An electromagnetic ultrasonic adaptive filtering method applied in 27.5kv / 50hz strong electromagnetic environment

Abstract

The application discloses an electromagnetic ultrasonic adaptive filtering method applied in a 27.5KV / 50Hz strong electromagnetic environment and belongs to the technical field of nondestructive testing and signal processing. Aiming at how to effectively suppress strong periodic power frequency interference and strong random pulse interference and how to extract electromagnetic ultrasonic echo signals from seriously distorted signals with high fidelity, the method comprises the following steps: firstly, signals of a main channel and a reference channel are synchronously collected, and when an instantaneous amplitude exceeds a preset dynamic threshold, it is determined that strong pulse interference exists through real-time amplitude detection; then, the repaired reference channel signal is taken as input, and periodic power frequency interference in the main channel is cancelled through a least mean square adaptive filter; finally, a method based on singular value decomposition and singular value differential spectrum peak is used to suppress residual broadband background noise. The application can extract electromagnetic ultrasonic echoes from seriously distorted signals with high fidelity, and effectively solves the problems of signal distortion under pulse interference and poor power frequency cancellation effect of a traditional method.

Description

Technical Field

[0001] This invention belongs to the field of nondestructive testing and signal processing technology, specifically relating to an electromagnetic ultrasonic adaptive filtering method applied in a strong electromagnetic environment of 27.5KV / 50Hz. Background Technology

[0002] Electromagnetic ultrasonic testing (EMAT) is widely used for non-destructive testing in harsh environments such as high temperature and high speed due to its significant advantages of being non-contact and requiring no coupling agent. In the field of electrified railways, this technology is particularly suitable for online inspection of rails and vehicle components. However, the operating environment of electrified railways is extremely unique, with overhead contact line voltages reaching 27.5 kV and frequencies of 50 Hz, and traction currents reaching thousands of amperes, creating an extremely strong power frequency electromagnetic field environment.

[0003] Even more seriously, the disconnection between the pantograph and the overhead contact line frequently generates high-intensity, wide-spectrum pulsed arc interference. The energy intensity of this pulse interference can be tens or even hundreds of times that of the useful electromagnetic ultrasonic signal (usually in the microvolt range), which can easily cause the receiving amplifier to saturate and completely drown out the useful signal.

[0004] Existing technologies have significant shortcomings in dealing with such complex interference: (1) Traditional hardware shielding and filtering methods are not only limited in effect, but also the large shielding body is not conducive to on-site installation and deployment; (2) Conventional digital filters such as FIR and IIR are effective against stable power frequency interference, but they are powerless against non-stationary sudden pulse interference, and may even cause pulse energy to spread, resulting in signal distortion; (3) Time-frequency analysis methods such as wavelet denoising are effective against white noise, but for strong pulse interference with highly concentrated energy, threshold processing can easily leave "holes" in the signal, destroying the continuity of the echo signal, and may even misjudge the strong pulse itself as a defect signal; (4) The synchronous averaging method requires a large number of repetitive periodic signals, cannot cope with non-periodic random pulses, and is inefficient and not suitable for high-speed online detection.

[0005] Therefore, there is an urgent need for a processing method that can simultaneously and effectively suppress strong periodic power frequency interference and strong random pulse interference, and can extract electromagnetic ultrasonic echo signals from severely distorted signals with high fidelity. Summary of the Invention

[0006] To address the challenges of effectively suppressing strong periodic power frequency interference and strong random pulse interference, and extracting electromagnetic ultrasonic echo signals with high fidelity from severely distorted signals, this invention provides an adaptive electromagnetic ultrasonic filtering method for use in a 27.5KV / 50Hz strong electromagnetic environment. This method employs a "layered processing, pulse-first, periodic-later" strategy, combined with pulse characteristic identification and repair, and adaptive cancellation technology based on reference noise, to achieve precise filtering.

