Ultrasonic phased array nondestructive testing method for strength of building steel structure

Abstract

The invention relates to the technical field of ultrasonic detection, in particular to a building steel structure strength ultrasonic phased array nondestructive testing method which comprises the following steps: acquiring morphology data of the surface of a building steel structure; analyzing coating thickness distribution and weld surface roughness of the surface of the building steel structure, and constructing a surface information matrix; automatically adjusting ultrasonic wave beam parameters of the ultrasonic phased array by using the self-adaptive parameter regulation and control model, and performing ultrasonic phased array detection to obtain a preliminary ultrasonic echo signal; adopting an echo signal compensation model to perform secondary compensation on the preliminary ultrasonic echo signal through the difference between the reference signal and the preliminary ultrasonic echo signal to obtain an optimized ultrasonic echo signal; fusing the optimized ultrasonic echo signal, the morphology data and the surface information matrix, and performing defect detection on the building steel structure by using the building steel structure detection model; evaluating a detection result, and dividing a confidence level; and carrying out local re-detection and parameter adjustment on the low-confidence region to realize closed-loop optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic testing, and particularly to an ultrasonic phased array non-destructive testing method for the strength of building steel structures. Background Art

[0002] Building steel structures are widely used in modern buildings due to their good strength, durability and seismic performance. However, over time, steel structures may be damaged due to long-term exposure to harsh environments or external impacts. Damage to steel structures, such as cracks, corrosion, fatigue and welding defects, may affect the load-bearing capacity and safety of the structure. Therefore, regular non-destructive testing of building steel structures to ensure structural safety and reliability is an essential and important part of modern building maintenance and management.

[0003] As an advanced non-destructive testing method, ultrasonic phased array technology emits and receives ultrasonic signals through a phased array probe, and uses multiple independent acoustic sensor arrays to achieve electronic control of the sound beam, so as to be able to perform real-time detection at multiple angles. Compared with traditional ultrasonic testing methods, ultrasonic phased array technology has higher detection accuracy, faster scanning speed and stronger signal processing ability, and can effectively detect defects in complex structures, especially suitable for large-scale, high-efficiency and high-precision non-destructive testing tasks. The application of ultrasonic phased array in building steel structures can obtain defect information inside the structure in real time and comprehensively, providing a scientific basis for the safety assessment of the structure.

[0004] Although ultrasonic phased array technology has obvious advantages in the detection of building steel structures, in practical applications, it still faces some challenges. First, the steel structure surface often has anti-corrosion coatings (such as paint, hot-dip galvanizing, aluminum spraying) or rust, and the thickness and uniformity of these coatings may affect the coupling effect of ultrasonic waves, resulting in poor coupling, energy loss and signal attenuation. Second, the surface of on-site welds may have welding slag, scale or uneven roughness, and these factors will affect the fit between the probe and the detection surface, resulting in fluctuations and errors in the detection signal. Therefore, the existing ultrasonic phased array technology still has the problem of unstable ultrasonic signals caused by surface defects of steel structures in actual detection, and there is an urgent need for an improved and optimized ultrasonic phased array technology to reduce the influence of steel structure surface defects on ultrasonic detection, improve the stability of ultrasonic signals, and thus improve the accuracy and reliability of building steel structure detection.

[0005] Therefore, an ultrasonic phased array non-destructive testing method for the strength of building steel structures is proposed. Summary of the Invention

[0006] The object of the present invention is to provide a non-destructive testing method for the strength of building steel structures by ultrasonic phased array, so as to reduce the influence of surface defects of steel structures on ultrasonic detection, improve the stability of ultrasonic signals, and thus enhance the accuracy and reliability of building steel structure detection. The method includes: obtaining the topography data of the surface of the building steel structure by using a high-precision laser scanner and electromagnetic induction technology; analyzing the coating thickness distribution and weld surface roughness on the surface of the building steel structure according to the topography data, and constructing a surface information matrix; according to the surface information matrix, automatically adjusting the ultrasonic beam parameters of the ultrasonic phased array by using an adaptive parameter regulation model, and performing ultrasonic phased array detection according to the ultrasonic beam parameters to obtain a preliminary ultrasonic echo signal; calculating the time-frequency domain difference between the preliminary ultrasonic echo signal and a reference signal by using an echo signal compensation model, and performing secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference to obtain an optimized ultrasonic echo signal; fusing the optimized ultrasonic echo signal, the topography data and the surface information matrix by using a pre-constructed and trained building steel structure detection model to construct a detection data set; performing defect detection on the building steel structure through the detection data set; evaluating the result of the defect detection, and dividing the confidence level according to a preset confidence threshold; performing local re-inspection and parameter adjustment according to the confidence level to achieve closed-loop optimization.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A non-destructive testing method for the strength of building steel structures by ultrasonic phased array, including:

[0009] Obtaining the topography data of the surface of the building steel structure by using a high-precision laser scanner and electromagnetic induction technology; analyzing the coating thickness distribution and weld surface roughness on the surface of the building steel structure according to the topography data, and constructing a surface information matrix;

[0010] According to the surface information matrix, automatically adjusting the ultrasonic beam parameters of the ultrasonic phased array by using an adaptive parameter regulation model, and performing ultrasonic phased array detection according to the ultrasonic beam parameters to obtain a preliminary ultrasonic echo signal;

[0011] Calculating the time-frequency domain difference between the preliminary ultrasonic echo signal and a reference signal by using an echo signal compensation model, and performing secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference to obtain an optimized ultrasonic echo signal;

[0012] Fusing the optimized ultrasonic echo signal, the topography data and the surface information matrix by using a pre-constructed and trained building steel structure detection model to construct a detection data set; performing defect detection on the building steel structure through the detection data set; evaluating the result of the defect detection, and dividing the confidence level according to a preset confidence threshold;

[0013] Perform local re-inspection and parameter adjustment according to the confidence level to achieve closed-loop optimization.

