CN111814872B - Power equipment environmental noise identification method based on time domain and frequency domain self-similarity - Google Patents
Power equipment environmental noise identification method based on time domain and frequency domain self-similarity Download PDFInfo
- Publication number
- CN111814872B CN111814872B CN202010646042.8A CN202010646042A CN111814872B CN 111814872 B CN111814872 B CN 111814872B CN 202010646042 A CN202010646042 A CN 202010646042A CN 111814872 B CN111814872 B CN 111814872B
- Authority
- CN
- China
- Prior art keywords
- similarity
- time domain
- frequency domain
- power equipment
- sample
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; 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
Abstract
A method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity comprises the steps of firstly collecting operation sound signals of the electric power equipment to be monitored; then, dividing the collected audio into minute-level recording samples, setting a proper frame length, performing framing processing on each sample, and extracting time domain and frequency domain characteristics of each frame; and performing similarity analysis on the features by using a cluster-based similarity analysis method, wherein samples which can only be clustered into a cluster are considered to have time domain and frequency domain self-similarity characteristics, and otherwise, the samples do not have similarity. When the recording sample has time domain and frequency domain self-similarity, the recording sample is reserved; otherwise, the recording sample is rejected. The method can effectively identify and eliminate the recording samples without time domain frequency domain self-similarity noise interference, screen out effective samples, and provide support for subsequent identification of the operating state of the power equipment based on the sound signals.
Description
Technical Field
The invention belongs to the field of on-line monitoring of the state of electric equipment based on sound signals, and particularly relates to a method for identifying environmental noise of electric equipment based on time domain and frequency domain self-similarity.
Background
At present, the maintenance and repair of the power equipment are transited from the traditional manual regular maintenance to the state maintenance based on the state online monitoring. After decades of development, various on-line monitoring technologies for temperature, oil gas and vibration at home and abroad are mature theoretically. In the state online monitoring technology applied to engineering at present, except for infrared temperature measurement and video monitoring, most of the state online monitoring technologies are contact state monitoring. The contact type on-line monitoring is carried out on high-voltage electrical equipment, the problem of high-voltage insulation is solved, the reliability problem caused by high temperature, frequent vibration and violent vibration during failure is also solved, the requirement on a contact type monitoring system is very high, and the popularization and the application of the contact type on-line monitoring technology are obviously restricted.
Since the new century, the non-contact measurement method is more and more outstanding in flexibility and stability, and is widely regarded. Some researchers have proposed using the sound generated when the device is running to judge whether the device has a fault, there are related cases of using sound signals to diagnose power devices such as transformers, reactors, capacitors, etc., and some methods of fault diagnosis based on sound are also proposed. In the power industry, teachers with abundant working experience can identify faults according to audible sound. In recent years, the development of artificial intelligence technology has brought forward a lot of voice-based innovative applications, and voice can be accurately identified on line. However, in the field of on-line monitoring of the state of the electrical equipment based on audible sound, because of the existence of strong environmental noise, the feasibility of exploring different technical routes is still in the development bottleneck period of overcoming the interference influence of background noise, the accuracy of the on-line monitoring of the state is influenced, and the application of the on-line monitoring of the state of the electrical equipment based on sound is limited.
Traditional noise elimination technique is mainly to the power equipment in the transformer substation, and environmental noise general interference is less or relatively fixed, mainly has two kinds of thinking: the first method adopts filtering to eliminate weak interference signals in strong signals, generally adopts wavelet analysis, combines soft and hard threshold filtering and other modes, and can obtain good noise elimination effect under the condition of weak noise. Audible background noise in the operating environment of the power equipment has strong uncertainty, filtering processing based on wavelet analysis is mainly used for eliminating weak noise or deterministic noise with mutually separated frequency bands of signals and noise, and is difficult to filter the strong background noise. The second is a sound source separation technique for recovering a radiation source signal from an observation signal, and there are many studies on separating a body vibration sound from an apparatus operation sound containing an environmental noise, but an actual environmental noise and a body vibration state may dynamically change, and this method has a problem of weakening effective data.
