Method for detecting a water supply system based on audio data and device therefor
By extracting multi-dimensional features and processing audio data from the water supply system using an SVM model, the system can identify equipment status and generate alarm information. This solves the problem of difficult and early detection of equipment failures in the water supply system, extends equipment life, and maintains efficient system operation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- THREE GORGES JINSHAJIANG CHUANYUN HYDROPOWER DEV CO LTD
- Filing Date
- 2023-06-15
- Publication Date
- 2026-07-14
Smart Images
Figure CN116741199B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a detection method and apparatus for a water supply system based on audio data. Background Technology
[0002] The water supply system of a hydroelectric power station mainly provides cooling water and lubrication water for key equipment of the turbine generator set. Sometimes, necessary pressure control equipment also requires the support of the technical water supply system.
[0003] In related technologies, various equipment and components in water supply systems may malfunction or be damaged, such as water pumps, pipes, and valves. Therefore, how to detect water supply system faults early, extend equipment lifespan, and maintain efficient system operation has become an important research direction. Summary of the Invention
[0004] This application aims to at least partially address one of the technical problems in related technologies. A first aspect of this application proposes a detection method for a water supply system based on audio data, comprising:
[0005] Obtain candidate audio data corresponding to the water supply system;
[0006] Multi-dimensional feature extraction is performed on candidate audio data to obtain target feature data, which includes feature data in multiple dimensions.
[0007] The target feature data is input into the target support vector machine (SVM) model, and the target SVM model performs binary classification on the target feature data to determine the working status of the water supply equipment.
[0008] An alarm message is generated in response to a faulty operating state.
[0009] A second aspect of this application provides a detection device for a water supply system based on audio data, comprising:
[0010] It includes a sound acquisition unit, a transmission unit, and a processing unit, wherein the transmission unit is connected to both the sound acquisition unit and the processing unit.
[0011] The sound acquisition unit is used to acquire candidate audio data corresponding to the water supply system;
[0012] The sending module includes a connected microprocessor chip and a network relay module. The microprocessor chip is used to package the candidate audio data using Transmission Control Protocol / Internet Protocol, and the network relay module is used to send the packaged candidate audio data to the processing unit.
[0013] The processing unit is used to extract features from candidate audio data in multiple dimensions, input target feature data into target support vector machine (SVM) model, and perform binary classification on target feature data by target SVM model to determine the working status of water supply equipment. In response to the working status being a fault state, alarm information is generated.
[0014] A third aspect of this application provides a detection device for a water supply system based on audio data, comprising:
[0015] The first acquisition module is used to acquire candidate audio data corresponding to the water supply system;
[0016] The second acquisition module is used to extract features from candidate audio data in multiple dimensions to obtain target feature data, which includes feature data in multiple dimensions.
[0017] The determination module is used to input the target feature data into the target support vector machine (SVM) model, and the target SVM model performs binary classification on the target feature data to determine the working status of the water supply equipment.
[0018] The alarm module is used to generate alarm information in response to a faulty working state.
[0019] A fourth aspect of this application provides an electronic device comprising:
[0020] At least one processor; and
[0021] A memory that is communicatively connected to at least one processor; wherein,
[0022] The memory stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the detection method for a water supply system based on audio data provided in the first aspect embodiment of this application.
[0023] A fifth aspect of this application provides a computer-readable storage medium having computer instructions stored thereon, wherein the computer instructions are used to cause a computer to perform a detection method for a water supply system based on audio data provided in a first aspect of this application.
[0024] A sixth aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the detection method for a water supply system based on audio data provided in the first aspect of this application.
[0025] This application can detect water supply system faults early, extend equipment life, and maintain efficient system operation. Attached Figure Description
[0026] Figure 1 This is a flowchart of a detection method for a water supply system based on audio data according to an embodiment of this application;
[0027] Figure 2 This is a flowchart of a detection method for a water supply system based on audio data according to an embodiment of this application;
[0028] Figure 3 This is a schematic diagram of a detection method for a water supply system based on audio data according to an embodiment of this application;
[0029] Figure 4 This is a structural diagram of a detection device for a water supply system based on audio data according to an embodiment of the present disclosure;
[0030] Figure 5 This is a structural diagram of a detection device for a water supply system based on audio data according to an embodiment of the present disclosure;
[0031] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation
[0032] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.
[0033] The following describes, with reference to the accompanying drawings, a method and apparatus for detecting audio data in a water supply system according to an embodiment of this application.
[0034] Figure 1 This is a flowchart of a detection method for a water supply system based on audio data according to an embodiment of this application, as follows: Figure 1 As shown, the method includes the following steps:
[0035] S101, Obtain candidate audio data corresponding to the water supply system.
[0036] In some implementations, the water supply system includes a water pump that can collect candidate audio data based on a microphone deployed on the water pump.
[0037] In some implementations, multiple microphones can be deployed within the water pump and pump room space to simultaneously collect multiple candidate audio data at the same time.
[0038] In some implementations, impedance transformation is performed on candidate audio data based on signal processing circuits. Impedance transformation can be used to convert mismatched impedances in the circuit into matched impedances, which is very important in signal transmission and power transfer because matched impedances can transmit signals to the maximum extent and reduce signal reflection and loss.
[0039] Audio signals may be too weak or mixed with noise. Audio data often contains various noises that are difficult to distinguish. Therefore, in some implementations, candidate audio data can be amplified and filtered based on signal processing circuits. Amplification can enhance the amplitude of the audio signal, making it clearer and easier to distinguish. Filtering can remove or suppress noise frequency components, thereby reducing noise interference on the audio signal and improving audio quality and clarity.
[0040] The candidate audio data undergoes analog-to-digital conversion and digital signal processing. This process converts analog audio signals into digital form and adjusts and optimizes the digital candidate audio data (e.g., improving audio quality, enhancing specific frequency ranges, reducing noise, channel balancing, echo cancellation, etc.). This makes the candidate audio data suitable for processing and analysis by digital systems.
[0041] S102, perform multi-dimensional feature extraction on the candidate audio data to obtain target feature data, which includes feature data in multiple dimensions.
[0042] In some implementations, the rectified mean of the candidate audio data can be extracted as a feature of one dimension in the target feature data. The rectified mean is the average value after rectifying the candidate audio data, which can reflect the overall energy level of the audio data. A higher rectified mean usually indicates that the audio data has higher energy.
[0043] In some implementations, the mean square value of the candidate audio data can be extracted as a feature of one dimension in the target feature data; the mean square value is the square root of the average of the sum of squares of the candidate audio data, which can measure the energy of the audio data, and a higher mean square value indicates that the audio data has higher energy.
[0044] In some implementations, the variance of the candidate audio data can be extracted as a feature of one dimension in the target feature data; variance is the average of the squares of the differences between the audio data and its mean, which can describe the dispersion of the audio data, and a larger variance indicates that the audio data is more volatile.
[0045] In some implementations, the skewness of the candidate audio data can be extracted as a feature of one dimension in the target feature data; skewness measures the asymmetry of the distribution of the candidate audio data. Positively skewed audio data has a longer tail on the right, while negatively skewed audio data has a longer tail on the left.
[0046] In some implementations, the kurtosis of the candidate audio data can be extracted as a feature of one dimension in the target feature data; kurtosis measures the sharpness of the audio data distribution. Higher kurtosis indicates that the audio data has sharper peaks, while lower kurtosis indicates that the audio data distribution is relatively flat.
[0047] In some implementations, waveform factors of candidate audio data can be extracted as a feature of one dimension in the target feature data. The waveform factor is the ratio of the mean square value to the rectified mean of the audio data, which can reflect the waveform shape of the audio data. A larger waveform factor indicates that the peak part of the audio data waveform is relatively high.
[0048] In some implementations, the peak factor of the candidate audio data can be extracted as a feature of one dimension in the target feature data; the peak factor is the ratio of the maximum value to the mean square value of the audio data, which can reflect the peak amplitude of the audio data. A larger peak factor indicates that the audio data contains a stronger peak component.
[0049] In some implementations, a margin factor can be extracted from the candidate audio data as a feature of one dimension in the target feature data. The margin factor is the ratio of the peak value to the rectified mean of the audio data, which can describe the peak-to-valley ratio of the audio data. A larger margin factor indicates that the peak-to-valley amplitude of the audio data is larger.
[0050] These statistical features can provide different characteristics of candidate audio data, including energy, dispersion, symmetry and sharpness of distribution, waveform shape, and amplitude of peaks and troughs.
[0051] In some implementations, multi-order Mel-frequency cepstral coefficients (MFCCs) of the candidate audio data can be extracted as features in multiple dimensions of the target feature data. For example, the 12th-order MFCCs of the candidate audio data can be extracted as 12-dimensional features of the target feature data. MFCCs primarily smooth the spectrum and eliminate the effects of harmonics, highlighting the formants of the candidate audio data.
[0052] In some implementations, multi-order Mel-Cepstral Difference Coefficients (MFCCs) of candidate audio data can be extracted as features in multiple dimensions of the target feature data. For example, the 12th-order MFCCs of candidate audio data can be extracted as 12-dimensional features of the target feature data. MFCCs can represent the frequency information, human auditory perception characteristics, and spectral envelope features of audio data, allowing for the extraction of effective audio features for sound analysis and speech recognition.
[0053] Optionally, in this embodiment of the application, the target feature data includes a total of 32 dimensions of features, including 12-dimensional Mel-Cepstral coefficients, 12-dimensional Mel-Cepstral difference coefficients, and 8-dimensional features such as rectified mean, mean square value, variance, skewness, kurtosis, waveform factor, peak factor, and margin factor.
[0054] S103, input the target feature data into the target support vector machine (SVM) model, and the target SVM model performs binary classification on the target feature data to determine the working status of the water supply equipment.
[0055] The target SVM model can separate target feature data of different categories using the optimal hyperplane. In this embodiment, the operating state of the water supply equipment can include a normal state and a fault state, which can be labeled as different categories. Furthermore, the optimal hyperplane is defined as the decision boundary that maximizes the interval between the two categories of feature data. This decision boundary is composed of support vectors, which are the sample points closest to the decision boundary. In other words, the target feature data is binary-classified based on the target SVM model to determine the category of the target feature data, and then the operating state of the water supply equipment is determined based on this category.
[0056] S104, in response to the fault state of the working state, generates alarm information.
[0057] In some implementations, the water supply equipment operates in a normal state, while in others it operates in a fault state. In these cases, in order to extend the lifespan of the equipment and maintain the efficient operation of the system, it is necessary to generate alarm information, such as light or sound alerts.
[0058] In this embodiment, multi-dimensional feature extraction is performed on candidate audio data to obtain target feature data, which includes feature data in multiple dimensions. The target feature data is then input into a target support vector machine (SVM) model, which performs binary classification on the target feature data to determine the operating status of the water supply equipment. In response to a faulty operating status, an alarm message is generated. This application can detect faults in the water supply system early, extend the lifespan of the equipment, and maintain the efficient operation of the system.
[0059] Figure 2 This is a flowchart of a detection method for a water supply system based on audio data according to an embodiment of this application, as follows: Figure 2 As shown, the training process of the target SVM model includes:
[0060] S201, acquire sample audio data of the water supply equipment under different states, wherein the reference working state corresponding to the sample audio data is a fault state or a normal state.
[0061] In some implementations, sample audio data of the water supply station equipment during normal operation is obtained as a positive sample for the reference working state as normal.
[0062] When a water supply system malfunctions, it may be due to issues with the water pump and its motor, such as low current or speed, abnormal outlet pressure, increased vibration, or insufficient flow. The causes may include pump or motor imbalance, misalignment, contact failure, air intake, foreign object intake, mechanical deformation, wear, cavitation, and pressure pulsation. In some implementations, the malfunction may be due to air pressure or volume not meeting operating requirements; increased resistance in the water supply pipeline; pipeline leakage; or contact failure. Optionally, in this embodiment, sample audio data of the water supply station equipment malfunction is obtained as a negative sample of the fault state, with the reference operating state as the fault state.
[0063] S202, for any sample audio data, perform signal processing on the sample audio data in N dimensions to obtain the sample feature data corresponding to the sample audio data.
[0064] For a description of step S202, please refer to the relevant content in the above embodiments, which will not be repeated here.
[0065] S203, the candidate SVM model performs binary classification on the sample feature data to determine the prediction status corresponding to the sample feature data.
[0066] In this embodiment, a Lagrangian function is constructed to solve for the optimal hyperplane, and the normal state and the fault state are respectively labeled as the categories on both sides of the optimal hyperplane. The mapping points of the sample feature data in high-dimensional space are obtained, and the target category corresponding to the sample feature data is determined from the categories on both sides based on the distance from the mapping points to the optimal hyperplane. The target category is determined as the predicted operating state of the water supply equipment.
[0067] It should be noted that the SVM model can transform the binary classification problem into solving for the hyperplane in the sample space. The optimal hyperplane is the one whose minimum distance to each sample point (sample feature data) is maximized.
[0068] In solving linearly separable problems, a hyperplane can be defined:
[0069] w T x+b=0
[0070] This equation represents the defined hyperplane, where w is the weight that determines the direction of the hyperplane, and x represents the sample. Expanding this equation, we get w1x1 + w2x2 + ... + w n x n +b = 0; w n It is the weight corresponding to the nth dimension feature in the sample feature data; x n represents the feature of the nth dimension in the sample feature data, where n is a positive integer greater than 2; b is the bias value, which determines the distance between the hyperplane and the origin; and T represents the transpose.
[0071] The feature data of the two classes of samples are completely separated. After normalizing the hyperplane, we can obtain:
[0072] y i (w t x+b)≥1
[0073] This formula is the result of normalizing the hyperplane, where y i This represents the target value corresponding to the feature data of the i-th sample.
[0074] Using the formula for the distance from a point in high-dimensional space to a hyperplane, the distance between the closest sample points at the separation interface between the two classes is obtained as follows:
[0075] In this formula, r represents the distance between the nearest sample points (support vectors) of the two separating hyperplanes; where g(x) = y(w T x+b);
[0076] like Figure 3 As shown, for example, in this embodiment of the application, the two separating hyperplanes represent two types of target feature data (“o” and “×” represent the two types of target feature data respectively), corresponding to the normal state and fault state of the water supply system. Optionally, w can be... T x+b=1 and w T x+b=-1 serves as the two separating hyperplanes, w T x+b=0 is the optimal hyperplane.
[0077] Therefore, the optimal hyperplane should satisfy:
[0078] This equation represents the constraint equation that the optimal hyperplane parameters must satisfy, requiring that... Maximizing is equivalent to minimizing ||w|| 2That is Make the expression satisfy y i (w t x+b)≥1.
[0079] By combining the two equations to construct the Lagrange function, the optimal solution can be obtained. Where a i Using the Lagrange operator, the optimal weight vector w can be derived. * and the optimal bias value b * Therefore, the optimal hyperplane is: w *T x+b * =0.
[0080] This formula represents the optimal hyperplane obtained, w * and b * These are the optimal weight vector and the optimal bias value, respectively.
[0081] When solving linearly inseparable problems, we can construct a nonlinear mapping z = φ(x). (This nonlinear mapping is used to solve linearly inseparable problems, making the mapping points in the high-dimensional space linearly separable.) This allows us to construct a high-dimensional hyperplane w. T φ(x) + b = 0 (is the constructed high-dimensional hyperplane). Then find a suitable kernel function to make it linearly separable in the high-dimensional space. If a suitable mapping relationship cannot be constructed, a relaxation term and a penalty factor can be introduced to minimize the classification error of the hyperplane.
[0082] like Figure 3 As shown in the embodiments of this application, the "o" category can be used to correspond to the normal operating state of the water supply system, at which time w *T x+b * If the value is >0, the water supply system's operating status can be categorized as faulty using the "×" category. In this case, w *T x+b * <0.
[0083] S204. Based on the reference working state and prediction working state of any sample audio data, the model parameters of the candidate SVM model are adjusted in reverse to obtain the target SVM model.
[0084] In this embodiment, for any sample audio data, the reference working state and the prediction working state are adjusted in reverse based on the loss function until the preset number of iterations is reached or the value of the loss function converges to a preset threshold range, thus determining that the model has completed training and obtaining the target SVM model.
[0085] This application presents a training process for a target SVM model of a water supply system. Sound sensors are deployed in the water supply pumping station, and candidate SVM models are established by acquiring sample audio data from these sensors. The training process for the candidate SVM model includes performing N-dimensional signal processing on any sample audio data to obtain corresponding sample feature data. The candidate SVM model then performs binary classification on each of these feature data to determine the predicted operating state. Based on the reference and predicted operating states of any sample audio data, the model parameters of the candidate SVM model are adjusted inversely to obtain the target SVM model. The N-dimensional signal processing utilizes time-domain features, Mel-frequency cepstral coefficients, and first-order difference coefficients. In sample classification, the SVM classification method achieves high accuracy. During unit operation, this model can identify fault states.
[0086] Figure 4 This is a structural diagram of a detection device for a water supply system based on audio data according to an embodiment of the present disclosure, as shown below. Figure 4 As shown, the detection device 400 for a water supply system based on audio data includes: a sound acquisition unit 410, a transmission unit 420, and a processing unit 430. The transmission unit 420 is connected to both the sound acquisition unit 410 and the processing unit 430.
[0087] The sound acquisition unit 410 is used to acquire candidate audio data corresponding to the water supply system.
[0088] The sending module 420 includes a connected microprocessor chip 421 and a network relay module 422. The microprocessor chip 421 is used to package the candidate audio data using Transmission Control Protocol / Internet Protocol (TCP / IP), and the network relay module 422 is used to send the packaged candidate audio data to the processing unit. In other words, the candidate audio data is packaged using TCP / IP by the microprocessor ARM chip, and the network relay module forwards the packaged data.
[0089] The processing unit 430 can be a data server, used to extract features from candidate audio data in multiple dimensions, input target feature data into target support vector machine (SVM) model, and perform binary classification on target feature data by target SVM model to determine the working status of water supply equipment. In response to the working status being a fault state, alarm information is generated.
[0090] In some embodiments, the sound acquisition unit 410 includes a connected microphone 411 and a signal processing module 412. The microphone 411 is deployed in the water supply pump or pump room of the water supply system for acquiring candidate audio data. The signal processing module 412 is used to perform impedance transformation, amplification and filtering on the candidate audio data, as well as analog-to-digital conversion and digital signal processing.
[0091] Figure 5 This is a structural diagram of a detection device for a water supply system based on audio data according to an embodiment of the present disclosure, as shown below. Figure 5 As shown, the detection device 600 for a water supply system based on audio data includes:
[0092] The first acquisition module 510 is used to acquire candidate audio data corresponding to the water supply system;
[0093] The second acquisition module 520 is used to extract features from candidate audio data in multiple dimensions and acquire target feature data, which includes feature data in multiple dimensions.
[0094] The determination module 530 is used to input the target feature data into the target support vector machine (SVM) model, and the target SVM model performs binary classification on the target feature data to determine the working status of the water supply equipment.
[0095] The alarm module 540 is used to generate alarm information in response to a fault state in the working state.
[0096] In some implementations, the second acquisition module 520 is further configured to:
[0097] Extract the rectified mean, mean square, variance, skewness, kurtosis, waveform factor, peak factor, and margin factor from the candidate audio data as features of multiple dimensions in the target feature data;
[0098] Extract the multi-order Mel-frequency cepstral coefficients from the candidate audio data as features of multiple dimensions in the target feature data;
[0099] Multi-order Mel-Cepstral Difference Coefficients of Candidate Audio Data are extracted as features of multiple dimensions in the target feature data.
[0100] In some implementations, the training process of the target SVM model includes:
[0101] Acquire sample audio data of water supply equipment under different states, where the reference working state corresponding to the sample audio data is either fault state or normal state;
[0102] For any sample audio data, perform signal processing in N dimensions on the sample audio data to obtain the sample feature data corresponding to the sample audio data;
[0103] The candidate SVM model performs binary classification on the sample feature data to determine the prediction status corresponding to the sample feature data.
[0104] The model parameters of the candidate SVM model are adjusted in reverse based on the reference working state and the prediction working state of any sample audio data to obtain the target SVM model.
[0105] In some implementations, the determining module 530 is further configured to:
[0106] Construct a Lagrangian function to solve for the optimal hyperplane, and label the normal state and the fault state as the categories on both sides of the optimal hyperplane, respectively;
[0107] Obtain the mapping points of the sample feature data in the high-dimensional space, and determine the target category corresponding to the sample feature data from the categories on both sides based on the distance from the mapping points to the optimal hyperplane;
[0108] The target category is determined as the predicted operating status of the water supply equipment.
[0109] In some implementations, the first acquisition module 510 is further configured to:
[0110] Impedance transformation, amplification, and filtering are performed on the candidate audio data; and / or
[0111] The candidate audio data undergoes analog-to-digital conversion and digital signal processing.
[0112] This application can detect water supply system faults early, extend equipment life, and maintain efficient system operation.
[0113] Based on the same concept, embodiments of this application also provide an electronic device.
[0114] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 6 As shown, the electronic device 600 includes a memory 601, a processor 602, and a computer program product stored in the memory 601 and capable of running on the processor 602. When the processor executes the computer program, it implements the aforementioned detection method for a water supply system based on audio data.
[0115] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0116] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0117] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0118] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0119] Based on the same concept, embodiments of this application also provide a computer-readable storage medium storing computer instructions thereon, wherein the computer instructions are used to cause a computer to execute the detection method of the water supply system based on audio data in the above embodiments.
[0120] Based on the same concept, this application also provides a computer program product, including a computer program that, when executed by a processor, provides the detection method for a water supply system based on audio data in the above embodiments.
[0121] It should be noted that any reference signs placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. This application can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.
[0122] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.
[0123] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0124] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of the invention. Therefore, if these modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.
Claims
1. A detection method for a water supply system based on audio data, characterized in that, include: Obtain candidate audio data corresponding to the water supply system; Multi-dimensional feature extraction is performed on the candidate audio data to obtain target feature data, which includes feature data in multiple dimensions. The target feature data is input into the target support vector machine (SVM) model, and the target SVM model performs binary classification on the target feature data to determine the working status of the water supply system. In response to the operating state being a fault state, an alarm message is generated; The feature extraction of the candidate audio data in multiple dimensions includes one or more of the following: The rectified mean, mean square, variance, skewness, kurtosis, waveform factor, peak factor, and margin factor of the candidate audio data are extracted as features of multiple dimensions in the target feature data. Extract the multi-order Mel-spectral coefficients of the candidate audio data as features of multiple dimensions in the target feature data; The multi-order Mel-frequency cepstral difference coefficients of the candidate audio data are extracted and used as features of multiple dimensions in the target feature data.
2. The method according to claim 1, characterized in that, The training process of the target SVM model includes: Acquire sample audio data of the water supply system under different states, wherein the reference working state corresponding to the sample audio data is a fault state or a normal state; For any sample audio data, perform signal processing on the sample audio data in N dimensions to obtain the sample feature data corresponding to the sample audio data; The candidate SVM model performs binary classification on the sample feature data to determine the prediction status corresponding to the sample feature data. The model parameters of the candidate SVM model are adjusted in reverse based on the reference working state and the prediction working state of any of the sample audio data to obtain the target SVM model.
3. The method according to claim 2, characterized in that, The step of performing binary classification on the sample feature data by the candidate SVM model to determine the prediction status corresponding to the sample feature data includes: Construct a Lagrangian function to solve for the optimal hyperplane, and label the normal state and the fault state as the categories on both sides of the optimal hyperplane, respectively; Obtain the mapping points of the sample feature data in a high-dimensional space, and determine the target category corresponding to the sample feature data from the categories on both sides based on the distance from the mapping points to the optimal hyperplane; The target category is determined as the predicted operating status of the water supply system.
4. The method according to claim 3, characterized in that, Before performing multi-dimensional signal processing on the candidate audio data, the method further includes: The candidate audio data is subjected to impedance transformation, amplification, and filtering; and / or The candidate audio data is subjected to analog-to-digital conversion and digital signal processing.
5. A detection device for a water supply system based on audio data, characterized in that, It includes a sound acquisition unit, a transmission unit, and a processing unit, wherein the transmission unit is connected to both the sound acquisition unit and the processing unit, and wherein: The sound acquisition unit is used to acquire candidate audio data corresponding to the water supply system; The sending unit includes a connected microprocessor chip and a network relay module. The microprocessor chip is used to package the candidate audio data using Transmission Control Protocol / Internet Protocol (TCP / IP), and the network relay module is used to send the packaged candidate audio data to the processing unit. The processing unit is used to extract features from the candidate audio data in multiple dimensions to obtain target feature data, wherein the target feature data includes feature data in multiple dimensions. The processing unit is further configured to input the target feature data into the target support vector machine (SVM) model, and the target SVM model performs binary classification processing on the target feature data to determine the working state of the water supply system. In response to the working state being a fault state, an alarm message is generated. The feature extraction of the candidate audio data in multiple dimensions includes one or more of the following: The rectified mean, mean square, variance, skewness, kurtosis, waveform factor, peak factor, and margin factor of the candidate audio data are extracted as features of multiple dimensions in the target feature data. Extract the multi-order Mel-spectral coefficients of the candidate audio data as features of multiple dimensions in the target feature data; The multi-order Mel-frequency cepstral difference coefficients of the candidate audio data are extracted and used as features of multiple dimensions in the target feature data.
6. The apparatus according to claim 5, characterized in that, The sound acquisition unit includes a connected microphone and a signal processing module, wherein: The microphones are deployed in the water supply pumps or pump rooms of the water supply system to collect candidate audio data; The signal processing module is used to perform impedance transformation, amplification, and filtering on the candidate audio data, as well as analog-to-digital conversion and digital signal processing.
7. A detection device for a water supply system based on audio data, characterized in that, include: The first acquisition module is used to acquire candidate audio data corresponding to the water supply system; The second acquisition module is used to extract features from the candidate audio data in multiple dimensions to obtain target feature data, wherein the target feature data includes feature data in multiple dimensions. The determination module is used to input the target feature data into the target support vector machine (SVM) model, and the target SVM model performs binary classification on the target feature data to determine the working status of the water supply system. The alarm module is used to generate alarm information in response to the operating state being a fault state; The feature extraction of the candidate audio data in multiple dimensions includes one or more of the following: The rectified mean, mean square, variance, skewness, kurtosis, waveform factor, peak factor, and margin factor of the candidate audio data are extracted as features of multiple dimensions in the target feature data. Extract the multi-order Mel-spectral coefficients of the candidate audio data as features of multiple dimensions in the target feature data; The multi-order Mel-frequency cepstral difference coefficients of the candidate audio data are extracted and used as features of multiple dimensions in the target feature data.
8. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-4.
9. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the steps of the method according to any one of claims 1-4.