CN110579534B

CN110579534B - Non-baseline detection and positioning method for defects of steel plate with welding seam based on reciprocity damage

Info

Publication number: CN110579534B
Application number: CN201910838511.3A
Authority: CN
Inventors: 周邵萍; 何天浩; 李沁霏; 李勇; 邢改兰
Original assignee: East China University of Science and Technology
Current assignee: East China University of Science and Technology
Priority date: 2019-09-05
Filing date: 2019-09-05
Publication date: 2022-06-07
Anticipated expiration: 2039-09-05
Also published as: CN110579534A

Abstract

The invention relates to a method for detecting and positioning defects of a welded steel plate without a base line based on reciprocity damage, which comprises the following steps: step 1: arranging a parallel linear sensor array, and acquiring reciprocity signals after setting steel ingot simulation defects; step 2: preprocessing the reciprocity signals, defining signal difference coefficients, and performing correlation quantitative analysis on each pair of preprocessed reciprocity signals; and step 3: obtaining a pair with the largest signal difference coefficient (reciprocity damage) in each group from a plurality of groups of pairs of reciprocity signals obtained by sequentially and interactively exciting and receiving each upper sensor and each lower sensor in a parallel linear sensor array, connecting the corresponding sensors to obtain a plurality of groups of connecting lines and corresponding intersection points; and 4, step 4: and finally determining the defect position by adopting a clustering algorithm aiming at the multiple groups of connecting lines and the corresponding intersection points. Compared with the prior art, the method gets rid of the dependence on the baseline data of the detected object without damage, realizes the baseline-free detection and positioning, and is closer to the practical engineering application.

Description

Non-baseline detection and positioning method for defects of steel plate with welding seam based on reciprocity damage

Technical Field

The invention relates to the technical field of ultrasonic guided wave nondestructive testing of a steel plate structure with a welding seam, in particular to a method for detecting and positioning defects of the steel plate structure with the welding seam without a base line based on reciprocity damage.

Background

The ultrasonic Lamb wave has the advantages of small attenuation, high sensitivity, no need of damaging the structure of the detected object, simple and convenient detection equipment and the like, and can realize large-scale detection, so that the detection method of the ultrasonic Lamb wave is widely applied to structural health monitoring in the field of nondestructive detection.

However, due to the inherent multi-modal characteristic and frequency dispersion characteristic of Lamb waves, signals are complex when Lamb waves propagate in complex structures (such as plates with welding seams), the signal to noise ratio is not high, the detection precision is reduced, and even false detection and missing detection are caused. At present, a large amount of research has been done by scholars at home and abroad aiming at the defect detection in plate structures by Lamb waves, for example, the positions of defects are accurately found by using some imaging algorithms such as a hyperbolic imaging algorithm, a full-focus imaging algorithm, a discrete ellipse imaging algorithm and the like; based on the minimum variance processing, the imaging precision is improved; and (4) automatically identifying the number of the defects and positioning by combining an intelligent algorithm in machine learning. However, most of the current research is based on baseline subtraction to obtain defect information, and the baseline (health) data is difficult to obtain in engineering practice, so that the current research has great limitations.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provide a method for detecting and positioning the defects of the welded steel plate without a base line based on damaged reciprocity.

The purpose of the invention can be realized by the following technical scheme:

a method for detecting and positioning defects of a welded steel plate without a base line based on reciprocity damage comprises the following steps:

step 1: arranging a parallel linear sensor array, and acquiring reciprocity signals after setting steel ingot simulation defects;

step 2: preprocessing the reciprocity signals, defining signal difference coefficients, and performing correlation quantitative analysis on each pair of preprocessed reciprocity signals;

and step 3: acquiring a pair with the largest signal difference coefficient in each group, namely a pair with the largest reciprocity damage, from a plurality of groups of pairs of reciprocity signals acquired by sequentially and interactively exciting and receiving each upper sensor and each lower sensor in a parallel linear sensor array, and connecting the corresponding sensors to acquire a plurality of groups of connecting lines and corresponding intersection points;

and 4, step 4: and finally determining the defect position by adopting a clustering algorithm aiming at the multiple groups of connecting lines and the corresponding intersection points.