[0014] Preferably, the morphology data includes surface three-dimensional topology information, reflection intensity data, and surface texture image data;

[0015] The specific process of analyzing the coating thickness distribution on the surface of the building steel structure and the surface roughness of the weld according to the morphology data includes: using point cloud reconstruction and digital image processing algorithms to perform differential and regression analysis on the surface three-dimensional topology information and the reflection intensity data to obtain the coating thickness distribution;

[0016] Extract the surface roughness of the weld based on the surface texture image data using statistical analysis methods.

[0017] Preferably, the adaptive parameter control model includes an ultrasonic propagation simulation unit and an ultrasonic beam control unit;

[0018] The ultrasonic propagation simulation unit calculates the influence of the coating thickness distribution and the surface roughness of the weld on the ultrasonic propagation path and energy attenuation based on the surface information matrix, and obtains the coating thickness distribution influence factor and the weld surface roughness influence factor;

[0019] The ultrasonic beam control unit dynamically adjusts the ultrasonic beam parameters of the ultrasonic phased array probe according to the coating thickness distribution influence factor and the weld surface roughness influence factor; the ultrasonic beam parameters include: beam angle, beam frequency, beam focal length, beam width, and wave speed.

[0020] Preferably, the echo signal compensation model includes: a reference signal acquisition unit, a signal difference analysis unit, and a signal correction unit;

[0021] The reference signal acquisition unit extracts a reference echo signal matching the current detection part and material from the standard defect-free steel structure sample library as the reference signal;

[0022] The signal difference analysis unit calculates the time-frequency domain difference between the reference signal and the preliminary ultrasonic echo signal through a wavelet transform algorithm;

[0023] The signal correction unit performs the secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference in combination with a preset compensation coefficient matrix to obtain the optimized ultrasonic echo signal.

[0024] Preferably, the formula for the time-frequency domain difference is:

[0025] W i (a,b) = w{S i (t)}(a,b);

[0026] W r (a,b) = w{S r (t)}(a,b);

[0027] △W(a,b) = W i (a,b) - W r (a,b);

[0028] Among them, W i (a,b) is the wavelet coefficient of the preliminary ultrasonic echo signal S i (t); W r (a,b) is the wavelet coefficient of the reference signal S r (t); △W(a,b) is the time-frequency domain difference; a is the scale parameter; b is the translation parameter; w{} is the wavelet transform;

[0029] The formula for the secondary compensation is:

[0030] S o (t) = S i (t) + w -1 {K(a,b)·△W(a,b)};

[0031] Among them, S o (t) is the optimized ultrasonic echo signal; K(a,b) is the preset compensation coefficient matrix; w -1 {} is the inverse wavelet transform.

[0032] Preferably, the building steel structure detection model includes: a multi-modal feature extraction unit, a defect identification unit, and a confidence evaluation unit;

[0033] The multi-modal feature extraction unit fuses the multi-modal features of the optimized ultrasonic echo signal, the morphology data, and the surface information matrix, and constructs a detection data set according to the multi-modal features; the formula for the detection data set is:

[0034] F = tanh(ω S ·S o ☉tanh(ω M ·M) + ω I ·I + b1);

[0035] D = σ(ω F ·F + b2);

[0036] Among them, F is the multi-modal feature; tanh() is the hyperbolic tangent activation function; ω S is the weight of the optimized ultrasonic echo signal; S o is the optimized ultrasonic echo signal; ⊙ is the element-wise and non-linear interaction; ω Mis the weight of the morphology data; M is the morphology data; ω I is the weight of the surface information matrix; I is the surface information matrix; D is the detection data set; σ() is the activation function; ω F is the multi-modal feature weight; b1 and b2 are the bias terms;

[0037] The defect identification unit identifies defects in the building steel structure based on the detection data set; the defects include: coating defects and weld defects; the coating defects include cracking and peeling of the coating; the weld defects include porosity, crater, undercut and crack on the weld surface;

[0038] The confidence evaluation unit evaluates the confidence of the defect detection result of the defect identification unit, and divides the confidence level according to the preset confidence threshold; the preset confidence threshold is accurately set according to different defects.

[0039] Preferably, local re-inspection and parameter adjustment are performed according to the confidence level, and the specific process is as follows:

[0040] For the area where the confidence is lower than the preset confidence threshold, the ultrasonic echo signal of this area is re-collected by using local high-resolution ultrasonic phased array detection for the local re-inspection;

[0041] Based on the result of the local re-inspection, the ultrasonic beam parameters of the ultrasonic phased array probe are dynamically adjusted by using the adaptive parameter regulation model to achieve the closed-loop optimization.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] 1. The present invention proposes an adaptive parameter regulation model to dynamically adjust the beam parameters of the ultrasonic phased array by using the surface information matrix, such as beam angle, frequency, focal length, width and wave velocity, which can adapt to variable factors such as the coating thickness on the surface of the building steel structure and the weld roughness in real time. This method can effectively improve the ultrasonic incidence and focusing effects, reduce the detection errors caused by surface non-uniformity, and thus improve the quality of the preliminary ultrasonic echo signal and the reliability of the subsequent overall detection.

[0044] 2. The present invention proposes an echo signal compensation model, which calculates the time-frequency domain difference between the preliminary signal and the reference signal through wavelet transform, and performs secondary compensation on the basis of the preliminary ultrasonic echo signal obtained by adjusting the beam parameters by the adaptive parameter regulation model, further correcting the time-frequency deviation caused by the surface characteristics of the coating and the weld. This compensation method can reduce noise and signal distortion, significantly improve the clarity and accuracy of the optimized echo signal, and provide a more reliable data basis for subsequent defect detection.

