 and name it CorrWeightMatrix. The system starts a loop and traverses from i=1 to N. In each loop, the system performs the following operations: i Fill in the i-th row and the first column of the matrix, that is, CorrWeightMatrix[i,0]=z i ; According to the depth z i As an index, query and retrieve the acoustic vibration delay correlation coefficient ρ corresponding to the deep reflection peak max，i ; The retrieved correlation coefficient value ρ max，i Fill in the i-th row and the second column of the matrix, that is, CorrWeightMatrix[i,1]=ρ max，i When the loop is complete, the CorrWeightMatrix is ​​fully populated. This matrix, as structured data, is formally output as a correlation weight matrix. The correlation weight matrix is ​​then passed to the multimodal fusion module to calculate the corresponding credibility weights for different pile length estimates, ensuring a complete closed-loop data flow.

[0126] The preferred embodiments of the present invention are described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the scope of protection of the present invention.

Claims

1. A method for estimating the length of an engineering pile, characterized in that: include: Acquire prior information of the stratum reflecting the pile hammer's impact response, as well as multi-source detection signals including acoustic and vibration components; Based on prior information of the formation, the acoustic wave components are adaptively analyzed to extract preliminary acoustic wave reflection characteristics; Using vibration components, preliminary acoustic wave reflection features are correlated and verified, false features are suppressed, and corrected reflection features are generated; Evaluate the signal quality of the corrected reflection characteristics, generate intelligent hammer control instructions based on this, execute enhanced signal excitation, fuse the information before and after enhancement, and calculate the estimated pile length.

2. The method according to claim 1, characterized in that The estimated pile length is calculated, including: According to the modified reflection characteristics, the attenuation state of the signal energy is evaluated to obtain the energy attenuation characteristics; Combining energy attenuation characteristics with preset preliminary pile length estimation, the intelligent hammer control parameters are optimized and generated; Execute directional hammer excitation according to intelligent hammer control parameters and simultaneously collect enhanced signals after resonance enhancement; The final pile length estimate is calculated by comprehensively modifying the reflection characteristics, enhancing the signal and the preset correlation weight matrix inherited from the acoustic-vibration coupling analysis, and performing multimodal fusion through the acoustic-vibration resonance model.

3. The method according to claim 2, characterized in that Intelligent hammer control parameters include at least the following three parameters: Based on the preliminary pile length estimate and the preset geotechnical physics model, the target resonance frequency in the target depth range is estimated. Combined with the energy attenuation characteristics, the optimized hammer frequency is calculated by sweeping around the target resonance frequency. Adaptively adjust and determine the adaptive hammering interval according to the signal energy strength reflected by the energy attenuation characteristics; Taking into account the target depth and energy attenuation characteristics, the dynamic hammer force is dynamically adjusted and determined.

4. The method according to claim 2, characterized in that The final pile length estimate is calculated by multi-modal fusion using the acoustic-vibration resonance model, including: Based on the modified reflection characteristics and enhanced signals, a multi-source pile length estimation value set with independent physical meanings is calculated and generated through the acoustic wave reflection, vibration mutation and acoustic-vibration resonance relationship model. According to the correlation weight matrix, the corresponding credibility weight is calculated for each estimated value in the multi-source pile length estimation value set to form a credibility weight set; The credibility weight set is used to perform weighted least squares fusion on the multi-source pile length estimation value set to obtain the final pile length estimate.

5. The method according to claim 1, wherein Generates corrected reflection features, including: Detect the spectrum mutation of the vibration component and extract its corresponding time-frequency features to form the vibration mutation feature; For each reflection peak in the preliminary acoustic wave reflection feature, calculate the acoustic vibration time delay correlation between it and the vibration mutation feature; Based on the magnitude of the acoustic vibration time delay correlation, the intensity of each reflection peak in the preliminary acoustic wave reflection feature is weighted and corrected to suppress false features and enhance true features to obtain the corrected reflection feature.

6. The method according to claim 5, characterized in that Calculate the acoustic vibration delay correlation, including: For each reflection peak, the theoretical propagation delay is calculated by integrating the real-time wave velocity that varies with depth along its depth path; A delay search window is set for each theoretical propagation delay, wherein the size of the delay search window is proportional to the value of the theoretical propagation delay; In the delay search window, the narrowband correlation coefficient is calculated in the characteristic frequency band of each reflection peak and vibration mutation pair, and energy-weighted synthesis is performed on it. The maximum value of the energy-weighted synthesis result is taken as the acoustic-vibration delay correlation.

7. The method according to claim 5, characterized in that The intensity of each reflection peak in the preliminary acoustic wave reflection feature is weighted and corrected to obtain a corrected reflection feature, including: For each reflection peak in the preliminary acoustic wave reflection characteristics, the pseudo peak discrimination index is calculated by combining its acoustic vibration time delay correlation, spectrum bandwidth and phase continuity to identify the pseudo peaks generated at the cascade processing boundary; Based on the correlation between pseudo peaks and acoustic vibration delay, an adaptive correction weight is generated for each reflection peak through an S-type nonlinear weight function. The original intensity of each reflection peak in the preliminary acoustic wave reflection feature is multiplied by the adaptive correction weight to obtain the corrected reflection feature.

8. The method according to claim 1, characterized in that Extract preliminary sound wave reflection features, including: Segment the acoustic wave components into segmented signal sequences with overlapping ends along the depth dimension; A local projection operator is configured for each segment of the signal, and cascade projection calculations are performed on the segmented signal sequence in sequence to obtain projection results; wherein the projection calculation of the latter segment of the signal includes the processing residual generated by the projection calculation of the previous segment; The weighted fusion method is used to smooth the overlapping areas of adjacent projection results to generate continuous projection signals; Multi-scale reflection information is extracted and assembled from continuous projection signals to form preliminary acoustic wave reflection features.

9. The method according to claim 8, characterized in that Generates continuous projection signals, including: For any data point in the overlapping area, a position-related weight is calculated based on the relative position of the data point in the overlapping area; the closer the data point is to the preset endpoint, the greater the weight of the segment result where the endpoint is located; The corresponding values ​​of adjacent projection results in the overlapping area are multiplied by the position-related weights and summed to obtain the fusion value of the data point in the continuous projection signal.

10. A system for estimating the length of engineering piles, characterized in that: include: a sensor assembly configured to acquire multi-source detection signals reflecting a pile hammer striking response; an electronically controlled hammering system configured to receive control instructions and execute a hammering action with adjustable parameters; A signal processing unit is operatively coupled to the sensor assembly and the electronically controlled hammer system, the signal processing unit being configured to: Extracting preliminary acoustic wave reflection characteristics based on acoustic wave components in multi-source detection signals obtained from the sensor assembly and pre-stored formation prior information; The vibration components in the multi-source detection signals are used to correlate and verify the preliminary acoustic wave reflection characteristics to generate the revised reflection characteristics; Evaluate and modify the reflection characteristics, and generate intelligent hammer control instructions based on them and send them to the electronically controlled hammer system to perform enhanced hammer excitation; The information before and after the enhanced hammering excitation is integrated to calculate the estimated final pile length.

Images