Different from the environment of a substation, the distribution equipment is generally placed in a ring main unit or a distribution room or even near a roadside telegraph pole and operates in an open severe environment, and the traditional method based on filtering or sound separation is difficult to adapt to complex and variable environmental noise. Because the interference of the environmental noise cannot be accurately identified, the application of the online monitoring technology of the state of the power distribution equipment based on the sound is greatly limited.
In summary, there is a need to develop a noise identification and labeling method that does not depend on filtering technology and has good adaptability to reduce the false alarm rate of the online monitoring system for the state of the power equipment based on sound.
Disclosure of Invention
The invention aims to provide a method for identifying environmental noise of electrical equipment based on time domain and frequency domain self-similarity aiming at the defects of the prior art. When the power equipment generally operates in an open environment and the operating state of the power equipment is monitored by using the operating sound signal of the power equipment, environmental noise interference is a main interference source, such as environmental noises including speaking sound, wind, rain, thunder, lightning sound, vehicle driving sound and the like, has strong uncertainty, and is difficult to predict a general rule. The sound monitoring sample is easy to be mistakenly judged to be in an abnormal state when mixed with environmental noise, and the invention researches how to identify and eliminate the environmental interference noise around the running power equipment.
The vibration-like and discharge-like failures caused by the aging of the electric power equipment is a long-term slow process in which abnormal sounds of vibration and discharge continue to exist. When the power equipment recording of operation is analyzed, as long as can effectively discern whether the recording data contains the ambient noise can get rid of this recording data, and only keep the recording data that does not contain the ambient noise and discern the fault anomaly.
On a time scale of a minute level, the fluctuation range of parameters such as voltage and current of the power equipment is limited, and the time domain and frequency domain characteristics of the emitted sound are small in change with time, so that the time domain and frequency domain characteristics have similarity. On the contrary, the general disturbing noise such as speaking sound, wind, rain, thunder, electricity, vehicle running sound, frog cry, etc. changes obviously with time in time domain and frequency domain, does not last for a long time, and generally has no self-similarity. After a sound sample at the level of minutes is divided into a plurality of segments by frames, similarity analysis can be carried out among the segments, when the sound does not have similarity in the time-frequency domain, the sample is judged to contain unstable noise and is judged to contain noise, and therefore the environmental interference noise can be identified.
In order to avoid the difficulty of parameter setting, the judgment of the self-similarity of the recording data can be carried out in a mode of not needing to preset cluster numbers. For recorded sound data, firstly, a minute-level sound recording can be cut into second-level sound recording segments in a windowing mode, then, the time domain characteristics and the frequency domain characteristics of each segment are extracted and input into a Clustering algorithm (Density-Based Spatial Clustering of Applications with Noise, abbreviated as DBSCAN) and the like which does not need to determine the number of clusters for Clustering analysis, wherein the Clustering algorithm is Based on Density-Based Clustering. The non-noisy device operating sound segments will be closely grouped into a cluster because of the similarity between the time domain and frequency domain features, while the noisy sound segments and the non-noisy segments have differences and will be identified as outliers or merged into a new cluster.
Therefore, the invention adopts the following technical scheme: a method for recognizing environmental noise of electrical equipment based on time domain and frequency domain self-similarity comprises the following steps:
step 1) sound sample collection: collecting sound signals in the operating environment of the power equipment to be detected;
when the sound signal is collected, the sound sensor is arranged near the target monitoring equipment for continuous collection or periodic collection;
step 2) dividing the collected sound signals into sound samples with time lengths in the order of minutes;
preferably, the minute-scale time period is 1-10 min. Unless otherwise specified herein, the minutes mentioned are all in the range of 1-10 min.
Step 3), sound sample clustering analysis: setting the frame length as T, respectively carrying out frame processing on each sample, extracting time domain and frequency domain characteristics of each frame to form a characteristic vector, and carrying out similarity analysis based on clustering; and if the clustering result of a sample only has one cluster, judging that the sample does not contain unstable noise, and if more than one cluster or outliers appear in the clustering analysis result of the sample, judging that the sample contains unstable noise.