[0045] 3. By fusing the optimized ultrasonic echo signal with the topography data and the surface information matrix to construct a detection data set and using this detection data set for defect detection, the present invention can comprehensively reflect the multi-dimensional characteristics of the steel structure surface. This fusion strategy makes full use of time-frequency information and surface topography information through data complementarity, improves the learning effect and discrimination ability of the building steel structure detection model, thereby achieving accurate positioning and reliable evaluation of defects, and further enhancing the closed-loop optimization effect of the overall non-destructive testing system. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flowchart of a method for ultrasonic phased array non-destructive testing of the strength of a building steel structure provided by an embodiment of the present invention;

[0047] Figure 2 is a working principle diagram of a method for ultrasonic phased array non-destructive testing of the strength of a building steel structure provided by an embodiment of the present invention;

[0048] Figure 3 is a working principle diagram of an adaptive parameter regulation model provided by an embodiment of the present invention;

[0049] Figure 4 is a working principle diagram of an echo signal compensation model provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0051] Building steel structures are widely used in modern buildings due to their good strength, durability and seismic performance. However, over time, steel structures may be damaged due to long-term exposure to harsh environments or external impacts. Damage to steel structures, such as cracks, corrosion, fatigue and welding defects, may affect the load-bearing capacity and safety of the structure. Therefore, regular non-destructive testing of building steel structures to ensure structural safety and reliability is an essential and important link in modern building maintenance and management.

[0052] The present invention provides a non-destructive testing method for the strength of building steel structures using ultrasonic phased arrays, aiming to reduce the influence of surface defects of steel structures on ultrasonic detection, improve the stability of ultrasonic signals, and thus enhance the accuracy and reliability of building steel structure detection. To illustrate that the method of the present invention can reduce the influence of surface defects of steel structures on ultrasonic detection and improve the stability of ultrasonic signals, the effectiveness of the present invention will be described below through two embodiments.

[0053] Embodiment 1

[0054] In the embodiment of the present application, the method proposed by the present invention is used to elaborate on the process of reducing the influence of surface defects of steel structures on ultrasonic detection, improving the stability of ultrasonic signals, and thus enhancing the accuracy and reliability of building steel structure detection. The embodiment of the present application focuses on the detection of the building steel structure in the hydrogenation workshop in the factory area of Company A. The hydrogenation workshop is a two-story steel frame structure building. The following will be based on Figure 1 the content to elaborate on the detection process of the building steel structure in the hydrogenation workshop in the factory area of Company A; among them, Figure 1 is the specific flow chart of the method proposed by the present invention, including: obtaining the topography data of the surface of the building steel structure using a high-precision laser scanner and electromagnetic induction technology; analyzing the coating thickness distribution and weld surface roughness on the surface of the building steel structure according to the topography data, and constructing a surface information matrix; automatically adjusting the ultrasonic beam parameters of the ultrasonic phased array according to the surface information matrix using an adaptive parameter regulation model, and performing ultrasonic phased array detection according to the ultrasonic beam parameters to obtain a preliminary ultrasonic echo signal; calculating the time-frequency domain difference between the preliminary ultrasonic echo signal and the reference signal using an echo signal compensation model, and performing secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference to obtain an optimized ultrasonic echo signal; fusing the optimized ultrasonic echo signal, the topography data, and the surface information matrix using a pre-constructed and trained building steel structure detection model to construct a detection data set; detecting the defects of the building steel structure through the detection data set; evaluating the results of the defect detection, and dividing the confidence level according to a preset confidence threshold; performing local re-inspection and parameter adjustment according to the confidence level to achieve closed-loop optimization. Figure 2 is the working principle diagram of the method proposed by the present invention. In combination with Figure 1 the content in, the following description is made:

[0055] A non-destructive testing method for the strength of building steel structures using ultrasonic phased arrays includes:

[0056] Obtaining the topography data of the surface of the building steel structure using a high-precision laser scanner and electromagnetic induction technology;

[0057] The topography data includes surface three-dimensional topology information, reflection intensity data, and surface texture image data;

[0058] Analyze the coating thickness distribution on the surface of the building steel structure and the surface roughness of the welds based on the described topography data, and construct a surface information matrix; the specific process includes:

[0059] Using point cloud reconstruction and digital image processing algorithms, perform differential and regression analyses on the three-dimensional coordinate data and the reflection intensity data to obtain the coating thickness distribution;

[0060] Extract the surface roughness of the welds based on the surface texture image data using statistical analysis methods.

[0061] Specifically, use a differential operator to calculate the height change and reflection intensity change of each point and the points in its neighborhood; according to the differential analysis results, select a set of sample data with known coating thicknesses, and obtain a regression model by fitting the relationship between the reflection intensity change and the actual thickness; apply this model to the entire three-dimensional coordinate data to obtain the coating thickness distribution at each detection point.

[0062] Use the gray-level co-occurrence matrix to extract the contrast, correlation, and uniformity of the surface texture image data, and fuse the contrast, correlation, and uniformity to obtain the surface roughness of the welds; the specific formula is:

[0063] R rough = k1·C + k2·(1 - Corr) + k3·(1 - H);

[0064]

[0065] where, R rough is the surface roughness of the welds; k1 is the contrast weight; C is the contrast; k2 is the correlation weight; Corr is the correlation; k2 is the uniformity weight; H is the uniformity; P(i,j) is the gray-level co-occurrence matrix; i,j = 0,1,...,N - 1 are the gray levels in the image; μ i and μ j respectively represent the means of the rows and columns of the gray-level co-occurrence matrix; σ i and σ j respectively represent the standard deviations of the rows and columns;

[0066] In the above formula, the greater the contrast, the higher the surface roughness of the welds; the higher the correlation, the more regular the image structure, so 1 - Corr is taken as the structural irregularity, and the larger its value, the higher the surface roughness of the welds; the greater the uniformity, the more uniform the surface texture, so 1 - H is taken as the texture non-uniformity, and the larger its value, the higher the surface roughness of the welds.

[0067] In the embodiments of the present application, high-precision laser scanners and electromagnetic induction technology are used to obtain the topography data of the surface of building steel structures, including surface three-dimensional topology information, reflection intensity data, and surface texture image data, which can comprehensively and accurately describe the characteristics of the structure surface. By using point cloud reconstruction and digital image processing algorithms, differential and regression analyses are performed on the three-dimensional coordinate data and reflection intensity data to obtain the coating thickness distribution; based on the surface texture image data, texture features such as contrast, correlation, and uniformity are extracted using the gray-level co-occurrence matrix, and these features are fused to obtain the surface roughness of the weld. The comprehensive application of the above technical means constructs a multi-dimensional surface information matrix, providing accurate basic data for subsequent ultrasonic phased array detection, enabling the detection system to achieve high-precision and high-reliability detection effects in subsequent parameter adaptive adjustment, signal compensation, and defect identification, thus significantly improving the accuracy and overall detection efficiency of building steel structure strength detection.

[0068] Preferably, according to the surface information matrix, an adaptive parameter regulation model is used to automatically adjust the ultrasonic beam parameters of the ultrasonic phased array, and ultrasonic phased array detection is performed according to the ultrasonic beam parameters to obtain a preliminary ultrasonic echo signal; refer to Figure 3 ;

[0069] The adaptive parameter regulation model includes an ultrasonic propagation simulation unit and an ultrasonic beam control unit;

[0070] Based on the surface information matrix, the ultrasonic propagation simulation unit calculates the influence of the coating thickness distribution and the surface roughness of the weld on the ultrasonic propagation path and energy attenuation, and obtains a coating thickness distribution influence factor and a weld surface roughness influence factor;

[0071] The ultrasonic beam control unit dynamically adjusts the ultrasonic beam parameters of the ultrasonic phased array probe according to the coating thickness distribution influence factor and the weld surface roughness influence factor; the ultrasonic beam parameters include: beam angle, beam frequency, beam focal length, beam width, and wave speed.

[0072] Specifically, the formula for the coating thickness distribution influence factor is:

[0073] α = f(d, v c , α c );

[0074] where α is the coating thickness distribution influence factor; f() is the coating thickness distribution influence factor function, which needs to be determined by experimental calibration; d is the coating thickness; v c is the propagation speed of ultrasonic waves in the coating; α c is the energy attenuation and the acoustic attenuation coefficient of the coating;

[0075] The formula for the weld surface roughness influence factor is as follows:

[0076] β = g(R a , R q );

[0077] Where β is the weld surface roughness influence factor; g() is the weld surface roughness influence factor function, which needs to be determined through experimental calibration; R a is the arithmetic mean roughness; R q is the root mean square roughness;

[0078] The formula for the beam angle is as follows:

[0079] θ = θ0 + k θ ·(α + β);

[0080] Where θ is the beam angle; θ0 is the initial beam angle; k θ is the beam angle adjustment coefficient;

[0081] The formula for the beam frequency is as follows:

[0082] f = f0·(1 + k f ·α)·(1 + k' f ·β);

[0083] Where f is the beam angle; f0 is the initial beam angle; k f is the adjustment coefficient of the coating thickness distribution on the beam frequency; k' f is the adjustment coefficient of the weld surface roughness on the beam frequency;

[0084] The formula for the beam focal length is as follows:

[0085] d = d0 + k d ·α;

[0086] Where d is the beam angle; d0 is the initial beam angle; k d is the beam angle adjustment coefficient;

[0087] The formula for the beam width is as follows:

[0088] ω = ω0·(1 + k ω ·β);

[0089] Where ω is the beam width; ω0 is the initial beam width; k ω is the beam width adjustment coefficient;

[0090] The formula for the wave velocity is as follows:

[0091] v = v0·(1 + k v ·α);

[0092] Among them, v is the wave velocity; v0 is the initial wave velocity; k v is the wave velocity adjustment coefficient;

[0093] Table 1 is a comparison table of each index of the ultrasonic echo signal before and after adjusting the ultrasonic beam parameters.

[0094] Table 1 Comparison Table of Each Index of the Ultrasonic Echo Signal Before and After Adjusting the Ultrasonic Beam Parameters

[0095] Index Before parameter adjustment After parameter adjustment Beam angle (°) 30.2 35.1 Beam frequency (MHz) 2.0 2.2 Beam focal length (mm) 50 55 Beam width (mm) 10 9 Wave velocity (m / s) 5000 5100 Signal-to-noise ratio (dB) 34 51

[0096] In the embodiment of the present application, an adaptive parameter regulation model is used to dynamically adjust the ultrasonic beam parameters according to the coating thickness distribution on the surface of the building steel structure and the surface roughness of the weld, including the beam angle, frequency, focal length, width, and wave velocity. This dynamic adjustment can optimize the propagation path and energy distribution of the ultrasonic wave, improve the stability of the ultrasonic signal, and thus enhance the sensitivity and resolution of the subsequent building steel structure detection. By precisely controlling the characteristics of the ultrasonic beam, it can provide a basis for more effectively detecting defects in complex structures subsequently, thereby improving the reliability and accuracy of the subsequent steel structure detection results.