Further, the method for acquiring the reciprocity signal in step 1 includes the following steps:

step 11: the method comprises the following steps that (1) parallel linear sensor arrays are symmetrically arranged on two sides of a weld joint of a steel plate with the weld joint, n sensors are arranged on each side, and the sensors on the same side are arranged at equal intervals;

step 12: exciting the upper sensor i and receiving the lower sensor j at proper excitation frequency to obtain a set of data S_ijExchanging the receiving sequence of the excitation of the upper and lower sensors to obtain another set of data S_ji，S_ijAnd S_jiForming a pair of reciprocity signals;

step 13: one of the sensors on the upper side of the welding seam and n sensors on the lower side of the welding seam exchange excitation receiving sequence in sequence, so that n pairs of reciprocity signals can be obtained; and (4) acquiring reciprocity signals of the n sensors on the upper side and the sensors on the lower side of the weld joint in turn according to the method, and finally acquiring n multiplied by n pairs of reciprocity signals.

Further, the step 2 comprises the following sub-steps:

step 21: preprocessing a reciprocity signal by using MATLAB, wherein the preprocessing comprises filtering, wavelet denoising and normalization;

step 22: calculating the correlation coefficient of each pair of reciprocity signals;

step 23: and further solving the corresponding signal difference coefficient according to the correlation coefficient result of each pair of reciprocity signals so as to judge the loss of the reciprocity of the signals.

Further, the correlation coefficient in step 22 is described by the formula:

in the formula, ρ_XYRepresenting the correlation coefficient, for characterizing the degree of similarity between two signals, i.e. the degree of coincidence of two signals in a time domain plot, the greater the degree of coincidenceThe higher the correlation coefficient, the larger Cov (X, Y) represents the covariance of signal X and signal Y, d (X) represents the variance of signal X, and d (Y) represents the variance of signal Y.

Further, the signal difference coefficient in step 23 is described by the formula:

SDC_XY＝1-ρ_XY

in the formula, SDC_XYRepresenting the signal difference coefficient, the greater the loss in reciprocity of the signal.

Further, the step 3 specifically includes: one of the sensors on the upper side of the welding seam and each sensor on the lower side are alternately switched to excite and receive sequences to obtain n pairs of reciprocity signals, a pair with the largest signal difference coefficient, namely a pair with the largest reciprocity damage, is selected from the reciprocity signals after correlation quantitative analysis, and the corresponding sensors are connected; and (4) carrying out reciprocity signal acquisition and quantitative analysis on the n sensors on the upper side and the sensors on the lower side of the weld joint in turn according to the processing, and finally obtaining connecting lines and corresponding intersection points of the n groups of sensors.

Further, the step 4 comprises the following sub-steps:

step 41: regions of a plurality of groups of connecting lines and corresponding intersection points are regions where defects possibly exist, and an ordered list decision diagram of the intersection points is generated by using an OPTICS algorithm through MATLAB;

step 42: judging the category number, namely the defect number according to the decision diagram, and selecting a proper neighborhood radius epsilon and a minimum domain point MinPts for clustering to obtain a corresponding cluster;

step 43: and acquiring a clustering center, namely the specific position of the defect, aiming at the cluster by using a k-means algorithm in an MATLAB toolbox.

In the technical scheme of the invention, n, i and j are all natural numbers.

Compared with the prior art, the invention has the following advantages:

(1) the method comprises the following steps of 2: preprocessing the reciprocity signals, defining signal difference coefficients, and performing correlation quantitative analysis on each pair of preprocessed reciprocity signals and step 3: the method comprises the steps of obtaining a pair with the largest signal difference coefficient in each group from a plurality of groups of pairs of reciprocity signals obtained by sequentially and interactively exciting and receiving each upper sensor and each lower sensor in a parallel linear sensor array, connecting the corresponding sensors to obtain a plurality of groups of connecting lines and corresponding intersection points, getting rid of dependence on baseline data under the condition that an object to be detected is not damaged, directly carrying out baseless detection and better meeting the actual requirements of engineering.