[0007] To achieve the above objectives, the present invention employs the following technical solutions:

[0008] An adaptive electromagnetic ultrasonic filtering method for use in a 27.5kV / 50Hz strong electromagnetic environment, the method comprising the following steps:

[0009] Step 1: Dual-channel synchronous signal acquisition: the main channel acquires the scrambled signal, and the reference channel acquires one or more reference noise signals. ;

[0010] The signal acquired by the main channel is an aliased signal received by the electromagnetic ultrasonic sensor, which includes ultrasonic echo signals. Periodic power frequency interference signals and random pulse interference signals ;

[0011] The reference noise signal acquired by the reference channel is obtained by a reference induction coil located near the electromagnetic ultrasonic sensor but not receiving ultrasonic echoes. This reference noise signal includes periodic power frequency interference signals related to the main channel. and random pulse interference signals It does not contain ultrasonic echo signals. ;

[0012] Step 2: Identification and Repair of Strong Pulse Interference: Real-time amplitude detection is performed on the signals acquired by the main channel and the reference channel. When the instantaneous amplitude of either signal exceeds a preset dynamic threshold, i.e. or The system determines that the signal at that moment contains strong pulse interference and precisely marks the pulse interference interval; among which... Main channel signal, For reference channel signal, For dynamic thresholds;

[0013] Then, an interpolation algorithm based on adaptive linear prediction is used to establish an autoregressive model using clean signal data before and after the pulse interference interval, to predict and repair the signal data within the pulse interference interval, so as to eliminate pulse spikes and restore the continuous shape of the signal.

[0014] The dynamic threshold in step 2 is set based on a multiple of the short-term root mean square value of the signal, using the following formula: ;

[0015] in, Indicates the threshold coefficient. This represents the short-term root mean square value of the signal.

[0016] In step 2, the autoregressive model is established using clean signal data before and after the pulse interference interval: ;

[0017] in, express Predicted signal value at any time (estimated signal for repair); express Keep moving forward The original clean signal of each sampling point (effective data before and after the interference interval); Represents the autoregressive coefficient; Indicates the signal sampling time; Indicates the sampling point offset (traversal) ); This indicates the order of the autoregressive model and represents the number of historical signal points involved in the prediction.

[0018] The signal data within the pulse interference range is predicted and repaired, and the repaired signal is: ;

[0019] in, This represents the original, undisturbed, normal signal; Indicates the start time of the pulse interference; This indicates the end time of the pulse interference.

[0020] Step 3: Adaptive cancellation of periodic power frequency interference: The reference noise signal of the reference channel after step 2 is processed... As a reference input, a minimum mean square adaptive filter is used to filter the aliased signal of the main channel after processing in step 2. This is achieved by continuously adjusting the weight coefficient vector of the minimum mean square adaptive filter. This makes the output of the minimum mean square adaptive filter... Periodic power frequency interference signals in the aliasing signal approaching the main channel Then the output of the least mean square adaptive filter Subtracting from the mixed signal in the main channel achieves adaptive cancellation of periodic power frequency interference;

[0021] The iterative formula for the least mean square adaptive filter in step 3 is: , ,

[0022] in, The error signal represents the desired pure ultrasonic signal output. This represents the confusion signal value of the main channel at the current moment. This is the reference noise signal vector of the reference channel at the current moment; This represents the weight coefficient vector at the current moment; To converge the step size; This represents the transpose of the weight coefficient vector estimate; This represents the updated weight coefficient vector at the next time step; This represents the weight coefficient vector at the current moment.

[0023] Step 4: Refined processing of residual noise: The signal processed in Step 3 is subjected to noise reduction based on singular value decomposition to reduce the noise in the one-dimensional signal. After constructing the Hankel matrix, singular value decomposition is performed. The number of effective singular values ​​is adaptively selected based on the peak value of the singular value difference spectrum for matrix reconstruction, thereby suppressing residual broadband background noise.

