A method of nonlinear filtering of a pulsed eddy current inspection signal
By employing a nonlinear filtering method for pulse eddy current detection signals, and performing logarithmic transformation and adaptive filtering on the time axis, the problem of noise interference in pulse eddy current detection is solved, thereby improving the filtering effect and evaluation accuracy of the signal.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GD POWER JIUQUAN GENERATION CO LTD
- Filing Date
- 2026-03-20
- Publication Date
- 2026-06-26
AI Technical Summary
In the prior art, pulsed eddy current detection signals in metal structures with cladding layers are easily affected by air white noise, power frequency noise and internal oscillations, resulting in insufficient accuracy in attenuation curve slope extraction and wall thickness determination. Furthermore, traditional filtering methods are difficult to balance noise suppression and signal fidelity.
A nonlinear filtering method is used to perform a logarithmic transformation on the time axis of the magnetic field decay voltage signal, and a fixed-length smoothing filtering time window is set on the nonlinearly compressed time axis. The filtering intensity is adaptively adjusted according to the time change, and the signal is processed by mean, weighted average or median filtering.
It improves the targeting and stability of signal filtering, effectively suppresses noise interference, reduces signal phase shift and high-frequency component distortion, and improves the accuracy of wall thickness and condition assessment.
Smart Images
Figure CN121899247B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of nondestructive testing technology, specifically relating to a nonlinear filtering method for pulsed eddy current detection signals. Background Technology
[0002] Steel structural components with cladding (such as pressure vessels, pressure pipelines, and boilers in the petroleum, chemical, and power industries) operate under harsh conditions such as high temperature, high pressure, and corrosion for extended periods, making them prone to defects such as thinning, corrosion, and cracking. Therefore, in-service non-destructive testing (NDT) is necessary to ensure safe operation. Pulsed eddy current testing (PECT) uses square wave excitation and features a rich spectrum and non-contact testing capabilities: low-frequency components have strong penetration capabilities, making them suitable for detecting wall thickness and deep defects beneath the cladding; high-frequency components are more sensitive to surface defects, allowing for the acquisition of multi-depth information in a single excitation. Therefore, PECT is widely used for the inspection of clad metal structural components.
[0003] When evaluating wall thickness based on the magnetic field attenuation curve of the received signal, the attenuated voltage signal collected by the receiving coil is easily coupled with interference from air white noise, power frequency noise, and internal oscillations. Even after hardware filtering, significant noise remains, affecting the accuracy of attenuation curve slope extraction and wall thickness determination. Existing technologies commonly use smoothing filters to suppress some noise, but while increasing the time window can improve the smoothing effect, it easily causes signal phase shift, affecting subsequent feature calculations. Bidirectional smoothing can reduce phase shift to some extent, but a longer window is still required to improve the signal-to-noise ratio in the later stages, which may cause distortion of high-frequency components. It is difficult to balance the filtering effect and waveform fidelity of early and later signals. Therefore, we propose a nonlinear filtering method for pulsed eddy current detection signals. Summary of the Invention
[0004] The present invention aims to at least solve one of the technical problems existing in the prior art, and provides a nonlinear filtering method for pulse eddy current detection signals.
[0005] This invention provides a nonlinear filtering method for pulsed eddy current detection signals, comprising the following steps:
[0006] S1: Acquire the magnetic field attenuation voltage signal of the metal structure with cladding layer;
[0007] S2: Perform a nonlinear compression transformation on the time axis corresponding to the magnetic field attenuation voltage signal;
[0008] S3: Set a smoothing filter time window of a preset length on the time axis after nonlinear compression;
[0009] S4: Determine the corresponding regular time interval using the smoothing filter time window as the unit;
[0010] S5: The magnetic field attenuation voltage signal falling within the conventional time interval is smoothed and filtered to obtain the filtered pulse eddy current detection signal.
[0011] Furthermore, in step S2, the nonlinear compression transformation is a logarithmic transformation of the time axis.