[0097] Preferably, a time-frequency domain difference between the preliminary ultrasonic echo signal and the reference signal is calculated by using an echo signal compensation model, and the preliminary ultrasonic echo signal is secondarily compensated according to the time-frequency domain difference to obtain an optimized ultrasonic echo signal; refer to Figure 4 ;

[0098] The echo signal compensation model includes: a reference signal acquisition unit, a signal difference analysis unit, and a signal correction unit;

[0099] The reference signal acquisition unit extracts a reference echo signal that matches the current detection part and material from a standard defect-free steel structure sample library as the reference signal;

[0100] The signal difference analysis unit calculates the time-frequency domain difference between the reference signal and the preliminary ultrasonic echo signal through a wavelet transform algorithm;

[0101] The signal correction unit performs the secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference in combination with a preset compensation coefficient matrix to obtain the optimized ultrasonic echo signal.

[0102] The embodiment of this application proposes an echo signal compensation model. By accurately calculating the differences between the preliminary ultrasonic echo signal and the reference signal in the time-frequency domain, and using a preset compensation coefficient matrix to perform secondary compensation on the preliminary signal, the quality and stability of the ultrasonic echo signal are fundamentally improved. By extracting a reference signal that matches the current detection part and material from a standard defect-free steel structure sample library, and using the wavelet transform algorithm to perform time-frequency domain analysis on the signal, this method can accurately capture the signal distortion and energy attenuation caused by uneven coating and weld defects, thereby realizing fine correction of the preliminary signal. This solution not only effectively reduces the noise and interference in the detection process, but also significantly improves the resolution and detection accuracy of ultrasonic phased array detection, providing more reliable signal data support for subsequent defect determination.

[0103] Preferably, the formula for the time-frequency domain difference is:

[0104] W i (a, b) = w{S i (t)}(a, b);

[0105] W r (a, b) = w{S r (t)}(a, b);

[0106] ΔW(a, b) = W i (a, b) - W r (a, b);

[0107] Where, W i (a, b) is the wavelet coefficient of the preliminary ultrasonic echo signal S i (t); W r (a, b) is the wavelet coefficient of the reference signal S r (t); ΔW(a, b) is the time-frequency domain difference; a is the scale parameter; b is the translation parameter; w{} is the wavelet transform;

[0108] The formula for the secondary compensation is:

[0109] S o (t) = S i (t) + w -1 {K(a, b)·ΔW(a, b)};

[0110] Where, S o (t) is the optimized ultrasonic echo signal; K(a, b) is the preset compensation coefficient matrix; w -1 {} is the inverse wavelet transform;

[0111] The preset compensation coefficient matrix is obtained by performing ultrasonic testing on a large number of standard defect-free and known-defect steel structure samples, collecting their echo signals, performing wavelet transform on these signals, and extracting time-frequency domain features at different scales and translation parameters; comparing these experimental data with reference signals, and solving the compensation coefficients at each scale parameter and translation parameter through optimization methods such as regression analysis, least squares fitting, or machine learning, so that after applying inverse wavelet transform for signal reconstruction, the error between the compensated signal and the true signal is minimized; through this series of steps, the set of coefficients obtained constitutes the preset compensation coefficient matrix.

[0112] Table 2 is a comparison table of the stability of ultrasonic echo signals before and after secondary compensation.

[0113] Table 2 Comparison table of the stability of ultrasonic echo signals before and after secondary compensation

[0114] Stability index Before secondary compensation After secondary compensation Signal-to-noise ratio (db) 51 63 Standard deviation of signal amplitude (V) 0.8 0.3 Stability index 0.85 0.92

[0115] In the embodiment of the present application, through the time-frequency domain difference formula, the coefficients of the preliminary ultrasonic echo signal and the reference signal at different scales and translation parameters are compared by using wavelet transform, and the subtle differences of the signal in the time domain and frequency domain are accurately quantified; by using the preset compensation coefficient matrix and the secondary compensation formula of inverse wavelet transform, these differences can be finely corrected to generate an optimized ultrasonic echo signal, thereby effectively eliminating signal distortion and energy attenuation caused by the surface state of the structure and material properties, significantly improving the clarity and stability of the signal, and further enhancing the reliability and overall detection ability of ultrasonic phased array non-destructive testing in the defect identification of building steel structures.

[0116] Preferably, the optimized ultrasonic echo signal, the morphology data, and the surface information matrix are fused by using a pre-constructed and trained building steel structure detection model to construct a detection data set; the building steel structure is defect-detected through the detection data set; the results of the defect detection are evaluated, and the confidence level is divided according to a preset confidence threshold;

[0117] The building steel structure detection model includes: a multi-modal feature extraction unit, a defect identification unit, and a confidence evaluation unit;

[0118] The multi-modal feature extraction unit fuses the multi-modal features of the optimized ultrasonic echo signal, the morphology data, and the surface information matrix, and constructs a detection data set according to the multi-modal features; the formula of the detection data set is:

[0119] F = tanh(ω S ·S o ☉tanh(ω M ·M) + ω I ·I + b1);

[0120] D = σ(ω F ·F + b2);

[0121] where F is the multi-modal feature; tanh() is the hyperbolic tangent activation function; ω S is the weight for optimizing the ultrasonic echo signal; S o is the optimized ultrasonic echo signal; ⊙ is the element-wise and non-linear interaction; ω M is the weight for the topography data; M is the topography data; ω I is the weight for the surface information matrix; I is the surface information matrix; D is the detection data set; σ() is the activation function; ω F is the weight for the multi-modal feature; b1 and b2 are the bias terms;

[0122] The defect recognition unit recognizes the defects in the building steel structure based on the detection data set; the defects include: coating defects and weld defects; the coating defects include cracking and peeling of the coating; the weld defects include porosity, crater, undercut and crack on the weld surface;

[0123] The confidence evaluation unit evaluates the confidence of the defect detection result of the defect recognition unit and divides the confidence level according to the preset confidence threshold; the preset confidence threshold is precisely set according to different defects.

