CN117110989B - Noise fault positioning detection method and system for power equipment - Google Patents
Noise fault positioning detection method and system for power equipment Download PDFInfo
- Publication number
- CN117110989B CN117110989B CN202311334263.1A CN202311334263A CN117110989B CN 117110989 B CN117110989 B CN 117110989B CN 202311334263 A CN202311334263 A CN 202311334263A CN 117110989 B CN117110989 B CN 117110989B
- Authority
- CN
- China
- Prior art keywords
- information
- screening
- fault
- acoustic
- sound source
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 238000012216 screening Methods 0.000 claims abstract description 71
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000013507 mapping Methods 0.000 claims abstract description 9
- 238000003909 pattern recognition Methods 0.000 claims abstract description 7
- 238000001228 spectrum Methods 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 21
- 238000004458 analytical method Methods 0.000 claims description 18
- 230000009466 transformation Effects 0.000 claims description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 230000003595 spectral effect Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 4
- 230000004807 localization Effects 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 2
- 230000002123 temporal effect Effects 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000003384 imaging method Methods 0.000 description 9
- 230000005236 sound signal Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 7
- 230000007246 mechanism Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000006722 reduction reaction Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
- G01S5/18—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using ultrasonic, sonic, or infrasonic waves
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- 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
-
- 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
- G10L25/18—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Acoustics & Sound (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Theoretical Computer Science (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Computational Linguistics (AREA)
- General Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention provides a noise fault location detection method and system for power equipment, and relates to a fault location technology, wherein the method comprises the steps of collecting sound source information of the power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and selecting the acoustic characteristics based on the relevance of the acoustic characteristics and pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics; mapping the screening characteristics into a target space, extracting multi-scale envelope information corresponding to the screening characteristics by combining time information corresponding to the screening characteristics, and creating a noise image corresponding to the sound source information by combining space grids corresponding to the target space; and carrying out pattern recognition and fault positioning on the noise image based on a preset fault positioning model, and determining the fault type and the fault position of the power equipment.
Description
Technical Field
The present invention relates to fault location technologies, and in particular, to a noise fault location detection method and system for an electrical device.
Background
Because the transformer station has a plurality of devices, and the transformer, the reactor and the high-voltage transmission line can generate noise with different degrees during normal operation, the noise source position of the transformer station is accurately identified and positioned, and the method has great significance in researching and managing the noise of the transformer station and optimizing and improving the devices in the station. And a microphone array is adopted to extract the sound source of the electrical equipment in the transformer substation under the condition of complex sound field in the transformer substation, so that the method is more accurate and more specific than the traditional noise measurement method, and the measurement result can be intuitively provided for measurement and analysis staff. However, since the noise of the electrical equipment is mainly low frequency, the noise sound wave is longer, and the diameter of the array must be increased to improve the positioning accuracy and high resolution of the sound source, the manufacturing cost of the array will be increased, and meanwhile, inconvenience is brought to transportation, debugging, measurement and popularization and application.
CN103235286B, a high-precision positioning method for electric noise source, uses a microphone array with a certain shape composed of several array elements to measure the sound field information near the noise source, firstly carries out routine processing to the information to obtain a preliminary acoustic imaging diagram, then subtracts the side lobes which are not contained in the acoustic imaging diagram, thus obtaining a series of clearer acoustic imaging diagrams composed of main lobes, and realizing high-precision positioning for the noise source. The invention is based on the conventional wave beam forming method, and carries out reconstruction of the acoustic imaging diagram after taking the maximum value of the sound intensity, thereby being capable of designing the width of the wave beam main lobe in the new acoustic imaging diagram and representing the sound source position with high resolution.