[0024] In step 4, the one-dimensional signal After constructing the Hankel matrix, singular value decomposition is performed. The Hankel matrix is... ,in, , , Represents a one-dimensional original signal Signal amplitude at different sampling points;

[0025] For matrix The singular value decomposition formula is as follows: ;

[0026] in, It is a singular value diagonal matrix;

[0027] The number of effective singular values ​​is adaptively selected based on the peak values ​​of the singular value difference spectrum. Take the front Reconstruct the matrix using the maximal singular values ​​to obtain the denoised Hankel matrix. Then, it is inversely transformed into a one-dimensional noise-reduced signal to suppress residual broadband background noise.

[0028] Step 5: Output the high signal-to-noise ratio electromagnetic ultrasonic signal after noise reduction.

[0029] Compared with the prior art, the present invention has the following advantages:

[0030] This invention addresses periodic power frequency interference and random pulse arc interference in a 27.5kV / 50Hz strong electromagnetic environment by proposing a "layered processing, pulse-first, periodic-later" strategy. First, through dual-channel synchronous acquisition and dynamic threshold detection, strong pulse interference is accurately identified. An interpolation algorithm based on adaptive linear prediction is then used to repair the pulse interval signal, avoiding signal "holes" or pulse misjudgments caused by traditional methods, and achieving high-fidelity recovery of the ultrasonic echo. Next, least mean square adaptive filtering is performed using the reference channel signal to effectively cancel periodic power frequency interference, eliminating the need for manual modeling and demonstrating strong adaptability. Finally, singular value decomposition is used to finely suppress residual broadband noise, further improving the signal-to-noise ratio. This method is computationally efficient, requires no long-term signal accumulation, is suitable for high-speed online detection, and is easier to deploy in engineering compared to traditional hardware shielding solutions, significantly improving the reliability and accuracy of electromagnetic ultrasonic testing in strong electromagnetic environments. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating an adaptive electromagnetic ultrasonic filtering method applied in a strong electromagnetic environment of 27.5KV / 50Hz. Detailed Implementation

[0032] To gain a deeper understanding of this invention, we will provide a comprehensive and detailed description. However, this invention has various implementations and is not limited to the specific examples listed herein. These examples are presented to enhance a full understanding of the disclosure of this invention.

[0033] An adaptive electromagnetic ultrasonic filtering method for use in a 27.5kV / 50Hz strong electromagnetic environment, the method comprising the following steps:

[0034] Step 1: Dual-channel synchronous signal acquisition: the main channel acquires the scrambled signal, and the reference channel acquires one or more reference noise signals. ;

[0035] The signal acquired by the main channel is an aliased signal received by the electromagnetic ultrasonic sensor, which includes ultrasonic echo signals. Periodic power frequency interference signals and random pulse interference signals ;

[0036] The reference noise signal acquired by the reference channel is obtained by a reference induction coil located near the electromagnetic ultrasonic sensor but not receiving ultrasonic echoes. This reference noise signal includes periodic power frequency interference signals related to the main channel. and random pulse interference signals It does not contain ultrasonic echo signals. ;

[0037] Step 2: Identification and Repair of Strong Pulse Interference: Real-time amplitude detection is performed on the signals acquired by the main channel and the reference channel. When the instantaneous amplitude of either signal exceeds a preset dynamic threshold, i.e. or The system determines that the signal at that moment contains strong pulse interference and precisely marks the pulse interference interval; among which... Main channel signal, For reference channel signal, For dynamic thresholds;

[0038] Then, an interpolation algorithm based on adaptive linear prediction is used to establish an autoregressive model using clean signal data before and after the pulse interference interval, to predict and repair the signal data within the pulse interference interval, so as to eliminate pulse spikes and restore the continuous shape of the signal.

[0039] The dynamic threshold in step 2 is set based on a multiple of the short-term root mean square value of the signal, using the following formula: ;

[0040] in, Indicates the threshold coefficient. This represents the short-term root mean square value of the signal.

[0041] In step 2, the autoregressive model is established using clean signal data before and after the pulse interference interval: ;

[0042] in, express Predicted signal value at any time (estimated signal for repair); express Keep moving forward The original clean signal of each sampling point (effective data before and after the interference interval); Represents the autoregressive coefficient; Indicates the signal sampling time; Indicates the sampling point offset (traversal) ); This indicates the order of the autoregressive model and represents the number of historical signal points involved in the prediction.