(2) The invention judges the number of the defects by an intelligent clustering algorithm OPTICS and realizes the accurate positioning of the specific positions of the defects by utilizing a k-means algorithm.

(3) The method is suitable for guided wave nondestructive detection in a plate structure (complex structure) with a welding seam.

(4) The invention improves the data acquisition and algorithm, has no special requirements on hardware, and can continue to use the original nondestructive testing equipment for testing.

(5) The invention has accurate positioning on the defects, clear imaging result and higher engineering application value.

Drawings

FIG. 1 is a schematic flow diagram of the present invention;

FIG. 2 is a schematic diagram of a welded steel plate and set defects according to an embodiment of the present invention;

FIG. 3 is a sensor array arrangement in an embodiment of the present invention;

FIG. 4 is a time domain diagram of the reciprocity signal after preprocessing in an embodiment of the invention. FIG. 4(a) is a time domain diagram of a reciprocity signal when a defect is located on a link between an excitation sensor and a reception sensor, and FIG. 4(b) is a time domain diagram of a reciprocity signal when a defect is located away from a link between an excitation sensor and a reception sensor;

FIG. 5 is a diagram of links and intersections of pairs of sensors in an embodiment of the present invention;

FIG. 6 is a diagram illustrating an ordered list decision made by the OPTIC algorithm in an embodiment of the present invention;

FIG. 7 is a schematic diagram of a clustering result extracted by the OPTIC algorithm in the embodiment of the present invention;

FIG. 8 is a diagram illustrating a cluster center result obtained by a k-means algorithm according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.

Examples

FIG. 1 is a flow chart of the present invention. As shown in fig. 1, the present embodiment provides a method for baseline-less defect detection and location of a welded steel plate based on reciprocity impairment, comprising the following steps:

step 1: collecting guided wave detection signals in a steel plate with a welding seam, and specifically comprising the following steps:

step 1-1: 18 (n-9) piezoelectric wafer sensors were arranged on a 1000mm x 3mm steel plate with a weld, the piezoelectric wafer having a diameter of 10mm and a thickness of 1 mm. The welding seam is arranged in the center of the steel plate, a rectangular coordinate system is established by taking the lower left corner of the steel plate as the origin of coordinates, a steel ingot is arranged between the sensor on one side and the welding seam, the diameter of the steel ingot is 10mm, the defect is simulated, the coordinates are (420, 360), and as shown in figure 2, (weld in the figure represents the welding seam, and defect represents the defect). The array type formed by the sensors is a parallel linear array, the sensors are symmetrically arranged on two sides of a welding seam, 9 sensors are arranged on each side, the distance between each sensor on the same side is 40mm, the distance between the upper sensor array and the lower sensor array is 400mm, the sensors on the upper side are numbered according to 1-9, the sensors on the lower side are numbered according to 1 'to 9', and the specific arrangement is shown in figure 3 (sensors in the figure represent sensors).

Step 1-2: a sine wave pulse signal modulated by a 5-period Hanning window is adopted as an excitation signal, and the center frequency is 250 kHz. Firstly, the No. 1 sensor is excited, and another sensor 1' at the lower side receives the excited signal to obtain a group of data S_11’Exchanging the receiving sequence of the excitation of the upper and lower sensors to obtain another set of data S_1’1，S_11’And S_1’1A pair of reciprocity signals is formed. No. 1 sensor and No. 1 '-9' sensors on the lower side of the welding seam are 9 sensors in total according to the methodBy exchanging the order of reception of the excitations in turn, 9 pairs of reciprocity signals can be obtained, named first large group of reciprocity signals. In the same way, the sensor No. 2 and the sensor No. 1 '-9' at the lower side also acquire reciprocity signals, and 9 pairs of reciprocity signals are obtained and named as a second large group of reciprocity signals. The number 1-9 sensors and the lower side sensor are sequentially subjected to reciprocity signal acquisition according to the method, and nine large groups of 81 pairs of reciprocity signals can be obtained.