[0012] Specifically, the regular time values on the time axis are denoted as... t i The logarithmic time value after logarithmic transformation is denoted as T i , ,in, t 0 is the reference time constant. lg It represents a logarithm with base 10.
[0013] Specifically, in step S3, the length of the smoothing filtering time window remains unchanged in the time axis after nonlinear compression.
[0014] Preferably, in step S4, the smoothing filtering time window corresponds to a time interval in the time axis before the nonlinear compression transformation, and the length of the time interval increases nonlinearly with time.
[0015] Specifically, in step S5, the smoothing filtering process includes: using a mean smoothing filter to take the average value of the magnetic field decay voltage signal within the time interval as the filtering result.
[0016] Furthermore, the smoothing filtering process is one of the following: weighted average filtering, median filtering, and moving average filtering.
[0017] Furthermore, in step S3, the value range of the smoothing filter time window is 0.01 to 0.5.
[0018] Furthermore, the smoothed and filtered magnetic field attenuation voltage signal is used to form a curve of the magnetic field attenuation voltage signal changing with time, and the rate of change parameter within a predetermined time interval is extracted from the curve.
[0019] Specifically, in the nonlinearly compressed time axis, the smoothing filtering time window is determined based on the time point corresponding to the magnetic field decay voltage signal.
[0020] The beneficial effects of this invention are as follows:
[0021] This invention performs nonlinear compression on the time axis corresponding to the pulsed eddy current detection signal and sets a smoothing filtering time window on the nonlinearly compressed time axis. This matches the filtering process with the time characteristics of the magnetic field attenuation signal, improving the targeting of the filtering process. A fixed-length smoothing filtering time window is used in the nonlinearly compressed time axis, forming a filtering time interval that changes with time in the conventional time axis. This allows the filtering intensity in the early and later stages of the signal to be adaptively adjusted, suppressing noise while reducing the impact on the effective signal characteristics. By extracting the variation characteristics of the filtered magnetic field attenuation voltage signal, the stability of subsequent analysis of the magnetic field attenuation signal is improved, which is beneficial for assessing the wall thickness or condition of clad metal structures. Attached Figure Description
[0022] Figure 1 This is a flowchart illustrating the steps of a nonlinear filtering method for pulsed eddy current detection signals according to a specific embodiment of the present invention.
[0023] Figure 2 The image shows an experimental signal after nonlinear compression and smoothing filtering, which is a specific embodiment of the nonlinear filtering method for pulse eddy current detection signals according to the present invention.
[0024] Figure 3 This is a schematic diagram illustrating the working principle of the nonlinear filtering method for pulse eddy current detection signals in a coated pipeline, according to a specific embodiment of the present invention, for excitation and reception of pulse eddy current detection signals.
[0025] Figure 4 This is a schematic diagram illustrating the principle of nonlinear filtering of pulse eddy current detection signals in a coated pipeline, according to a specific embodiment of the present invention.
[0026] The components are: 1. Pipeline; 2. Covering filler; 3. Covering skin; 4. Processing module. Detailed Implementation
[0027] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0028] like Figure 1 , Figure 3 , Figure 4 As shown in the figure, a nonlinear filtering method for pulsed eddy current detection signals is provided by a specific embodiment of the present invention. The nonlinear filtering method is used for receiving and processing pulsed eddy current detection signals of clad metal structural components, and includes the following steps:
[0029] S1: Acquire the magnetic field attenuation voltage signal of the metal structure with cladding layer;
[0030] S2: Perform nonlinear compression transformation on the time axis corresponding to the magnetic field decay voltage signal;
[0031] S3: Set a smoothing filter time window of a preset length on the nonlinearly compressed time axis;
[0032] S4: Determine the corresponding regular time interval using the smoothing filter time window as the unit;
[0033] S5: The magnetic field decay voltage signal falling within the normal time interval is smoothed and filtered to obtain the filtered pulse eddy current detection signal.
