Environmental noise suppression method, device and equipment applied to traction transformer in heavy haul railway, storage medium and program product

Abstract

The invention relates to an environmental noise suppression method, device and equipment applied to a traction transformer in a heavy haul railway, a storage medium and a program product, and relates to the technical field of power equipment fault diagnosis. By adopting the method, the noise reduction effect of the environmental noise can be improved. The method comprises the following steps: performing voiceprint feature extraction on a collected environmental noise digital signal to obtain a voiceprint feature of the environmental noise digital signal, and constructing a stable environmental noise filter based on the voiceprint feature; inputting the collected traction transformer sample sound signal into a stable environment noise filter to carry out noise preliminary suppression so as to obtain a noise reduction signal; noise interference is added to the noise reduction signal for data enhancement to obtain an enhanced target noise reduction signal, and a deep noise reduction model is trained through the target noise reduction signal to obtain a non-stationary complex noise suppression model; and carrying out noise suppression on the collected current sound signal of the traction transformer through a stationary environment noise filter and a non-stationary complex noise suppression model.

Description

Technical Field

[0001] This application relates to the field of power equipment fault diagnosis technology, and in particular to an environmental noise suppression method, apparatus, computer equipment, computer-readable storage medium and computer program product applied to traction transformers in heavy-haul railways. Background Technology

[0002] Traction transformers, as crucial equipment in railway power supply, play a vital role in the transmission and voltage conversion of electricity in electrified railways; their failures can lead to enormous losses. In recent years, acoustic fault diagnosis technology for traction transformers has developed rapidly, offering advantages such as high accuracy and non-contact real-time monitoring. However, the condition monitoring accuracy of this technology is significantly affected by environmental noise.

[0003] The environmental noise interference in traction substations is diverse and intense. Traditional methods for suppressing transformer environmental noise mainly employ wavelet threshold denoising, empirical mode decomposition, and blind source separation. However, these methods are limited by the type and number of noise sources, making them unsuitable for the complex noise environment within traction substations and resulting in poor noise reduction performance. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for suppressing environmental noise in traction transformers of heavy-haul railways, in response to the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for suppressing environmental noise in traction transformers used in heavy-haul railways, comprising:

[0006] The environmental noise digital signal is collected in the area along the heavy-haul railway. The voiceprint feature of the environmental noise digital signal is extracted. Based on the voiceprint feature, a stable environmental noise filter is constructed.

[0007] A sample sound signal of a traction transformer containing stable ambient noise is collected, and the sample sound signal of the traction transformer is input into the stable ambient noise filter for preliminary noise suppression to obtain a noise-reduced signal;

[0008] Noise interference is added to the denoised signal to perform data augmentation and obtain an enhanced target denoised signal. The deep denoising model is then trained using the target denoised signal to obtain a non-stationary complex noise suppression model.

[0009] The current sound signal of the traction transformer in the area along the heavy-haul railway is collected, and noise suppression is performed on the current sound signal of the traction transformer through the stationary environmental noise filter and the non-stationary complex noise suppression model.

[0010] In one embodiment, the step of extracting the voiceprint features of the environmental noise digital signal includes:

[0011] The environmental noise digital signal is preprocessed using a preset window function, and a Fast Fourier Transform is performed on the preprocessed environmental noise digital signal to obtain the transformed spectrum. The noise energy spectrum is obtained based on the mean value of the spectrum, and the stopband frequency of the noise energy spectrum is determined. Based on the stopband frequency, the lower cutoff frequency and upper cutoff frequency of the noise energy spectrum are obtained as the voiceprint feature.

[0012] In one embodiment, before constructing a stable environmental noise filter based on the voiceprint features, the method further includes: determining a first preset index corresponding to the upper cutoff frequency and a second preset index corresponding to the lower cutoff frequency;

[0013] The step of constructing a stable environmental noise filter based on the voiceprint features includes: determining the filter order of the stable environmental noise filter according to the voiceprint features, the first preset index, and the second preset index, and constructing the stable environmental noise filter based on the stopband frequency and the filter order.