Step 2: preprocessing signals, and specifically comprises the following steps:

step 2-1: in order to eliminate the interference caused by boundary reflection and mode conversion, signals before the mode wave packet of the first section A0 is intercepted and preprocessed. And setting a band-pass filter by using a button function and a filter function in the MATLAB, and filtering the acquired reciprocity signal to remove unnecessary noise signals. The center frequency of the band-pass filter is 250kHz, and the bandwidth is 100 kHz. After bandpass filtering, the reciprocity signal was further wavelet denoised with a db40 wavelet using the wden function in the MATLAB toolbox.

Step 2-2: since lamb waves attenuate as the distance traveled increases and where structural discontinuities are encountered, the amplitude of the signals received from different pairs of sensors will vary significantly. For better data comparison, normalization processing is performed to convert the data into values in [0, 1] interval collectively. The transformation used for normalization is:

wherein x is_iAnd y_iRespectively, an initial signal at any time in a pair of reciprocity signals;

and

respectively, after normalization in a pair of reciprocity signals; x is the number of_maxAnd x_minAre the maximum and minimum values in one of the initial sets of signals. Fig. 4 is a time domain diagram of a pair of reciprocity signals after the preprocessing, which is represented by a solid line and a dashed line on the same diagram. FIGS. 4(a) and 4(b) are time domain graphs of reciprocity signals after preprocessing, respectively, where the defect is on the line connecting the excitation and reception sensors and the defect is away from the line connecting the excitation and reception sensors, (where Amplitude represents Amplitude).

And step 3: and carrying out correlation analysis on the preprocessed signals, wherein the correlation analysis comprises the following specific steps:

step 3-1: and (3) solving the correlation coefficient of each pair of reciprocity signals by using MATLAB, wherein the expression is as follows:

in the formula, ρ_XYAnd the correlation coefficient is used for representing the similarity degree between the two signals, namely the coincidence degree of the two signals in a time domain diagram is represented, the higher the coincidence degree is, the larger the correlation coefficient is, Cov (X, Y) represents the covariance of the signal X and the signal Y, D (X) represents the variance of the signal X, and D (Y) represents the variance of the signal Y.

The correlation coefficient can represent the similarity between two signals, which is intuitively the coincidence degree of the two signals on a time domain diagram, and the higher the coincidence degree is, the larger the correlation coefficient is.

Step 3-2: after the signal correlation coefficients are found in step 3-1, to quantify the difference between the reciprocal signals, signal difference coefficients are then defined, which are expressed as:

SDC_XY＝1-ρ_XY

And 4, step 4: according to the signal difference coefficient, correspondingSensor lines and their intersections. The reciprocity principle refers to the inverse symmetry between the field emission source and the receiver when the elastic wave propagates in the material medium, if the Lamb wave propagation path region is linear elastic, then the signal S_ijAnd S_jiThere is theoretically no difference, i.e. a pair of perfectly reciprocal reciprocity signals. Because the welding seam can be regarded as a nonlinear area, if the connecting line of the two sensors ij has defects (steel ingots), the amplitudes of the guided waves when reaching the welding seam when the sensors i and j are respectively used as excitation sensors are different, the welding seam can attenuate the guided waves, and because the attenuation degrees of nonlinear effects are different, the amplitudes of the finally received waves are different, namely, the reciprocity is damaged. The total number of the sets of the reciprocity signals is nine as described in step 1-2, a pair of data with the largest signal difference coefficient, that is, a pair of signals with the largest damaged reciprocity, is respectively found out from the reciprocity signals of each set, and the corresponding sensor is connected, so that the defect is likely to be located on the connection. A total of 9 pairs of sensor links and their intersections are obtained, the intersection of the links being likely to be at or near the location of the defect. The sensor connecting lines and the intersection points obtained by the experiment and the data analysis are shown in fig. 5.