[0034] Specifically, in step S1, the magnetic field attenuation voltage signal is an induced voltage signal that decays over time, obtained by the receiving coil after the excitation signal is turned off during the pulse eddy current detection process. Pulse eddy current detection is a non-destructive testing method based on the principle of electromagnetic induction. By applying a pulse current to the excitation coil, eddy currents are excited in the pipe with the cladding layer being tested, and the attenuation characteristics of the eddy current magnetic field are detected to detect the pipe wall thickness or defect status under the cladding layer. The clad metal structure generally includes a pipe 1, a cladding layer filler 2, and a cladding layer skin 3. The detection signal is generally sent to the processing module 4 for processing.
[0035] Furthermore, the magnetic field attenuation voltage signal includes effective signal components generated by the attenuation of eddy currents inside the workpiece, as well as noise components, including environmental noise, power frequency interference, and internal noise of the detection system.
[0036] Furthermore, the attenuation characteristics of the magnetic field attenuation voltage signal are related to the wall thickness and internal defect state of the clad metal structure.
[0037] Based on the above basic implementation method, in step S2, the nonlinear compression transformation is performed as a logarithmic transformation on the time axis; the conventional time values on the time axis are denoted as... t i ,and t i >0, the logarithmic time value after logarithmic transformation is denoted as T i , ;
[0038] in, t 0 is the reference time constant. lg Represents the logarithm to base 10. T i The time value is a dimensionless logarithmic value.
[0039] Specifically, by performing a logarithmic transformation on the time axis, the time resolution of the magnetic field decay voltage signal is improved in the early time period, while the time span is compressed in the later time period, thus matching the exponential decay characteristics of the magnetic field decay signal.
[0040] In one specific implementation, in step S3, the length of the preset length of the smoothing filtering time window remains unchanged in the nonlinearly compressed time axis; in step S4, the length of the time interval corresponding to the smoothing filtering time window in the conventional time axis increases nonlinearly with time.
[0041] In this embodiment, in the early stage of the signal, since the corresponding normal time interval is short, the number of data points participating in the smoothing filter is small, thereby reducing the impact on the high-frequency components of the signal; in the later stage of the signal, since the corresponding normal time interval is long, the number of data points participating in the smoothing filter increases, thereby enhancing the noise suppression capability.
[0042] Specifically, the same length of smoothing filter time window corresponds to different lengths of regular time intervals at different time positions, so that the filter intensity changes adaptively with time.
[0043] In another specific embodiment, in step S5, the smoothing filtering process includes: using a mean smoothing filter to take the average value of the magnetic field decay voltage signal within the time interval as the filtering result; the smoothing filtering process can also be one of weighted average filtering, median filtering, and moving average filtering.
[0044] Specifically, the smoothing filtering method can be selected according to the detection object, noise level, and signal characteristics, so as to maintain the effective characteristics of the magnetic field attenuation voltage signal while suppressing noise.
[0045] In another specific embodiment, in step S3, the value range of the smoothing filtering time window is 0.01 to 0.5; the smoothed magnetic field decay voltage signal is used to form a curve of the magnetic field decay voltage signal changing with time, and the rate of change parameter within a predetermined time interval is extracted from the curve; the smoothing filtering time window is set with the time point corresponding to the magnetic field decay voltage signal as the center in the nonlinearly compressed time axis.
[0046] Furthermore, the rate of change parameter is used to characterize the trend of the magnetic field decay curve within a predetermined time interval; the rate of change parameter can be used for subsequent wall thickness analysis or condition assessment of clad metal structural components.
[0047] Furthermore, in step S3, a smoothing filter time window of length L is set on the nonlinearly compressed time axis, where L ranges from 0.01 to 0.5. Let any logarithmic time value on the nonlinearly compressed time axis be...T Then there is In the formula, T For logarithmic time values, t These are standard time values on a standard time axis. t 0 is the reference time constant (preferably 1s), and the logarithmic time value is used as the reference time constant. T The smoothing filter time window centered on [ T -L / 2, T +L / 2], its corresponding time interval on the regular time axis is [10 lgT–L / 2 10 lgT+L / 2 Therefore, the length of this time interval varies with the logarithmic time value. T The increase is not linear.
[0048] In another specific embodiment, the test specimens used in this embodiment are Q235 flat plate specimens of different thicknesses. Q235 carbon structural steel flat plate specimens of different thicknesses are selected as the test objects to simulate the actual testing scenario of clad metal structural components. The Q235 flat plate specimens have different thicknesses to reflect the changing characteristics of the magnetic field attenuation signal under different wall thicknesses. The surface of the specimen is covered with a non-conductive cladding layer to simulate the anti-corrosion layer or insulation layer of pipelines or structural components in actual engineering. The pulsed eddy current testing platform used in the experiment includes a pulsed eddy current meter, an excitation coil, a receiving coil, a data acquisition module, and a signal processing module. The excitation coil and the receiving coil are integrated into the same probe, which is positioned above the surface of the specimen, separated from the specimen by the cladding layer. In the experiment, the pulsed eddy current excitation current is set to 2A, the bipolar square wave duty cycle is 50%, and the excitation frequency is 1Hz.
[0049] Further, during the detection process, a pulsed excitation current is applied to the excitation coil, and the excitation signal is a square wave pulse signal. After the excitation signal is turned off, eddy currents are generated inside the Q235 flat plate specimen. The eddy currents decay over time and form a secondary magnetic field. The receiving coil induces a decaying magnetic field voltage signal that decays over time. The magnetic field decaying voltage signal is sampled by the data acquisition module to obtain the corresponding time series data; the sampling frequency is 50kHz, and the total duration of a falling curve interval is 0.25s = 250ms. The effective signal before 0.1s = 100ms is extracted, resulting in 5000 data points. Each voltage value of the received signal is... v i Corresponding standard time value t i Take the logarithm to obtain the logarithmic time value. T i ,in If the time unit is milliseconds (ms), then a reference time constant is used. t0 = 1ms. Signals from 1ms (inclusive) to 100ms (exclusive) are filtered, corresponding to a logarithmic time interval T ∈ [0, 2), a sampling period of 0.02ms, and 4950 data points. With a logarithmic time window length L = 0.1, the logarithmic time value at any center point is... T i Its logarithmic time window is [ T i -L / 2, T i +L / 2], the corresponding regular time interval is [10 lgTi–L / 2 10 lgTi+L / 2 The number of data points N within the interval satisfies: N = [(10)] lgTi–L / 2 -10 lgTi+L / 2 [0.02]+1;
[0050] When logarithmic time value T i When the logarithmic time value is 0.05, the logarithmic time window is [0, 0.1], the corresponding regular time interval is [1, 1.2589] ms, and the number of data points within the interval is [(1.2589-1) / 0.02]+1=13; when the logarithmic time value T i When the logarithmic time value is 0.5, the corresponding normal time interval is [2.8184, 3.5481] ms, and the number of data points within the interval is 37; when the logarithmic time value T i When the value is 1.5, the corresponding normal time interval is [28.1838, 35.4813] ms, and the number of data points within the interval is 365. This shows that when using a smoothing filter time window of the same length on the logarithmic time axis, the number of data points corresponding to it on the normal time axis increases with time, thus achieving adaptive variation of the filter strength over time.
[0051] The magnetic field decay voltage signal falling within a normal time interval is subjected to smoothing filtering. In this embodiment, the smoothing filtering adopts the mean smoothing filtering method, that is, the average value of the magnetic field decay voltage signal within the time interval is used as the filtering result; in other embodiments, weighted average filtering, median filtering, or moving average filtering can also be used, and the result is as follows. Figure 2 As shown.
[0052] Furthermore, after the aforementioned nonlinear filtering process, a filtered pulsed eddy current detection signal is obtained. The filtered magnetic field attenuation voltage signal is then used to form a curve showing the change of the magnetic field attenuation voltage signal over time, and the rate of change parameter of the curve is extracted within a predetermined time interval. Experimental results show that the rate of change parameters of the magnetic field attenuation curves for Q235 flat plate specimens of different thicknesses differ, and the extracted rate of change parameter exhibits higher stability after employing the nonlinear filtering method described in this invention, which is beneficial for distinguishing specimens of different thicknesses. It can be seen that this filtering method effectively suppresses noise interference in pulsed eddy current detection, and the filtering effect is greatly improved.
[0053] In summary, the embodiments disclosed herein have at least the following technical effects:
[0054] This invention addresses the exponential decay of the magnetic field attenuation voltage signal over time in pulsed eddy current detection by performing nonlinear compression processing on the time axis corresponding to the signal and setting a smoothing filtering time window on the nonlinearly compressed time axis to match the filtering process with the time characteristics of the magnetic field attenuation signal, thereby improving the targeting and effectiveness of the filtering process.
[0055] By using a fixed-length smoothing filtering time window in the nonlinearly compressed time axis, a filtering time interval that increases nonlinearly with time is formed in the conventional time axis. This results in a smaller filtering window in the early stage of the signal and a gradually increasing filtering window in the later stage, realizing the adaptive change of filtering intensity with time. This effectively suppresses noise components in the later signal without significantly weakening the effective signal in the early stage.
[0056] The nonlinear filtering method described in this invention can smooth the magnetic field attenuation voltage signal while reducing the signal phase shift and high-frequency component distortion caused by the fixed window length in traditional smoothing filtering methods, thereby better preserving the original characteristics of the magnetic field attenuation signal.
[0057] By generating a curve of the filtered magnetic field attenuation voltage signal as a function of time and extracting the rate of change parameter within a predetermined time interval, the stability of the magnetic field attenuation curve change feature extraction is improved, which is beneficial for subsequent analysis and evaluation of the wall thickness or state of clad metal structural components.
[0058] The nonlinear filtering method of this invention is simple to implement and can be completed by software algorithm without relying on complex hardware modifications. It is applicable to different types of pulse eddy current detection systems and has good versatility and engineering application value.
[0059] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.
Claims
1. A nonlinear filtering method for pulsed eddy current detection signals, characterized in that, Includes the following steps: S1: Acquire the magnetic field attenuation voltage signal of the metal structure with cladding layer; S2: Perform a nonlinear compression transformation on the time axis corresponding to the magnetic field attenuation voltage signal; S3: Set a smoothing filtering time window of a preset length on the nonlinearly compressed time axis. The length of the smoothing filtering time window remains unchanged in the nonlinearly compressed time axis. Furthermore, the smoothing filtering time window is determined based on the time point corresponding to the magnetic field attenuation voltage signal in the nonlinearly compressed time axis. S4: Determine the corresponding regular time interval using the smoothing filter time window as the unit. The smoothing filter time window corresponds to the time interval in the time axis before the nonlinear compression transformation. The length of the time interval increases nonlinearly with time. S5: The magnetic field attenuation voltage signal falling within the conventional time interval is smoothed and filtered to obtain the filtered pulse eddy current detection signal.
2. The nonlinear filtering method for pulsed eddy current detection signals according to claim 1, characterized in that, In step S2, the nonlinear compression transformation is a logarithmic transformation of the time axis.
3. The nonlinear filtering method for pulsed eddy current detection signals according to claim 2, characterized in that, The regular time values on the time axis are denoted as t i The logarithmic time value after logarithmic transformation is denoted as T i , ,in, t 0 is the reference time constant. lg It represents a logarithm with base 10.
4. The nonlinear filtering method for pulsed eddy current detection signals according to claim 1, characterized in that, In step S5, the smoothing filtering process includes: using a mean smoothing filter to take the average value of the magnetic field decay voltage signal within the time interval as the filtering result.
5. The nonlinear filtering method for pulsed eddy current detection signals according to claim 1, characterized in that, In step S3, the value range of the smoothing filter time window is 0.01 to 0.
5.
6. The nonlinear filtering method for pulsed eddy current detection signals according to claim 1, characterized in that, The smoothed and filtered magnetic field attenuation voltage signal is used to form a curve of the magnetic field attenuation voltage signal changing with time, and the rate of change parameter within a predetermined time interval is extracted from the curve.