Self-adaptive acoustic imaging method and system suitable for complex environment

Abstract

The invention discloses a self-adaptive acoustic imaging method and system suitable for a complex environment, and the method comprises the steps: obtaining a first target sound signal, and carrying out the first preprocessing of the first target sound signal, and obtaining a first sound source distribution image; and pre-training a first imaging model, and carrying out acoustic imaging on the first sound source distribution image based on the first imaging model. By introducing an attention mechanism, the algorithm can pay more attention to the key features of the sound source signal, so that the positioning precision of the abnormal point is improved. By using a negative feedback mechanism, the algorithm can adjust parameters in the acoustic imaging process according to real-time feedback so as to adapt to changes in a complex environment, and the real-time performance and accuracy of detection are enhanced. According to the method and the system, when the low-frequency signal is processed, the detection coverage degree of the low-frequency band can be effectively improved through the optimized preprocessing step and the imaging model, so that more comprehensive sound source detection is realized in a complex environment.

Description

Technical Field

[0001] The present invention relates to the technical field of adaptive acoustic imaging, and particularly to an adaptive acoustic imaging method and system applicable to complex environments. Background Art

[0002] An acoustic imaging device is a device that combines array signal processing technology and sound visualization technology. This device is equipped with an acoustic imaging algorithm capable of visualizing and locating abnormal sound sources. It mainly detects abnormal vibration sound waves generated in the object to be measured, locates the sound source position by analyzing the signals received by the acoustic sensor array through algorithms, measures the corresponding characteristic signals in the sound signal, and combines the real-time acquired images through the visible light imaging device set on the device panel to obtain a spatial distribution map of the signals, ultimately achieving the purpose of helping users locate the abnormal sound position of the device.

[0003] Existing acoustic imaging algorithms usually first assume that the sound source map is a linear combination of the point spread function PSF and point sound sources in space. By compensating for the time difference generated during the reception process for each channel microphone to synchronize the sound signals of each channel, and then performing weighted summation to output the maximum value. According to the microphone array structure and the received data, signals in the direction or position of interest are filtered out under a certain criterion, and signal interference from other directions is suppressed to obtain a preliminary sound source distribution map. This result is expressed in the form of a convolution of the point spread function and the actual sound source, and deconvolution calculation is performed on the preliminary sound source distribution result to obtain the sound source distribution image.

[0004] However, since only the analysis and processing of sound signals are involved in the acoustic imaging positioning process, there are inevitably a large amount of noises during the acoustic detection process, making it difficult to accurately locate abnormal points. Moreover, the existing acoustic imaging algorithms currently available are difficult to cover a large frequency range. When detecting the sound source, they cannot accurately analyze the frequency band where the sound is located and perform adaptive frequency range detection on it, especially the detection coverage of the low frequency band is relatively low and the real-time performance is poor. To improve the above problems, this algorithm improves the existing acoustic imaging algorithm, improves the recognition resolution and accuracy, and introduces the attention mechanism and negative feedback mechanism in machine learning, enabling the algorithm to perform adaptive sound source recognition. Summary of the Invention

[0005] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title. However, such simplifications or omissions shall not be used to limit the scope of the present invention.

[0006] In view of the above existing problems, the present invention is proposed.

[0007] Therefore, the present invention provides an adaptive acoustic imaging method and system applicable to complex environments, which can solve the problems mentioned in the background art.

[0008] To solve the above technical problems, the present invention provides the following technical solutions:

[0009] In a first aspect, the present invention provides an adaptive acoustic imaging method applicable to complex environments, including:

[0010] Obtain a first target sound signal, and perform a first preprocessing on the first target sound signal to obtain a first sound source distribution image;

[0011] Pre-train a first imaging model, where the first imaging model includes at least a first mechanism and a second mechanism;

[0012] Perform acoustic imaging on the first sound source distribution image based on the first imaging model.

[0013] As a preferred solution of the adaptive acoustic imaging method applicable to complex environments according to the present invention, wherein: the performing a first preprocessing on the first target sound signal includes:

[0014] Obtain a first target sound signal, and combine it with a first delay compensation strategy to obtain a first compensation signal;

[0015] Perform a first weighting operation on the first compensation signal to obtain a first weighted signal;

[0016] Obtain a first sound source distribution image according to the first weighted signal.

[0017] As a preferred solution of the adaptive acoustic imaging method applicable to complex environments according to the present invention, wherein: the first imaging model includes:

[0018] The first imaging model is any model with the first sound source distribution image as the input and the acoustic imaging result or relevant parameters that can directly or indirectly obtain the acoustic imaging result as the output.

[0019] As a preferred solution of the adaptive acoustic imaging method applicable to complex environments according to the present invention, wherein: the performing acoustic imaging on the first sound source distribution image based on the first imaging model includes:

[0020] The first imaging model is used to identify a fault sound source;

[0021] After performing acoustic imaging on the first sound source distribution image, a second sound source distribution image is obtained, and the second sound source distribution image is used to directly or indirectly display the fault sound source.

[0022] As a preferred solution of the adaptive acoustic imaging method applicable to complex environments according to the present invention, wherein: the first target sound signal includes sound signals acquired by an array composed of a plurality of microphones.

[0023] As a preferred solution of the adaptive acoustic imaging method applicable to complex environments according to the present invention, wherein: the first delay compensation strategy includes:

[0024] Determine a reference microphone, compare the signals of other microphones with the signal of the reference microphone, and calculate the time difference;

[0025] Calculate the time difference between the signals received by each microphone channel in the array and the signal of the reference microphone;

[0026] According to the calculated time difference, perform corresponding delay compensation on the signals of each microphone channel.

[0027] As a preferred solution of the adaptive acoustic imaging method applicable to complex environments according to the present invention, wherein: the first weighting operation includes:

[0028] Determine the weights of each channel of each microphone in the microphone array;

[0029] For each time point, perform weighted summation on the signals of each channel according to the set weights;

[0030] After performing weighted summation, obtain an output signal of the weighted sum, select the channel with the maximum value as the final output, and the channel corresponding to the maximum value is the direction where the sound source is located.

[0031] In a second aspect, the present invention provides an adaptive acoustic imaging system applicable to complex environments, including:

[0032] A data acquisition module, configured to acquire a first target sound signal and perform a first preprocessing on the first target sound signal to obtain a first sound source distribution image;

[0033] A model training module, configured to pre-train a first imaging model, and the first imaging model at least includes a first mechanism and a second mechanism;

[0034] An identification module, configured to perform acoustic imaging on the first sound source distribution image based on the first imaging model.

[0035] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the method described above are implemented.

[0036] Fourthly, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are implemented.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes an adaptive acoustic imaging method and system applicable to complex environments, obtains a first target sound signal, and performs a first preprocessing on the first target sound signal to obtain a first sound source distribution image; pre-trains a first imaging model, where the first imaging model at least includes a first mechanism and a second mechanism; performs acoustic imaging on the first sound source distribution image based on the first imaging model. By introducing an attention mechanism, the algorithm can pay more attention to the key features of the sound source signal, thereby improving the positioning accuracy of abnormal points. Using a negative feedback mechanism, the algorithm can adjust the parameters in the acoustic imaging process according to real-time feedback to adapt to changes in complex environments and enhance the real-time performance and accuracy of detection. When processing low-frequency signals, through optimized preprocessing steps and imaging models, the method and system can effectively improve the detection coverage in the low-frequency band, thereby achieving more comprehensive sound source detection in complex environments. The adaptive acoustic imaging method and system of the present application can significantly improve the efficiency and accuracy of acoustic detection in practical applications, and are particularly suitable for sound source monitoring and analysis in fields such as industry, environmental protection, and medical care. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. Among them:

[0039] Figure 1 It is a flowchart of a method for an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0040] Figure 2 It is a flowchart of a feasible specific embodiment method for an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0041] Figure 3 It is a flowchart of microphone array reception and delay compensation for an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0042] Figure 4 It is a flowchart of weighted summation to output the maximum value and filter out signals in the direction of interest for an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0043] Figure 5 Flowchart for obtaining the sound source map of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0044] Figure 6 Flowchart of the negative feedback mechanism of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0045] Figure 7 Frequency adjustment interface of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0046] Figure 8 Flowchart of the attention mechanism of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0047] Figure 9 Flowchart of the attention scoring function of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0048] Figure 10 Output spectrum of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention. (a) is the output spectrum before introducing the mechanism, and (b) is the output spectrum after introducing the mechanism;

[0049] Figure 11 Time-frequency diagram of partial discharge signals of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0050] Figure 12 Histograms of the main frequency distributions of four different partial discharge acoustic signals of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention;

[0051] Figure 13 Internal structure diagram of a computer device of an adaptive acoustic imaging method and system applicable to complex environments provided by an embodiment of the present invention. Detailed implementation manners

[0052] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the detailed implementation manners of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0053] Embodiment 1

[0054] Referring to Figures 1 - 13 , which is the first embodiment of the present invention. This embodiment provides an adaptive acoustic imaging method and system applicable to complex environments, including:

[0055] In the existing related technologies, there are some problems. For example, during the acoustic imaging process, due to the influence of environmental noise, the accuracy of sound source localization is often not ideal enough.

[0056] This application provides a method that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to elaborate in detail how to implement the adaptive acoustic imaging method applicable to complex environments;

[0057] Figure 1 The flowchart of a method for an adaptive acoustic imaging method and system applicable to complex environments is shown, including:

[0058] S100, acquire a first target sound signal, and perform a first preprocessing on the first target sound signal to obtain a first sound source distribution image;

[0059] In an optional embodiment, the first target sound signal can be acquired through a sensor array. By designing the number and arrangement form of the array elements of the sensor array, the best acoustic acquisition effect can be achieved on the hardware, and the sound signal is acquired through a multi-channel microphone array;

[0060] In an optional embodiment, the first target sound signal can also be acquired in other ways. For example, it is received from a remote device through wireless transmission technology, or acquired from other computer systems through a network interface. The acquired sound signal will then undergo preprocessing to ensure the quality and accuracy of the signal, providing a reliable data basis for subsequent acoustic imaging.

[0061] In an optional embodiment, the first preprocessing is to accurately locate the sound source of the acquired first target sound signal. Therefore, the first preprocessing can include multiple steps, such as filtering, amplification, digital conversion, etc., to ensure the purity and clarity of the signal. These steps help reduce noise interference and improve the accuracy of sound source localization.

[0062] In the embodiments of the present application, a signal acquisition module is designed. The signal acquisition module includes two parts: acoustic imaging and visible light imaging. Acoustic imaging is mainly realized by a sensor array. By designing the number and arrangement form of the array elements of the sensor array, the best acoustic acquisition effect is achieved in terms of hardware, and the sound signal is collected through a multi-channel microphone array. When using the microphone array for acoustic imaging, due to the limited speed of sound propagation, there will be a slight time difference in the signals received by different microphones from the same sound source. This time difference can be used to locate the direction of the sound source. To accurately locate the sound source, it is necessary to compensate for the time difference of the signals in different microphone channels.

[0063] In the embodiments of the present application, the first preprocessing of the first target sound signal includes:

[0064] Obtain the first target sound signal, and combine it with the first delay compensation strategy to obtain the first compensation signal;

[0065] Perform a first weighting operation on the first compensation signal to obtain the first weighted signal;

[0066] Obtain the first sound source distribution image according to the first weighted signal.

[0067] In an optional embodiment, the first delay compensation strategy is any strategy for delay compensation through the first target sound signal, which can be designed according to the actual needs of relevant technical personnel. For example, a fixed delay value can be set to compensate the signal, or a more advanced algorithm, such as a model-based prediction method, can be used to dynamically calculate the delay compensation value for each microphone channel.

[0068] It should be noted that this dynamic calculation method usually needs to consider the propagation speed of sound waves in different media, as well as the influence of environmental factors such as temperature and humidity on the speed of sound. By comprehensively considering these factors, the delay compensation value for each channel can be calculated more accurately, thereby improving the accuracy of sound source localization.

[0069] In an optional embodiment, the first delay compensation strategy can also be designed in other ways. For example, machine learning algorithms can be used to predict the best delay compensation strategy. Through the training data set, the algorithm can learn how to adjust the delay compensation to obtain the best sound source localization effect under different environmental conditions. This method can further improve the adaptability and accuracy of the acoustic imaging system.

[0070] In the embodiments of the present application, the first target sound signal includes the sound signals obtained by an array composed of several microphones.

[0071] In the embodiments of the present application, the first delay compensation strategy includes:

[0072] Determine a reference microphone, compare the signals of other microphones with the signal of the reference microphone, and calculate the time difference;

[0073] Calculate the time difference between the signals received by each microphone channel in the array and the signal of the reference microphone;

[0074] According to the calculated time difference, perform corresponding delay compensation on the signals of each microphone channel.

[0075] Exemplarily, first, a reference microphone needs to be determined. Usually, the central microphone in the array or a certain microphone at a fixed position is selected as the reference. The signals of other microphones will be compared with the signal of the reference microphone to calculate the time difference.

[0076] Furthermore, calculate the time difference between the signals received by each microphone channel in the array and the signal of the reference microphone. The cross - correlation function can calculate the similarity of two signals at different delays and find the delay that maximizes the similarity, which is the time difference.

[0077] Furthermore, according to the calculated time difference, perform corresponding delay compensation on the signals of each microphone channel. The way of delay compensation is usually to store the received signal in a buffer after receiving it, and then move the signal in the buffer according to the calculated delay amount to achieve signal synchronization.

[0078] In the embodiment of the present application, the first weighting operation includes:

[0079] Determine the weights of each channel of each microphone in the microphone array;

[0080] For each time point, perform weighted summation on the signals of each channel according to the set weights;

[0081] After performing the weighted summation, an output signal of the weighted sum is obtained. Select the channel with the maximum value as the final output. The channel corresponding to the maximum value is the direction where the sound source is located.

[0082] It should be noted that obtaining the first target sound signal and performing the first pre - processing on the first target sound signal to obtain the first sound source distribution image can provide a clear and accurate sound source positioning basis for subsequent acoustic imaging, thereby improving the detection accuracy and reliability of the entire system.

[0083] S200, pre - train the first imaging model, and the first imaging model includes at least a first mechanism and a second mechanism;

[0084] In the embodiment of the present application, the first imaging model includes:

[0085] The first imaging model is any model that takes the first sound source distribution image as input and outputs the acoustic imaging result or relevant parameters that can directly or indirectly obtain the acoustic imaging result.

[0086] In an optional embodiment, the first imaging model can be constructed using deep learning techniques, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs). These models can learn the complex features of sound signals and optimize the model parameters through the training process to achieve more accurate acoustic imaging.

[0087] In an optional embodiment, the first imaging model can also be constructed using other machine learning algorithms, such as support vector machines (SVMs) or random forests. Although these algorithms may not be as efficient as CNNs or RNNs in processing sequential data, they may provide better generalization ability and interpretability in certain specific scenarios.

[0088] In an optional embodiment, the first imaging model can also use a hybrid model that combines the advantages of deep learning techniques and traditional machine learning algorithms to adapt to acoustic imaging tasks of different complexities.

[0089] In the embodiments of the present application, the first imaging model is designed using a neural network, including multiple hidden layers. Each hidden layer consists of multiple neurons, and these neurons are connected by weights and can learn the complex features of the input data.

[0090] In an optional embodiment, the first mechanism and the second mechanism can include an attention mechanism, a negative feedback mechanism, and other possible mechanisms, such as data augmentation and regularization techniques, to further improve the performance and generalization ability of the model. The attention mechanism is used to enhance the model's focus on the key features of the sound source signal, thereby improving the localization accuracy of abnormal points. The negative feedback mechanism is used to adjust the parameters in the acoustic imaging process according to real-time feedback to adapt to changes in complex environments and improve the real-time performance and accuracy of detection.

[0091] In an optional embodiment, the attention mechanism can be implemented in various ways. For example, it can be implemented using an attention network. The attention network can learn the importance of different parts of the input data and allocate different weights accordingly. In acoustic imaging, this means that the model can pay more attention to the signal features that have a greater impact on sound source localization and imaging quality, thereby improving the overall imaging effect.

[0092] In an optional embodiment, the implementation of the negative feedback mechanism may include real-time monitoring of various parameters during the acoustic imaging process, such as signal-to-noise ratio, signal intensity, etc., and dynamically adjusting the parameters of the imaging model according to the real-time feedback of these parameters. For example, if a decrease in the signal-to-noise ratio is detected, the system can automatically increase the amplification factor of the signal or adjust the filter settings to ensure that the imaging quality is not affected by environmental noise. This mechanism enables the system to adapt to changing environmental conditions and maintain stable detection performance.

[0093] In the embodiment of the present application, the first mechanism may be an attention mechanism, which can enable the model to pay more attention to the key features of the sound source signal, thereby improving the positioning accuracy of abnormal points. The attention mechanism assigns different weights to different parts of the input data, enabling the model to focus on processing the information that is most important for the current task.

[0094] In the embodiment of the present application, the second mechanism may be a negative feedback mechanism, which can adjust the parameters during the acoustic imaging process according to real-time feedback to adapt to changes in complex environments and enhance the real-time performance and accuracy of detection. The negative feedback mechanism compares the difference between the model output and the expected output and feeds this difference back into the model to adjust the model parameters, making the model output closer to the real situation.

[0095] It should be noted that pre-training the first imaging model, where the first imaging model includes at least the first mechanism and the second mechanism, can provide a powerful foundation for the acoustic imaging system, enabling it to still maintain high-precision and high-efficiency imaging capabilities when facing complex environments.

[0096] S300, perform acoustic imaging on the first sound source distribution image based on the first imaging model.

[0097] In the embodiment of the present application, performing acoustic imaging on the first sound source distribution image based on the first imaging model includes:

[0098] The first imaging model is used to identify faulty sound sources;

[0099] After performing acoustic imaging on the first sound source distribution image, a second sound source distribution image is obtained, and the second sound source distribution image is used to directly or indirectly display the faulty sound source.

[0100] In an optional embodiment, the second sound source distribution image can be displayed through visualization techniques. For example, by methods such as color coding and brightness adjustment, the intensity and position information of the sound source can be intuitively presented to the user. This visualization technique helps technicians quickly identify and locate faulty sound sources, thereby performing effective maintenance and repair work.

[0101] In an alternative embodiment, the output of acoustic imaging can be used for various applications, such as health monitoring of industrial equipment, analysis of environmental noise, acoustic positioning systems, etc. By accurately locating the sound source, timely maintenance of the equipment can be carried out to avoid potential failures and accidents.

[0102] In an alternative embodiment, acoustic imaging technology can also be used in combination with other sensor data, such as temperature sensors, vibration sensors, etc., to provide more comprehensive equipment status information. This method of multi-modal data fusion can further improve the accuracy and reliability of fault detection.

[0103] In an alternative embodiment, corresponding operations can also be performed according to the acoustic imaging results.

[0104] In an alternative embodiment, performing corresponding operations according to the acoustic imaging results includes:

[0105] Analyze the acoustic imaging results to identify potential fault sound sources;

[0106] Based on the identified fault sound sources, formulate corresponding maintenance or repair plans;

[0107] Execute the maintenance or repair plan to ensure the normal operation and safety of the equipment.

[0108] In an alternative embodiment, the analysis of acoustic imaging results can be done manually by professional technicians or achieved through automated data analysis software. Automated analysis can improve efficiency and reduce human errors, especially when dealing with a large amount of data.

[0109] In an alternative embodiment, when formulating maintenance or repair plans, various factors can be considered, such as the importance of the equipment, the severity of the fault, the repair cost, etc. By comprehensively evaluating these factors, the most appropriate maintenance strategy can be formulated.

[0110] In an alternative embodiment, when executing the maintenance or repair plan, various methods can be adopted, such as replacing damaged components, adjusting equipment settings, performing software upgrades, etc. These methods can be flexibly selected according to the actual situation to achieve the best maintenance effect.

[0111] It should be noted that the adaptive acoustic imaging method and system applicable to complex environments provided in this application can significantly improve the detection accuracy and maintenance efficiency of equipment by accurately locating the sound source and performing effective acoustic imaging, thus providing strong technical support for fields such as industrial production and environmental monitoring.

[0112] In summary, the present invention proposes an adaptive acoustic imaging method applicable to complex environments, which acquires a first target sound signal and performs a first preprocessing on the first target sound signal to obtain a first sound source distribution image; pre-trains a first imaging model, where the first imaging model at least includes a first mechanism and a second mechanism; and performs acoustic imaging on the first sound source distribution image based on the first imaging model. By introducing an attention mechanism, the algorithm can pay more attention to the key features of the sound source signal, thereby improving the positioning accuracy of abnormal points. Using a negative feedback mechanism, the algorithm can adjust the parameters in the acoustic imaging process according to real-time feedback to adapt to changes in complex environments and enhance the real-time performance and accuracy of detection. When processing low-frequency signals, this method and system can effectively improve the detection coverage in the low-frequency band through optimized preprocessing steps and imaging models, so as to achieve more comprehensive sound source detection in complex environments. The adaptive acoustic imaging method and system of the present application can significantly improve the efficiency and accuracy of acoustic detection in practical applications, and are particularly suitable for sound source monitoring and analysis in fields such as industry, environmental protection, and medical treatment.

[0113] Embodiment 2

[0114] In a preferred embodiment, the specific implementation process of the present application may include the following processes, as Figure 2 shown:

[0115] S1: The signal acquisition module includes two parts: acoustic imaging and visible light imaging. Acoustic imaging is mainly implemented by a sensor array. By designing the number and arrangement form of the array elements of the sensor array, the best acoustic acquisition effect is achieved on the hardware, and the sound signal is collected through a multi-channel microphone array. When using a microphone array for acoustic imaging, due to the limited speed of sound propagation, there will be a small time difference in the signals received by different microphones from the same sound source. This time difference can be used to locate the direction of the sound source. In order to accurately locate the sound source, it is necessary to compensate for the time difference of the signals in different microphone channels.

[0116] S2: The weights for weighted summation are designed according to factors such as the sound source localization target and the array geometry. Generally, the microphone channels aligned with the sound source direction are given higher weights, while the channels perpendicular to the sound source direction are given lower weights to suppress noise from other directions.

[0117] S3: According to the channel corresponding to the output maximum value and the known array geometry, the position of the sound source can be initially determined or an image of the sound field can be generated through a sound source localization algorithm or an image generation algorithm. The obtained preliminary sound source distribution result is expressed in the form of the convolution of a point spread function and the actual sound source, and deconvolution calculation is performed on the preliminary sound source distribution result to obtain a preliminary sound source distribution image.

[0118] S4: Introduce a real-time feedback mechanism, which can improve the detection accuracy of the acoustic imaging algorithm in the low-frequency band. According to the actual detection situation, the parameters of the acoustic imaging algorithm are adjusted in real time, enabling the algorithm to better adapt to the low-frequency signal characteristics in different environments, thereby improving the detection accuracy. According to the real-time feedback information, the signal-to-noise ratio processing strategy of the acoustic imaging algorithm is adaptively adjusted.

[0119] S5: In acoustic imaging, introducing a spatial attention mechanism can make the algorithm pay more attention to the regions of interest, focusing on the possible positions of the sound sources while ignoring background noise or unimportant information. By dynamically adjusting the degree of attention of the algorithm to different regions, the accuracy of imaging can be improved. The temporal attention mechanism can enable the algorithm to better understand the temporal characteristics of the signal and adjust the degree of attention to different time periods as needed. This helps to improve the performance of the acoustic imaging algorithm in the analysis of sound signal characteristics.

[0120] During the acoustic imaging process, first, the target sound source signal needs to be received through a specifically arranged microphone array, and the signals of different microphone channels are compensated for the time difference. The receiving process mainly includes the following steps, as Figure 3 shown:

[0121] S101: First, a reference microphone needs to be determined. Usually, the central microphone in the array or a certain microphone at a fixed position is selected as the reference. The signals of other microphones will be compared with the signal of the reference microphone to calculate the time difference.

[0122] S102: Calculate the time difference between the signals received by each microphone channel in the array and the signal of the reference microphone. The cross-correlation function can calculate the similarity of two signals at different time delays and find the time delay that maximizes the similarity, which is the time difference.

[0123] S103: According to the calculated time difference, perform corresponding delay compensation on the signals of each microphone channel. The way of delay compensation is usually to store the received signal in a buffer after receiving it, and then move the signal in the buffer according to the calculated delay amount to achieve signal synchronization.

[0124] In S2, weighted summation is performed on the received signals to output the maximum value and filter out the signals in the direction of interest, including the following steps, as Figure 4 shown:

[0125] S201: When performing weighted summation, first determine the form of the microphone array and analyze the selected microphone form to determine the weights of each channel.

[0126] S202: For each time point, the signals of each channel are weighted and summed according to the set weights. These weights are usually calculated by a beamforming algorithm to maximize or minimize the sound signal in a specific direction.

[0127] S203: After the weighted summation, an output signal of the weighted sum is obtained. Usually, this output signal will undergo some form of normalization process, and then the channel with the maximum value is selected as the final output. The channel corresponding to this maximum value is the direction where the sound source is located.

[0128] Step S3 includes the following steps, as Figure 5 shown:

[0129] S301: The signals of a single microphone are added with the time delays generated by the assumed sound source position, so that the sum of the amplitudes of the signals is attenuated to a certain extent. The output at the true sound source position is often larger in amplitude than other parts because the delay correction is completely correct. The direct delay summation of the time-domain signal, and its output formula is as follows:

[0130]

[0131] where y(ω,r) is the output signal, t is the sampling time; r is the output position; M is the number of microphones; is the distance between the output position and the i-th microphone; P i is the signal value of the i-th microphone; v is the speed of sound; R i (r) / v is the time delay of the i-th microphone relative to the assumed sound source position.

[0132] S302: Through the FFT transform, the obtained time-domain signal is converted to the frequency domain.

[0133] S303: The sound source map is regarded as a linear combination of the PSF point spread function and the point sound source in space. The obtained sound source distribution result after time difference weighted compensation is deconvolved to process the intermediate result. The aperture of the array, the number and arrangement of microphones, the detection frequency, etc. affect the array response, and they are the main factors restricting the positioning resolution. This algorithm converts the influence of these factors into a PSF model, and then determines the specific position of the sound source through the deconvolution inverse operation.

[0134] The application premise of this algorithm is that the sound sources are non-coherent sources or weakly correlated sources, then the sound source covariance matrix is a diagonal matrix as shown in the formula

[0135]

[0136] The cross-spectrum matrix in the absence of noise interference is gk represents the array vector of the k-th detection point on the detection plane, and s is the intensity of the sound source. Assume that the sound sources are independent:

[0137]

[0138] The preliminary sound source imaging result can be expressed in the form of the convolution of the point spread function and the actual sound source, where the point spread function is as follows. In the formula, W n is the weight vector of beamforming:

[0139]

[0140] Assume that the sound power on each scanning grid is unknown and is expressed in vector form. The point spread function matrix is expressed as:

[0141] y = Ax

[0142] The expanded form is as follows:

[0143]

[0144] Use the Gauss-Seidel iteration method to perform the first iteration to solve the sound source position. The specific iteration process is as follows:

[0145]

[0146] After the above iteration program, the actual sound source is finally calculated. This method greatly improves the positioning accuracy and spatial resolution.

[0147] The process in S4 includes the following steps, as Figure 6 shown:

[0148] S401: After introducing the attention mechanism, more information about the sound source points is obtained. By analyzing the sound source information, a negative feedback mechanism is introduced to adaptively adjust the imaging parameters. The principle of the real-time negative feedback mechanism is based on the negative feedback control principle in control theory. Negative feedback control compares the system output with the expected value, calculates the error, and adjusts the system input according to the magnitude and direction of the error so that the system output can approach the expected value as much as possible. The real-time negative feedback mechanism continuously monitors the output during the system operation, discovers and corrects the deviation in a timely manner, so that the system can work more stably and reliably.

[0149] S402: According to the actual detection situation, the bandwidth, sensitivity threshold, etc. of the filter are adjusted in real time, as follows Figure 7 The frequency adjustment interface on the left. In actual operation, the sensitive frequency range can be manually dragged to detect the signals in this range more precisely. This can make the algorithm better adapt to the low-frequency signal characteristics in different environments, thereby improving the detection accuracy.

[0150] S403: Dynamically adjust the detection area of the acoustic imaging algorithm according to real-time feedback information. When a signal with specific frequency components is detected, automatically adjust the detection range of the algorithm. As shown in the figure above, the range of the sound source point is positioned more precisely and concentrated more on the area where the target may exist, thereby improving the detection accuracy.

[0151] The process in S5 includes the following steps, as Figure 8 shown:

[0152] S501: The attention mechanism is a technology that mimics the human visual or cognitive process and is commonly used in the fields of machine learning and deep learning. Use a feature extraction network to extract features from sound signals and image data, classify the features of the sound signals obtained under different faults, and extract features such as the sound center frequency, root mean square frequency, sound intensity distribution, rectified average value, variance, standard deviation, mean square value, and root mean square value. Introduce the attention mechanism to "score" the input information and obtain the corresponding attention distribution through the softmax function. The attention distribution can be regarded as a probability distribution, which represents the importance of different input information for the task or target. Fuse the calculated attention signal with the original features to obtain the weighted features, as Figure 9 shown:

[0153] S502: For a feature map in acoustic imaging, introducing the spatial attention mechanism can make the algorithm pay more attention to the area of interest, focus on analyzing the possible positions of the sound source, and ignore background noise or unimportant information. By dynamically adjusting the degree of attention of the algorithm to different regions, the accuracy of imaging can be improved. As shown in the figure below, in acoustic detection, by introducing the spatial attention mechanism, a higher weight amplitude is given to the possible fault location, a k = ∑W hk ×b h , W hk is the input weight, b k is the input value from the input to the output node, and a k is the final output value of the neural network. Figure 10 The left side is before the introduction of the mechanism, Figure 10 and the right side is the output spectrum after the introduction of the mechanism.

[0154] S503: Acoustic signals have temporality. Introducing the temporal attention mechanism can enable the algorithm to better understand the temporal characteristics of the signals and adjust the degree of attention to different time periods as needed. This helps to improve the performance of the acoustic imaging algorithm in a dynamic environment. The frequency distribution and signal intensity of the sound signal are different at different times, as follows Figure 11As shown in the figure below, it is a type of time-frequency diagram of partial discharge signals. It can be seen that the frequency signal intensity of 20 kHz is relatively high at 0.015 s and 0.03 s. By introducing an attention mechanism, a larger weight is assigned to this type of signal, so as to better identify the fault signal.

[0155] S504: Acoustic signals usually have components of different frequencies, and these frequency components may have different importance during the imaging process. Introducing a spectral attention mechanism can make the algorithm pay more attention to the frequency components with higher energy or more important information, thereby improving the resolution and accuracy of imaging. Taking four different partial discharge acoustic signals as an example, as Figure 12 shown, through experiments, it is measured that the frequencies of the four partial discharge acoustic signals are concentrated in the 20 kHz frequency range under the experimental conditions. Therefore, when performing partial discharge fault detection, a larger weight can be assigned to the signals in this frequency band to achieve the purpose of improving the resolution.

[0156] Embodiment 3

[0157] In this embodiment, an adaptive acoustic imaging system applicable to complex environments is also provided, including:

[0158] A data acquisition module, configured to acquire a first target sound signal and perform a first preprocessing on the first target sound signal to obtain a first sound source distribution image;

[0159] A model training module, configured to pre-train a first imaging model, where the first imaging model includes at least a first mechanism and a second mechanism;

[0160] An identification module, configured to perform acoustic imaging on the first sound source distribution image based on the first imaging model.

[0161] The above-mentioned unit modules can be embedded in the processor of the computer device in a hardware form or be independent of it, or can be stored in the memory of the computer device in a software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0162] This embodiment also provides a computer device, which can be a terminal, and its internal structure diagram can be as Figure 13As shown in the figure. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an 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 medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements an adaptive acoustic imaging method applicable to complex environments. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a button, trackball, or touchpad set on the computer device housing, or an external keyboard, touchpad, or mouse, etc.

[0163] This embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0164] Obtain a first target sound signal, and perform a first preprocessing on the first target sound signal to obtain a first sound source distribution image;

[0165] Pre-train a first imaging model, where the first imaging model includes at least a first mechanism and a second mechanism;

[0166] Perform acoustic imaging on the first sound source distribution image based on the first imaging model.

[0167] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0168] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0169] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce a means for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0170] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction means that implements the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0171] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0172] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0173] Obviously, those skilled in the art can make various changes and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. An adaptive acoustic imaging method suitable for complex environments, characterized in that: include: Acquiring a first target sound signal, and performing a first preprocessing on the first target sound signal to obtain a first sound source distribution image; Pre-training a first imaging model, wherein the first imaging model includes at least a first mechanism and a second mechanism; Acoustic imaging is performed on the first sound source distribution image based on the first imaging model.

2. The adaptive acoustic imaging method for complex environments according to claim 1, characterized in that: The performing a first preprocessing on the first target sound signal comprises: Acquire a first target sound signal, and combine it with a first delay compensation strategy to obtain a first compensation signal; Performing a first weighted operation on the first compensation signal to obtain a first weighted signal; A first sound source distribution image is acquired according to the first weighted signal.

3. The adaptive acoustic imaging method for complex environments according to claim 2, characterized in that: The first imaging model comprises: The first imaging model is any model whose input is the first sound source distribution image and whose output is the acoustic imaging result or the related parameters of the acoustic imaging result that can be directly or indirectly obtained.

4. The adaptive acoustic imaging method for complex environments according to claim 3, characterized in that: The performing acoustic imaging on the first sound source distribution image based on the first imaging model includes: The first imaging model is used to identify the fault sound source; After performing acoustic imaging on the first sound source distribution image, a second sound source distribution image is obtained, and the second sound source distribution image is used to directly or indirectly display the fault sound source.

5. The adaptive acoustic imaging method for complex environments according to claim 4, characterized in that: The first target sound signal includes a sound signal acquired by an array consisting of a plurality of microphones.

6. The adaptive acoustic imaging method for complex environments according to claim 5, characterized in that: The first delay compensation strategy includes: Determine a reference microphone, compare the signals of other microphones with the reference microphone's signal, and calculate the time difference; Calculate the time difference between the signal received by each microphone channel in the array and the reference microphone signal; According to the calculated time difference, the signal of each microphone channel is delayed and compensated accordingly.

7. The adaptive acoustic imaging method for complex environments according to claim 6, characterized in that: The first weighting operation includes: Determine the weight of each channel of each microphone in the microphone array; For each time point, the signals of each channel are weighted and summed according to the set weights; After weighted summation, a weighted sum output signal is obtained, and the channel with the maximum value is selected as the final output. The channel corresponding to the maximum value is the direction of the sound source.

8. An adaptive acoustic imaging system suitable for complex environments, characterized in that: include: A data acquisition module, configured to acquire a first target sound signal, and perform a first preprocessing on the first target sound signal to obtain a first sound source distribution image; A model training module, used for pre-training a first imaging model, wherein the first imaging model includes at least a first mechanism and a second mechanism; A recognition module is used to perform acoustic imaging on the first sound source distribution image based on the first imaging model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images