[0124] Specifically, the defect recognition unit identifies the probability value of each defect type in the steel structure through a pre-trained CNN model;

[0125] The confidence evaluation unit calculates the confidence according to the probability value of each defect type; the formula for the confidence is:

[0126] C = max(P1, P2,..., P i ,..., P n );

[0127] where C is the confidence; P i is the probability value of the i-th defect type; n is the total number of defect types;

[0128] The preset confidence threshold is: for critical structural parts, such as welds, a higher confidence threshold is set; in this embodiment, the weld confidence threshold is set to 0.8; for the steel frame coating part, the coating confidence threshold is set to 0.6.

[0129] In the embodiment of the present application, a detection data set is constructed by fusing and optimizing ultrasonic echo signals, topography data, and surface information matrices, and then a pre-trained deep learning model is combined to detect defects in building steel structures, which can accurately identify various defects in coatings and welds. The CNN model is used to extract the probability values of each defect type, and the confidence evaluation unit calculates the probability and divides the threshold for the detection results, realizing differential detection of key parts (such as welds) and secondary parts (such as coatings). A higher confidence threshold is set for the weld area to ensure the strictness of detection, while the threshold for the coating area is set to 0.6 to balance detection accuracy and error tolerance. This method not only improves the accuracy and reliability of detection, reduces the phenomena of false detection and missed detection, but also provides a scientific basis for subsequent defect repair and safety assessment, significantly improving the overall efficiency and application value of non-destructive testing of building steel structures.

[0130] Preferably, local re-inspection and parameter adjustment are performed according to the confidence level to achieve closed-loop optimization. The specific process is as follows:

[0131] For the area where the confidence is lower than the preset confidence threshold, the ultrasonic echo signal of this area is re-collected by using local high-resolution ultrasonic phased array detection for the local re-inspection;

[0132] Based on the results of the local re-inspection, the ultrasonic beam parameters of the ultrasonic phased array probe are dynamically adjusted by using the adaptive parameter regulation model to achieve the closed-loop optimization.

[0133] Table 3 is a comparison table of detection performance before and after local re-inspection and parameter adjustment.

[0134] Table 3 Comparison table of detection performance before and after local re-inspection and parameter adjustment

[0135] Detection performance index Before local re-inspection After local re-inspection Defect detection accuracy rate (%) 92.9 96.8 Omission rate (%) 15.7 7.2 False detection rate (%) 6.3 4.9 Mean confidence level 0.65 0.76

[0136] In the embodiment of the present application, by performing local re-inspection on the area where the confidence is lower than the preset threshold, further compensation for the insufficient part of the preliminary detection result is realized. The echo signal of the target area is re-collected by using high-resolution ultrasonic phased array detection, enabling the system to capture more detailed information, and then combined with the dynamic adjustment of the ultrasonic beam parameters by the adaptive parameter regulation model, thereby compensating for the signal deviation caused by local environmental changes and / or insufficient detection conditions. This process not only improves the accuracy and stability of the detection signal, but also realizes real-time feedback and parameter optimization, further reducing the risk of missed detection, ensuring the efficiency and reliability of defect detection of building steel structures, and providing solid data support for engineering safety assessment.

[0137] In the embodiments of the present application, by accurately acquiring surface information, adaptively adjusting detection parameters, compensating for signal distortion, and closed-loop optimizing the detection process, the accuracy and reliability of non-destructive testing of building steel structures have been significantly improved. First, a high-precision laser scanner and electromagnetic induction technology are used to obtain the topographic data of the surface of the building steel structure, including the coating thickness distribution and the surface roughness of the weld, and a surface information matrix is constructed. This enables the detection system to accurately grasp the surface characteristics of the structure to be measured and provides accurate input parameters for subsequent ultrasonic testing. Secondly, based on the surface information matrix, an adaptive parameter regulation model is adopted to automatically adjust the ultrasonic beam parameters of the ultrasonic phased array, such as beam angle, frequency, focal length, width, and wave velocity. This dynamic adjustment ensures the best propagation path and energy distribution of ultrasonic waves under complex surface conditions and improves the quality of the initial ultrasonic echo signal. Then, through the echo signal compensation model, the time-frequency domain differences between the initial ultrasonic echo signal and the reference signal are calculated, these differences are analyzed using wavelet transform, and the initial echo signal is secondarily compensated in combination with a preset compensation coefficient matrix to obtain an optimized ultrasonic echo signal. This process effectively corrects the signal distortion caused by surface characteristics and enhances the accuracy of the signal. Finally, the optimized ultrasonic echo signal, topographic data, and surface information matrix are fused to construct a detection data set, and a pre-trained building steel structure detection model is used for defect detection. After the detection results are evaluated, local re-inspection and parameter adjustment are performed for low-confidence regions, realizing the closed-loop optimization of the detection process.

[0138] Embodiment 2

[0139] In Embodiment 1, the method proposed by the present invention has successfully achieved reducing the influence of surface defects of steel structures on ultrasonic detection, improving the stability of ultrasonic signals, and thus enhancing the accuracy and reliability of building steel structure detection. To further verify the effectiveness of the present invention, steel structure detection was also carried out on another steel frame structure building B in the embodiments of the present application.

[0140] A method for non-destructive testing of the strength of building steel structures using ultrasonic phased arrays, comprising:

[0141] Using a high-precision laser scanner and electromagnetic induction technology to obtain the topographic data of the surface of the building steel structure;

[0142] The topographic data includes surface three-dimensional topological information, reflection intensity data, and surface texture image data;

[0143] Analyzing the coating thickness distribution and the surface roughness of the weld on the surface of the building steel structure according to the topographic data, and constructing a surface information matrix; the specific process includes:

[0144] Using point cloud reconstruction and digital image processing algorithms to perform differential and regression analysis on the three-dimensional coordinate data and the reflection intensity data to obtain the coating thickness distribution;

[0145] Extract the surface roughness of the weld seam by using a statistical analysis method based on the surface texture image data.

[0146] Preferably, according to the surface information matrix, use an adaptive parameter regulation model to automatically adjust the ultrasonic beam parameters of the ultrasonic phased array, and perform ultrasonic phased array detection according to the ultrasonic beam parameters to obtain a preliminary ultrasonic echo signal;

[0147] The adaptive parameter regulation model includes an ultrasonic propagation simulation unit and an ultrasonic beam control unit;

[0148] The ultrasonic propagation simulation unit calculates the influence of the coating thickness distribution and the surface roughness of the weld seam on the ultrasonic propagation path and energy attenuation based on the surface information matrix, and obtains a coating thickness distribution influence factor and a weld seam surface roughness influence factor;

[0149] The ultrasonic beam control unit dynamically adjusts the ultrasonic beam parameters of the ultrasonic phased array probe according to the coating thickness distribution influence factor and the weld seam surface roughness influence factor; the ultrasonic beam parameters include: beam angle, beam frequency, beam focal length, beam width, and wave speed.

[0150] Preferably, use an echo signal compensation model to calculate the time-frequency domain difference between the preliminary ultrasonic echo signal and the reference signal, and perform secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference to obtain an optimized ultrasonic echo signal;

[0151] The echo signal compensation model includes: a reference signal acquisition unit, a signal difference analysis unit, and a signal correction unit;

[0152] The reference signal acquisition unit extracts a reference echo signal that matches the current detection part and material from a standard defect-free steel structure sample library as the reference signal;

[0153] The signal difference analysis unit calculates the time-frequency domain difference between the reference signal and the preliminary ultrasonic echo signal through a wavelet transform algorithm;

[0154] The signal correction unit performs the secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference in combination with a preset compensation coefficient matrix to obtain the optimized ultrasonic echo signal.

[0155] Preferably, the formula for the time-frequency domain difference is:

[0156] W i (a,b) = w{S i (t)}(a,b);

[0157] Wr (a,b) = w{S r (t)}(a,b);

[0158] △W(a,b) = W i (a,b) - W r (a,b);

[0159] Among them, W i (a,b) is the wavelet coefficient of the preliminary ultrasonic echo signal S i (t); W r (a,b) is the wavelet coefficient of the reference signal S r (t); △W(a,b) is the time-frequency domain difference; a is the scale parameter; b is the translation parameter; w{} is the wavelet transform;

[0160] The formula for the secondary compensation is:

[0161] S o (t) = S i (t) + w -1 {K(a,b)·△W(a,b)};

[0162] Among them, S o (t) is the optimized ultrasonic echo signal; K(a,b) is the preset compensation coefficient matrix; w -1 {} is the inverse wavelet transform.

[0163] Preferably, the optimized ultrasonic echo signal, the topography data and the surface information matrix are fused by using a pre-constructed and trained building steel structure detection model to construct a detection data set, and the building steel structure is defect-detected through the detection data set; the result of the defect detection is evaluated, and the confidence level is divided according to the preset confidence threshold;

[0164] The building steel structure detection model includes: a multi-modal feature extraction unit, a defect identification unit and a confidence evaluation unit;

[0165] The multi-modal feature extraction unit fuses the multi-modal features of the optimized ultrasonic echo signal, the topography data and the surface information matrix, and constructs a detection data set according to the multi-modal features; the formula of the detection data set is:

[0166] F = tanh(ω S ·S o ☉tanh(ω M ·M) + ω I ·I + b1);

[0167] D = σ(ω F ·F + b2);

[0168] Among them, F is the multi-modal feature; tanh() is the hyperbolic tangent activation function; ω S is the weight for optimizing the ultrasonic echo signal; S o is the optimized ultrasonic echo signal; ⊙ is the element-wise and non-linear interaction; ω M is the weight for the topography data; M is the topography data; ω I is the weight for the surface information matrix; I is the surface information matrix; D is the detection data set; σ() is the activation function; ω F is the weight for the multi-modal feature; b1 and b2 are the bias terms;

[0169] The defect recognition unit identifies the defects in the building steel structure based on the detection data set; the defects include: coating defects and weld defects; the coating defects include cracking and peeling of the coating; the weld defects include porosity, crater, undercut and crack on the weld surface;

[0170] The confidence evaluation unit evaluates the confidence of the defect detection result of the defect recognition unit and divides the confidence level according to a preset confidence threshold; the preset confidence threshold is precisely set according to different defects.

[0171] Preferably, local re-inspection and parameter adjustment are performed according to the confidence level to achieve closed-loop optimization; the specific process is as follows:

[0172] For the area in the confidence level that is lower than the preset confidence threshold, the ultrasonic echo signal of this area is re-acquired by using local high-resolution ultrasonic phased array detection for the local re-inspection;

[0173] Based on the result of the local re-inspection, the ultrasonic beam parameters of the ultrasonic phased array probe are dynamically adjusted by using the adaptive parameter regulation model to achieve the closed-loop optimization.

[0174] Table 4 gives the overall performance optimization result table.

[0175] Table 4 Overall Performance Optimization Result Table

[0176] Optimization steps Signal-to-noise ratio (dB) Stability index Detection accuracy rate (%) Original ultrasonic echo signal 29 0.64 76.3％ Initial ultrasonic echo signal 53 0.81 87.5％ Optimized ultrasonic echo signal 68 0.90 93.7％ Signal after closed-loop optimization 72 0.93 97.1％

[0177] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for nondestructive testing of building steel structure strength by ultrasonic phased array, characterized in that: include: Use high-precision laser scanners and electromagnetic induction technology to obtain topographic data on the surface of building steel structures; Analyzing the coating thickness distribution and weld surface roughness of the building steel structure surface according to the topographic data, and constructing a surface information matrix; According to the surface information matrix, an adaptive parameter control model is used to automatically adjust ultrasonic beam parameters of an ultrasonic phased array, and ultrasonic phased array detection is performed according to the ultrasonic beam parameters to obtain a preliminary ultrasonic echo signal; The echo signal compensation model is used to calculate the time-frequency domain difference between the preliminary ultrasonic echo signal and the reference signal, and the preliminary ultrasonic echo signal is secondary compensated according to the time-frequency domain difference to obtain an optimized ultrasonic echo signal; The optimized ultrasonic echo signal, the morphological data and the surface information matrix are fused using a pre-built and trained building steel structure detection model to construct a detection data set; Perform defect detection on the building steel structure using the detection data set; Evaluating the defect detection results and dividing the confidence levels according to a preset confidence threshold; Local re-inspection and parameter adjustment are performed according to the confidence level to achieve closed-loop optimization.

2. The ultrasonic phased array nondestructive testing method for building steel structure strength according to claim 1 is characterized by: The topographic data includes surface three-dimensional topological information, reflection intensity data and surface texture image data; The specific process of analyzing the coating thickness distribution and weld surface roughness of the building steel structure surface according to the topographic data includes: using point cloud reconstruction and digital image processing algorithms to perform differential and regression analysis on the surface three-dimensional topological information and the reflection intensity data to obtain the coating thickness distribution; The surface roughness of the weld is extracted using a statistical analysis method based on the surface texture image data.

3. The ultrasonic phased array nondestructive testing method for building steel structure strength according to claim 1 is characterized by: The adaptive parameter control model includes an ultrasonic propagation simulation unit and an ultrasonic beam control unit; the ultrasonic propagation simulation unit calculates the influence of the coating thickness distribution and the weld surface roughness on the ultrasonic propagation path and energy attenuation based on the surface information matrix, and obtains the coating thickness distribution influence factor and the weld surface roughness influence factor; the ultrasonic beam control unit dynamically adjusts the ultrasonic beam parameters of the ultrasonic phased array probe according to the coating thickness distribution influence factor and the weld surface roughness influence factor; the ultrasonic beam parameters include: beam angle, beam frequency, beam focal length, beam width and wave speed.

4. The ultrasonic phased array nondestructive testing method for building steel structure strength according to claim 1 is characterized by: The echo signal compensation model includes: a reference signal acquisition unit, a signal difference analysis unit and a signal correction unit; the reference signal acquisition unit extracts a reference echo signal that matches the current detection position and material from a standard defect-free steel structure sample library as the reference signal; the signal difference analysis unit calculates the time-frequency domain difference between the reference signal and the preliminary ultrasonic echo signal through a wavelet transform algorithm; the signal correction unit performs the secondary compensation on the preliminary ultrasonic echo signal according to the time-frequency domain difference in combination with a preset compensation coefficient matrix to obtain the optimized ultrasonic echo signal.

5. The ultrasonic phased array nondestructive testing method for building steel structure strength according to claim 1 is characterized by: The formula for the time-frequency domain difference is: W i (a,b)=w{S i (t)}(a,b); W r (a,b)=w{S r (t)}(a,b); △W(a,b)=W i (a,b)-W r (a,b); Among them, W i (a, b) are the preliminary ultrasonic echo signals S i (t) wavelet coefficient; W r (a, b) is the reference signal S r (t); △W(a,b) is the difference in time and frequency domain; a is the scale parameter; b is the translation parameter; w{} is the wavelet transform; The formula for the secondary compensation is: S o (t)=S i (t)+w -1 {K(a,b)·△W(a,b)}; Among them, S o (t) is the optimized ultrasonic echo signal; K(a,b) is the preset compensation coefficient matrix; w -1 {} is the inverse wavelet transform.

6. The ultrasonic phased array nondestructive testing method for building steel structure strength according to claim 1 is characterized by: The building steel structure detection model includes: a multimodal feature extraction unit, a defect recognition unit and a confidence assessment unit; the multimodal feature extraction unit fuses the multimodal features of the optimized ultrasonic echo signal, the morphological data and the surface information matrix, and constructs a detection data set according to the multimodal features; the formula of the detection data set is: F=tanh(ω S ·S o ☉tanh(ω M ·M)+ω I ·I+b1); D=σ(ω F ·F+b2); Among them, F is the multimodal feature; tanh() is the hyperbolic tangent activation function; ω S To optimize the weight of ultrasonic echo signal; S o is to optimize the ultrasonic echo signal; ⊙ is the element and nonlinear interaction; ω M is the weight of the morphological data; M is the morphological data; ω I is the weight of the surface information matrix; I is the surface information matrix; D is the detection data set; σ() is the activation function; ω F is the multimodal feature weight; b1 and b2 are bias terms; The defect recognition unit recognizes defects in the building steel structure based on the detection data set; the defects include: coating defects and weld defects; the coating defects include cracking and shedding of the coating; the weld defects include pores, arc pits, undercuts and cracks on the weld surface; The confidence evaluation unit evaluates the confidence of the defect detection result of the defect recognition unit, and divides the confidence level according to the preset confidence threshold; the preset confidence threshold is accurately set according to different defects.

7. The ultrasonic phased array nondestructive testing method for building steel structure strength according to claim 1 is characterized by: The low confidence area in the confidence level is locally rechecked and parameter adjusted. The specific process is as follows: For the area whose confidence level is lower than the preset confidence threshold, the ultrasonic echo signal of the area is re-collected by using local high-resolution ultrasonic phased array detection to perform the local re-inspection; based on the result of the local re-inspection, the ultrasonic beam parameters of the ultrasonic phased array probe are dynamically adjusted by using the adaptive parameter control model to achieve the closed-loop optimization.

Images