[0043] The signal data within the pulse interference range is predicted and repaired, and the repaired signal is: ;

[0044] in, This represents the original, undisturbed, normal signal; Indicates the start time of the pulse interference; This indicates the end time of the pulse interference.

[0045] Step 3: Adaptive cancellation of periodic power frequency interference: The reference noise signal of the reference channel after step 2 is processed... As a reference input, a minimum mean square adaptive filter is used to filter the aliased signal of the main channel after processing in step 2. This is achieved by continuously adjusting the weight coefficient vector of the minimum mean square adaptive filter. This makes the output of the minimum mean square adaptive filter... Periodic power frequency interference signals in the aliasing signal approaching the main channel Then the output of the least mean square adaptive filter Subtracting from the mixed signal in the main channel achieves adaptive cancellation of periodic power frequency interference;

[0046] The iterative formula for the least mean square adaptive filter in step 3 is: , ,

[0047] in, The error signal represents the desired pure ultrasonic signal output. This represents the confusion signal value of the main channel at the current moment. This is the reference noise signal vector of the reference channel at the current moment; This represents the weight coefficient vector at the current moment; To converge the step size; This represents the transpose of the weight coefficient vector estimate; This represents the updated weight coefficient vector at the next time step; This represents the weight coefficient vector at the current moment.

[0048] Step 4: Refined processing of residual noise: The signal processed in Step 3 is subjected to noise reduction based on singular value decomposition to reduce the noise in the one-dimensional signal. After constructing the Hankel matrix, singular value decomposition is performed. The number of effective singular values ​​is adaptively selected based on the peak value of the singular value difference spectrum for matrix reconstruction, thereby suppressing residual broadband background noise.

[0049] In step 4, the one-dimensional signal After constructing the Hankel matrix, singular value decomposition is performed. The Hankel matrix is... ,in, , , Represents a one-dimensional original signal Signal amplitude at different sampling points;

[0050] For matrix The singular value decomposition formula is as follows: ;

[0051] in, It is a singular value diagonal matrix;

[0052] The number of effective singular values ​​is adaptively selected based on the peak values ​​of the singular value difference spectrum. Take the front Reconstruct the matrix using the maximal singular values ​​to obtain the denoised Hankel matrix. Then, it is inversely transformed into a one-dimensional noise-reduced signal to suppress residual broadband background noise.

[0053] Step 5: Output the high signal-to-noise ratio electromagnetic ultrasonic signal after noise reduction.

[0054] Contents not described in detail in this specification are prior art known to those skilled in the art. Although illustrative specific embodiments of the invention have been described above to facilitate understanding by those skilled in the art, it should be understood that the invention is not limited to the scope of the specific embodiments. Various modifications are readily apparent to those skilled in the art as long as they fall within the spirit and scope of the invention as defined and determined by the appended claims, and all inventions utilizing the concept of this invention are protected.

