CN112634945A - Intelligent water leakage sound identification method based on cloud platform - Google Patents
Intelligent water leakage sound identification method based on cloud platform Download PDFInfo
- Publication number
- CN112634945A CN112634945A CN202011472638.7A CN202011472638A CN112634945A CN 112634945 A CN112634945 A CN 112634945A CN 202011472638 A CN202011472638 A CN 202011472638A CN 112634945 A CN112634945 A CN 112634945A
- Authority
- CN
- China
- Prior art keywords
- water leakage
- model
- noise
- feature
- sound
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 59
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000010801 machine learning Methods 0.000 claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 12
- 230000003595 spectral effect Effects 0.000 claims description 12
- 238000013145 classification model Methods 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 8
- 238000002790 cross-validation Methods 0.000 claims description 6
- 230000002068 genetic effect Effects 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 description 17
- 230000005236 sound signal Effects 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- QVRVXSZKCXFBTE-UHFFFAOYSA-N n-[4-(6,7-dimethoxy-3,4-dihydro-1h-isoquinolin-2-yl)butyl]-2-(2-fluoroethoxy)-5-methylbenzamide Chemical compound C1C=2C=C(OC)C(OC)=CC=2CCN1CCCCNC(=O)C1=CC(C)=CC=C1OCCF QVRVXSZKCXFBTE-UHFFFAOYSA-N 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F17—STORING OR DISTRIBUTING GASES OR LIQUIDS
- F17D—PIPE-LINE SYSTEMS; PIPE-LINES
- F17D5/00—Protection or supervision of installations
- F17D5/02—Preventing, monitoring, or locating loss
- F17D5/06—Preventing, monitoring, or locating loss using electric or acoustic means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Acoustics & Sound (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Mathematical Physics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mechanical Engineering (AREA)
- Medical Informatics (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Examining Or Testing Airtightness (AREA)
Abstract
An intelligent water leakage sound identification method based on a cloud platform comprises the following steps: step 1: extracting acoustic feature information from the early warning platform; step 2: selecting sound characteristics useful for classification discrimination analysis as input of a machine learning model; and step 3: training the machine learning model to obtain an optimized recognition model, and storing the model; and 4, step 4: using the stored model, performing classification calculation on target noise audio according to a water leakage identification algorithm to judge whether the sound characteristics are water leakage sound, and storing the noise text into a cloud platform noise database and classifying the noise text into corresponding categories after the sound characteristics are confirmed to be noise; and 5: after the noise database is updated, the machine learning model is retrained again to optimize the model. According to the method, the machine learning model is learned based on the noise database, the model can be continuously optimized, a feedback type training mode is formed, and the accuracy of water leakage sound identification can be continuously improved along with the richness of the database.
Description
Technical Field
The invention relates to the field of water leakage sound identification, in particular to an intelligent water leakage sound identification method based on a cloud platform.
Background
Leakage is an important challenge to be faced by each urban water supply network, and leakage can be timely discovered through water leakage identification. The common detection means is mainly a manual inspection method, judgment is carried out according to a visual observation result, influence of human factors is large, and the problems of low efficiency and poor accuracy exist.
There are some water leakage identification devices or water leakage monitoring systems, and the detector automatically judges according to the received sound signal of the acquisition device without depending on the experience of personnel, so as to improve the accuracy and reliability of monitoring, such as the pipe water leakage detector of application number CN201320051823.8 and the pipe network water leakage monitoring system with the detector. However, the following problems are often present: the detector cannot be automatically updated and learned, and the accuracy and objectivity of identification still need to be improved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide an intelligent water leakage sound identification method based on a cloud platform, a machine learning model learns based on a noise database, the model is continuously optimized, a feedback type training mode is formed, and the accuracy of water leakage sound identification is continuously improved along with the richness of the database.
An intelligent water leakage sound identification method based on a cloud platform comprises the following steps:
step 1: extracting acoustic feature information from the early warning platform;
step 2: selecting sound characteristics useful for classification discrimination analysis as input of a machine learning model;
and step 3: training the machine learning model to obtain an optimized recognition model, and storing the model;
and 4, step 4: using the stored model, performing classification calculation on target noise audio according to a water leakage identification algorithm to judge whether the sound characteristics are water leakage sound, and storing the noise text into a cloud platform noise database and classifying the noise text into corresponding categories after the sound characteristics are confirmed to be noise;
and 5: after the noise database is updated, the machine learning model is retrained again to optimize the model.
The method comprises the steps of carrying out classification calculation on target noise audios through a trained and learned machine learning model, and outputting the probability of water leakage at the position corresponding to a noise signal; after the authenticity of the leak points is confirmed manually, storing the noise file into a water leakage noise database and classifying the noise file into a corresponding category; after the noise database is updated, the machine learning model is trained again to optimize the model, so that a feedback type machine learning identification mechanism is formed.
Preferably, in step 2, the acoustic feature selection is performed by a ReliefF algorithm, which includes the following steps:
step 2-1: performing dimension reduction processing on the acoustic feature information extracted in the step 1, and forming a plurality of feature subsets;
step 2-2: classifying the feature subsets by using a classifier to evaluate the feature subsets with the optimal classification accuracy;
step 2-3: and verifying the effectiveness of the optimal feature subset through a cross-validation mode, and determining the optimal feature subset as an input feature subset of the machine learning model under the condition that the verification result is effective.
Preferably, step 2-2, during classification, a bayesian classifier is used for classification, wherein the bayesian classifier is based on a probability density function for calculating each feature in a gaussian model and is established by classification accuracy rate-variance.
Preferably, in step 3, the machine learning model includes an SVM classification model, and the training of the SVM classification model specifically includes the following steps:
step 3-1: selecting a kernel function of the expression
；
Step 3-2: calculating a hyper-parameter related to the kernel function, wherein the hyper-parameter is as follows: penalty of C-objective functionThe factor(s) is (are),
-a width coefficient of the kernel function;
step 3-3: using the optimum C,
And selecting the characteristic subset, training the SVM classification model by adopting a cross validation method to obtain an identification model containing the optimal hyperplane, and storing the model for a water leakage identification algorithm.
Preferably, in step 3-2, the C and C with the highest accuracy are calculated by using a genetic algorithm
The value is obtained.
Preferably, step 3-0 is also provided before step 3-1: and carrying out normalization processing on the selected feature vectors.
Preferably, the water leakage identification algorithm includes the following steps:
step 4-1: extracting acoustic features in the target noise audio;
step 4-2: carrying out data normalization processing on the acoustic features;
step 4-3: calling a training model, carrying out recognition and classification by combining a decision function, and judging whether the sound characteristics are water leakage sound, wherein the decision function is as follows:
，
b is the normal form distance, is calculated and obtained when the model is trained, namely the offset,
is a support vector; and, after inputting the feature vector, if
If =1, it is judged as a water leakage sound, and if so, it is judged as a water leakage sound
And if the signal is not equal to or less than 1, judging the signal to be non-water leakage sound.
Preferably, in step 1, feature extraction is performed from three aspects, namely, a time domain, a frequency domain and a high-order domain.
Preferably, the time-domain class features include a mean, a variance, a short-time energy, an energy entropy, a peak coefficient, a shape parameter, a pulse factor, and a kurtosis factor of the signal;
the frequency domain characteristics comprise effective bandwidth, spectral peak, spectral entropy and spectral shape parameters of the signal;
the high-order domain features comprise approximate entropy, spectral moments and HHT singular values of the signals.
In conclusion, the invention has the following beneficial effects:
firstly, extracting features from different domains, characterizing leakage sound in a multi-dimensional way, and ensuring classification accuracy;
automatically acquiring an optimal feature subset by using a feature selection algorithm and automatically acquiring an optimal super parameter value by using a genetic algorithm, and converting an SVM (support vector machine) originally belonging to a supervised machine learning algorithm into a semi-supervised algorithm, so that the consumption of resources in engineering use is reduced;
according to the method, the machine learning model is learned based on the noise database, the model can be continuously optimized, a feedback type training mode is formed, and the accuracy of water leakage sound identification can be continuously improved along with the richness of the database.
Drawings
FIG. 1 is a flow chart of the method.
Detailed Description
The invention will be further explained by means of specific embodiments with reference to the drawings.
Example 1: as shown in fig. 1, an intelligent water leakage sound identification method based on a cloud platform comprises the following steps,
step 1: extracting acoustic feature information from the early warning platform, specifically extracting the acoustic feature information through a feature extraction algorithm, and preferably extracting the features from three aspects of a time domain, a frequency domain and a high-order domain, wherein the time domain type features comprise an average value, a variance, short-time energy, energy entropy, a peak coefficient, a shape parameter, a pulse factor and a kurtosis factor of a signal; the frequency domain characteristics comprise effective bandwidth, spectral peak, spectral entropy and spectral shape parameters of the signal; the high-order domain features comprise approximate entropy, spectral moments and HHT singular values of the signals. The reason for this extraction of acoustic features is because: when the pipeline leaks water, when high-pressure water in the pipeline is sprayed outwards from a leakage position, impact friction between the water and media such as the pipeline wall and soil outside the pipeline can be caused, sound signals with the frequency band ranging from tens of hertz to thousands of hertz are generated, sound vibration at the leakage position of the pipeline is caused, the sound signals can be picked up on the pipeline or the ground, and then the leakage detection can be realized by correspondingly processing and analyzing the sound signals, but the waveform of the leakage sound signals is extremely complex, the sound signals are the combination of various component signals, belong to time-varying non-stable random signals and have no obvious distinguishable characteristics visually; the leakage signal is wide in distribution frequency band (mainly concentrated on 200 Hz-2500 Hz) in spectrum view. Therefore, it is difficult to characterize the leaking acoustic signal with a single characteristic parameter. In the embodiment, features are extracted from different domains, leakage sound is represented in a multi-dimensional mode, and classification accuracy is guaranteed;
step 2: selecting sound characteristics useful for classification discrimination analysis as input of a machine learning model; this step exists because: in the steps, because redundant information may be contained in a large number of extracted characteristic values, the performance of the classifier is greatly influenced, the more the number of the characteristics is, dimensionality disaster is easily caused, and the generalization capability of the classifier is reduced, so that the acoustic characteristics are selectively selected as the input of the machine learning model, the redundant characteristics and irrelevant characteristics contained in the complex information are removed, the complexity of subsequent machine learning training is reduced, and the performance of the recognition model is improved;
and step 3: training the machine learning model to obtain an optimized recognition model, and storing the model;
and 4, step 4: using the stored model, performing classification calculation on target noise audio according to a water leakage identification algorithm to judge whether the sound characteristics are water leakage sound, and storing the noise text into a cloud platform noise database and classifying the noise text into corresponding categories after the sound characteristics are confirmed to be noise;
and 5: after the noise database is updated, the machine learning model is retrained again to optimize the model.
The method comprises the steps of carrying out classification calculation on target noise audios through a trained and learned machine learning model, and outputting the probability of water leakage at the position corresponding to the noise signal; after the authenticity of the leak points is confirmed manually, storing the noise file into a water leakage noise database and classifying the noise file into a corresponding category; after the noise database is updated, the machine learning model is trained again to optimize the model, so that a feedback type machine learning identification mechanism is formed.
In this embodiment, specifically, in step 2, the acoustic feature selection is performed through a ReliefF algorithm, where the ReliefF algorithm is a feature weight algorithm, different weights are given to features according to the correlations between the features and categories, and features whose weights are smaller than a certain threshold are removed; the running time of the Relieff algorithm increases linearly with the increase of the sampling times of the samples and the number of the original characteristics, so that the running efficiency is very high, and the algorithm comprises the following steps:
step 2-1: performing dimension reduction processing on the acoustic feature information extracted in the step 1, and forming a plurality of feature subsets; a plurality of acoustic features are present in the subset of features;
step 2-2: classifying the feature subsets by using a classifier to evaluate the feature subsets with the optimal classification accuracy;
step 2-3: and verifying the effectiveness of the optimal feature subset through a cross-validation mode, and determining the optimal feature subset as an input feature subset of the machine learning model under the condition that the verification result is effective.
The information features which are useful for classification discrimination analysis can be selected through a feature selection algorithm and then used as input of machine learning, namely, the most effective information features are selected from the original information features, an effective subset of the information features is obtained by compressing a high-dimensional feature space to a low-dimensional feature space, redundant features and irrelevant features contained in complex information are removed, complexity of subsequent machine learning training is reduced, and performance of a recognition model is improved.
Preferably, step 2-2, a Bayesian classifier is used for classification during classification, the Bayesian classifier is used for classification, the accuracy is high, the Bayesian classifier is based on a probability density function for calculating each feature in a Gaussian model and is established by classification accuracy-variance, the occurrence of overfitting can be reduced during classification, a CCR variance threshold is set for the classifier before classification, the variance value of the feature subset is calculated during classification, and when the variance value reaches the threshold, the feature subset with the optimal accuracy is found.
In this embodiment, specifically, in step 3, the machine learning model includes an SVM classification model, the SVM is a two-class classification model, a basic model of the SVM is defined as a linear classifier with a maximum interval in a feature space, and a main learning strategy is to maximize an inter-class interval and convert the inter-class interval into a solution of a convex quadratic programming problem; the SVM model is adopted in the machine learning of the embodiment because: the SVM has the greatest advantages that aiming at a nonlinear irreparable data set, the SVM can classify the nonlinear separable data set, and can also classify the nonlinear irreparable data set, and the influence on the accuracy of the classifier is only the support vector in the sample, so that other samples are directly filtered when the weight is calculated, the running time is greatly saved, and the farthest distance from the nearest sample point of the boundary to the hyperplane is finally obtained; because the classification characteristic quantity set is nonlinear, the nonlinear SVM is adopted in the invention, namely a Kernel function (Kernel function) is selected to map nonlinear training data to a high-dimensional space; the SVM firstly completes calculation in a low-dimensional space, then maps an input space to a high-dimensional feature space through a kernel function, and constructs an optimal separation hyperplane in the high-dimensional feature space, so that nonlinear data which are not well separated on the plane are separated. The main idea of the SVM classifier is as follows:
setting the defined hyperplane as:
3.2.1 
wherein
Is to use kernel function to characteristic quantity
Performing high-dimensional mapping;
in order to be a function of the relaxation variable,
is the weight; b is an offset; c is a penalty factor which is obtained by calculating the average value of the parameters,
is an actual target value;
secondly, by using an SMO algorithm, Lagrangian duality is applied, saddle points are obtained by solving the duality problem,
3.2.2 
are respectively paired
The partial derivatives are obtained:
3.2.3 
thirdly, converting the formula 3.2.1 into an optimization problem for solving the formula 3.2.4, namely solving
To pair
Is determined to be the maximum value of (c),
3.2.4 
fourthly, the equation in the formula 3.2.4 is obtained
Of (2) an optimal solution
According to the formula 3.2.3, the result is
Of (2) an optimal solution
，
3.2.5 
Calculating b (offset), selecting not 0
Substituting into the formula 3.2.6 to obtain the b value,
3.2.6
in this embodiment, the training of the SVM classification model specifically includes the following steps:
step 3-1: selecting a kernel function of the expression
；
Step 3-2: calculating a hyper-parameter related to the kernel function, wherein the hyper-parameter is as follows: c-a penalty factor of the objective function,
-a width coefficient of the kernel function;
step 3-3: using the optimum C,
And selecting the characteristic subset, training the SVM classification model by adopting a cross validation method to obtain the SVM classification modelAnd (4) an optimal hyperplane identification model is provided, and the model is stored for a water leakage identification algorithm.
Preferably, in step 3-2, C and C with the highest accuracy are calculated by using a genetic algorithm
The value of C and C with the maximum accuracy can be calculated more efficiently through a genetic algorithm
The value is obtained.
Preferably, step 3-0 is also provided before step 3-1: and carrying out normalization processing on the selected feature vectors. Normalization processing, namely feature set normalization processing, wherein the main function of the operation is to avoid that the attribute of a large numerical range overgoverns the attribute of a small numerical range; another advantage is that numerical complexity in the calculation process can be avoided, and each attribute is linearly scaled within the range of [0, 1] in the invention.
In this embodiment, specifically, the water leakage identification algorithm includes the following steps:
step 4-1: extracting acoustic features in the target noise audio;
step 4-2: carrying out data normalization processing on the acoustic features;
step 4-3: calling a training model, carrying out recognition and classification by combining a decision function, and judging whether the sound characteristics are water leakage sound, wherein the decision function is as follows:
，
b is the normal form distance, is calculated and obtained when the model is trained, namely the offset,
is a support vector; and, after inputting the feature vector, if
If =1, it is judged as a water leakage sound, and if so, it is judged as a water leakage sound
And if the signal is not equal to or less than 1, judging the signal to be non-water leakage sound.
The above-described embodiments are merely illustrative of the preferred embodiments of the present invention and do not limit the spirit and scope of the present invention. Various modifications and improvements of the technical solutions of the present invention may be made by those skilled in the art without departing from the design concept of the present invention, and the technical contents of the present invention are all described in the claims.
Claims (9)
1. An intelligent water leakage sound identification method based on a cloud platform is characterized by comprising the following steps:
step 1: extracting acoustic feature information from the early warning platform;
step 2: selecting sound characteristics useful for classification discrimination analysis as input of a machine learning model;
and step 3: training the machine learning model to obtain an optimized recognition model, and storing the model;
and 4, step 4: using the stored model, performing classification calculation on target noise audio according to a water leakage identification algorithm to judge whether the sound characteristics are water leakage sound, and storing the noise text into a cloud platform noise database and classifying the noise text into corresponding categories after the sound characteristics are confirmed to be noise;
and 5: after the noise database is updated, the machine learning model is retrained again to optimize the model.
2. The intelligent water leakage sound identification method based on the cloud platform according to claim 1, wherein in the step 2, sound feature selection is performed through a Relieff algorithm, and the algorithm comprises the following steps:
step 2-1: performing dimension reduction processing on the acoustic feature information extracted in the step 1, and forming a plurality of feature subsets;
step 2-2: classifying the feature subsets by using a classifier to evaluate the feature subsets with the optimal classification accuracy;
step 2-3: and verifying the effectiveness of the optimal feature subset through a cross-validation mode, and determining the optimal feature subset as an input feature subset of the machine learning model under the condition that the verification result is effective.
3. The intelligent water leakage sound identification method based on the cloud platform as claimed in claim 2, wherein in step 2-2, classification is performed by using a bayesian classifier during classification, wherein the bayesian classifier is based on a probability density function of each feature calculated in a gaussian model and is established by classification accuracy-variance.
4. The intelligent water leakage sound recognition method based on the cloud platform according to claim 1, wherein in the step 3, the machine learning model comprises an SVM classification model, and the training of the SVM classification model specifically comprises the following steps:
step 3-1: selecting a kernel function of the expression
；
Step 3-2: calculating a hyper-parameter related to the kernel function, wherein the hyper-parameter is as follows: c-a penalty factor of the objective function,
-a width coefficient of the kernel function;
step 3-3: using the optimum C,
And selecting the characteristic subset, training the SVM classification model by adopting a cross validation method to obtain an identification model containing the optimal hyperplane, and storing the model for a water leakage identification algorithm.
5. A cloud-based system according to claim 4The intelligent water leakage sound identification method of the platform is characterized in that in step 3-2, C and C with the highest accuracy are calculated by using a genetic algorithm
The value is obtained.
6. The intelligent water leakage sound identification method based on the cloud platform according to claim 4, wherein a step 3-0 is further provided before the step 3-1: and carrying out normalization processing on the selected feature vectors.
7. The intelligent water leakage sound identification method based on the cloud platform according to claim 1, wherein the water leakage identification algorithm comprises the following steps:
step 4-1: extracting acoustic features in the target noise audio;
step 4-2: carrying out data normalization processing on the acoustic features;
step 4-3: calling a training model, carrying out recognition and classification by combining a decision function, and judging whether the sound characteristics are water leakage sound, wherein the decision function is as follows:
8. The intelligent water leakage sound identification method based on the cloud platform according to claim 1, wherein in the step 1, feature extraction is performed from three aspects of time domain, frequency domain and high-order domain.
9. The intelligent water leakage sound identification method based on the cloud platform according to claim 8, wherein the time-domain class features include a mean value, a variance, a short-time energy, an energy entropy, a peak coefficient, a shape parameter, a pulse factor and a kurtosis factor of a signal;
the frequency domain characteristics comprise effective bandwidth, spectral peak, spectral entropy and spectral shape parameters of the signal;
the high-order domain features comprise approximate entropy, spectral moments and HHT singular values of the signals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011472638.7A CN112634945A (en) | 2020-12-15 | 2020-12-15 | Intelligent water leakage sound identification method based on cloud platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011472638.7A CN112634945A (en) | 2020-12-15 | 2020-12-15 | Intelligent water leakage sound identification method based on cloud platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112634945A true CN112634945A (en) | 2021-04-09 |
Family
ID=75313120
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011472638.7A Pending CN112634945A (en) | 2020-12-15 | 2020-12-15 | Intelligent water leakage sound identification method based on cloud platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112634945A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113505997A (en) * | 2021-07-13 | 2021-10-15 | 同济大学 | Building wall leakage water risk level assessment method based on machine learning |
| CN117235661A (en) * | 2023-08-30 | 2023-12-15 | 广州怡水水务科技有限公司 | AI-based direct drinking water quality monitoring method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109284777A (en) * | 2018-08-28 | 2019-01-29 | 内蒙古大学 | Recognition methods is leaked based on signal time-frequency characteristics and the water supply line of support vector machines |
| CN109555977A (en) * | 2018-11-23 | 2019-04-02 | 水联网技术服务中心（北京）有限公司 | The equipment and recognition methods of leak noise measuring |
| CN110044554A (en) * | 2019-04-16 | 2019-07-23 | 重庆大学 | A kind of online test method of the metal pressure container leakage based on acoustic emission signal |
| CN110185939A (en) * | 2019-05-16 | 2019-08-30 | 西北工业大学 | Gas pipe leakage recognition methods based on convolutional neural networks |
| CN111709299A (en) * | 2020-05-19 | 2020-09-25 | 哈尔滨工程大学 | Underwater sound target identification method based on weighting support vector machine |
-
2020
- 2020-12-15 CN CN202011472638.7A patent/CN112634945A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109284777A (en) * | 2018-08-28 | 2019-01-29 | 内蒙古大学 | Recognition methods is leaked based on signal time-frequency characteristics and the water supply line of support vector machines |
| CN109555977A (en) * | 2018-11-23 | 2019-04-02 | 水联网技术服务中心（北京）有限公司 | The equipment and recognition methods of leak noise measuring |
| CN110044554A (en) * | 2019-04-16 | 2019-07-23 | 重庆大学 | A kind of online test method of the metal pressure container leakage based on acoustic emission signal |
| CN110185939A (en) * | 2019-05-16 | 2019-08-30 | 西北工业大学 | Gas pipe leakage recognition methods based on convolutional neural networks |
| CN111709299A (en) * | 2020-05-19 | 2020-09-25 | 哈尔滨工程大学 | Underwater sound target identification method based on weighting support vector machine |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113505997A (en) * | 2021-07-13 | 2021-10-15 | 同济大学 | Building wall leakage water risk level assessment method based on machine learning |
| CN117235661A (en) * | 2023-08-30 | 2023-12-15 | 广州怡水水务科技有限公司 | AI-based direct drinking water quality monitoring method |
| CN117235661B (en) * | 2023-08-30 | 2024-04-12 | 广州怡水水务科技有限公司 | AI-based direct drinking water quality monitoring method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103868690B (en) | Rolling bearing state automatic early warning method based on extraction and selection of multiple characteristics | |
| EP3680639A1 (en) | Abnormal sound detecting device, abnormality model learning device, abnormality detecting device, abnormal sound detecting method, abnormal sound generating device, abnormal data generating device, abnormal sound generating method, and program | |
| Sharma et al. | Brain tumor segmentation using genetic algorithm and artificial neural network fuzzy inference system (ANFIS) | |
| CN112634945A (en) | Intelligent water leakage sound identification method based on cloud platform | |
| CN112560596B (en) | Radar interference category identification method and system | |
| CN109919241A (en) | Hyperspectral unknown class target detection method based on probability model and deep learning | |
| KR102387887B1 (en) | Apparatus for refining clean labeled data for artificial intelligence training | |
| Esfahanian et al. | On contour-based classification of dolphin whistles by type | |
| CN112738092A (en) | Log data enhancement method, classification detection method and system | |
| CN114444620A (en) | Indicator diagram fault diagnosis method based on generating type antagonistic neural network | |
| CN113884844A (en) | Transformer partial discharge type identification method and system | |
| CN112949954B (en) | Method for establishing financial fraud recognition model based on recognition learning | |
| CN115035913A (en) | Sound abnormity detection method | |
| Li et al. | Intelligent fault diagnosis of aeroengine sensors using improved pattern gradient spectrum entropy | |
| CN108508297A (en) | A kind of fault arc detection method based on change coefficient and SVM | |
| CN113343123B (en) | Training method and detection method for generating confrontation multiple relation graph network | |
| Li et al. | Learning stage-wise gans for whistle extraction in time-frequency spectrograms | |
| CN117056678B (en) | Machine pump equipment operation fault diagnosis method and device based on small sample | |
| CN112763155A (en) | Pressure container gas leakage detection method | |
| Fadchar et al. | Design and development of a neural network—based coconut maturity detector using sound signatures | |
| CN110701487B (en) | KPCA and Cas-SVDD-based multi-working-condition pipeline leakage detection method | |
| CN116772122A (en) | Natural gas pipeline leakage fault diagnosis method, system, equipment and medium | |
| CN116150687A (en) | Fluid pipeline leakage identification method based on multi-classification G-WLSTSVM model | |
| CN110555457A (en) | Engine lubricating oil wear signal characteristic processing system and method | |
| CN112378942B (en) | White spirit grade classification and identification method based on nuclear magnetic resonance fingerprint |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210409 |