Water leakage sound automatic recognition system and method based on artificial intelligence

Abstract

The present disclosure relates to an automatic water leakage sound recognition system and method based on artificial intelligence. The method includes: using a sound sensor to collect sound signals; using a high-pass filter, a low-pass filter and a programmable amplifier to respectively perform high-pass filtering, low-pass filtering and amplification processing on the sound signal to obtain a pre-processed sound signal; using an audio interface chip to convert the pre-processed sound signal into a sound signal digital signal modal data; using an audio conversion chip to convert the sound signal digital signal modal data into an analog signal to obtain a sound signal analog modal data; using a wireless communication module to transmit the sound signal analog modal data to a server; and using a server to process the sound signal analog modal data to determine whether there is a leak. In this way, the automatic recognition and positioning of the sound of water leakage can be realized, thereby improving the accuracy and efficiency of water leakage point detection, and providing more reliable technical support for water pipe maintenance and management.

Description

Technical Field

[0001] The present disclosure relates to the technical field of automatic sound recognition, and in particular, to an automatic water leakage sound recognition system and method based on artificial intelligence. Background Art

[0002] When searching for leaks in underground water pipes, it is usually necessary to locate the leaks with the help of a leak detector. The function of a leak detector is to capture the sound signal generated by the leak through a sound sensor, transmit it to the headphones after amplification, and display the amplitude and spectrum of the sound on the screen. Such equipment can help people judge the sound of water leakage and locate the leak point based on the change in sound amplitude. However, in actual work, the sound of water leakage is often very weak, and it requires experienced inspectors to accurately identify it. In other words, experienced professionals can identify the sound of water leakage and find the leak point with their years of accumulated experience. On the contrary, it is difficult for inexperienced people to accurately identify the sound of water leakage, which makes it difficult to detect the leak point.

[0003] Therefore, an automatic water leakage sound recognition system based on artificial intelligence is desired. Summary of the invention

[0004] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0005] In a first aspect, the present disclosure provides an automatic water leakage sound recognition system based on artificial intelligence, the system comprising:

[0006] Sound sensor, high-pass filter, low-pass filter, programmable amplifier, audio interface chip, audio conversion chip, wireless communication module and server, among which,

[0007] The sound sensor is used to collect sound signals;

[0008] The high-pass filter, the low-pass filter and the programmable amplifier are used to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal respectively to obtain a pre-processed sound signal;

[0009] The audio interface chip is used to convert the pre-processed sound signal into sound signal digital signal modal data;

[0010] The audio conversion chip is used to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data;

[0011] The wireless communication module is used to transmit the sound signal simulation modal data to the server; and

[0012] The server is used to process the sound signal simulation modal data to determine whether there is a leak.

[0013] Optionally, the server includes: a sound simulation signal wavelet transformation module, which is used to use wavelet transform to process the sound signal simulation modal data to obtain a two-dimensional time-frequency diagram of the sound signal; a sound signal two-dimensional time-frequency diagram grayscale processing module, which is used to grayscale the sound signal two-dimensional time-frequency diagram to obtain a sound signal time-frequency grayscale diagram; a sound signal time-frequency super-pixel block semantic association coding module, which is used to perform pixel block association semantic coding processing based on super-pixel segmentation on the sound signal time-frequency grayscale diagram to obtain a sequence of context sound signal time-frequency grayscale super-pixel block semantic coding feature vectors; a sound signal important content semantic highlighting module, which is used to pass the sequence of context sound signal time-frequency grayscale super-pixel block semantic coding feature vectors through an important content attention integration network to obtain a sound signal important content semantic highlighting coding representation vector; a leak detection module, which is used to determine whether there is a leak based on the sound signal important content semantic highlighting coding representation vector.

[0014] Optionally, the sound signal time-frequency superpixel block semantic association coding module includes: a sound signal time-frequency grayscale superpixel segmentation unit, which is used to use the superpixel segmentation module to perform image segmentation on the sound signal time-frequency grayscale image to obtain a set of sound signal time-frequency grayscale superpixel blocks; a sound signal time-frequency grayscale superpixel block context semantic coding unit, which is used to pass the set of sound signal time-frequency grayscale superpixel blocks through a pixel block association semantic encoder including a superpixel block embedding encoder to obtain a sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors.

[0015] Optionally, the module for highlighting the semantics of important content in sound signals comprises: an autocorrelation semantic metric coefficient calculation unit, which is used to calculate the autocorrelation semantic metric coefficients between superpixel blocks of each context sound signal time-frequency grayscale superpixel block semantic coding feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors to obtain a sequence of autocorrelation semantic metric coefficients between superpixel blocks; a normalization processing unit, which is used to normalize the sequence of autocorrelation semantic metric coefficients between superpixel blocks to obtain a sequence of autocorrelation attention weight coefficients between superpixel blocks; and a vector-by-vector weighted sum calculation unit, which is used to calculate the vector-by-vector weighted sum of the sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors using the sequence of autocorrelation attention weight coefficients between superpixel blocks as weights to obtain the semantically highlighted coding representation vector of important content in the sound signal.

[0016] Optionally, the autocorrelation semantic metric coefficient calculation unit includes: a matrix product calculation subunit, which is used to respectively calculate the matrix products between each weight coefficient matrix and the corresponding semantic coding feature vector of each context sound signal time-frequency grayscale super pixel block to obtain a sequence of context sound signal time-frequency grayscale super pixel block semantic weighted feature vectors; a vector addition subunit, which is used to add the corresponding bias weight vector to each context sound signal time-frequency grayscale super pixel block semantic weighted feature vector in the sequence of context sound signal time-frequency grayscale super pixel block semantic weighted feature vectors to obtain the context sound signal. A sequence of time-frequency grayscale superpixel block semantic bias feature vectors; an activation subunit, used to activate the sequence of context sound signal time-frequency grayscale superpixel block semantic bias feature vectors to obtain a sequence of context sound signal time-frequency grayscale superpixel block semantic activation feature vectors; a vector product subunit, used to calculate the product between each context sound signal time-frequency grayscale superpixel block semantic activation feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic activation feature vectors and the transposed vector of a preset reference feature vector to obtain a sequence of autocorrelation semantic measurement coefficients between the superpixel blocks.

[0017] Optionally, the activation subunit is used to: use a Selu activation function to activate the sequence of context sound signal time-frequency grayscale superpixel block semantic bias feature vectors to obtain the sequence of context sound signal time-frequency grayscale superpixel block semantic activation feature vectors.

[0018] Optionally, the leak detection module is used to: pass the semantically highlighted encoding representation vector of the important content of the sound signal through a classifier-based leak detector to obtain a detection result, and the detection result is used to indicate whether a leak exists.

[0019] Optionally, it also includes a training module for training the pixel block association semantic encoder including the superpixel block embedding encoder, the important content attention integration network and the classifier-based missing point detector.

[0020] Optionally, the training module includes: a training sound sensor unit, which is used to collect the training sound signal and determine whether there is a real value of a leakage point; a training preprocessing unit, in which the high-pass filter, the low-pass filter and the programmable amplifier are respectively used to perform high-pass filtering, low-pass filtering and amplification processing on the training sound signal to obtain a training preprocessed sound signal; a training audio interface chip unit, which is used to convert the training preprocessed sound signal into a training sound signal digital signal modal data; a training audio conversion chip unit, which is used to convert the training sound signal digital signal modal data into a training analog signal to obtain a training sound signal analog modal data; a training sound analog signal wavelet change unit, which is used to process the training sound signal analog modal data using wavelet transform to obtain a two-dimensional time-frequency diagram of the training sound signal; a training sound signal two-dimensional time-frequency diagram grayscale processing unit, which is used to grayscale the training sound signal two-dimensional time-frequency diagram to obtain a training sound signal time-frequency grayscale diagram; a training sound signal time-frequency superpixel block An inter-semantic association coding unit is used to perform inter-pixel association semantic coding processing based on superpixel segmentation on the time-frequency grayscale image of the training sound signal to obtain a sequence of time-frequency grayscale superpixel block semantic coding feature vectors of the training context sound signal; a training sound signal important content semantic highlighting unit is used to pass the sequence of time-frequency grayscale superpixel block semantic coding feature vectors of the training context sound signal through the important content attention integration network to obtain the training sound signal important content semantic highlighting coding representation vector; a training classification unit is used to pass the training sound signal important content semantic highlighting coding representation vector through the classifier-based leak detector to obtain a detection result; a training classification loss function value calculation unit is used to calculate the cross entropy loss function value between the detection result and the true value to obtain a classification loss function value; a training unit is used to train the inter-pixel association semantic encoder including the superpixel block embedding encoder, the important content attention integration network and the classifier-based leak detector based on the classification loss function value.

[0021] In a second aspect, the present disclosure provides a method for automatic recognition of water leakage sound based on artificial intelligence, the method comprising:

[0022] Use sound sensors to collect sound signals;

[0023] Use a high-pass filter, a low-pass filter and a programmable amplifier to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal respectively to obtain a pre-processed sound signal;

[0024] Using an audio interface chip to convert the pre-processed sound signal into sound signal digital signal modal data;

[0025] Using an audio conversion chip to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data;

[0026] Using a wireless communication module to transmit the sound signal simulation modal data to a server; and

[0027] The server is used to process the sound signal simulation modal data to determine whether there is a leak.

[0028] The above technical solution is adopted, by using a sound sensor to collect sound signals; using a high-pass filter, a low-pass filter and a programmable amplifier to respectively perform high-pass filtering, low-pass filtering and amplification processing on the sound signals to obtain pre-processed sound signals; using an audio interface chip to convert the pre-processed sound signals into sound signal digital signal modal data; using an audio conversion chip to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data; using a wireless communication module to transmit the sound signal analog modal data to a server; and using a server to process the sound signal analog modal data to determine whether there is a leak. In this way, the automatic recognition and positioning of the sound of water leakage can be realized, thereby improving the accuracy and efficiency of water leakage point detection, and providing more reliable technical support for water pipe maintenance and management.

[0029] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings:

[0031] Figure 1 The present invention is a schematic diagram showing a leak detector implementing artificial intelligence identification based on the Internet of Things according to an exemplary embodiment.

[0032] Figure 2 The figure is a schematic diagram of the structure of an intelligent digital leak detector according to an exemplary embodiment.

[0033] Figure 3 It is a block diagram of an automatic water leakage sound recognition system based on artificial intelligence according to an exemplary embodiment.

[0034] Figure 4 It is a block diagram of the server of the automatic water leakage sound recognition system based on artificial intelligence according to an exemplary embodiment.

[0035] Figure 5The present invention is a flowchart of a method for automatically identifying water leakage sound based on artificial intelligence according to an exemplary embodiment.

[0036] Figure 6 It is a block diagram of an electronic device according to an exemplary embodiment.

[0037] Figure 7 It is an application scenario diagram of an automatic water leakage sound recognition system based on artificial intelligence according to an exemplary embodiment. DETAILED DESCRIPTION

[0038] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0039] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0040] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0041] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0042] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0043] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0044] When searching for leaks in underground water pipes, it is often necessary to use a leak detector to find the leaks. The function of the leak detector is to amplify the sound signal of the leak obtained by the sound sensor, send it to the earphone, and display the amplitude and spectrum of the sound on the screen, so as to judge the sound of the leak and find the leak according to the change in the amplitude of the sound of the leak. During work, the sound of the leak is often very small. Due to different experiences, experienced people can distinguish the sound of the leak by experience and find the leak. But for inexperienced people, it is difficult to distinguish the sound of the leak. In this regard, the present invention is based on the Internet of Things and combines big data with artificial intelligence to automatically identify the sound of the leak and quickly find the leak.

[0045] The present invention effectively reduces the misjudgment caused by human factors and greatly improves work efficiency. Since the results of each measurement are uploaded to the server, the data is continuously improved, and the judgment accuracy is continuously improved, and finally reaches a 100% accuracy level.

[0046] Specifically, Figure 1 As shown, the sound sensor converts the sound signal into an electrical signal, which is amplified by high-pass and low-pass filters, and then converted into a digital signal through an audio interface chip and transmitted to the central processor. The central processor processes the digital signal and stores it in a memory, and then transmits it to the earphone through an audio conversion chip, so that the inspector can hear the sound of water leakage. The inspector can record and save it. The leak detector has an onboard WAPI or WIFI module. Users can upload the collected sound data to the server in real time through the wireless network. The server automatically determines whether there is a leak based on the data provided by the user and the historical experience data, and sends the result to the leak detector in real time, and displays the characteristic spectrum data at the same time. Until the exact location of the leak is found. At the same time, the collected data and analysis results are classified and saved. In this way, as time goes by, the user's detection data continues to increase, and the accuracy of artificial intelligence detection and judgment will gradually approach 100%.

[0047] In one embodiment of the present disclosure, an intelligent digital leak detector is provided. Figure 2 As shown, the intelligent digital leak detector is a very effective test equipment for detecting water leakage points. By connecting sensors, it can collect water leakage noise transmitted along the pipe wall and soil, and help users accurately locate the water leakage point by amplifying the water leakage sound and filtering the signal.

[0048] Its functions include: 1. Automatic filtering function: The device can automatically adjust the filtering range according to the acquired sound signal spectrum, without manual frequency selection, and the operation is simpler. 2. Spectrum analysis function: The filtering range of the continuous measurement mode of the device is 0-6000Hz, and it has spectrum analysis function. The operator can select the appropriate filtering range in different test scenarios through spectrum analysis. 3. Accurate positioning: After confirming the approximate location of the leak, multi-point leak detection and measurement of the pipeline can be performed to more accurately locate the leak. 4. Smartphone software: The device can connect to the mobile phone software via Bluetooth to realize the transmission of location, leak sound, and test data, and the leak point can be marked on the software map.

[0049] The objects to be tested can be underground water supply pipes, fire protection pipes, heating pipes, etc.

[0050] In order to solve the above problems, the present disclosure provides an automatic recognition system and method for water leakage sound based on artificial intelligence, which collects sound signals by using sound sensors; uses high-pass filters, low-pass filters and programmable amplifiers to perform high-pass filtering, low-pass filtering and amplification processing on sound signals respectively to obtain pre-processed sound signals; uses an audio interface chip to convert the pre-processed sound signals into sound signal digital signal modal data; uses an audio conversion chip to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data; uses a wireless communication module to transmit the sound signal analog modal data to a server; and uses a server to process the sound signal analog modal data to determine whether there is a leak. In this way, the automatic recognition and positioning of water leakage sounds can be realized, thereby improving the accuracy and efficiency of water leakage point detection and providing more reliable technical support for water pipe maintenance and management.

[0051] The specific embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0052] Figure 3 FIG. 1 is a block diagram of an automatic water leakage sound recognition system based on artificial intelligence according to an exemplary embodiment. Figure 3 As shown, the water leakage sound automatic identification system 100 includes:

[0053] Sound sensor 1, high pass filter 2, low pass filter 3, programmable amplifier 4, audio interface chip 5, audio conversion chip 6, wireless communication module 7 and server 8, wherein:

[0054] The sound sensor 1 is used to collect sound signals;

[0055] The high-pass filter 2, the low-pass filter 3 and the programmable amplifier 4 are used to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal respectively to obtain a pre-processed sound signal;

[0056] The audio interface chip 5 is used to convert the pre-processed sound signal into sound signal digital signal modal data;

[0057] The audio conversion chip 6 is used to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data;

[0058] The wireless communication module 7 is used to transmit the sound signal simulation modal data to the server 8; and

[0059] The server 8 is used to process the sound signal simulation modal data to determine whether there is a leak.

[0060] Figure 4 FIG. 1 is a block diagram of a server of a water leakage sound automatic recognition system based on artificial intelligence according to an exemplary embodiment. Figure 4 As shown, the server 8 includes:

[0061] The sound analog signal wavelet transformation module 101 is used to process the sound signal analog modal data using wavelet transformation to obtain a two-dimensional time-frequency diagram of the sound signal;

[0062] The sound signal two-dimensional time-frequency graph grayscale processing module 102 is used to perform grayscale processing on the sound signal two-dimensional time-frequency graph to obtain a sound signal time-frequency grayscale graph;

[0063] The sound signal time-frequency super-pixel block semantic association coding module 103 is used to perform pixel block association semantic coding processing based on super-pixel segmentation on the sound signal time-frequency grayscale image to obtain a sequence of context sound signal time-frequency grayscale super-pixel block semantic coding feature vectors;

[0064] The sound signal important content semantic highlighting module 104 is used to pass the sequence of semantic coding feature vectors of the context sound signal time-frequency grayscale super pixel block through the important content attention integration network to obtain the sound signal important content semantic highlighting coding representation vector;

[0065] The missing point detection module 105 is used to determine whether there is a missing point based on the semantic highlighting encoding representation vector of the important content of the sound signal.

[0066] In view of the above technical problems, in the technical solution of the present invention, an automatic water leakage sound recognition system based on artificial intelligence is proposed, which includes: a sound sensor, a high-pass filter, a low-pass filter, a programmable amplifier, an audio interface chip, an audio conversion chip (ARM in the figure), a wireless communication module and a server, wherein the sound sensor is used to collect sound signals; the high-pass filter, the low-pass filter and the programmable amplifier are used to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal respectively to obtain a pre-processed sound signal; the audio interface chip is used to convert the pre-processed sound signal into sound signal digital signal modal data; the audio conversion chip is used to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data; the wireless communication module is used to transmit the sound signal analog modal data to the server; the server is used to process the sound signal analog modal data to determine whether there is a leak.

[0067] Accordingly, in the above-mentioned automatic water leakage sound recognition system based on artificial intelligence, after a series of processing is performed on the sound signal collected by the sound sensor and converted into sound signal simulation modal data, a signal processing and analysis algorithm based on artificial intelligence and deep learning is introduced on the server side to analyze the sound signal simulation modal data, so as to determine whether there is a leak. In this way, the automatic recognition and positioning of water leakage sound can be realized, avoiding the problems of low efficiency and low precision brought by traditional manual detection methods, thereby improving the accuracy and efficiency of water leakage point detection and providing more reliable technical support for water pipe maintenance and management.

[0068] Specifically, in the technical solution of the present disclosure, in the server, first, the sound signal simulation modal data is processed using wavelet transform to obtain a two-dimensional time-frequency diagram of the sound signal. It should be understood that since the sound signal is a time domain signal, and the wavelet transform has good resolution characteristics in the time and frequency domain, it can provide good time resolution and frequency resolution at the same time, which enables the wavelet transform to more accurately capture the instantaneous characteristics and frequency characteristics of the signal in the sound signal processing, which is conducive to a detailed analysis of the sound signal. That is to say, in the sound signal processing, the sound of water leakage may have specific frequency components and time domain characteristics, and these characteristics can be better extracted by wavelet transform to help identify the sound of water leakage. Therefore, the obtained two-dimensional time-frequency diagram of the sound signal can intuitively show the characteristic changes of the sound signal in time and frequency, which helps to more accurately analyze the characteristics of the sound signal. In addition, the wavelet transform can also effectively suppress the noise in the signal and improve the signal-to-noise ratio. In the recognition of water leakage sound, since the sound of water leakage is often weak and is disturbed by environmental noise, the wavelet transform can help reduce the influence of noise on signal analysis and improve the detection accuracy of water leakage sound.

[0069] Then, since the two-dimensional time-frequency diagram of the sound signal contains a large amount of various semantic information about the sound signal, in order to highlight the brightness and contrast in the two-dimensional time-frequency diagram and make the main features of the sound in the image more obvious, in the technical solution disclosed in the present invention, the two-dimensional time-frequency diagram of the sound signal is further grayed to obtain a sound signal time-frequency grayscale diagram. By graying the time-frequency diagram, the characteristic changes of the sound signal in the time-frequency domain can be better displayed, which helps to highlight important sound features. In other words, the grayscale image is easier to analyze and understand, and can intuitively display the time-frequency information of the sound signal. In addition, the grayscale processing can convert the color image of the two-dimensional time-frequency diagram of the sound signal into a grayscale image, reducing the amount of calculation for data processing and improving processing efficiency. After the sound signal time-frequency diagram is grayed, subsequent analysis work such as feature extraction and pattern recognition can be more conveniently performed, which is conducive to the automatic recognition and detection of water leakage sound.

[0070] In one embodiment of the present disclosure, the sound signal time-frequency superpixel block semantic association coding module includes: a sound signal time-frequency grayscale superpixel segmentation unit, which is used to use the superpixel segmentation module to perform image segmentation on the sound signal time-frequency grayscale image to obtain a set of sound signal time-frequency grayscale superpixel blocks; a sound signal time-frequency grayscale superpixel block context semantic coding unit, which is used to pass the set of sound signal time-frequency grayscale superpixel blocks through an inter-pixel block association semantic encoder including a superpixel block embedding encoder to obtain a sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors.

[0071] It should be understood that since the sound signal time-frequency grayscale image contains the frequency components and time domain features of different sounds, in order to be able to perform pixel-level segmentation and classification on the information of these different sound components, so as to segment the time-frequency image into regions with similar features, so as to facilitate the subsequent extraction and analysis of the time-frequency semantic features of different sound components, in the technical solution disclosed in the present invention, the sound signal time-frequency grayscale image is further segmented using a superpixel segmentation module to obtain a set of sound signal time-frequency grayscale superpixel blocks. The sound signal time-frequency grayscale image can be segmented into multiple compact regions with similar features through superpixel segmentation to reduce the computational complexity in subsequent processing steps. Therefore, superpixel segmentation can help extract local features of the image so that the pixels inside each superpixel block have higher similarity. This helps to distinguish regions with different frequency components and features in the sound signal, thereby more accurately capturing local sound time-frequency features with higher similarity in the sound signal time-frequency image, and providing a more accurate basis for subsequent sound signal time-frequency feature analysis and leak detection.

[0072] Further, considering that in each sound signal time-frequency grayscale superpixel block in the set of the sound signal time-frequency grayscale superpixel block, each pixel block has its own semantic feature information, and there is mutual correlation and contextual relationship between the semantic features of these pixel blocks. Based on this, in order to be able to perform semantic analysis and contextual coding for each pixel block in the sound signal time-frequency grayscale map, in the technical solution disclosed in the present invention, the set of the sound signal time-frequency grayscale superpixel block is further passed through a pixel block correlation semantic encoder including a superpixel block embedding encoder to obtain a sequence of contextual sound signal time-frequency grayscale superpixel block semantic coding feature vectors. In particular, the superpixel block embedding encoder can perform embedded coding of superpixel blocks by adopting a fully connected layer to capture the embedded semantic features in each superpixel block. In addition, the pixel block correlation semantic encoder also captures the contextual correlation information between the semantic features of each pixel block by adopting a Transformer structure. Therefore, the pixel block correlation semantic encoder including the superpixel block embedding encoder can effectively integrate the contextual semantics and correlation information between each superpixel block, helping the system to better understand the semantic correlation relationship between different superpixel blocks in the sound signal time-frequency map. This helps improve the system's overall understanding of sound signals, thereby helping to better understand the overall meaning and characteristics of sound signals and provide richer information for leak detection.

[0073] Next, considering that each contextual sound signal time-frequency grayscale superpixel block semantic coding feature vector in the sequence of the contextual sound signal time-frequency grayscale superpixel block semantic coding feature vector respectively contains the superpixel block contextual semantic features of different sound components in the sound signal time-frequency grayscale map, the superpixel block semantic features of these different sound components have correlation and mutual influence. At the same time, the superpixel block contextual semantics of different sound components have different importance in the pipeline leak detection task. Based on this, in order to help the system focus on the most important part of the sound signal and highlight the key information to perform leak detection more accurately, in the technical solution of the present disclosure, the sequence of the contextual sound signal time-frequency grayscale superpixel block semantic coding feature vector is further passed through the important content attention integration network to obtain the sound signal important content semantic highlight coding representation vector. Through the processing of the important content attention integration network, it is possible to use the attention mechanism to reduce the attention to the secondary information in the sound signal in the process of superpixel block semantic feature fusion of various sounds, thereby reducing the interference of noise on the sound signal analysis, so as to highlight the key features in the sound signal and help the system better understand the important semantic content of the sound signal for subsequent pipeline leak detection.

[0074] In one embodiment of the present disclosure, the module for semantically highlighting important content of sound signals includes: an autocorrelation semantic metric coefficient calculation unit, which is used to calculate the autocorrelation semantic metric coefficients between superpixel blocks of each context sound signal time-frequency grayscale superpixel block semantic coding feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors to obtain a sequence of autocorrelation semantic metric coefficients between superpixel blocks; a normalization processing unit, which is used to normalize the sequence of autocorrelation semantic metric coefficients between superpixel blocks to obtain a sequence of autocorrelation attention weight coefficients between superpixel blocks; and a vector-by-vector weighted sum calculation unit, which is used to calculate the vector-by-vector weighted sum of the sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors using the sequence of autocorrelation attention weight coefficients between superpixel blocks as weights to obtain the semantically highlighting coding representation vector of important content of the sound signal.

[0075] Furthermore, in one embodiment of the present disclosure, the autocorrelation semantic metric coefficient calculation unit includes: a matrix product calculation subunit, which is used to respectively calculate the matrix product between each weight coefficient matrix and the corresponding semantic coding feature vector of each context sound signal time-frequency grayscale super pixel block to obtain a sequence of context sound signal time-frequency grayscale super pixel block semantic weighted feature vectors; a vector addition subunit, which is used to add the corresponding bias weight vector to each context sound signal time-frequency grayscale super pixel block semantic weighted feature vector in the sequence of context sound signal time-frequency grayscale super pixel block semantic weighted feature vectors to obtain A sequence of context sound signal time-frequency grayscale superpixel block semantic bias feature vectors; an activation subunit, used to activate the sequence of context sound signal time-frequency grayscale superpixel block semantic bias feature vectors to obtain a sequence of context sound signal time-frequency grayscale superpixel block semantic activation feature vectors; a vector product subunit, used to calculate the product between each context sound signal time-frequency grayscale superpixel block semantic activation feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic activation feature vectors and the transposed vector of a preset reference feature vector to obtain a sequence of autocorrelation semantic measurement coefficients between the superpixel blocks.

[0076] Furthermore, in one embodiment of the present disclosure, the activation subunit is used to: use the Selu activation function to activate the sequence of semantic bias feature vectors of the context sound signal time-frequency grayscale superpixel block to obtain the sequence of semantic activation feature vectors of the context sound signal time-frequency grayscale superpixel block.

[0077] Specifically, in one embodiment of the present disclosure, the sequence of the semantic coding feature vectors of the context sound signal time-frequency grayscale superpixel block is processed by the important content attention integration network with the following important content semantic highlighting formula to obtain the important content semantic highlighting coding representation vector of the sound signal;

[0078] The formula for highlighting the important content semantics is:

[0079]

[0080]

[0081]

[0082] Among them, h i is the i-th context sound signal time-frequency grayscale superpixel block semantic coding feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors, and W i Represent the preset reference feature vector and weight coefficient matrix, B i is the bias weight vector, Selu(·) represents the Selu function, e i is the attention score value of the semantic coding feature vector of the i-th time-frequency grayscale superpixel block of the context sound signal, λ and α are both hyperparameters, softmax(·) represents the softmax function, t is the number of vectors in the sequence of semantic coding feature vectors of the time-frequency grayscale superpixel block of the context sound signal, and V is the semantic highlight coding representation vector of the important content of the sound signal.

[0083] Then, the semantically highlighted encoding representation vector of the important content of the sound signal is passed through a leak detector based on a classifier to obtain a detection result, and the detection result is used to indicate whether there is a leak. In other words, the semantic fusion features of the important content in the sound signal are used for classification processing to determine whether there is a leak. In this way, the automatic recognition and positioning of the sound of water leakage can be achieved, avoiding the problems of low efficiency and low precision brought about by traditional manual detection methods, thereby improving the accuracy and efficiency of water leakage detection and providing more reliable technical support for water pipe maintenance and management.

[0084] In one embodiment of the present disclosure, the leakage point detection module is used to: pass the semantically highlighted encoding representation vector of the important content of the sound signal through a classifier-based leakage point detector to obtain a detection result, and the detection result is used to indicate whether there is a leakage point.

[0085] Furthermore, in one embodiment of the present disclosure, the artificial intelligence-based automatic water leakage sound recognition system also includes a training module for training the pixel block association semantic encoder including the superpixel block embedding encoder, the important content attention integration network and the classifier-based leakage point detector. The training module includes: a training sound sensor unit, which is used to collect the training sound signal and the real value of whether there is a leakage point; a training preprocessing unit, in which the high-pass filter, the low-pass filter and the programmable amplifier are respectively used to perform high-pass filtering, low-pass filtering and amplification processing on the training sound signal to obtain the training preprocessed sound signal; a training audio interface chip unit, which is used to convert the training preprocessed sound signal into a training sound signal digital signal modal data; a training audio conversion chip unit, which is used to convert the training sound signal digital signal modal data into a training analog signal to obtain the training sound signal analog modal data; a training sound analog signal wavelet change unit, which is used to use wavelet transform to process the training sound signal analog modal data to obtain a two-dimensional time-frequency diagram of the training sound signal; a training sound signal two-dimensional time-frequency diagram grayscale processing unit, which is used to grayscale the training sound signal two-dimensional time-frequency diagram to obtain a training sound signal time-frequency grayscale diagram; a training sound signal time-frequency superpixel block interlinguistic A semantic association coding unit is used to perform pixel block association semantic coding processing based on superpixel segmentation on the time-frequency grayscale image of the training sound signal to obtain a sequence of semantic coding feature vectors of the time-frequency grayscale superpixel blocks of the training context sound signal; a training sound signal important content semantic highlighting unit is used to pass the sequence of semantic coding feature vectors of the time-frequency grayscale superpixel blocks of the training context sound signal through the important content attention integration network to obtain the important content semantic highlighting coding representation vector of the training sound signal; a training classification unit is used to pass the important content semantic highlighting coding representation vector of the training sound signal through the classifier-based leak detector to obtain a detection result; a training classification loss function value calculation unit is used to calculate the cross entropy loss function value between the detection result and the true value to obtain a classification loss function value; a training unit is used to train the pixel block association semantic encoder including the superpixel block embedding encoder, the important content attention integration network and the classifier-based leak detector based on the classification loss function value.

[0086] In the technical solution disclosed in the present invention, each training context sound signal time-frequency grayscale superpixel block semantic coding feature vector in the sequence of training context sound signal time-frequency grayscale superpixel block semantic coding feature vectors respectively represents the local time-frequency context semantic association features of the training sound detection signal. When the sequence of training context sound signal time-frequency grayscale superpixel block semantic coding feature vectors is passed through the important content attention integration network, the important content attention integration network measures the semantic importance of each training context sound signal time-frequency grayscale superpixel block semantic coding feature vector and fuses the sequence of training context sound signal time-frequency grayscale superpixel block semantic coding feature vectors through an attention mechanism to obtain a significant expression of the global time-frequency semantic association features of the training sound signal. However, considering that each training sound signal time-frequency grayscale superpixel block in the set of the training sound signal time-frequency grayscale superpixel blocks has source domain content differences in the time-frequency domain, in the process of measuring semantic importance, not only the effective information components of each training sound signal time-frequency grayscale superpixel block are considered, but also the noise is taken into account, which makes the significant expression of the global time-frequency semantic association features of the training sound signal, that is, the semantic highlighting encoding representation vector of the important content of the training sound signal, have poor overall distribution regression constraints on the classifier-based leakage detector.

[0087] In a preferred example, the important content semantic highlighting encoding representation vector of the training sound signal is passed through a classifier-based leak detector to obtain a detection result, specifically comprising the following steps:

[0088] The training sound signal important content semantic salient coding representation vector is multiplied by the square root of the length of the training sound signal important content semantic salient coding representation vector and the reciprocal of the square root of the second norm of the training sound signal important content semantic salient coding representation vector to obtain the training sound signal important content semantic rotation offset feature vector;

[0089] Calculating an exponential function of the important content semantic rotation shift feature vector of the training sound signal with a natural constant as the base to obtain the important content semantic class shift prediction feature vector of the training sound signal;

[0090] Performing a dot multiplication of the training sound signal important content semantic salient coding representation vector with a norm and a weight hyperparameter of the training sound signal important content semantic salient coding representation vector to obtain a training sound signal important content semantic boundary constraint feature vector;

[0091] The training sound signal important content semantic class offset prediction feature vector is interpolated with the training sound signal important content semantic boundary constraint feature vector to obtain an optimized sound signal important content semantic salient coding representation vector; and

[0092] The optimized sound signal important content semantic highlighting encoding representation vector is passed through the classifier-based leak detector to obtain the detection result.

[0093] In the above preferred example, the structured norm representation of the training sound signal important content semantically salient coding representation vector is used as the local canonical coordinates for each eigenvalue of the training sound signal important content semantically salient coding representation vector to determine the class rotation offset of the vector overall distribution of the training sound signal important content semantically salient coding representation vector relative to the eigenvalue for the class offset prediction direction of each eigenvalue of the training sound signal important content semantically salient coding representation vector as the center, and the bounding box of the vector eigenvalue distribution of the training sound signal important content semantically salient coding representation vector is used to constrain the eigenvalues ​​to improve the constraint of the training sound signal important content semantically salient coding representation vector under the overall regression distribution, thereby improving the training speed of the model and the accuracy of the detection results obtained by the training sound signal important content semantically salient coding representation vector through the classifier-based leak detector. In this way, pipeline leak detection can be performed more accurately, avoiding the problems of low efficiency and low accuracy brought about by traditional manual detection methods, and providing more reliable technical support for water pipe maintenance and management.

[0094] In summary, the above scheme is adopted to convert the sound signal collected by the sound sensor into sound signal simulation modal data through a series of processing, and then introduce a signal processing and analysis algorithm based on artificial intelligence and deep learning on the server side to analyze the sound signal simulation modal data to determine whether there is a leak. In this way, the automatic recognition and positioning of the sound of water leakage can be realized, avoiding the problems of low efficiency and low precision brought by traditional manual detection methods, thereby improving the accuracy and efficiency of water leakage point detection and providing more reliable technical support for water pipe maintenance and management.

[0095] Figure 5 is a flow chart of a method for automatic recognition of water leakage sound based on artificial intelligence according to an exemplary embodiment, such as Figure 5 As shown, the method includes:

[0096] Step 201: Collect sound signals using a sound sensor;

[0097] Step 202: Use a high-pass filter, a low-pass filter and a programmable amplifier to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal to obtain a pre-processed sound signal;

[0098] Step 203: using an audio interface chip to convert the pre-processed sound signal into sound signal digital signal modal data;

[0099] Step 204: using an audio conversion chip to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data;

[0100] Step 205: Use a wireless communication module to transmit the sound signal simulation modal data to a server; and

[0101] Step 206: Use the server to process the sound signal simulation modal data to determine whether there is a leak.

[0102] Reference below Figure 6 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present disclosure. The terminal device in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0103] like Figure 6 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0104] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 6 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0105] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0106] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0107] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0108] The computer-readable medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0109] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0110] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0111] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a module does not limit the module itself in some cases. For example, a test parameter acquisition module may also be described as a "module for acquiring device test parameters corresponding to a target device".

[0112] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0113] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0114] Figure 7 FIG. 1 is an application scenario diagram of an automatic water leakage sound recognition system based on artificial intelligence according to an exemplary embodiment. Figure 7 As shown, in this application scenario, first, a sound signal is collected (for example, Figure 7 Then, the acquired sound signal is input to a server (for example, Figure 7 In the S) shown in , the server can process the sound signal based on an artificial intelligence-based automatic recognition algorithm for water leakage sounds to determine whether there is a leak.

[0115] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0116] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0117] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. An automatic water leakage sound recognition system based on artificial intelligence, characterized in that: include: Sound sensor, high-pass filter, low-pass filter, programmable amplifier, audio interface chip, audio conversion chip, wireless communication module and server, among which, The sound sensor is used to collect sound signals; The high-pass filter, the low-pass filter and the programmable amplifier are used to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal respectively to obtain a pre-processed sound signal; The audio interface chip is used to convert the pre-processed sound signal into sound signal digital signal modal data; The audio conversion chip is used to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data; The wireless communication module is used to transmit the sound signal simulation modal data to the server; and The server is used to process the sound signal simulation modal data to determine whether there is a leak; Wherein, the server comprises: A sound simulation signal wavelet transformation module is used to process the sound signal simulation modal data using wavelet transformation to obtain a two-dimensional time-frequency diagram of the sound signal; A sound signal two-dimensional time-frequency graph grayscale processing module, used for grayscale processing the sound signal two-dimensional time-frequency graph to obtain a sound signal time-frequency grayscale graph; A sound signal time-frequency super-pixel block semantic association coding module is used to perform pixel block association semantic coding processing based on super-pixel segmentation on the sound signal time-frequency grayscale image to obtain a sequence of context sound signal time-frequency grayscale super-pixel block semantic coding feature vectors; The module for highlighting the important contents of the sound signal is used to pass the sequence of semantically coded feature vectors of the time-frequency grayscale superpixel block of the contextual sound signal through an important content attention integration network to obtain a semantically highlighted coding representation vector of the important contents of the sound signal; A leak detection module, used to determine whether there is a leak based on the semantic highlighting encoding representation vector of the important content of the sound signal; The sound signal time-frequency super-pixel block semantic association coding module includes: A sound signal time-frequency grayscale superpixel segmentation unit, used for performing image segmentation on the sound signal time-frequency grayscale image using a superpixel segmentation module to obtain a set of sound signal time-frequency grayscale superpixel blocks; A sound signal time-frequency grayscale super-pixel block context semantic coding unit, used for passing the set of the sound signal time-frequency grayscale super-pixel blocks through an inter-pixel block association semantic encoder including a super-pixel block embedding encoder to obtain a sequence of context sound signal time-frequency grayscale super-pixel block semantic coding feature vectors; The module for highlighting the important semantic contents of the sound signal includes: An autocorrelation semantic metric coefficient calculation unit is used to calculate the inter-superpixel semantic autocorrelation semantic metric coefficients of each context sound signal time-frequency grayscale superpixel block semantic coding feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic coding feature vectors to obtain a sequence of inter-superpixel semantic autocorrelation semantic metric coefficients; A normalization processing unit, used for normalizing the sequence of semantic autocorrelation semantic measurement coefficients between the superpixel blocks to obtain a sequence of semantic autocorrelation attention weight coefficients between the superpixel blocks; The vector-by-vector weighted sum calculation unit is used to calculate the vector-by-vector weighted sum of the sequence of semantic coding feature vectors of the time-frequency grayscale super-pixel blocks of the context sound signal using the sequence of autocorrelation attention weight coefficients between the semantics of the super-pixel blocks as weights to obtain the semantically highlighted coding representation vector of the important content of the sound signal.

2. The water leakage sound automatic recognition system based on artificial intelligence according to claim 1 is characterized in that: The autocorrelation semantic metric coefficient calculation unit includes: A matrix product calculation subunit, used to calculate the matrix products between each weight coefficient matrix and the corresponding semantic coding feature vector of each context sound signal time-frequency grayscale super pixel block to obtain a sequence of semantic weighted feature vectors of context sound signal time-frequency grayscale super pixel blocks; A vector addition subunit, used for adding a corresponding bias weight vector to each context sound signal time-frequency grayscale super pixel block semantic weighted feature vector in the sequence of context sound signal time-frequency grayscale super pixel block semantic weighted feature vectors to obtain a sequence of context sound signal time-frequency grayscale super pixel block semantic bias feature vectors; An activation subunit, used to activate the sequence of context sound signal time-frequency grayscale super pixel block semantic bias feature vectors to obtain a sequence of context sound signal time-frequency grayscale super pixel block semantic activation feature vectors; The vector product subunit is used to calculate the product between each context sound signal time-frequency grayscale superpixel block semantic activation feature vector in the sequence of context sound signal time-frequency grayscale superpixel block semantic activation feature vectors and the transposed vector of the preset reference feature vector to obtain a sequence of autocorrelation semantic measurement coefficients between the superpixel blocks.

3. The water leakage sound automatic recognition system based on artificial intelligence according to claim 2 is characterized in that: The activation subunit is used to: use the Selu activation function to activate the sequence of semantic bias feature vectors of the context sound signal time-frequency grayscale super pixel block to obtain the sequence of semantic activation feature vectors of the context sound signal time-frequency grayscale super pixel block.

4. The water leakage sound automatic recognition system based on artificial intelligence according to claim 3 is characterized in that: The leak detection module is used to: pass the semantically highlighted encoding representation vector of the important content of the sound signal through a leak detector based on a classifier to obtain a detection result, and the detection result is used to indicate whether there is a leak.

5. The water leakage sound automatic recognition system based on artificial intelligence according to claim 4 is characterized in that: It also includes a training module for training the pixel block association semantic encoder including the superpixel block embedding encoder, the important content attention integration network and the classifier-based missing point detector.

6. The water leakage sound automatic recognition system based on artificial intelligence according to claim 5 is characterized in that: The training module comprises: A training sound sensor unit is used to collect the training sound signal and the real value of whether there is a leak point; A training preprocessing unit, wherein the high-pass filter, the low-pass filter and the programmable amplifier are respectively used to perform high-pass filtering, low-pass filtering and amplification processing on the training sound signal to obtain a training preprocessed sound signal; A training audio interface chip unit, used for converting the training pre-processed sound signal into training sound signal digital signal modal data; A training audio conversion chip unit, used for converting the training sound signal digital signal modal data into a training analog signal to obtain training sound signal analog modal data; A training sound analog signal wavelet transform unit, used for processing the training sound signal analog modal data using wavelet transform to obtain a two-dimensional time-frequency diagram of the training sound signal; A training sound signal two-dimensional time-frequency graph grayscale processing unit, used for grayscale processing the training sound signal two-dimensional time-frequency graph to obtain a training sound signal time-frequency grayscale graph; A training sound signal time-frequency superpixel block semantic association coding unit is used to perform pixel block association semantic coding processing based on superpixel segmentation on the training sound signal time-frequency grayscale image to obtain a sequence of training context sound signal time-frequency grayscale superpixel block semantic coding feature vectors; A training sound signal important content semantic highlighting unit is used to pass the sequence of the training context sound signal time-frequency grayscale superpixel block semantic coding feature vectors through the important content attention integration network to obtain the training sound signal important content semantic highlighting coding representation vector; A training classification unit, used for passing the semantically salient encoding representation vector of the important content of the training sound signal through the classifier-based leak detector to obtain a detection result; A training classification loss function value calculation unit is used to calculate the cross entropy loss function value between the detection result and the true value to obtain a classification loss function value; A training unit is used to train the pixel block association semantic encoder including the super pixel block embedding encoder, the important content attention integration network and the classifier-based missing point detector based on the classification loss function value.

7. A method for automatically identifying water leakage sounds based on artificial intelligence, using the automatic water leakage sound identification system based on artificial intelligence according to claim 1, characterized in that: include: Using sound sensors to collect sound signals; Use a high-pass filter, a low-pass filter and a programmable amplifier to perform high-pass filtering, low-pass filtering and amplification processing on the sound signal respectively to obtain a pre-processed sound signal; Using an audio interface chip to convert the pre-processed sound signal into sound signal digital signal modal data; Using an audio conversion chip to convert the sound signal digital signal modal data into an analog signal to obtain sound signal analog modal data; Using a wireless communication module to transmit the sound signal simulation modal data to a server; as well as The server is used to process the sound signal simulation modal data to determine whether there is a leak.

Images