CN103336266A, a portable substation noise imaging positioning detection device, comprising an imaging display mechanism and a noise detection mechanism which are connected through a coaxial cable, wherein the noise detection mechanism comprises a noise sensor and a signal preprocessing circuit which are arranged on a quadric surface structure in a matrix manner; the imaging display mechanism comprises a display, a processor and a data memory connected with the processor; the processor is provided with a sound intensity mapping processing module which is used for driving and displaying a sound intensity mapping graph to rapidly judge the noise source. According to the method, a cubic data model of a quadric surface is established according to the position of equipment to be tested of a transformer substation, a three-dimensional coordinate system is established based on the model, related noise signals are preprocessed and displayed, noise imaging is carried out by using pseudo image information, and quick fault positioning is realized by using color ring belt analysis.
Disclosure of Invention
The embodiment of the invention provides a noise fault positioning detection method and system for power equipment, which can at least solve part of problems in the prior art.
In a first aspect of an embodiment of the present invention,
provided is a noise fault location detection method for an electrical device, including:
acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and selecting the acoustic characteristics based on the relevance of the acoustic characteristics and the pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics, wherein the acoustic characteristics comprise at least one of frequency spectrum characteristics, time domain characteristics and energy characteristics;
mapping the screening characteristics into a target space, extracting multi-scale envelope information corresponding to the screening characteristics by combining time information corresponding to the screening characteristics, and creating a noise image corresponding to the sound source information by combining space grids corresponding to the target space;
and carrying out pattern recognition and fault positioning on the noise image based on a preset fault positioning model, and determining the fault type and the fault position of the power equipment.
In an alternative embodiment of the present invention,
when the acoustic signature is a spectral signature,
extracting features of the sound source information, wherein determining acoustic features corresponding to the sound source information comprises:
converting the sound source information into a frequency domain signal through a fast Fourier transform algorithm, and determining amplitude and phase information of different frequency components from the frequency domain signal as frequency spectrum information;
and carrying out smoothing processing on the spectrum information through an average sliding window, converting the smoothed spectrum information into analysis information by combining Hilbert transformation, extracting the amplitude of the analysis information as envelope information, and taking the envelope information as spectrum characteristics.
In an alternative embodiment of the present invention,
performing feature selection on the acoustic features based on the relevance of the acoustic features and the pre-acquired historical target features, and obtaining screening features corresponding to the acoustic features comprises:
respectively determining historical target features corresponding to all the acoustic features, and determining first relevance between all the features and the historical target features and second relevance between all the features;
and determining a sorting score of each feature based on the ratio of the first relevance to the second relevance, and taking the acoustic feature with the sorting score exceeding a preset screening threshold as a screening feature.
In an alternative embodiment of the present invention,
mapping the screening feature into a target space, extracting multi-scale envelope information corresponding to the screening feature by combining time information corresponding to the screening feature, and creating a noise image corresponding to the sound source information by combining a space grid corresponding to the target space, wherein the step of creating the noise image comprises the following steps:
according to the screening characteristics, combining time information corresponding to the screening characteristics, decomposing the time information into a plurality of time-scale sub-signals through wavelet transformation, performing Hilbert transformation on each time-scale sub-signal to obtain a plurality of analysis signals, extracting amplitude information of each analysis signal as envelope information of the current scale, and obtaining multi-scale envelope information in a weighted average mode;
dividing space grids based on the dimension and the range of the target space, wherein each space grid is used for indicating pixels of a noise image, and distributing space weights to the screening features based on the position correlation between the multi-scale envelope information and the target space;
and distributing the corresponding screening characteristics to the space grid of the target space based on the space weights of the screening characteristics, and creating a noise image corresponding to the sound source information.
In an alternative embodiment of the present invention,
the method further includes training a fault localization model:
obtaining a failure training data set, wherein samples of the failure training data set comprise a plurality of failure sample tags;
initializing network weight and bias parameters of a fault location model to be trained, inputting a fault training data set into the fault location model to be trained, determining a model output of the fault location model and a prediction deviation of a sample label corresponding to a sample of the fault training data set;
and determining the gradient of the network weight and the bias parameter of the fault location model to be trained on the loss function through a back propagation algorithm based on the loss function of the fault location model in combination with the prediction deviation, iteratively and automatically adjusting the learning rate and updating the network weight and the bias parameter of the fault location model to be trained so as to minimize the loss value of the loss function.
In an alternative embodiment of the present invention,
constructing a loss function includes:
；
wherein,LOSSrepresenting the loss value corresponding to the loss function,i,jrespectively represent the firstiSeed failure sample tag and the firstjA seed failure sample tag is used to identify,Nrepresenting the number of samples to be taken,w i represent the firstiThe failure weight corresponding to the failure sample tag,p i 、y i respectively represent the firstiModel output and model number corresponding to seed fault sample labeliA seed failure sample tag is used to identify,r ij represent the firstiSeed failure sample tag and the firstjThe correlation weight corresponding to the failure sample tag,y j represent the firstjA failure sample tag.
In a second aspect of an embodiment of the present invention,
there is provided a noise fault location detection system for an electrical device, comprising:
the device comprises a first unit, a second unit and a third unit, wherein the first unit is used for acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and selecting the acoustic characteristics based on the relevance of the acoustic characteristics and pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics, wherein the acoustic characteristics comprise at least one of frequency spectrum characteristics, time domain characteristics and energy characteristics;
a second unit, configured to map the screening feature to a target space, extract multi-scale envelope information corresponding to the screening feature in combination with time information corresponding to the screening feature, and create a noise image corresponding to the sound source information in combination with a spatial grid corresponding to the target space;
and the third unit is used for carrying out pattern recognition and fault location on the noise image based on a preset fault location model and determining the fault type and the fault position of the power equipment.
In a third aspect of an embodiment of the present invention,
there is provided an electronic device including:
a processor;
a memory for storing processor-executable instructions;
wherein the processor is configured to invoke the instructions stored in the memory to perform the method described previously.
In a fourth aspect of an embodiment of the present invention,
there is provided a computer readable storage medium having stored thereon computer program instructions which, when executed by a processor, implement the method as described above.
The beneficial effects of the embodiments of the present invention may refer to the effects corresponding to technical features in the specific embodiments, and are not described herein.
Drawings
Fig. 1 is a flow chart of a noise fault location detection method for an electrical device according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a noise fault location detection system for an electrical device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The technical scheme of the invention is described in detail below by specific examples. The following embodiments may be combined with each other, and some embodiments may not be repeated for the same or similar concepts or processes.
Fig. 1 is a flow chart of a noise fault location detection method for an electrical device according to an embodiment of the present invention, as shown in fig. 1, where the method includes:
s101, acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and selecting the acoustic characteristics based on the relevance of the acoustic characteristics and pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics, wherein the acoustic characteristics comprise at least one of frequency spectrum characteristics, time domain characteristics and energy characteristics;
the sensor array is arranged in the target area, so that the coverage range of the sensor covers the sound source information of the power equipment, the sound signal data in the sensor array are collected in real time, and the high quality and the real-time performance of the data are ensured. Preprocessing the collected sound signal data, including noise reduction, filtering, normalization and the like, so as to improve the signal-to-noise ratio and the data quality. Spectral features (e.g., fourier transforms), time domain features (e.g., time domain waveform parameters), energy features (e.g., instantaneous energy), etc., are extracted from the preprocessed data, and the selection of the appropriate feature extraction method depends on the particular problem and data characteristics.
Target characteristics associated with the status of the electrical device are obtained from the historical data and may include information associated with the operational status of the device such as current, voltage, temperature, etc. The correlation analysis is carried out on the acoustic characteristics and the historical target characteristics, and the relationship between the acoustic characteristics and the historical target characteristics can be determined by using methods such as pearson correlation coefficients, mutual information and the like. A suitable feature selection algorithm, such as Recursive Feature Elimination (RFE) or a relevance-based feature selection method, is selected to determine which acoustic features have a strong relevance to the historical target features. Using the results of the correlation analysis, a correlation model may be established, which may be used to predict the power device state. A suitable machine learning algorithm, such as linear regression, support Vector Machine (SVM), random forest, etc., may be selected, a relevance model is built, and supervised learning is performed using historical target features as tag data. When the sensor array collects new sound source information, the new sound source information is input into a trained relevance model, and the state of the power equipment is predicted. And monitoring the prediction result in real time, finding out potential power equipment problems in time, and taking corresponding maintenance or treatment measures.