Claims

1. An adaptive electromagnetic ultrasonic filtering method applied in a 27.5KV / 50Hz strong electromagnetic environment, characterized in that, The method includes the following steps: Step 1: Dual-channel synchronous signal acquisition: the main channel acquires the scrambled signal, and the reference channel acquires one or more reference noise signals. ; The signal acquired by the main channel is an aliased signal received by the electromagnetic ultrasonic sensor, which includes ultrasonic echo signals. Periodic power frequency interference signals and random pulse interference signals ; The reference noise signal acquired by the reference channel is obtained by a reference induction coil located near the electromagnetic ultrasonic sensor but not receiving ultrasonic echoes. This reference noise signal includes periodic power frequency interference signals related to the main channel. and random pulse interference signals It does not contain ultrasonic echo signals. ; Step 2: Identification and Repair of Strong Pulse Interference: Real-time amplitude detection is performed on the signals acquired by the main channel and the reference channel. When the instantaneous amplitude of either signal exceeds a preset dynamic threshold, i.e. or The system determines that the signal at that moment contains strong pulse interference and precisely marks the pulse interference interval; among which... Main channel signal, For reference channel signal, Dynamic threshold; Then, an interpolation algorithm based on adaptive linear prediction is used to establish an autoregressive model using clean signal data before and after the pulse interference interval, to predict and repair the signal data within the pulse interference interval, so as to eliminate pulse spikes and restore the continuous shape of the signal. Step 3: Adaptive cancellation of periodic power frequency interference: The reference noise signal of the reference channel after step 2 is processed... As a reference input, a minimum mean square adaptive filter is used to filter the aliased signal of the main channel after processing in step 2. This is achieved by continuously adjusting the weight coefficient vector of the minimum mean square adaptive filter. This makes the output of the minimum mean square adaptive filter... Periodic power frequency interference signals in the aliasing signal approaching the main channel Then the output of the least mean square adaptive filter Subtracting from the mixed signal in the main channel achieves adaptive cancellation of periodic power frequency interference; Step 4: Refined processing of residual noise: The signal processed in Step 3 is subjected to noise reduction based on singular value decomposition to reduce the noise in the one-dimensional signal. After constructing the Hankel matrix, singular value decomposition is performed. The number of effective singular values ​​is adaptively selected based on the peak value of the singular value difference spectrum for matrix reconstruction, thereby suppressing residual broadband background noise. Step 5: Output the high signal-to-noise ratio electromagnetic ultrasonic signal after noise reduction.

2. The electromagnetic ultrasonic adaptive filtering method applied in a 27.5KV / 50Hz strong electromagnetic environment according to claim 1, characterized in that, The dynamic threshold in step 2 is set based on a multiple of the short-term root mean square value of the signal, using the following formula: ; in, Represents the threshold coefficient. This represents the short-term root mean square value of the signal.

3. The electromagnetic ultrasonic adaptive filtering method applied in a 27.5KV / 50Hz strong electromagnetic environment according to claim 2, characterized in that, In step 2, the autoregressive model is established using clean signal data before and after the pulse interference interval: ; in, express Predict signal values ​​at any time; express Keep moving forward The original clean signal of each sampling point; Represents the autoregressive coefficient; Indicates the signal sampling time; Indicates the sampling point offset; This indicates the order of the autoregressive model and represents the number of historical signal points involved in the prediction. The signal data within the pulse interference range is predicted and repaired, and the repaired signal is: ； in, This represents the original, undisturbed, normal signal; Indicates the start time of the pulse interference; This indicates the end time of the pulse interference.

4. The electromagnetic ultrasonic adaptive filtering method applied in a 27.5KV / 50Hz strong electromagnetic environment according to claim 3, characterized in that, The iterative formula for the least mean square adaptive filter in step 3 is: ， ； in, The error signal represents the desired pure ultrasonic signal output. This represents the confusion signal value of the main channel at the current moment. The reference noise signal for the reference channel at the current moment; This represents the weight coefficient vector at the current moment; To converge the step size; This represents the transpose of the weight coefficient vector estimate; This represents the updated weight coefficient vector at the next time step; This represents the weight coefficient vector at the current moment.

5. The electromagnetic ultrasonic adaptive filtering method applied in a 27.5KV / 50Hz strong electromagnetic environment according to claim 4, characterized in that, In step 4, the one-dimensional signal After constructing the Hankel matrix, singular value decomposition is performed. The Hankel matrix is... ,in, , , Represents a one-dimensional original signal Signal amplitude at different sampling points; For matrix The singular value decomposition formula is as follows: ; in, It is a singular value diagonal matrix; The number of effective singular values ​​is adaptively selected based on the peak values ​​of the singular value difference spectrum. Take the front Reconstruct the matrix using the maximal singular values ​​to obtain the denoised Hankel matrix. Then, it is inversely transformed into a one-dimensional noise-reduced signal to suppress residual broadband background noise.

Images