And 5: and (4) quantifying and positioning the defects by using an OPTIC and k-means algorithm according to the data points obtained in the step (4), wherein the method comprises the following specific steps:

step 5-1: an ordered list decision graph of the above intersections is generated using the OPTICS algorithm using MATLAB. The OPTICS algorithm is insensitive to the input parameter neighborhood radius epsilon and the minimum domain point number MinPts, so that MinPts is generally selected to be 4 and epsilon is infinity according to experience, and objects in the data set are sorted to obtain an ordered object list. From this ordered list, a decision graph can be obtained, as shown in fig. 6.

Step 5-2: it can be seen from fig. 6 that the pattern is approximately a recessed valley, and the cluster is expressed as a valley in the coordinate axis, so that it is determined that the data has only one category, i.e., only one defect in the steel plate, which is consistent with the actual situation. And then selecting a proper neighborhood radius epsilon to extract clusters, wherein epsilon is selected to be 14, and the data points with the vertical coordinate less than 14 in the graph are classified into one class. The clusters extracted after removing the noise points by using MATLAB and using the OPTICS algorithm are shown in FIG. 7, wherein circles are represented as noise points (noise), cross points are valid points (cluster), and OPTICS Clustering represents the processing result of the OPTICS algorithm.

Step 5-3: and (5) solving the cluster center of the cluster class extracted in the step 5-2 by using MATLAB and a k-means algorithm. The extracted cluster center is regarded as the position of the defect, as shown in fig. 8, wherein the black solid circle is the actual position of the defect (actual defect), the white square is the position of the defect (cluster center) located according to the algorithm, and the white circle represents the sensor position (sensor). As can be seen from FIG. 8, the positioning result is very close to the actual result, and it can be considered that the position of the defect in the steel plate with the weld joint can be accurately positioned by combining the baseline-free method and the intelligent algorithm.

While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for detecting and positioning defects of a welded steel plate without a base line based on reciprocity damage is characterized by comprising the following steps:

step 1: arranging a horizontal parallel linear sensor array, and acquiring reciprocity signals after setting steel ingot simulation defects;

and step 3: acquiring a pair with the largest signal difference coefficient in each group, namely a pair with the largest reciprocity damage, from a plurality of groups of pairs of reciprocity signals acquired by sequentially exciting and receiving upper and lower sensors in a parallel linear sensor array, connecting the corresponding sensors to obtain a plurality of groups of connecting lines and corresponding intersection points;

2. The method for baseline-less defect detection and location of a welded steel plate based on reciprocity damage according to claim 1, wherein the method for obtaining the reciprocity signal in step 1 comprises the following steps:

step 11: the method comprises the following steps that (1) horizontal parallel linear sensor arrays are symmetrically arranged on two sides of a weld joint of a steel plate with the weld joint, n sensors are arranged on each side, and the sensors on the same side are arranged at equal intervals;

3. The method for baseline-free defect detection and location of a welded steel plate based on reciprocity damage according to claim 1, wherein the step 2 comprises the following sub-steps:

4. The method for baseline-less detection and location of defects of a welded steel plate based on reciprocity impairment of claim 3, wherein the correlation coefficient in step 22 is described by the formula:

5. The method for baseline-less detection and location of defects of a welded steel plate based on reciprocity impairment of claim 3, wherein the signal difference coefficient in step 23 is described by the formula:

SDC_XY＝1-ρ_XY

6. The method for baseline-free defect detection and location of a welded steel plate based on reciprocity damage according to claim 1, wherein the step 3 specifically comprises: one of the sensors on the upper side of the welding seam and each sensor on the lower side are alternately switched to excite and receive sequences to obtain n pairs of reciprocity signals, a pair with the largest signal difference coefficient, namely a pair with the largest reciprocity damage, is selected from the reciprocity signals after correlation quantitative analysis, and the corresponding sensors are connected; and (4) carrying out reciprocity signal acquisition and quantitative analysis on the n sensors on the upper side and the sensors on the lower side of the weld joint in turn according to the processing, and finally obtaining connecting lines and corresponding intersection points of the n groups of sensors.

7. The method for baseline-free defect detection and location of a welded steel plate based on reciprocity damage according to claim 1, wherein the step 4 comprises the following sub-steps: