DAS-based sound source event positioning method, system and electronic device

Abstract

The application relates to the technical field of optical fiber sensing, in particular to a sound source event positioning method and system based on DAS and electronic equipment. The method comprises the following steps: acquiring an audio feature image according to an abnormal sound source signal collected by DAS; performing feature extraction on the audio feature image to obtain an audio feature vector; acquiring an event feature vector according to the audio feature vector, and identifying an abnormal sound source event according to the event feature vector; performing feature splicing on the event feature vector and the audio feature vector to obtain a spliced feature vector, and processing the spliced feature vector to obtain a sound source distance. Compared with the prior art, the application performs the tasks of predicting an abnormal sound source event and a sound source distance in parallel based on the audio feature vector, and thus simultaneously realizes identification and position positioning of an optical cable abnormal sound source event.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optical fiber sensing, and more particularly to a DAS-based sound source event positioning method and system and electronic equipment. BACKGROUND

[0002] With the continuous development of optical cable technology, optical cable operation and maintenance is facing great challenges. The main content of optical cable operation and maintenance includes, when the optical cable is abnormal, collecting the optical cable data of the abnormal position, analyzing the collected optical cable data, determining the event type of the abnormality, and calculating the occurrence position of the abnormality. In the prior art, different methods are used to determine the abnormal sound source event and calculate the occurrence position of the abnormality, so two different methods and devices are usually needed to process, which greatly increases the cost of operation and maintenance. SUMMARY

[0003] The present application provides a DAS-based sound source event positioning method, system and electronic equipment, which can realize the identification and position positioning of the abnormal sound source event of the optical cable through the audio data collected by the optical cable.

[0004] According to a first aspect of the present application, a DAS-based sound source event positioning method is provided, the method comprising:

[0005] obtaining an audio feature image according to the abnormal sound source signal collected by the DAS;

[0006] performing feature extraction on the audio feature image to obtain an audio feature vector;

[0007] obtaining an event feature vector according to the audio feature vector, and identifying an abnormal sound source event according to the event feature vector;

[0008] performing feature splicing on the event feature vector and the audio feature vector to obtain a spliced feature vector, and processing the spliced feature vector to obtain a sound source distance.

[0009] By acquiring the audio feature image of the optical cable position point of the abnormal sound source signal, it is convenient to better extract the audio feature vector based on the audio feature image by using mature image deep learning algorithm; after obtaining the audio feature vector, two classification prediction tasks are performed based on the audio feature vector, one classification prediction task is to directly predict the type of the abnormal sound source event through the audio feature vector, and the other classification prediction task will splice the event feature vector identified according to the audio feature vector with the audio feature vector, and obtain the sound source distance according to the obtained spliced feature vector, the two classification prediction tasks are processed in parallel, and then the abnormal sound source event and the sound source distance can be obtained at the same time, and the identification and position positioning of the optical cable abnormal sound source event are realized at the same time. One of the classification prediction tasks splices the event feature vector with the audio feature vector, so that the model can combine the type of the abnormal sound source event when predicting the sound source distance based on the audio feature vector, realize the fusion of the two types of features, and then more accurately obtain the sound source distance.

[0010] Optionally, the audio feature image is acquired according to the abnormal sound source signal collected by the DAS, comprising:

[0011] The optical cable is divided into a plurality of audio channels according to a preset optical cable interval length;

[0012] According to the optical cable position point of the abnormal sound source signal collected by the DAS, the audio digital signals of a preset number of audio channels in the optical cable are acquired;

[0013] Each of the audio digital signals is processed to obtain the audio feature image.

[0014] By pre-dividing the optical cable into a plurality of audio channels, the influence area of the abnormal sound source event on the optical cable can be better determined, and the audio channels corresponding to the influence area are determined, and then the corresponding audio digital signals are collected through the corresponding audio channels to generate the audio feature image.

[0015] Optionally, the audio digital signals of a preset number of audio channels in the optical cable are acquired according to the optical cable position point of the abnormal sound source signal collected by the DAS, comprising:

[0016] The audio channel where the optical cable position point is located is acquired as a first audio channel;

[0017] According to the preset number, a plurality of audio channels before and after the first audio channel are acquired as second audio channels; the audio digital signals of the first audio channel and each of the second audio channels are acquired respectively.

[0018] The audio channel adjacent to the optical cable position point is affected by the abnormal sound source event similar to the optical cable position point, so that the audio digital signals of the first audio channel and the second audio channel are obtained, and more information of the abnormal sound source event is contained in the generated audio feature image.

[0019] Optionally, the processing of the audio digital signals to obtain the audio feature image comprises:

[0020] The obtained audio digital signals are normalized, and the normalized audio digital signals are stacked according to a preset channel order, and the audio feature image is obtained according to the stacked audio digital signals.

[0021] The abnormal sound source event includes a main sound source event and a sub-sound source event.

[0022] The main sound source event is an abnormal sound source event type of the abnormal sound source signal.

[0023] The sub-sound source event is a vibration state of the main sound source event, and the sub-sound source event at least includes a vibration starting state, a vibration state and a vibration ending state.

[0024] Optionally, the feature extraction of the audio feature image to obtain an audio feature vector comprises:

[0025] The audio feature image is extracted by a plurality of sequentially connected convolution units.

[0026] The extracted features are normalized by a regularization unit.

[0027] The features after the normalization are processed by a feature output unit to obtain the audio feature vector.

[0028] Optionally, the event feature vector is obtained according to the audio feature vector, and the abnormal sound source event is identified according to the event feature vector, comprising:

[0029] The audio feature vector is linearly processed by a first linear unit to obtain the event feature vector.

[0030] The event feature vector is identified by a category output unit to obtain the abnormal sound source event.

[0031] Optionally, the event feature vector and the audio feature vector are spliced to obtain a spliced feature vector, and the spliced feature vector is processed to obtain a sound source distance, comprising:

[0032] The event feature vector and the audio feature vector are feature-spliced by a feature splicing unit to obtain a spliced feature vector.

[0033] The spliced feature vector is processed by a second linear unit, and the processed spliced feature vector is feature-recognized by a distance output unit to obtain the sound source distance.

[0034] According to a second aspect of the present application, a sound source event positioning system based on DAS is provided, and the system comprises:

[0035] An audio image acquisition module is configured to acquire an audio feature image according to an abnormal sound source signal collected by the DAS.

[0036] An audio vector acquisition module is configured to extract features from the audio feature image to obtain an audio feature vector.

[0037] An event type acquisition module is configured to acquire an event feature vector according to the audio feature vector, and identify an abnormal sound source event according to the event feature vector.

[0038] A sound source distance acquisition module is configured to perform feature splicing on the event feature vector and the audio feature vector to obtain a spliced feature vector, and process the spliced feature vector to obtain a sound source distance.

[0039] According to a third aspect of the present application, an electronic device is provided, and the electronic device comprises:

[0040] A memory is configured to store one or more computer programs.

[0041] A processor is configured to implement the sound source event positioning method based on DAS according to the first aspect when the one or more computer programs are executed by the processor.

[0042] According to a fourth aspect of the present application, a computer readable storage medium is provided, and the computer readable storage medium stores computer instructions, which are configured to enable a processor to implement the sound source event positioning method based on DAS according to the first aspect when executed by the processor.

[0043] Based on any one of the above aspects, the DAS-based sound source event positioning method, system, electronic device and computer readable storage medium provided by the embodiments of the present application obtain the audio feature image of the optical cable position point of the abnormal sound source signal, so as to better extract the audio feature vector based on the audio feature image using the mature image deep learning algorithm; after obtaining the audio feature vector, two classification prediction tasks are performed based on the audio feature vector, one classification prediction task is to directly predict the type of the abnormal sound source event through the audio feature vector, and the other classification prediction task is to splice the event feature vector identified according to the audio feature vector with the audio feature vector, obtain the sound source distance according to the obtained spliced feature vector, and the two classification prediction tasks are processed in parallel, so that the abnormal sound source event and the sound source distance can be obtained at the same time, and the identification and position positioning of the optical cable abnormal sound source event are realized at the same time. The event feature vector and the audio feature vector are spliced in one of the classification prediction tasks, so that the model can combine the type of the abnormal sound source event when predicting the sound source distance based on the audio feature vector, realize the fusion of the two types of features, and further accurately obtain the sound source distance.

[0044] Meanwhile, before analyzing the optical cable, the optical cable is divided into a plurality of audio channels in advance, so that the influence range of the abnormal sound source event can be determined according to the optical cable position point of the abnormal sound source signal when data extraction and analysis are performed subsequently; meanwhile, the extraction of the audio digital signal can be better realized based on the audio channel, and the audio feature image containing the abnormal sound source event information can be better obtained, and the prediction of the abnormal sound source event and the sound source distance can be realized. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0046] Figure 1 The step flowchart of the positioning method provided by the present embodiment.

[0047] Figure 2 The step flowchart of the audio feature image acquisition provided by the present embodiment.

[0048] Figure 3 The step flowchart of the audio digital signal acquisition provided by the present embodiment.

[0049] Figure 4 The network connection structure in the audio event analysis model provided by the present embodiment.

[0050] Figure 5 A unit connection structure schematic diagram in the audio event analysis model provided for the embodiment.

[0051] Figure 6 A system structure schematic diagram of the positioning system provided for the embodiment.

[0052] Figure 7 A device structure schematic diagram of the electronic device provided for the embodiment.

[0053] FIG. 1 is a schematic diagram of an audio event analysis model according to an embodiment of the present application; FIG. 2 is a schematic diagram of a positioning system according to an embodiment of the present application; FIG. 3 is a schematic diagram of an electronic device according to an embodiment of the present application; and FIG. 4 is a schematic diagram of a system structure of the electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0054] The drawings attached hereto are only used for illustrative purposes, and cannot be understood as a limitation of the present application. In order to better illustrate the following embodiments, some components in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size; it is understandable for those skilled in the art that some well-known structures and their descriptions in the drawings may be omitted.

[0055] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the scope of protection of the present application.

[0056] It is to be understood that the terms "first", "second", and the like used in the description and the claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0057] With the continuous development of optical cable technology, optical cable operation and maintenance is facing great challenges. The main content of optical cable operation and maintenance includes collecting optical cable data at the abnormal position after the optical cable is abnormal, continuing to analyze the collected optical cable data, determining the event type of the abnormality, and calculating the occurrence position of the abnormality.

[0058] In the prior art, the identification of abnormal sound source events is usually determined according to the characteristic identification of the scattered light formed when the abnormal sound source events affect the optical cable, and the calculation of the abnormal occurrence position is usually obtained according to the propagation time difference and distance of the scattered light corresponding to the abnormal sound source events. Therefore, the identification of abnormal sound source events and the calculation of abnormal occurrence positions use different ways, so two different methods and devices are usually needed for processing, which greatly increases the cost of operation and maintenance.

[0059] The embodiment provides a technical solution that can solve the above problems, and the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0060] As shown in Figure 1 The embodiment provides a sound source event positioning method based on DAS, which can include the following steps:

[0061] S1: obtaining an audio feature image according to the abnormal sound source signal collected by DAS;

[0062] S2: performing feature extraction on the audio feature image to obtain an audio feature vector;

[0063] S3: obtaining an event feature vector according to the audio feature vector, and identifying an abnormal sound source event according to the event feature vector;

[0064] S4: performing feature splicing on the event feature vector and the audio feature vector to obtain a spliced feature vector, and processing the spliced feature vector to obtain a sound source distance.

[0065] In the embodiment, when an abnormal sound source event occurs around the optical cable, the optical cable receives the sound wave signal generated by the abnormal sound source event, and the optical cable generates corresponding scattered light when receiving the sound wave signal. The scattered light is received by the DAS device and analyzed to obtain an audio analog signal. The audio analog signal is converted into a corresponding audio digital signal by pulse code modulation as the abnormal sound source signal.

[0066] It can be understood that the optical cable position point can be a sampling point on the optical cable that first receives the sound wave signal. Since the sound wave signal of the abnormal sound source event is propagated to the surrounding, a certain range of the optical cable is affected by the sound wave signal, and corresponding audio digital signals are generated in the range.

[0067] Therefore, in an optional embodiment, as shown in Figure 2 the acquisition of the audio feature image in step S1 can include:

[0068] S11: dividing the optical cable into a plurality of audio channels according to a preset optical cable interval length;

[0069] S12: acquiring audio digital signals of a preset number of the audio channels in the optical cable according to the optical cable position point of the abnormal sound source signal collected by the DAS;

[0070] S13: processing each of the audio digital signals to obtain the audio feature image.

[0071] It can be understood that in the embodiment, the optical cable can be divided into a plurality of audio channels in advance, and the length of each audio channel is the optical cable interval length. For example, for a 5km optical cable, the optical cable interval length can be set to 1km, and the 5km optical cable can be divided into 5 audio channels. The division of the audio channels can be performed according to actual optical cable monitoring requirements, so as to more comprehensively cover each region passed by the optical cable.

[0072] As described above, since the sound wave signal of the abnormal sound source event acts on a certain range of the optical cable, a plurality of abnormal sound source signals appear on the optical cable. In order to better identify the abnormal sound source event and calculate the distance of the abnormal sound source event, a preset number of corresponding audio digital signals of the audio channels are acquired based on the optical cable position point receiving the abnormal sound source signal for identification.

[0073] In an optional embodiment, as shown in Figure 3 the acquisition of the audio digital signals can include:

[0074] S121: Obtain the audio channel where the optical cable position point is located as a first audio channel;

[0075] S122: According to the preset number, obtain a plurality of audio channels before and after the first audio channel as second audio channels;

[0076] S123: Obtain the audio digital signal of the first audio channel and each second audio channel respectively.

[0077] It can be understood that in the embodiment, the first audio channel is the audio channel where the optical cable position point is located, and a plurality of audio channels adjacent to the first audio channel are selected as second audio channels with the first audio channel as the center. For example, assuming that the optical cable is divided into 10 audio channels, the preset number is 5, and the optical cable position point is located in the 4th audio channel, the 4th audio channel is selected as the first audio channel, and 4 audio channels are selected as the second audio channel. The 2nd, 3rd, 5th and 6th audio channels adjacent to the 4th audio channel can be selected as the second audio channel. By selecting a plurality of audio channels, the audio digital signal containing the feature of the abnormal sound source event can be obtained, and the recognition of the abnormal sound source event and the calculation of the sound source distance can be better realized.

[0078] In an optional embodiment, the acquisition of the audio feature image can include:

[0079] The acquired audio digital signal is normalized, and the normalized audio digital signal is stacked according to a preset channel order, and the audio feature image is acquired according to the stacked audio digital signal.

[0080] In the embodiment, the audio digital signals of each audio channel are stacked, which is equivalent to forming a two-dimensional audio digital matrix, and then the image processing tool is used to convert the audio digital matrix into a corresponding grayscale image, and the grayscale image is used as the audio feature image.

[0081] It can be understood that since the first audio channel and the second audio channel are audio channels affected by the abnormal sound source event, the audio digital signal acquired thereon contains the feature of the abnormal sound source event, therefore, the audio feature image obtained by stacking contains the corresponding feature of the abnormal sound source event, which can be effectively used to realize the recognition of the abnormal sound source event.

[0082] The conventional abnormality occurrence position calculation process usually only focuses on the type of abnormal sound source event, such as construction site construction and road repair, but the abnormal sound source event includes the whole process of the event activity from the pre-vibration to the end of the vibration; therefore, the whole process of the event activity can be divided into multiple vibration intervals, and each vibration interval can be regarded as a vibration state corresponding to the abnormal sound source event.

[0083] In the whole event activity, the signal-to-noise ratio in the audio digital signal obtained in the pre-vibration state and the vibration attenuation state (end vibration state) is high, and the signal-to-noise ratio in the audio digital signal obtained in other vibration states such as the background noise, continuous audio and saturated audio state is low, which will interfere with the distance prediction of the abnormal sound source event. Therefore, in the distance prediction process of the abnormal sound source event, if different vibration states are not distinguished, a large error will be caused in the distance prediction of the abnormal sound source event due to the existence of many interference states.

[0084] In a preferred embodiment of the present embodiment, the abnormal sound source event can include a main sound source event and a sub-sound source event; wherein the main sound source event is the abnormal sound source event type of the abnormal sound source signal, such as the construction site construction and road repair described above; the sub-sound source event is the vibration state corresponding to the main sound source event, and in the present embodiment, the sub-sound source event at least includes the pre-vibration state, the vibration state and the end vibration state, and it can be understood that the sub-sound source event can also include the saturated vibration state and the background noise vibration state.

[0085] By taking the vibration state corresponding to the abnormal sound source signal as the sub-sound source event, the event feature vector obtained by the feature recognition not only contains the features of the abnormal sound source event itself, but also contains the features of the vibration state corresponding to the abnormal sound source event, and then when the spliced feature vector is processed, the sound source distance can be obtained on the basis of the abnormal sound source event and further combined with the vibration state under the abnormal sound source event, so that the present embodiment can better focus on the sound source data in the pre-vibration state and the end vibration state of the abnormal sound source event, and effectively resist the interference of other vibration states.

[0086] In an optional embodiment, the method can include:

[0087] processing the audio feature image through the audio event analysis model to obtain the abnormal sound source event and the sound source distance corresponding to the abnormal sound source signal;

[0088] In the present embodiment, the audio event analysis model is a pre-trained model, such as Figure 4As shown, the audio event analysis model can include a feature extraction network 100, an event type prediction network 200, and an event distance prediction network 300.

[0089] In an optional implementation, the above step S2 can be performed by the feature extraction network 100, that is, the feature extraction network 100 is used to perform feature extraction on the audio feature image to obtain an audio feature vector.

[0090] In an implementation, as shown in the figure, Figure 5 As shown, the feature extraction network 100 can include a regularization unit 102, a feature output unit 103, and a plurality of convolution units 101 connected in sequence, and the step S2 can include:

[0091] The audio feature image is subjected to feature extraction by the plurality of convolution units 101, the extracted features are subjected to regularization processing by the regularization unit 102, and the features after the regularization processing are subjected to feature processing by the feature output unit 103 to obtain the audio feature vector.

[0092] The convolution unit 101 can be set as a convolutional neural network, and a ReLU (Rectified Linear Unit) activation function is set after the convolutional neural network. The number of convolution units 101 can be set to 4, wherein the convolutional neural network is used to extract the features of the audio feature image, and the ReLU is used to introduce nonlinearity to the features of the audio feature image extracted by the convolutional neural network to enhance the expression ability of the convolutional neural network.

[0093] The regularization unit 102 can include a regularization network Dropout, a Linear network, and a ReLU activation function connected in sequence.

[0094] In this embodiment, the regularization network is used to perform regularization processing on the features extracted by each convolution unit 101 to avoid overfitting of the audio event analysis model during processing; the Linear network is used to input the features therein, and the connection weights between each neuron in the Linear network and all neurons in the previous network layer are subjected to matrix multiplication and bias addition operations to realize linear processing of the input features. Each neuron of the Linear network is connected to all neurons of the previous network layer, and the weights and biases of each neuron are optimized by a backpropagation algorithm so that the output result can better fit the data. The ReLU activation function is used to introduce a nonlinearity factor to the features after the linear processing by the Linear network, so that the network can learn more complex patterns. Among them, the Linear network and the ReLU are used together to complete the feature processing from input to output.

[0095] The feature output unit 103 can be set as a Linear network and a ReLU activation function; it can be understood that the Linear network of the feature output unit 103 performs feature processing on the features output by the regular unit 102.

[0096] In an optional embodiment, the above step S3 can be performed by the event type prediction network 200, that is, by the event type prediction network 200, an event feature vector is obtained according to the audio feature vector, and an abnormal sound source event is identified according to the event feature vector.

[0097] In an embodiment, as shown in the figure, Figure 5 The event type prediction network 200 can include a first linear unit 201 and a category output unit 202, and the step S3 can include:

[0098] The first linear unit 201 performs linear processing on the audio feature vector to obtain the event feature vector;

[0099] The category output unit 202 performs feature identification on the event feature vector to obtain the abnormal sound source event.

[0100] The first linear unit 201 can be set as a Linear network, and the category output unit 202 can be set as a Linear network and a ReLU activation function.

[0101] In an optional embodiment, the above step S4 can be performed by the event distance prediction network 300, that is, by the event distance prediction network 300, the event feature vector and the audio feature vector are spliced to obtain a spliced feature vector, and the spliced feature vector is processed to obtain a sound source distance.

[0102] In an embodiment, as shown in the figure, Figure 5 The event distance prediction network 300 includes a feature splicing unit 301, a second linear unit 302, and a distance output unit 303, and the step S4 can include:

[0103] The feature splicing unit 301 splices the event feature vector and the audio feature vector to obtain a spliced feature vector, the second linear unit 302 performs linear processing on the spliced feature vector, and the distance output unit 303 performs feature identification on the feature-processed spliced feature vector to obtain the sound source distance.

[0104] The second linear unit 302 can be configured as a Linear network, and the distance output unit 303 can be configured as a Linear network and a ReLU activation function.

[0105] It can be understood that the audio event analysis model can identify and output the abnormal sound source event and the sound source distance by pre-training.

[0106] In this embodiment, the training of the audio event analysis model can include:

[0107] A plurality of audio feature training images are obtained, and event feature labels and distance feature labels are added to the audio feature training images. The event feature labels represent the abnormal sound source events corresponding to the audio feature training images. The distance feature labels include valid distance feature labels and invalid distance feature labels. The valid distance labels and the invalid distance labels are used to guide the learning of the audio event analysis model, wherein the valid distance labels learn to output valid sound source distances for audio feature images of target abnormal sound source events, and the invalid distance labels learn to output invalid sound source distances for audio feature images of non-target abnormal sound source events.

[0108] Therefore, when the abnormal sound source event of the audio feature training image belongs to a target sound source event, the valid distance feature label is added to the audio feature training image; if it does not belong to the target sound source event, the invalid distance feature label is added to the audio feature training image.

[0109] As described above, the abnormal sound source event can include a main sound source event and a sub-sound source event, and therefore in an implementation, the target sound source event can include a main target event and a sub-target event, wherein the main target event represents a type of target abnormal sound source event, and the sub-target event includes a vibration start state and a vibration end state.

[0110] Adding the distance feature label to the audio feature image can include:

[0111] If the main sound source event belongs to the main target event, and the sub-sound source event is in the vibration start state or the vibration end state, the valid distance feature label is added to the audio feature training image; otherwise, the invalid distance feature label is added to the audio feature training image.

[0112] The audio event analysis model to be trained is trained by adding an event feature label and a distance feature label to a plurality of audio feature training images. In the training process, a predicted sound source event and a predicted sound source distance of the audio feature training images are obtained, the loss of the predicted sound source event and the event feature label and the loss of the predicted sound source distance and the distance feature label are calculated respectively, the audio event analysis model to be trained is updated according to the loss, and a trained audio event analysis model is obtained.

[0113] By adding the event feature label, the audio event analysis model can more accurately identify the abnormal sound source event of the abnormal sound source signal. At the same time, by adding the distance feature label, the model can directly predict the effective sound source distance and the invalid sound source distance based on the abnormal sound source event, such as when the sub-sound source event is in the vibration state, an invalid sound source distance is output, and the abnormal sound source event can be directly filtered by the audio event analysis model, so that the embodiment can pay more attention to the target sound source event and the data of the vibration start state and the vibration end state of the target sound source event.

[0114] In a specific example, when the DAS collects an abnormal sound source signal, the first audio channel and the second audio channel are determined according to the optical cable position point at which the abnormal sound source signal is collected, the audio digital signals about the abnormal sound source signal in the first audio channel and each second audio channel are collected, and the audio digital signals are processed to obtain the audio feature images.

[0115] Then, the obtained audio feature images are input into the audio event analysis model, the audio feature vectors of the audio feature images are extracted by the feature extraction network 100, the extracted audio feature vectors are subjected to feature recognition by the event type prediction network 200, the event feature vectors are extracted, and the abnormal sound source event type and the vibration state of the abnormal sound source signal are recognized according to the event feature vectors.

[0116] The event feature vectors and the audio feature vectors are input into the event distance prediction network 300. In the event distance prediction network 300, the event feature vectors and the audio feature vectors are spliced by the feature splicing unit 301 to obtain the spliced feature vectors. If the main sound source event of the abnormal sound source event belongs to the main target event of the target sound source event and the vibration state is the vibration start state or the vibration attenuation state, the event distance prediction network 300 will output an effective sound source distance according to the spliced feature vectors. Otherwise, the event distance prediction network 300 will output an invalid sound source distance according to the spliced feature vectors.

[0117] The embodiment also provides an acoustic source event positioning system of an optical cable, as shown in the figure. Figure 6 The system can include:

[0118] An audio image acquisition module 11 is configured to acquire an audio feature image according to the abnormal acoustic source signal collected by the DAS.

[0119] In this embodiment, the audio image acquisition module 11 can be configured to perform step S1 as shown in the figure. Figure 1 The specific description of the audio image acquisition module 11 can refer to the description of step S1.

[0120] An audio vector acquisition module 12 is configured to perform feature extraction on the audio feature image to acquire an audio feature vector.

[0121] In this embodiment, the audio vector acquisition module 12 can be configured to perform step S2 as shown in the figure. Figure 1 The specific description of the audio vector acquisition module 12 can refer to the description of step S2.

[0122] An event type acquisition module 13 is configured to acquire an event feature vector according to the audio feature vector, and identify an abnormal acoustic source event according to the event feature vector.

[0123] In this embodiment, the event type acquisition module 13 can be configured to perform step S3 as shown in the figure. Figure 1 The specific description of the event type acquisition module 13 can refer to the description of step S3.

[0124] An acoustic source distance acquisition module 14 is configured to perform feature splicing on the event feature vector and the audio feature vector to obtain a spliced feature vector, and process the spliced feature vector to obtain an acoustic source distance.

[0125] In this embodiment, the acoustic source distance acquisition module 14 can be configured to perform step S4 as shown in the figure. Figure 1 The specific description of the acoustic source distance acquisition module 14 can refer to the description of step S4.

[0126] In an implementation, the system can further include a model construction module 15 configured to construct the audio event analysis model, and the audio vector acquisition module 12, the event type acquisition module 13 and the acoustic source distance acquisition module 14 perform corresponding steps through corresponding units of the constructed audio event analysis model respectively.

[0127] The embodiment provides an electronic device, and a structure thereof is shown in the figure. Figure 7

[0128] ​The electronic device includes a memory 21, a processor 22, a communication module 23, and an input / output interface 24, etc. Optionally, the memory 21, the processor 22, the communication module 23, and the input / output interface 24 can be connected and communicated through a bus 25.

[0129] The memory 21 is configured to store one or more computer programs and transmit codes of the computer programs to the processor 22; when the one or more computer programs are executed by the processor 22, the acoustic source event positioning method of the optical cable in the embodiments of the present application is implemented.

[0130] Optionally, the electronic device can be connected to a network through the communication module 23 to communicate with other devices such as terminals or servers through the network to realize the interaction of data. The electronic device can be various forms of digital computers, such as desktop computers, servers, workstations, mainframe computers, or other types of computers. The electronic device can also be various forms of mobile terminals, such as smartphones, tablet computers, wearable devices (such as helmets, glasses, watches, etc.), and other similar mobile terminals.

[0131] Optionally, the electronic device can connect the required input / output devices such as keyboards, display devices, etc. through the input / output interface 24, and the electronic device itself can have a display device and can also be externally connected to other display devices through the input / output interface 24. Optionally, the storage device such as a hard disk can also be connected through the input / output interface 24, so that the data in the electronic device can be stored in the storage device or read from the storage device, and the data in the storage device can also be stored in the memory 21. It can be understood that the input / output interface 24 can be a wired interface or a wireless interface. According to different actual application scenarios, the devices connected to the input / output interface 24 can be a component of the electronic device, or an external device connected to the electronic device when needed.

[0132] Optionally, the memory 21 can be a volatile memory and / or a non-volatile memory. The volatile memory can be a random access memory, etc. The non-volatile memory can be a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, or a flash memory, etc.

[0133] Optionally, the computer program stored in the processor 22 can be divided into one or more modules, which are stored in the memory 21 and executed by the processor 22 to complete the method provided by the embodiment. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the electronic device.

[0134] Optionally, the processor 22 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 22 include, but are not limited to, a central processing unit, a graphics processing unit, a digital signal processor, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, and any appropriate controller, microcontroller, processor, etc. The processor 22 executes various methods and processes of the embodiment, exemplarily, a sound source event positioning method of an optical cable of the embodiment.

[0135] Optionally, the bus 25 can include a channel for transmitting information. According to different functions, the bus 25 can be divided into an address bus, a data bus, a control bus, etc.

[0136] In an optional implementation, the embodiment further provides a computer storage medium having a computer program stored thereon, which enables a computer to execute the method of the method embodiment when executed by the computer. Part or all of the computer program can be loaded and / or installed on the memory 21 of the electronic device. When the computer program is executed by the processor 22, one or more steps of the sound source event positioning method of the optical cable of the embodiment can be executed.

[0137] Optionally, the computer readable storage medium can be a random access memory, a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, etc.

[0138] Obviously, the above embodiments of the present application are only examples for clearly illustrating the technical solutions of the present application, and are not intended to limit the specific embodiments of the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the claims of the present application shall be included in the protection scope of the claims of the present application.

Claims

1. A DAS-based acoustic source event localization method, characterized in that, The method comprises: obtaining an audio feature image according to an abnormal sound source signal collected by a DAS; extracting features of the audio feature image to obtain an audio feature vector; obtaining an event feature vector according to the audio feature vector, and identifying an abnormal sound source event according to the event feature vector; splicing the event feature vector and the audio feature vector to obtain a spliced feature vector, and processing the spliced feature vector to obtain a sound source distance. The method comprises: dividing the optical cable into a plurality of audio channels according to a preset interval length of the optical cable; obtaining audio digital signals of a preset number of the audio channels in the optical cable according to a position point of the abnormal sound source signal collected by the DAS; processing the audio digital signals to obtain the audio feature image.

2. The DAS-based acoustic source event localization method of claim 1, wherein, The method comprises: obtaining the audio channel where the position point is located as a first audio channel; obtaining a plurality of audio channels before and after the first audio channel as second audio channels according to the preset number; obtaining audio digital signals of the first audio channel and each of the second audio channels.

3. The DAS-based acoustic source event localization method of claim 1, wherein, The method comprises: normalizing the obtained audio digital signals, stacking the normalized audio digital signals according to a preset channel order, and obtaining the audio feature image according to the stacked audio digital signals.

4. The DAS-based acoustic source event localization method of claim 1, wherein, The abnormal sound source event comprises a main sound source event and a sub-sound source event; The main sound source event is an abnormal sound source event type of the abnormal sound source signal; The sub-sound source event is a vibration state of the main sound source event, and the sub-sound source event at least comprises a vibration starting state, a vibration state, and a vibration ending state.

5. The DAS-based acoustic source event localization method according to any one of claims 1-4, characterized in that, The method comprises: extracting features of the audio feature image through a plurality of convolution units connected in sequence to obtain convolution features; regularizing the convolution features through a regularization unit to obtain regularized features; processing the regularized features through a feature output unit to obtain the audio feature vector.

6. The DAS-based acoustic source event localization method according to any one of claims 1-4, characterized in that, The method comprises: linearly processing the audio feature vector through a first linear unit to obtain the event feature vector; processing the event feature vector through a category output unit to obtain the abnormal sound source event.

7. The DAS-based acoustic source event localization method according to any one of claims 1-4, characterized in that, The method comprises: splicing the event feature vector and the audio feature vector through a feature splicing unit to obtain a spliced feature vector; The spliced feature vector is processed by a second linear unit, and the feature-processed spliced feature vector is recognized by a distance output unit to obtain the sound source distance.

8. A DAS-based acoustic source event location system, characterized by The system comprises: An audio image acquisition module is configured to acquire an audio feature image based on the abnormal sound source signal collected by the DAS. An audio vector acquisition module is configured to extract features from the audio feature image to obtain an audio feature vector. An event type acquisition module is configured to acquire an event feature vector based on the audio feature vector, and identify an abnormal sound source event based on the event feature vector. A sound source distance acquisition module is configured to splice the event feature vector and the audio feature vector to obtain a spliced feature vector, and process the spliced feature vector to obtain a sound source distance. The audio feature image is acquired based on the abnormal sound source signal collected by the DAS, and comprises: The optical cable is divided into a plurality of audio channels according to a preset optical cable interval length. The audio digital signals of a preset number of the audio channels in the optical cable are acquired based on the optical cable position points of the abnormal sound source signal collected by the DAS. The audio digital signals are processed to obtain the audio feature image.

9. The DAS-based acoustic source event localization system of claim 8, wherein, The audio digital signals of a preset number of the audio channels in the optical cable are acquired based on the optical cable position points of the abnormal sound source signal collected by the DAS, and comprise: The audio channel where the optical cable position point is located is acquired as a first audio channel. A plurality of audio channels before and after the first audio channel are acquired as second audio channels based on the preset number. The audio digital signals of the first audio channel and each of the second audio channels are acquired respectively.

10. An electronic device, comprising: It comprises: A memory is configured to store one or more computer programs. A processor is configured to implement the DAS-based sound source event positioning method according to any one of claims 1-7 when the one or more computer programs are executed by the processor.

Images