[0014] In one embodiment, the noise suppression of the current sound signal of the traction transformer using the stationary environmental noise filter and the non-stationary complex noise suppression model includes:

[0015] The current sound signal of the traction transformer is input into the stable environmental noise filter for stable environmental noise suppression to obtain the traction transformer sound signal with initial noise reduction; the traction transformer sound signal with initial noise reduction is input into the non-stationary complex noise suppression model for non-stationary complex noise suppression to obtain the traction transformer sound signal with secondary noise reduction.

[0016] In one embodiment, training the deep denoising model using the target denoised signal to obtain a non-stationary complex noise suppression model includes:

[0017] The target denoising signal is input into the deep denoising model, which learns the noise distribution and denoising mapping through the target denoising signal to obtain a pre-trained deep denoising model. The pre-trained deep denoising model is then optimized by minimizing the reconstruction error and the perceptual loss function to obtain the non-stationary complex noise suppression model.

[0018] In one embodiment, the method further includes:

[0019] Obtain the filter feedback data of the stationary environmental noise filter and the model feedback data of the non-stationary complex noise suppression model; update the stationary environmental noise filter based on the filter feedback data; and update the non-stationary complex noise suppression model based on the model feedback data.

[0020] The step of suppressing noise in the current sound signal of the traction transformer using the stationary environmental noise filter and the non-stationary complex noise suppression model includes: suppressing noise in the current sound signal of the traction transformer using the updated stationary environmental noise filter and the updated non-stationary complex noise suppression model.

[0021] Secondly, this application also provides an environmental noise suppression device for traction transformers in heavy-haul railways, comprising:

[0022] The feature extraction module is used to collect digital environmental noise signals along the heavy-haul railway line, extract the voiceprint features of the digital environmental noise signals to obtain the voiceprint features of the digital environmental noise signals, and construct a stable environmental noise filter based on the voiceprint features.

[0023] The signal processing module is used to acquire the traction transformer sample sound signal containing steady ambient noise, input the traction transformer sample sound signal into the steady ambient noise filter for preliminary noise suppression, and obtain a noise-reduced signal.

[0024] The model training module is used to add noise interference to the denoised signal to perform data augmentation and obtain an enhanced target denoised signal. The target denoised signal is then used to train the deep denoising model to obtain a non-stationary complex noise suppression model.

[0025] The noise suppression module is used to collect the current sound signal of the traction transformer in the area along the heavy-haul railway, and to suppress the noise of the current sound signal of the traction transformer through the stationary environmental noise filter and the non-stationary complex noise suppression model.

[0026] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0027] A digital environmental noise signal is collected within the area along a heavy-haul railway. Voiceprint features are extracted from the digital environmental noise signal to obtain its voiceprint characteristics. Based on these voiceprint characteristics, a stationary environmental noise filter is constructed. Sample sound signals from traction transformers containing stationary environmental noise are collected and input into the stationary environmental noise filter for initial noise suppression, resulting in a denoised signal. Noise interference is added to the denoised signal for data augmentation to obtain an enhanced target denoised signal. A deep denoising model is trained using the target denoised signal to obtain a non-stationary complex noise suppression model. Finally, the current sound signal from the traction transformers within the area along the heavy-haul railway is collected and subjected to noise suppression using the stationary environmental noise filter and the non-stationary complex noise suppression model.

[0028] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0029] A digital environmental noise signal is collected within the area along a heavy-haul railway. Voiceprint features are extracted from the digital environmental noise signal to obtain its voiceprint characteristics. Based on these voiceprint characteristics, a stationary environmental noise filter is constructed. Sample sound signals from traction transformers containing stationary environmental noise are collected and input into the stationary environmental noise filter for initial noise suppression, resulting in a denoised signal. Noise interference is added to the denoised signal for data augmentation to obtain an enhanced target denoised signal. A deep denoising model is trained using the target denoised signal to obtain a non-stationary complex noise suppression model. Finally, the current sound signal from the traction transformers within the area along the heavy-haul railway is collected and subjected to noise suppression using the stationary environmental noise filter and the non-stationary complex noise suppression model.