When the operating state of the power equipment is monitored by using the sound signal, the ambient noise interference around the power equipment is a main interference source, such as animal sound such as speaking sound, vehicle driving sound, insect whistling and the like, wind, rain, thunder and electricity and the like, most of the noise is unstable and discontinuous, and has no continuous and stable characteristic in a time domain and a frequency domain, and the sound emitted by the power equipment in a minute level is stable and continuous no matter the power equipment is in a normal state or an abnormal state in the operation, so that the power equipment has the self-similar characteristic in the time domain and the frequency domain. Therefore, the difference of the self-similarity can be utilized, and clustering analysis is carried out in a clustering algorithm without determining the cluster number, such as DBSCAN, so as to identify whether the recording data contains environmental noise.
Preferably, the above-mentioned frame length T is taken from 1 to 10s, preferably 2 s.
The time domain features mentioned above are one or more combinations of mean, peak, root mean square value, variance, standard deviation, peak factor, kurtosis, form factor, pulse index or skewness; the frequency domain features comprise one or more combinations of the amplitude of the frequency in the frequency spectrum after fast Fourier transform, the amplitude statistical feature of the power spectral density, the shape statistical feature, Mel coefficients obtained after MFCC feature extraction, or wavelet coefficients obtained by wavelet or wavelet packet decomposition or energy features of different frequency bands.
The above mentioned similarity analysis method based on clustering is a conventional technique in the art, and any algorithm that does not need to determine the number of clusters is suitable for the present invention, such as DBSCAN clustering algorithm, affinity propagation clustering (AP clustering for short), mean shift clustering algorithm (MeanshiftClustering for short), and the like. When clustering is carried out, the power equipment without noise is generally in a stable operation state within a minute level, vibration is very stable, characteristic changes such as time domain, frequency domain and the like are small, similarity exists, the power equipment can be identified into a cluster based on the similarity clustering, and common environmental noise such as noise of vehicle driving, speaking, wind blowing, rain and the like does not have self-similarity and can be identified into a plurality of clusters.
The method utilizes a similarity analysis method of time domain and frequency domain characteristics to identify and mark the environmental noise interference when the power equipment is monitored based on the audible sound. Firstly, the collected audio is segmented into samples at a clock level, then proper framing length is set, the samples are framed, time domain and frequency domain characteristics of sound of each frame are extracted, similarity analysis is carried out on each frame after the samples are framed by using a similarity analysis method based on DBSCAN clustering and the like, and because the sound samples without noise have similarity among frames, the distance in the space is small, a tight cluster can be formed during clustering, and noises such as speaking sound, vehicle running sound, wind blowing and the like are extremely unstable without self-similarity, the unstable noise interference can be easily taken as outliers or more clusters can be formed, so that whether the samples contain unstable noise can be judged through a clustering result, effective samples are screened out by identifying the interference noise of the surrounding environment, and support is provided for monitoring the running state of subsequent power equipment based on the sound signals. Of course, the method of the present invention is also applicable to other continuously operating rotary equipment such as cement mixing, water pumps, etc.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention.
Fig. 2 is a typical 3-layer wavelet packet decomposition diagram.
Fig. 3 is a time domain waveform of noise-free and noise-containing samples and a clustering result in an operating state.
Wherein, a is a normal sample without noise, b is a fault sample without noise, c is a normal sample with speaking noise, and d is a normal sample with wind; a. the upper graph in b, c and d is a waveform, and the lower graph is a cluster.
Detailed Description
The invention provides a method for identifying environmental noise of electrical equipment based on time domain and frequency domain self-similarity, which comprises the following steps:
step 1) sound sample collection: collecting sound signals in the operating environment of the power equipment to be detected;
step 2) dividing the collected sound signals into sound samples with time length of minute level, wherein the time length of each sample is 1-10 min;
step 3), sound sample clustering analysis: setting the frame length as T, respectively carrying out frame processing on each sample, extracting time domain and frequency domain characteristics of each frame to form a characteristic vector, and carrying out similarity analysis based on clustering; if the clustering result of a sample only has one cluster, the sample is indicated to contain no unstable noise, and if more than one cluster or outliers appear in the clustering analysis result of the sample, the sample is indicated to contain unstable noise.
Preferably, the above-mentioned frame length T is taken from 1 to 10s, preferably 2 s.
The time domain features mentioned above are one or more combinations of mean, peak, root mean square value, variance, standard deviation, peak factor, kurtosis, form factor, pulse index or skewness; the frequency domain features comprise one or more combinations of the amplitude of the frequency in the frequency spectrum after fast Fourier transform, the amplitude statistical feature of the power spectral density, the shape statistical feature, Mel coefficients obtained after MFCC feature extraction, or wavelet coefficients obtained by wavelet or wavelet packet decomposition or energy features of different frequency bands.
During clustering (clustering without determining the number of clusters), the power equipment without noise is generally in a stable operation state within a minute-level time length of 1-10min, vibration is very stable, characteristic changes such as time domain, frequency domain and the like are small, similarity exists, the power equipment can be identified into one cluster based on the similar clustering, and common noise such as noise of vehicle driving, speaking, wind blowing, rain falling and the like can be identified into a plurality of clusters due to no similarity.
Example 1
Continuously recording audio data of a distribution transformer in a distribution room by using a sound sensor, wherein the sampling frequency of the sensor is 48kHz, and dividing the collected audio into sound samples according to the length of 60 s; the method comprises the steps of performing framing processing on each sample by using a Hamming window, wherein due to the fact that the length of a framing window is small, details are amplified excessively, the similarity of samples with the original similarity is reduced, the details are smoothed due to the fact that the size of the framing window is too large, noise interference is easily ignored, a large amount of experimental analysis is integrated, the length of the framing window is set to be 2s, the frame is moved to be 1s, the samples are divided into 59 frames, time domain and frequency domain characteristics of each frame are extracted, similarity analysis based on DBSCAN clustering is performed, and the method is combined with the reference of fig. 1.
Examples herein use mean, peak, root mean square, variance, peak factor, kurtosis, form factor, pulse index, or skewness as the time domain index, let the signal be { x }i(i is 1-N, N is the number of sampling points), the temporal feature expression is as follows:
1) the mean value is defined as
2) A peak value; will { xiDividing N sampling points into N segments, finding N peak values { X } in each segmentpjJ is 1 to n), then { x }iThe peak index of the } is:
3) variance; is used for reflecting the overall fluctuation condition and is defined as
4) A root mean square value; the RMS value is time-averaged to reflect the energy of the signal and is defined as
5) A crest factor; used for reflecting mutation points and is defined as
The total energy of the generated pulse waveform is not large, but the peak degree of the waveform is obvious, and the crest factor is suitable for diagnosing the faults.
6) A kurtosis; used to reflect the mutational components of the waveform, are defined as:
to reflect the presence of abrupt components in the waveform.
7) A form factor; the form factor is defined as the ratio of the root mean square value to the absolute mean, defined as:
when in use
When the value is too large, it indicates that a mutation is possible;
when the value is too small, the stability is relatively stable
8) Pulse index
9) Skewness; the data distribution symmetry measure of the skewness response signal is defined as:
in this embodiment, different frequency band energies obtained after wavelet packet decomposition are selected as frequency domain characteristics. FIG. 2 is a typical 3-layer wavelet packet decomposition tree, where a signal with a bandwidth f undergoes 3-layer decomposition to obtain 2 at layer 33Each node corresponds to one frequency band, and it can be seen that the wavelet packet equally divides the whole bandwidth into 8 sub-bands for analysis.
In general, a feature extraction method is commonly used to extract energy of each sub-band in the last layer and use the energy ratio as a feature vector. The wavelet packet energy of node k on the ith layer is:
wherein x is(j,k)(j＝0,1,2…,2i-1; k-0, 1,2.., N) is the reconstruction signal fi,jThe discrete point amplitude of (a); n represents the number of discrete points of the signal.
The total frequency band energy of the signal and the ratio of the frequency band energy occupied by each node are respectively:
the decomposition layer number is set to be 3, the wavelet base is db10, the whole frequency band is divided into 8 sub-frequency bands, namely [ 0-2500 ], [ 2500-5000 ], [ 5000-7500 ], [ 7500-10000 ], [ 10000-12500 ], [ 12500-15000 ], [ 15000-17500 ], [ 17500-20000 ] (Hz), and the proportion of each frequency band is used as a feature vector.
And combining the obtained time domain characteristics and frequency domain characteristic vectors and then carrying out similarity analysis, wherein the method adopts a similarity analysis method based on DBSCAN to carry out similarity analysis among frames.
Respectively combining the time domain and frequency domain characteristics of normal samples without noise, fault samples and normal samples with noise into a characteristic vector, inputting the characteristic vector into a DBSCAN clustering algorithm for clustering analysis, recording clustering results, and recording the appearance of outliers into a cluster for convenient recording. As a result, referring to fig. 3, it can be seen that the samples with noise cannot be grouped into one type and thus do not have similarity, while the normal and fault operation sound samples without noise can be grouped into one type at different frame lengths and thus have similarity, the sound samples with similarity are retained, and the sound samples with noise are removed.
Claims (9)
1. A method for recognizing environmental noise of electrical equipment based on time domain and frequency domain self-similarity is characterized by comprising the following steps:
step 1) sound sample collection: collecting sound signals in the operating environment of the power equipment to be detected;
step 2) sound sample segmentation: dividing the collected sound signal into sound samples with time length of the order of minutes;
step 3), sound sample clustering analysis: setting the frame length as T, respectively carrying out frame processing on each divided sample, extracting time domain characteristics and frequency domain characteristics of each frame to form a characteristic vector, carrying out similarity analysis based on clustering, wherein if the clustering result of a sample only has one cluster, the sample does not contain unstable noise, and if the clustering result of a sample has more than one cluster or contains outliers, the sample contains unstable noise.
2. The method for recognizing the environmental noise of the electric power equipment based on the time domain and frequency domain self-similarity as claimed in claim 1, wherein the time length of the minute level in the step 2) is 1-10 min.
3. The method for recognizing the environmental noise of the electrical equipment based on the self-similarity of the time domain and the frequency domain as claimed in claim 1, wherein the value range of the framing length T in the step 3) is 1-10 s.
4. The method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity according to claim 3, wherein the framing length T is 2 s.
5. The method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity as claimed in claim 1, wherein the time domain features in the step 3) are one or more combinations of mean, peak, variance, root mean square value, peak factor, kurtosis, form factor, pulse index or skewness.
6. The method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity as claimed in claim 1, wherein the frequency domain features in step 3) are one or more combinations of the amplitude of frequency in the frequency spectrum after fast fourier transform, the amplitude statistical feature of power spectral density, the shape statistical feature, Mel coefficients obtained after MFCC feature extraction, or wavelet coefficients obtained by wavelet or wavelet packet decomposition or energy features of different frequency bands.
7. The method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity according to claim 1, wherein the similarity analysis based on clustering in the step 3) is an algorithm which does not need to determine the number of clusters in advance.
8. The method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity as claimed in claim 1, wherein in step 1), a sound sensor is used for collecting sound signals.
9. The method for recognizing environmental noise of electric power equipment based on time domain and frequency domain self-similarity as claimed in claim 1, wherein the collection of the sound signal in step 1) is continuous collection or periodic collection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010646042.8A CN111814872B (en) | 2020-07-07 | 2020-07-07 | Power equipment environmental noise identification method based on time domain and frequency domain self-similarity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010646042.8A CN111814872B (en) | 2020-07-07 | 2020-07-07 | Power equipment environmental noise identification method based on time domain and frequency domain self-similarity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111814872A CN111814872A (en) | 2020-10-23 |
| CN111814872B true CN111814872B (en) | 2022-03-29 |
Family
ID=72841798
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010646042.8A Active CN111814872B (en) | 2020-07-07 | 2020-07-07 | Power equipment environmental noise identification method based on time domain and frequency domain self-similarity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111814872B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112484843B (en) * | 2020-11-23 | 2023-04-07 | 国网北京市电力公司 | State analysis method and device of transformer substation and electronic equipment |
| CN112863546A (en) * | 2021-01-21 | 2021-05-28 | 安徽理工大学 | Belt conveyor health analysis method based on audio characteristic decision |
| CN113793626A (en) * | 2021-08-09 | 2021-12-14 | 邹平市供电有限公司 | Power equipment running state monitoring system and method |
| CN113724735A (en) * | 2021-09-01 | 2021-11-30 | 广州博冠信息科技有限公司 | Voice stream processing method and device, computer readable storage medium and electronic equipment |
| CN113901999B (en) * | 2021-09-29 | 2023-09-29 | 国网四川省电力公司电力科学研究院 | Fault diagnosis method and system for high-voltage shunt reactor |
| CN114155879B (en) * | 2021-12-06 | 2022-07-01 | 哈尔滨工程大学 | Abnormal sound detection method for compensating abnormal perception and stability by using time-frequency fusion |
| CN115691509A (en) * | 2022-12-29 | 2023-02-03 | 杭州兆华电子股份有限公司 | Interference identification method suitable for abnormal sound detection of industrial equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650576A (en) * | 2016-09-22 | 2017-05-10 | 中国矿业大学 | Mining equipment health state judgment method based on noise characteristic statistic |
| CN110596539A (en) * | 2019-09-24 | 2019-12-20 | 长沙理工大学 | Power distribution network fault positioning method based on transient waveform correlation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10198697B2 (en) * | 2014-02-06 | 2019-02-05 | Otosense Inc. | Employing user input to facilitate inferential sound recognition based on patterns of sound primitives |
-
2020
- 2020-07-07 CN CN202010646042.8A patent/CN111814872B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650576A (en) * | 2016-09-22 | 2017-05-10 | 中国矿业大学 | Mining equipment health state judgment method based on noise characteristic statistic |
| CN110596539A (en) * | 2019-09-24 | 2019-12-20 | 长沙理工大学 | Power distribution network fault positioning method based on transient waveform correlation |
Non-Patent Citations (4)
| Title |
|---|
| 一种基于近邻传播聚类的语音端点检测方法;林琴等;《安徽大学学报(自然科学版)》;20190515(第03期);全文 * |
| 分布式变电站设备音频指纹特征采集和诊断分析系统;姚明等;《电器工业》;20190605(第06期);全文 * |
| 基于模式识别的盲分离语音信号获取方法;徐舜等;《电声技术》;20061217(第12期);全文 * |
| 无人值守变电站电力设备音频监测及故障诊断系统;易琳等;《计算机测量与控制》;20171125(第11期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111814872A (en) | 2020-10-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111814872B (en) | Power equipment environmental noise identification method based on time domain and frequency domain self-similarity | |
| CN102426835B (en) | Method for identifying local discharge signals of switchboard based on support vector machine model | |
| JP7199608B2 (en) | Methods and apparatus for inspecting wind turbine blades, and equipment and storage media therefor | |
| CN109616140B (en) | Abnormal sound analysis system | |
| CN102163427B (en) | Method for detecting audio exceptional event based on environmental model | |
| CN109599120B (en) | Abnormal mammal sound monitoring method based on large-scale farm plant | |
| CN109034046B (en) | Method for automatically identifying foreign matters in electric energy meter based on acoustic detection | |
| CN104167207B (en) | A kind of equipment sound identification method based on Intelligent Mobile Robot | |
| CN108169639B (en) | Method for identifying switch cabinet fault based on parallel long-time and short-time memory neural network | |
| CN106546892A (en) | The recognition methodss of shelf depreciation ultrasonic audio and system based on deep learning | |
| Venter et al. | Automatic detection of African elephant (Loxodonta africana) infrasonic vocalisations from recordings | |
| CN112857767B (en) | Hydro-turbo generator set rotor fault acoustic discrimination method based on convolutional neural network | |
| CN113566948A (en) | Fault audio recognition and diagnosis method for robot coal pulverizer | |
| CN111508517A (en) | Intelligent micro-motor product control method based on noise characteristics | |
| WO2023245990A1 (en) | Classification and abnormality detection method for electrical devices of power plant on basis of multi-information fusion | |
| CN114325256A (en) | Power equipment partial discharge identification method, system, equipment and storage medium | |
| CN115081473A (en) | Multi-feature fusion brake noise classification and identification method | |
| CN113345399A (en) | Method for monitoring sound of machine equipment in strong noise environment | |
| CN114352486A (en) | Wind turbine generator blade audio fault detection method based on classification | |
| CN116318457B (en) | Radio signal monitoring method and system | |
| CN111341343B (en) | Online updating system and method for abnormal sound detection | |
| CN114242085A (en) | Fault diagnosis method and device for rotating equipment | |
| CN116244600A (en) | Method, system and equipment for constructing GIS intermittent discharge mode identification model | |
| CN106160944B (en) | A kind of variable rate coding compression method of ultrasonic wave local discharge signal | |
| CN114157023B (en) | Distribution transformer early warning information acquisition method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |