Adaptive UHF partial discharge signal extraction device and method based on dual-modal noise reference

Abstract

This invention discloses an adaptive UHF partial discharge signal extraction device and method based on dual-modal noise reference, relating to the field of communication technology. It includes: a dual-sensor collaborative acquisition system comprising a directional UHF sensor group and a noise feature learning module; an adaptive noise cancellation device connected to the dual-sensor collaborative acquisition system; and an intelligent signal identification system connected to the adaptive noise cancellation device. The auxiliary sensor in the directional UHF sensor group specifically collects ambient background noise, and the noise feature learning module uses Hilbert-Huang transform to extract the time-frequency domain feature spectrum of the background noise, enabling the device to understand the changes in noise characteristics in real time. Simultaneously, the dynamic threshold adjustment unit in the adaptive noise cancellation device automatically adjusts the signal trigger threshold according to the noise energy level, achieving automatic tracking of noise changes and greatly improving the device's adaptability to complex noise environments.

Description

Technical Field

[0001] This invention relates to the field of communication technology, specifically to an adaptive UHF partial discharge signal extraction device and method based on dual-mode noise reference. Background Technology

[0002] In the monitoring of UHF partial discharge signals in the field of communications, past technologies have faced numerous challenges. With changes in the environment, UHF partial discharge signals in the field are subject to varying degrees of noise interference, and the intensity and characteristics of the noise also change accordingly. For example, in industrial settings, the operation of electrical equipment generates various types of electromagnetic noise, which superimposed on the UHF partial discharge signal, making the monitored signal more complex.

[0003] Traditional online partial discharge (PD) monitoring technologies struggle to automatically and accurately track noise changes. On one hand, they lack effective adaptive mechanisms to handle noise of varying frequencies and intensities, failing to adjust monitoring parameters promptly to adapt to changes in the noise environment. For example, when high-intensity impulse noise suddenly appears in the environment, traditional monitoring equipment may misinterpret the noise as a PD signal because it cannot adjust thresholds in time. On the other hand, traditional techniques have limitations in noise reduction. Existing filters typically only process specific types of noise, making it difficult to achieve comprehensive and effective noise reduction for complex and variable noise. For instance, while comb filters can suppress periodic interference, they are ineffective at handling non-stationary signals and sudden impulse noise, resulting in the extracted PD signal still containing a significant amount of noise, affecting the signal's accuracy and reliability.

[0004] Furthermore, traditional technologies lack efficient methods for distinguishing between partial discharge (PD) signals and interference signals. Because typical PD signals and industrial interference signals share certain similarities in characteristics, relying solely on simple feature analysis can easily lead to misjudgments, thereby reducing the accuracy of the monitoring system. For example, the phase distribution spectrum and bandgap energy ratio of some industrial interference signals are quite similar to those of PD signals, making it difficult for traditional methods to accurately distinguish between the two, resulting in biased monitoring results. These problems severely affect the accuracy and reliability of UHF PD signal monitoring, making it difficult to achieve ideal conditions for assessing electrical equipment, and consequently impacting the safe and stable operation of the equipment. Summary of the Invention

[0005] To address the aforementioned technical problems, an adaptive UHF partial discharge signal extraction device and method based on dual-mode noise reference are provided. This technical solution solves the problems mentioned above.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] An adaptive UHF partial discharge signal extraction device based on dual-mode noise reference includes:

[0008] A dual-sensor collaborative acquisition system, comprising a directional ultra-high frequency sensor group and a noise feature learning module;

[0009] An adaptive noise cancellation device is connected to the dual-sensor collaborative acquisition system;

[0010] The intelligent signal identification system is connected to the adaptive noise cancellation device;

[0011] The directional ultra-high frequency sensor group adopts an array antenna design, including a main sensor (UHF1) and an auxiliary sensor (UHF2). The main sensor (UHF1) is used to receive electromagnetic wave signals from the device body in a directional manner, and the auxiliary sensor (UHF2) is used to collect ambient background noise. The noise feature learning module is used to extract the time-frequency domain feature spectrum of the background noise through Hilbert-Huang transform.

[0012] Preferably, the adaptive noise cancellation device includes a dynamic threshold adjustment unit and a three-stage composite filter chain;

[0013] The dynamic threshold adjustment unit automatically adjusts the signal trigger threshold based on the noise energy level, and the sensitivity range of the dynamic threshold adjustment unit is 0.01-100mV;

[0014] The three-stage composite filter chain consists of a comb filter, a wavelet packet denoising module, and a neural network filter.

[0015] Preferably, the comb filter is used to suppress periodic interference, the wavelet packet denoising module is used to process non-stationary signals, and the neural network filter is used to identify burst noise.

[0016] Preferably, the intelligent signal identification system includes a multi-dimensional feature library and an online pattern recognition engine;

[0017] The multi-dimensional feature library contains feature parameters of more than 50 typical partial discharge signals and 30 types of industrial interference signals.

[0018] The online pattern recognition engine uses the DTW (Dynamic Time Warping) algorithm to align time-domain waveforms and combines it with a convolutional neural network for discharge pattern classification.

[0019] Preferably, the specific steps of the Hilbert-Huang transform include: performing empirical mode decomposition on the acquired background noise signal to obtain multiple intrinsic mode functions, and then performing Hilbert transform on each intrinsic mode function to obtain the time-frequency domain feature spectrum of the background noise.

[0020] Preferably, the formula for the dynamic threshold adjustment unit to automatically adjust the signal trigger threshold is:

[0021] T = k × E + b

[0022] Where T is the adjusted signal trigger threshold, E is the noise energy level, and k and b are preset coefficients, which are determined according to the actual application scenario.

[0023] Preferably, the formula for calculating the distance between two time-domain waveforms in the DTW dynamic time warping algorithm is as follows:

[0024]

[0025] Where x(i) and y(j) are the amplitudes of two time-domain waveforms at times i and j, respectively, and D(i,j) is the cumulative distance between the two waveforms at times i and j.

[0026] An adaptive UHF partial discharge signal extraction method based on dual-mode noise reference includes:

[0027] The main sensor (UHF1) of the directional ultra-high frequency sensor group receives electromagnetic wave signals from the device body in a directional manner, while the auxiliary sensor (UHF2) collects ambient background noise.

[0028] The noise feature learning module is used to extract the time-frequency domain feature map of background noise through Hilbert-Huang transform;

[0029] The acquired signal is transmitted to an adaptive noise cancellation device, which automatically adjusts the signal trigger threshold based on the noise energy level through a dynamic threshold adjustment unit and filters the signal through a three-stage composite filter chain. The filtered signal is then transmitted to an intelligent signal identification system, which uses an online pattern recognition engine to align the time-domain waveform with the DTW dynamic time warping algorithm and combines it with a convolutional neural network to classify the discharge mode of the signal, thereby extracting the UHF partial discharge signal.

[0030] Preferably, in the step of the adaptive noise cancellation device filtering the signal, the comb filter first suppresses periodic interference in the signal, then the wavelet packet noise reduction module processes the non-stationary signal, and finally the neural network filter identifies and eliminates pulse burst noise.

[0031] Preferably, the filtering formula of the comb filter is:

[0032]

[0033] In the formula, M is the filter order and z is a complex variable;

[0034] The wavelet packet denoising module uses the wavelet packet decomposition formula, specifically:

[0035]

[0036] In the formula, h(n) and g(n) are the coefficients of the low-pass and high-pass filters, respectively, j is the number of decomposition layers, k is the node number, and t is the time variable.

[0037] Compared with existing technologies, the advantages of this invention are as follows: Auxiliary sensors in the directional ultra-high frequency sensor array specifically collect ambient background noise, and the noise feature learning module uses Hilbert-Huang transform to extract the time-frequency domain feature spectrum of the background noise. This enables the device to understand the changes in noise characteristics in real time. Simultaneously, the dynamic threshold adjustment unit in the adaptive noise cancellation device automatically adjusts the signal trigger threshold according to the noise energy level, achieving automatic tracking of noise changes and greatly improving the device's adaptability to complex noise environments.

[0038] In terms of noise reduction, the comb filter, wavelet packet denoising module and neural network filter work together to process periodic interference, non-stationary signals and impulse burst noise respectively, which comprehensively and effectively eliminates various types of noise, improves the purity of the signal and makes the extracted UHF partial discharge signal more accurate and reliable.

[0039] To address the difficulty in distinguishing between partial discharge signals and interference signals, a multi-dimensional feature library containing characteristic parameters of various typical partial discharge signals and industrial interference signals was established. Furthermore, the DTW dynamic time warping algorithm combined with a convolutional neural network was used for discharge mode classification. This significantly improved the ability to identify different signals, reduced the false alarm rate, ensured the accuracy of monitoring results, provided a reliable basis for the condition assessment of electrical equipment, and guaranteed the safe and stable operation of the equipment. Attached Figure Description

[0040] Figure 1 This is a diagram illustrating the device assembly of the present invention;

[0041] Figure 2 This is a flowchart of the present invention. Detailed Implementation

[0042] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0043] Reference Figure 1 and Figure 2 As shown, the adaptive UHF partial discharge signal extraction device based on dual-mode noise reference includes:

[0044] A dual-sensor collaborative acquisition system, comprising a directional ultra-high frequency sensor group and a noise feature learning module;

[0045] An adaptive noise cancellation device is connected to the dual-sensor collaborative acquisition system;

[0046] The intelligent signal identification system is connected to the adaptive noise cancellation device;

[0047] The directional ultra-high frequency sensor group adopts an array antenna design, including a main sensor (UHF1) and an auxiliary sensor (UHF2). The main sensor (UHF1) is used to receive electromagnetic wave signals from the device body in a directional manner, and the auxiliary sensor (UHF2) is used to collect ambient background noise. The noise feature learning module is used to extract the time-frequency domain feature spectrum of the background noise through Hilbert-Huang transform.

[0048] Specifically, the device mainly consists of a dual-sensor collaborative acquisition system, an adaptive noise cancellation device, and an intelligent signal identification system. The directional UHF sensor group in the dual-sensor collaborative acquisition system receives electromagnetic wave signals emitted by the device itself through the main sensor (UHF1), which accurately acquires signals related to partial discharge (PD). The auxiliary sensor (UHF2) is responsible for collecting ambient background noise. The noise feature learning module uses the Hilbert-Huang transform to process the background noise collected by the auxiliary sensor, decomposing the complex noise signal and extracting time-frequency domain feature spectra to analyze the noise characteristics. The adaptive noise cancellation device and the intelligent signal identification system respectively perform noise reduction and classification processing on the collected signals, ultimately achieving the extraction of UHF PD signals. The dual-sensor collaborative acquisition system has a clear division of labor, making the acquisition of both device signals and background noise more targeted. Obtaining the noise time-frequency domain feature spectra through the Hilbert-Huang transform provides an important basis for subsequent noise cancellation and signal processing, helping to improve the adaptability of the entire device to complex noise environments and ensuring the accuracy of UHF PD signal extraction.

[0049] The adaptive noise cancellation device includes a dynamic threshold adjustment unit and a three-stage composite filter chain;

[0050] The dynamic threshold adjustment unit automatically adjusts the signal trigger threshold based on the noise energy level, and the sensitivity range of the dynamic threshold adjustment unit is 0.01-100mV;

[0051] The three-stage composite filter chain consists of a comb filter, a wavelet packet denoising module, and a neural network filter.

[0052] Specifically, the dynamic threshold adjustment unit automatically adjusts the signal trigger threshold based on changes in noise energy level, with sensitivity adjustable between 0.01-100mV. When noise energy is high, the trigger threshold is appropriately increased to avoid misjudgment due to noise interference; when noise energy is low, the trigger threshold is decreased to ensure that even weak partial discharge signals can be captured. The three-stage composite filter chain consists of a comb filter, a wavelet packet denoising module, and a neural network filter, processing different types of noise. The dynamic threshold adjustment unit improves the device's response to noise changes, ensuring the effectiveness of signal acquisition. The combination of the three-stage composite filter chain achieves comprehensive suppression of various types of noise, greatly improving signal quality and making subsequent processing of UHF partial discharge signals more reliable.

[0053] The comb filter is used to suppress periodic interference, the wavelet packet denoising module is used to process non-stationary signals, and the neural network filter is used to identify burst noise.

[0054] Specifically, comb filters utilize their filtering characteristics to suppress periodic interference in signals. They can identify and remove noise signals with periodic patterns. Wavelet packet denoising modules excel at handling non-stationary signals, whose statistical characteristics change over time. By decomposing the signal into multiple layers and dividing it into different frequency bands, wavelet packet denoising modules effectively remove non-stationary noise. Neural network filters utilize the powerful pattern recognition capabilities of neural networks to identify and eliminate burst noise. These three types of filters each perform their specific functions and work together to comprehensively eliminate noise with different characteristics, resulting in a purer filtered signal, improved extraction accuracy of UHF partial discharge signals, and reduced noise interference in signal analysis.

[0055] The intelligent signal identification system includes a multi-dimensional feature library and an online pattern recognition engine;

[0056] The multi-dimensional feature library contains feature parameters of more than 50 typical partial discharge signals and 30 types of industrial interference signals.

[0057] The online pattern recognition engine uses the DTW (Dynamic Time Warping) algorithm to align time-domain waveforms and combines it with a convolutional neural network for discharge pattern classification.

[0058] Specifically, the intelligent signal identification system consists of a multi-dimensional feature library and an online pattern recognition engine. The multi-dimensional feature library stores feature parameters for over 50 typical partial discharge signals and 30 types of industrial interference signals, covering aspects such as the signal's phase distribution spectrum, pulse rise time, and frequency band energy ratio. The online pattern recognition engine employs the DTW (Dynamic Time Warping) algorithm to align the acquired time-domain waveform with standard waveforms in the feature library, resolving the difficulty in comparison due to inconsistent signal time scales. Subsequently, combined with the powerful image recognition and classification capabilities of convolutional neural networks, the system classifies the signal by discharge mode, determining whether it is a partial discharge signal or an interference signal. The multi-dimensional feature library provides rich reference data for signal classification, and the combination of the DTW algorithm and convolutional neural network significantly improves the accuracy of signal identification, effectively avoiding misclassification of interference signals as partial discharge signals and enhancing the reliability of UHF partial discharge signal extraction.

[0059] The specific steps of the Hilbert-Huang transform include: performing empirical mode decomposition on the acquired background noise signal to obtain multiple intrinsic mode functions, and then performing Hilbert transform on each intrinsic mode function to obtain the time-frequency domain feature spectrum of the background noise.

[0060] Specifically, the Hilbert-Huang transform first performs empirical mode decomposition (EMD) on the acquired background noise signal. EMD decomposes a complex signal into several intrinsic mode functions (IMFs), which represent the signal's characteristics at different time scales. Then, a Hilbert transform is applied to each IMF to convert the time-domain signal to the frequency domain, ultimately obtaining the time-frequency domain feature map of the background noise, clearly showing the noise's distribution characteristics in time and frequency. This method enables in-depth analysis of background noise characteristics, providing detailed noise information for adaptive noise cancellation devices and intelligent signal identification systems, helping these devices better adapt to noisy environments and improving their ability to extract and analyze UHF partial discharge signals.

[0061] The formula for the dynamic threshold adjustment unit to automatically adjust the signal trigger threshold is:

[0062] T = k × E + b

[0063] Where T is the adjusted signal trigger threshold, E is the noise energy level, and k and b are preset coefficients, which are determined according to the actual application scenario.

[0064] Specifically, through this automatic adjustment mechanism, the device can flexibly adapt to environments with different noise intensities, avoid false triggering in high-noise environments, and not miss weak partial discharge signals in low-noise environments, thereby improving the accuracy and reliability of signal acquisition.

[0065] The formula for calculating the distance between two time-domain waveforms in the DTW dynamic time warping algorithm is as follows:

[0066]

[0067] Where x(i) and y(j) are the amplitudes of two time-domain waveforms at times i and j, respectively, and D(i,j) is the cumulative distance between the two waveforms at times i and j.

[0068] Specifically, the algorithm finds the optimal time warping path to better align the two waveforms in time, facilitating subsequent comparison and classification. This solves the problem of inconsistent time scales between waveforms in different time domains, enabling accurate matching of the acquired signals with standard waveforms in the feature library. This provides a more reliable data foundation for convolutional neural networks to classify discharge patterns, improving the accuracy of signal classification.

[0069] An adaptive UHF partial discharge signal extraction method based on dual-mode noise reference includes:

[0070] The main sensor (UHF1) of the directional ultra-high frequency sensor group receives electromagnetic wave signals from the device body in a directional manner, while the auxiliary sensor (UHF2) collects ambient background noise.

[0071] The noise feature learning module is used to extract the time-frequency domain feature map of background noise through Hilbert-Huang transform;

[0072] The acquired signal is transmitted to an adaptive noise cancellation device, which automatically adjusts the signal trigger threshold based on the noise energy level through a dynamic threshold adjustment unit and filters the signal through a three-stage composite filter chain. The filtered signal is then transmitted to an intelligent signal identification system, which uses an online pattern recognition engine to align the time-domain waveform with the DTW dynamic time warping algorithm and combines it with a convolutional neural network to classify the discharge mode of the signal, thereby extracting the UHF partial discharge signal.

[0073] Specifically, the method first utilizes the main and auxiliary sensors of a directional UHF sensor array to collect electromagnetic wave signals from the device itself and ambient background noise. Next, a noise feature learning module performs a Hilbert-Huang transform on the background noise to obtain its feature spectrum. The collected signals are transmitted to an adaptive noise cancellation device, where they undergo dynamic threshold adjustment and filtering through a three-stage composite filter chain. Finally, the filtered signals enter an intelligent signal identification system, where discharge mode classification is performed using the DTW (Dynamic Time Warping) algorithm and a convolutional neural network to extract the UHF partial discharge signal. This method's steps are closely interconnected, forming a complete UHF partial discharge signal extraction process from signal acquisition, noise analysis, noise reduction processing to signal classification, ensuring high efficiency and accuracy in signal extraction.

[0074] In the step of the adaptive noise cancellation device filtering the signal, the comb filter first suppresses periodic interference in the signal, then the wavelet packet denoising module processes the non-stationary signal, and finally the neural network filter identifies and eliminates pulse burst noise.

[0075] The filtering formula for the comb filter is:

[0076]

[0077] In the formula, M is the filter order and z is a complex variable;

[0078] The wavelet packet denoising module uses the wavelet packet decomposition formula, specifically:

[0079]

[0080] In the formula, h(n) and g(n) are the coefficients of the low-pass and high-pass filters, respectively, j is the number of decomposition layers, k is the node number, and t is the time variable.

[0081] Specifically, when the adaptive noise cancellation device filters the signal, the comb filter first suppresses periodic interference and removes periodic noise components. Then, the wavelet packet denoising module further processes the signal after the comb filter, decomposing it into different frequency bands for noise reduction, taking into account the characteristics of non-stationary signals. Finally, the neural network filter identifies and eliminates any potential burst noise in the signal. These formulas also provide a theoretical basis for filter design and optimization, enabling the comb filter and wavelet packet denoising module to process noise more accurately, improving the overall performance of the noise cancellation device and facilitating more accurate extraction of UHF partial discharge signals.

[0082] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. An adaptive UHF partial discharge signal extraction device based on dual-mode noise reference, characterized in that, include: A dual-sensor collaborative acquisition system, comprising a directional ultra-high frequency sensor group and a noise feature learning module; An adaptive noise cancellation device is connected to the dual-sensor collaborative acquisition system; The intelligent signal identification system is connected to the adaptive noise cancellation device; The directional ultra-high frequency sensor group adopts an array antenna design, including a main sensor (UHF1) and an auxiliary sensor (UHF2). The main sensor (UHF1) is used to receive electromagnetic wave signals from the device body in a directional manner, and the auxiliary sensor (UHF2) is used to collect ambient background noise. The noise feature learning module is used to extract the time-frequency domain feature spectrum of the background noise through Hilbert-Huang transform.

2. The adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to claim 1, characterized in that, The adaptive noise cancellation device includes a dynamic threshold adjustment unit and a three-stage composite filter chain; The dynamic threshold adjustment unit automatically adjusts the signal trigger threshold based on the noise energy level, and the sensitivity range of the dynamic threshold adjustment unit is 0.01-100mV; The three-stage composite filter chain consists of a comb filter, a wavelet packet denoising module, and a neural network filter.

3. The adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to claim 2, characterized in that, The comb filter is used to suppress periodic interference, the wavelet packet denoising module is used to process non-stationary signals, and the neural network filter is used to identify burst noise.

4. The adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to claim 1, characterized in that, The intelligent signal identification system includes a multi-dimensional feature library and an online pattern recognition engine; The multi-dimensional feature library contains feature parameters of more than 50 typical partial discharge signals and 30 types of industrial interference signals. The online pattern recognition engine uses the DTW (Dynamic Time Warping) algorithm to align time-domain waveforms and combines it with a convolutional neural network for discharge pattern classification.

5. The adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to claim 1, characterized in that, The specific steps of the Hilbert-Huang transform include: performing empirical mode decomposition on the acquired background noise signal to obtain multiple intrinsic mode functions, and then performing Hilbert transform on each intrinsic mode function to obtain the time-frequency domain feature spectrum of the background noise.

6. The adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to claim 2, characterized in that, The formula for the dynamic threshold adjustment unit to automatically adjust the signal trigger threshold is: T = k × E + b Where T is the adjusted signal trigger threshold, E is the noise energy level, and k and b are preset coefficients, which are determined according to the actual application scenario.

7. The adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to claim 4, characterized in that, The formula for calculating the distance between two time-domain waveforms in the DTW dynamic time warping algorithm is as follows: Where x(i) and y(j) are the amplitudes of two time-domain waveforms at times i and j, respectively, and D(i,j) is the cumulative distance between the two waveforms at times i and j.

8. An adaptive UHF partial discharge signal extraction method based on dual-mode noise reference, comprising the adaptive UHF partial discharge signal extraction device based on dual-mode noise reference according to any one of claims 1-7, characterized in that, include: The main sensor (UHF1) of the directional ultra-high frequency sensor group receives electromagnetic wave signals from the device body in a directional manner, while the auxiliary sensor (UHF2) collects ambient background noise. The noise feature learning module is used to extract the time-frequency domain feature map of background noise through Hilbert-Huang transform; The acquired signal is transmitted to an adaptive noise cancellation device, which automatically adjusts the signal trigger threshold based on the noise energy level through a dynamic threshold adjustment unit, and filters the signal through a three-stage composite filter chain. The filtered signal is transmitted to an intelligent signal identification system. The intelligent signal identification system uses an online pattern recognition engine to align the time-domain waveform with the DTW dynamic time warping algorithm and combines it with a convolutional neural network to classify the discharge mode of the signal, thereby extracting the UHF partial discharge signal.

9. The adaptive UHF partial discharge signal extraction method based on dual-mode noise reference according to claim 8, characterized in that, In the step of the adaptive noise cancellation device for filtering the signal, the comb filter first suppresses periodic interference in the signal, then the wavelet packet denoising module processes the non-stationary signal, and finally the neural network filter identifies and eliminates pulse burst noise.

10. The adaptive UHF partial discharge signal extraction method based on dual-mode noise reference according to claim 9, characterized in that, The filtering formula for the comb filter is: In the formula, M is the filter order and z is a complex variable; The wavelet packet denoising module uses the wavelet packet decomposition formula, specifically: In the formula, h(n) and g(n) are the coefficients of the low-pass and high-pass filters, respectively, j is the number of decomposition layers, k is the node number, and t is the time variable.

Images