[0030] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0031] A digital environmental noise signal is collected within the area along a heavy-haul railway. Voiceprint features are extracted from the digital environmental noise signal to obtain its voiceprint characteristics. Based on these voiceprint characteristics, a stationary environmental noise filter is constructed. Sample sound signals from traction transformers containing stationary environmental noise are collected and input into the stationary environmental noise filter for initial noise suppression, resulting in a denoised signal. Noise interference is added to the denoised signal for data augmentation to obtain an enhanced target denoised signal. A deep denoising model is trained using the target denoised signal to obtain a non-stationary complex noise suppression model. Finally, the current sound signal from the traction transformers within the area along the heavy-haul railway is collected and subjected to noise suppression using the stationary environmental noise filter and the non-stationary complex noise suppression model.

[0032] The aforementioned method, apparatus, computer equipment, computer-readable storage medium, and computer program product for suppressing environmental noise in traction transformers of heavy-haul railways extracts the voiceprint features of the collected digital environmental noise signals. Based on these voiceprint features, a stationary environmental noise filter is constructed. The collected sample sound signals from the traction transformer are then input into the stationary environmental noise filter for initial noise suppression to obtain a denoised signal. Noise interference is added to the denoised signal for data enhancement to obtain an enhanced target denoised signal. The target denoised signal is used to train a deep denoising model to obtain a non-stationary complex noise suppression model. Finally, the stationary environmental noise filter and the non-stationary complex noise suppression model are used to suppress the noise in the collected current sound signal from the traction transformer. This application enables in-depth extraction of voiceprint features from collected digital environmental noise signals, thereby further exploring the sources and characteristics of environmental noise that affect the accuracy of identifying the operating characteristic signals of traction power supply equipment in heavy-haul railways. It also allows for the use of different noise suppression methods for different types of environmental noise, unaffected by the type or number of noise sources, and can effectively suppress complex environmental noise within traction substations, thus significantly improving noise reduction. Furthermore, this application reduces the impact of environmental noise on the time-frequency distribution of the acoustic vibration signals of the power supply equipment itself, and enhances the effectiveness and reliability of traction power supply equipment operating status diagnosis based on voiceprint recognition technology, reducing the probability of misjudgment or missed judgment in the diagnostic system. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is an application environment diagram of an environmental noise suppression method for traction transformers in heavy-haul railways, as shown in one embodiment.

[0035] Figure 2 This is a flowchart illustrating an environmental noise suppression method applied to a traction transformer in a heavy-haul railway, as described in one embodiment.

[0036] Figure 3 This is a flowchart illustrating the voiceprint feature extraction steps in one embodiment;

[0037] Figure 4 This is a flowchart illustrating an environmental noise suppression method applied to traction transformers in heavy-haul railways in a specific embodiment.

[0038] Figure 5 This is a structural block diagram of an environmental noise suppression device applied to a traction transformer in a heavy-haul railway, as shown in one embodiment.

[0039] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0041] The environmental noise suppression method for traction transformers in heavy-haul railways provided in this application embodiment can be applied to a set of systems including, for example, Figure 1 The system shown comprises acoustic measurement equipment, front-end data processing equipment, and a back-end server. The acoustic measurement equipment communicates with the front-end data processing equipment via a network, and the front-end data processing equipment communicates with the back-end server via a network. Figure 1 In the application environment shown, the front-end data processing device can be, but is not limited to, various personal computers, laptops, smartphones, and tablets. The back-end server can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0042] In one embodiment, such as Figure 2 As shown, an environmental noise suppression method for traction transformers in heavy-haul railways is provided. Taking the application of this method to the above system as an example, the method includes the following steps:

[0043] Step S201: Collect digital environmental noise signals along the heavy-haul railway line, extract voiceprint features from the digital environmental noise signals to obtain voiceprint features, and construct a stable environmental noise filter based on the voiceprint features.

[0044] Voiceprint features include, but are not limited to, the frequency distribution, time-domain characteristics, and voiceprint patterns of environmental noise digital signals.

[0045] Specifically, the system collects digital environmental noise signals along the heavy-haul railway line using acoustic measurement equipment. Then, the digital signal of environmental noise. Voiceprint features are extracted to obtain the voiceprint features of the digital signal of environmental noise, and then a stable environmental noise filter is designed and constructed based on the voiceprint features.

[0046] Step S202: Collect the traction transformer sample sound signal containing stable ambient noise, input the traction transformer sample sound signal into the stable ambient noise filter for preliminary noise suppression, and obtain the noise-reduced signal.

[0047] Among them, the types of stable environmental noise include, but are not limited to, airborne noise, solid-borne noise, white noise, blue noise, and pink noise.

[0048] Specifically, the system acquires sample sound signals of the traction transformer containing steady ambient noise using acoustic measurement equipment. The characteristics of environmental noise are analyzed, and different noise suppression methods are adopted for different types of environmental noise to finally obtain a noise-reduced signal. .

[0049] Step S203: Add noise interference to the denoised signal to perform data augmentation and obtain the enhanced target denoised signal. Train the deep denoising model with the target denoised signal to obtain a non-stationary complex noise suppression model.

[0050] Specifically, the system reduces the noise signal. Add noise interference Data augmentation is performed to obtain the enhanced target denoising signal, which is then used to improve the deep denoising model. (in The parameters of the deep noise reduction model are trained, and noise suppression algorithms with different characteristics are integrated to obtain a non-stationary complex noise suppression model, thereby achieving multi-angle and multi-stage environmental noise suppression. For short-term stable or slowly changing environmental noise source signals, spectral subtraction filtering is used as a preliminary noise reduction algorithm for the traction power supply equipment's sound signal. For environmental noise that does not overlap with the frequency band of the target traction power supply equipment's sound and vibration signals, the spectral characteristics of the environmental noise can be statistically analyzed, and corresponding noise reduction filters can be designed to filter out noise interference in the frequency domain. For complex environmental noise, data synthesis is used to add additional noise interference to the target traction power supply equipment's sound data according to a set signal-to-noise ratio, thereby obtaining a large amount of enhanced data for training and training the deep noise reduction network model to separate and suppress noise.

[0051] Step S204: Collect the current sound signal of the traction transformer in the area along the heavy-haul railway, and suppress the noise of the current sound signal of the traction transformer by using a stable environmental noise filter and a non-stationary complex noise suppression model.

[0052] Specifically, the system collects the current sound signals of traction transformers along the heavy-haul railway line using acoustic measurement equipment. Furthermore, the current sound signal of the traction transformer is analyzed using a stable environmental noise filter and a non-stationary complex noise suppression model. Noise suppression is performed.

[0053] The above-mentioned environmental noise suppression method applied to traction transformers in heavy-haul railways involves extracting the acoustic features of the collected digital environmental noise signals. Based on these features, a stationary environmental noise filter is constructed. The collected sample sound signals from the traction transformers are then input into the stationary environmental noise filter for initial noise suppression to obtain a denoised signal. Noise interference is added to the denoised signal for data enhancement to obtain an enhanced target denoised signal. The target denoised signal is used to train a deep denoising model to obtain a non-stationary complex noise suppression model. Finally, the stationary environmental noise filter and the non-stationary complex noise suppression model are used to suppress the noise in the collected current sound signals from the traction transformers. This application enables in-depth extraction of voiceprint features from collected digital environmental noise signals, thereby further exploring the sources and characteristics of environmental noise that affect the accuracy of identifying the operating characteristic signals of traction power supply equipment in heavy-haul railways. It also allows for the use of different noise suppression methods for different types of environmental noise, unaffected by the type or number of noise sources, and can effectively suppress complex environmental noise within traction substations, thus significantly improving noise reduction. Furthermore, this application reduces the impact of environmental noise on the time-frequency distribution of the acoustic vibration signals of the power supply equipment itself, and enhances the effectiveness and reliability of traction power supply equipment operating status diagnosis based on voiceprint recognition technology, reducing the probability of misjudgment or missed judgment in the diagnostic system.

[0054] In one embodiment, such as Figure 3 As shown, in step S201 above, the voiceprint features of the environmental noise digital signal are extracted, which specifically includes the following steps:

[0055] Step S301: The environmental noise digital signal is preprocessed using a preset window function, and the preprocessed environmental noise digital signal is subjected to a fast Fourier transform to obtain the transformed spectrum.

[0056] Step S302: Obtain the noise energy spectrum based on the mean value of the spectrum, and determine the stopband frequency of the noise energy spectrum.

[0057] Step S303: Based on the stopband frequency, obtain the lower cutoff frequency and upper cutoff frequency of the noise energy spectrum, which are used as voiceprint features.

[0058] The preset window function can be the hanning window function.

[0059] Specifically, it may include the following steps:

[0060] Step 1: Digital signal analysis of ambient noise Multiply by the Hanning window function:

[0061]

[0062] In the above formula, M is the window size of the window function.

[0063] Step 2: Perform a Fast Fourier Transform on the signal multiplied by the window function to obtain the transformed spectrum. Spectrum Length N:

[0064]

[0065] In the above formula, N is the number of transformation points of the discrete Fourier transform.

[0066] Step 3: Subsequent statistics The mean value is used to obtain the noise energy spectrum. Complete the spectral subtraction construction.

[0067] Step 4: Observe the frequency distribution of the noise spectrum to obtain the stopband frequency. Calculate the filter order N according to the requirements. and They are the passbands at and The target indicators to be achieved:

[0068]

[0069] In the above formula, and These represent the lower and upper cutoff frequencies of the passband.

[0070] Step 5: Design a stable ambient noise filter based on N. Complete the filter design:

[0071]

[0072] In one embodiment, before constructing a stable ambient noise filter based on voiceprint features, the method of this application further includes the following steps:

[0073] Determine the first preset indicator corresponding to the upper limit cutoff frequency and the second preset indicator corresponding to the lower limit cutoff frequency;

[0074] In step S201 above, a stable environmental noise filter is constructed based on voiceprint features, specifically including the following steps:

[0075] Based on the voiceprint characteristics, the first preset index, and the second preset index, the filter order of the stable environmental noise filter is determined, and a stable environmental noise filter is constructed based on the stopband frequency and the filter order.

[0076] Specifically, the system determines the first preset index corresponding to the upper cutoff frequency and the second preset index corresponding to the lower cutoff frequency. Then, based on the voiceprint characteristics, the first preset index and the second preset index, the filter order of the stable environmental noise filter is determined, and a stable environmental noise filter is constructed based on the stopband frequency and the filter order.

[0077] In one embodiment, step S204 above involves noise suppression of the current sound signal of the traction transformer using a stationary environmental noise filter and a non-stationary complex noise suppression model, specifically including the following steps:

[0078] The current sound signal of the traction transformer is input into a stable environmental noise filter for stable environmental noise suppression to obtain the traction transformer sound signal with initial noise reduction; the traction transformer sound signal with initial noise reduction is input into a non-stationary complex noise suppression model for non-stationary complex noise suppression to obtain the traction transformer sound signal with secondary noise reduction.

[0079] Specifically, the system first inputs the current sound signal of the traction transformer into a stable ambient noise filter for stable ambient noise suppression, resulting in a traction transformer sound signal with initial noise reduction; then, it inputs the traction transformer sound signal with initial noise reduction into a non-stationary complex noise suppression model for non-stationary complex noise suppression, resulting in a traction transformer sound signal with secondary noise reduction, ultimately achieving the goal of noise reduction.

[0080] In one embodiment, step S203 above involves training a non-stationary complex noise suppression model using the target denoising signal, specifically including the following steps:

[0081] The target denoising signal is input into the deep denoising model, which learns the noise distribution and denoising mapping through the target denoising signal to obtain a pre-trained deep denoising model. The pre-trained deep denoising model is then optimized by minimizing the reconstruction error and the perceptual loss function to obtain a non-stationary complex noise suppression model.

[0082] Minimizing the reconstruction error can be the mean-square error (MSE), and the perceptual loss function can be the contrastive loss.

[0083] Specifically, the system inputs the target denoising signal into the deep denoising model, which learns the noise distribution and denoising mapping through the target denoising signal to obtain a pre-trained deep denoising model. Then, the pre-trained deep denoising model is optimized by minimizing the reconstruction error and the perceptual loss function to obtain a non-stationary complex noise suppression model for noise separation.

[0084] In one embodiment, the method of this application further includes the following steps:

[0085] Obtain the filter feedback data of the stationary environmental noise filter and the model feedback data of the non-stationary complex noise suppression model. Update the stationary environmental noise filter based on the filter feedback data and update the non-stationary complex noise suppression model based on the model feedback data.

[0086] In step S204 above, noise suppression is performed on the current sound signal of the traction transformer using a stationary environmental noise filter and a non-stationary complex noise suppression model. Specifically, this includes the following steps:

[0087] Noise suppression is achieved by using an updated stationary ambient noise filter and an updated non-stationary complex noise suppression model to suppress the current sound signal of the traction transformer.

[0088] Among them, the filter feedback data can be historical output data obtained after processing by a stationary environmental noise filter, and the model feedback data can be historical output data obtained after processing by a non-stationary complex noise suppression model.

[0089] Specifically, the system acquires the filter feedback data of the stationary environmental noise filter and the model feedback data of the non-stationary complex noise suppression model. Based on the filter feedback data, the stationary environmental noise filter is updated, and based on the model feedback data, the non-stationary complex noise suppression model is updated. The current sound signal of the traction transformer is then sequentially input into the updated stationary environmental noise filter and the updated non-stationary complex noise suppression model for noise suppression.

[0090] In one embodiment, such as Figure 4 As shown, a specific embodiment of an environmental noise suppression method for traction transformers in heavy-haul railways is provided, which includes the following steps:

[0091] Step S401: Collect digital environmental noise signals within the area along the heavy-haul railway line; preprocess the digital environmental noise signals using a preset window function; perform a Fast Fourier Transform on the preprocessed digital environmental noise signals to obtain the transformed spectrum; obtain the noise energy spectrum based on the mean value of the spectrum and determine the stopband frequency of the noise energy spectrum; based on the stopband frequency, obtain the lower cutoff frequency and upper cutoff frequency of the noise energy spectrum as acoustic signature features.

[0092] Step S402: Determine the first preset index corresponding to the upper cutoff frequency and the second preset index corresponding to the lower cutoff frequency; determine the filter order of the stable environmental noise filter based on the voiceprint characteristics, the first preset index and the second preset index, and construct the stable environmental noise filter based on the stopband frequency and the filter order.

[0093] Step S403: Collect the traction transformer sample sound signal containing stable ambient noise, input the traction transformer sample sound signal into the stable ambient noise filter for preliminary noise suppression, and obtain the noise-reduced signal; add noise interference to the noise-reduced signal for data enhancement to obtain the enhanced target noise-reduced signal.

[0094] Step S404: Input the target denoising signal into the deep denoising model. The deep denoising model learns the noise distribution and denoising mapping through the target denoising signal to obtain a pre-trained deep denoising model. Optimize the pre-trained deep denoising model by minimizing the reconstruction error and the perceptual loss function to obtain a non-stationary complex noise suppression model.

[0095] Step S405: Collect the current sound signal of the traction transformer in the area along the heavy-haul railway line, input the current sound signal of the traction transformer into a stable environmental noise filter for stable environmental noise suppression, and obtain the traction transformer sound signal with initial noise reduction; input the traction transformer sound signal with initial noise reduction into a non-stationary complex noise suppression model for non-stationary complex noise suppression, and obtain the traction transformer sound signal with secondary noise reduction.

[0096] The beneficial effects of the above embodiments are as follows:

[0097] 1) This application addresses environmental noise sources whose frequency bands do not overlap with or change slowly with the acoustic and vibration signals of the target traction power supply equipment. Based on the spectral subtraction filtering method, a corresponding stationary environmental noise filter is designed to initially filter out noise interference in the frequency domain. For non-stationary and complex environmental noise, an open-source dataset of random non-stationary environmental noise is used to add additional noise interference to the sound data of the target traction power supply equipment according to a set signal-to-noise ratio, thereby obtaining a large amount of training data. Subsequently, a deep denoising neural network for denoising the audio signal of the traction power supply equipment is constructed, and the deep denoising neural network is trained to extract the target audio from the audio of the power supply equipment mixed with complex environmental noise. After training, the final non-stationary complex noise suppression model is obtained, thereby achieving noise suppression of the input transformer sound signal.

[0098] 2) When performing environmental noise identification, this application can further classify the sources of interference to the acoustic signals of heavy-haul railway traction power supply equipment, and adopt corresponding noise reduction algorithms for different types of environmental noise characteristics.

[0099] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0100] Based on the same inventive concept, this application also provides an environmental noise suppression device for traction transformers in heavy-haul railways, used to implement the aforementioned method for suppressing environmental noise in traction transformers in heavy-haul railways. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the environmental noise suppression device for traction transformers in heavy-haul railways provided below can be found in the limitations of the environmental noise suppression method for traction transformers in heavy-haul railways described above, and will not be repeated here.

[0101] In one exemplary embodiment, such as Figure 5 As shown, an environmental noise suppression device for traction transformers in heavy-haul railways is provided, comprising:

[0102] The feature extraction module 501 is used to collect digital environmental noise signals in the area along the heavy-haul railway, extract the voiceprint features of the digital environmental noise signals, and construct a stable environmental noise filter based on the voiceprint features.

[0103] Signal processing module 502 is used to acquire the traction transformer sample sound signal containing steady ambient noise, input the traction transformer sample sound signal into the steady ambient noise filter for preliminary noise suppression, and obtain the noise-reduced signal;

[0104] The model training module 503 is used to add noise interference to the denoised signal to perform data augmentation and obtain the enhanced target denoised signal. The deep denoising model is trained using the target denoised signal to obtain a non-stationary complex noise suppression model.

[0105] The noise suppression module 504 is used to collect the current sound signal of the traction transformer in the area along the heavy-haul railway and suppress the noise of the current sound signal of the traction transformer through a stable environmental noise filter and a non-stationary complex noise suppression model.

[0106] In one embodiment, the feature extraction module 501 is further configured to preprocess the environmental noise digital signal using a preset window function, and perform a fast Fourier transform on the preprocessed environmental noise digital signal to obtain the transformed spectrum; obtain the noise energy spectrum based on the mean value of the spectrum, and determine the stopband frequency of the noise energy spectrum; and obtain the lower cutoff frequency and upper cutoff frequency of the noise energy spectrum based on the stopband frequency, as voiceprint features.

[0107] In one embodiment, the environmental noise suppression device applied to the traction transformer in a heavy-haul railway further includes an index determination module, which is used to determine a first preset index corresponding to the upper cutoff frequency and a second preset index corresponding to the lower cutoff frequency; the feature extraction module 501 is also used to determine the filter order of the smooth environmental noise filter based on the acoustic features, the first preset index and the second preset index, and to construct a smooth environmental noise filter based on the stopband frequency and the filter order.

[0108] In one embodiment, the noise suppression module 504 is further configured to input the current sound signal of the traction transformer into a stable environmental noise filter for stable environmental noise suppression to obtain a traction transformer sound signal with initial noise reduction; and input the traction transformer sound signal with initial noise reduction into a non-stationary complex noise suppression model for non-stationary complex noise suppression to obtain a traction transformer sound signal with secondary noise reduction.

[0109] In one embodiment, the model training module 503 is further configured to input the target denoising signal into the deep denoising model, and the deep denoising model learns the noise distribution and denoising mapping through the target denoising signal to obtain a pre-trained deep denoising model; and optimize the pre-trained deep denoising model according to minimizing the reconstruction error and the perceptual loss function to obtain a non-stationary complex noise suppression model.

[0110] In one embodiment, the environmental noise suppression device applied to the traction transformer in a heavy-haul railway further includes a model update module, which is used to acquire filter feedback data of a stationary environmental noise filter and model feedback data of a non-stationary complex noise suppression model, update the stationary environmental noise filter according to the filter feedback data, and update the non-stationary complex noise suppression model according to the model feedback data; the noise suppression module 504 is also used to suppress the noise of the current sound signal of the traction transformer through the updated stationary environmental noise filter and the updated non-stationary complex noise suppression model.

[0111] The various modules in the environmental noise suppression device applied to traction transformers in heavy-haul railways can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0112] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 6As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements an environmental noise suppression method for traction transformers in heavy-haul railways. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0113] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0114] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0115] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0116] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0117] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0118] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0119] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0120] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for suppressing environmental noise in traction transformers used in heavy-haul railways, characterized in that, The method includes: The environmental noise digital signal is collected in the area along the heavy-haul railway. The voiceprint feature of the environmental noise digital signal is extracted. Based on the voiceprint feature, a stable environmental noise filter is constructed. A sample sound signal of a traction transformer containing stable ambient noise is collected, and the sample sound signal of the traction transformer is input into the stable ambient noise filter for preliminary noise suppression to obtain a noise-reduced signal; Noise interference is added to the denoised signal to perform data augmentation and obtain an enhanced target denoised signal. The deep denoising model is then trained using the target denoised signal to obtain a non-stationary complex noise suppression model. The current sound signal of the traction transformer in the area along the heavy-haul railway is collected, and noise suppression is performed on the current sound signal of the traction transformer through the stationary environmental noise filter and the non-stationary complex noise suppression model.

2. The method according to claim 1, characterized in that, The step of extracting the voiceprint features of the environmental noise digital signal includes: The environmental noise digital signal is preprocessed by a preset window function, and the preprocessed environmental noise digital signal is then subjected to a fast Fourier transform to obtain the transformed spectrum. The noise energy spectrum is obtained based on the mean value of the spectrum, and the stopband frequency of the noise energy spectrum is determined. Based on the stopband frequency, the lower cutoff frequency and upper cutoff frequency of the noise energy spectrum are obtained, which are used as the voiceprint features.

3. The method according to claim 2, characterized in that, Before constructing a stable environmental noise filter based on the acoustic signature features, the method further includes: Determine the first preset index corresponding to the upper limit cutoff frequency and the second preset index corresponding to the lower limit cutoff frequency; The construction of a stable environmental noise filter based on the acoustic signature features includes: Based on the voiceprint characteristics, the first preset index, and the second preset index, the filter order of the stable environmental noise filter is determined, and the stable environmental noise filter is constructed based on the stopband frequency and the filter order.

4. The method according to claim 1, characterized in that, The noise suppression of the current sound signal of the traction transformer using the stable environmental noise filter and the non-stationary complex noise suppression model includes: The current sound signal of the traction transformer is input into the stable environmental noise filter for stable environmental noise suppression to obtain the traction transformer sound signal with initial noise reduction. The traction transformer sound signal with initial noise reduction is input into the non-stationary complex noise suppression model for non-stationary complex noise suppression to obtain the traction transformer sound signal with secondary noise reduction.

5. The method according to claim 1, characterized in that, The step of training a deep denoising model using the target denoised signal to obtain a non-stationary complex noise suppression model includes: The target noise reduction signal is input into the deep noise reduction model, and the deep noise reduction model learns the noise distribution and the mapping of noise reduction through the target noise reduction signal to obtain a preliminarily trained deep noise reduction model. The pre-trained deep noise reduction model is optimized by minimizing the reconstruction error and the perceptual loss function to obtain the non-stationary complex noise suppression model.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Obtain the filter feedback data of the stationary environmental noise filter and the model feedback data of the non-stationary complex noise suppression model; update the stationary environmental noise filter based on the filter feedback data; and update the non-stationary complex noise suppression model based on the model feedback data. The noise suppression of the current sound signal of the traction transformer using the stable environmental noise filter and the non-stationary complex noise suppression model includes: The noise suppression of the current sound signal of the traction transformer is achieved by using an updated stationary environmental noise filter and an updated non-stationary complex noise suppression model.

7. An environmental noise suppression device for traction transformers in heavy-haul railways, characterized in that, The device includes: The feature extraction module is used to collect digital environmental noise signals along the heavy-haul railway line, extract the voiceprint features of the digital environmental noise signals to obtain the voiceprint features of the digital environmental noise signals, and construct a stable environmental noise filter based on the voiceprint features. The signal processing module is used to acquire the traction transformer sample sound signal containing steady ambient noise, input the traction transformer sample sound signal into the steady ambient noise filter for preliminary noise suppression, and obtain a noise-reduced signal. The model training module is used to add noise interference to the denoised signal to perform data augmentation and obtain an enhanced target denoised signal. The target denoised signal is then used to train the deep denoising model to obtain a non-stationary complex noise suppression model. The noise suppression module is used to collect the current sound signal of the traction transformer in the area along the heavy-haul railway, and to suppress the noise of the current sound signal of the traction transformer through the stationary environmental noise filter and the non-stationary complex noise suppression model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images