In an alternative embodiment of the present invention,
acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, and determining acoustic characteristics corresponding to the sound source information, wherein the acoustic characteristics comprise at least one of frequency spectrum characteristics, time domain characteristics and energy characteristics, and performing characteristic selection on the acoustic characteristics to obtain screening characteristics corresponding to the acoustic characteristics;
for example, an array of sensors may be installed at multiple locations in the target area, ensuring that its frequency response range is appropriate for the sound of the power equipment; and ensures that the sampling frequency of the sensor array is high enough to capture high noise. The sensor array may be connected to a data acquisition system to acquire sound data in real time. The acquisition interval is defined, typically in milliseconds, to control the frequency of data sampling, during which the acquired sound signals are recorded and stored in a computer or data storage device.
The collected sound data is noise reduced by using a signal processing technology to reduce the influence of environmental noise, wherein methods such as a digital filter, wavelet transformation and the like can be used. The sound signal is filtered to remove irrelevant frequency components, leaving the signal associated with the power equipment. Amplification or gain adjustment is performed to ensure that the sound signal has the proper amplitude range for subsequent analysis.
Wherein the feature extraction can be performed on the sound source information, the corresponding spectral features of the sound source information can be determined, in particular,
a fast fourier transform algorithm may be used to transform the sound source information into a frequency domain signal, which will produce spectral information, including amplitude and phase information of different frequency components in the sound source information; extracting envelope information from the spectrum to capture the overall spectral characteristics of the sound signal may be implemented using low pass filters or envelope detection techniques.
Illustratively, the spectrum is smoothed to reduce the influence of high-frequency noise, and a filter or an average sliding window or the like may be used to smooth the spectrum, and the smoothed spectrum helps to more accurately capture the overall characteristics of the sound signal. Extracting envelope information from the smoothed spectrum by using an envelope detection technology, wherein the object of the envelope detection is to identify peaks and troughs of the spectrum so as to capture rising and falling trends of the spectrum, converting the spectrum into an analytic signal by using a Hilbert transform, and then extracting the amplitude of the analytic signal as an envelope; the extracted envelope is further processed, such as smoothed or denoised, to ensure that the envelope information captures the main features of the sound signal, while removing noise and unwanted fluctuations.
In an alternative embodiment of the present invention,
performing feature selection on the acoustic features based on the relevance of the acoustic features and the pre-acquired historical target features, and obtaining screening features corresponding to the acoustic features comprises:
respectively determining historical target features corresponding to all the acoustic features, and determining first relevance between all the features and the historical target features and second relevance between all the features;
and determining a sorting score of each feature based on the ratio of the first relevance to the second relevance, and taking the acoustic feature with the sorting score exceeding a preset screening threshold as a screening feature.
Illustratively, for each acoustic feature, calculating its pearson correlation coefficient with the historical target feature to obtain a first correlation coefficient matrix representing the correlation of the respective feature with the historical target feature; and calculating the Pearson correlation coefficient between the acoustic features to obtain a second correlation coefficient matrix, wherein the second correlation coefficient matrix represents the correlation between the features. And for each acoustic feature, calculating the ratio of the first relevance and the second relevance of the acoustic features to obtain a sorting score, and taking the acoustic feature with the sorting score exceeding a preset screening threshold as a screening feature.
An association matrix between the features and the historical target features is established, wherein rows represent acoustic features and columns represent the historical target features. And determining possible influencing factors corresponding to each acoustic feature, namely historical target features by using knowledge of domain experts or a data analysis method.
The characteristic with the greatest influence on the state of the power equipment is determined by analyzing the relevance between the acoustic characteristic and the historical target characteristic, so that the use of redundant or irrelevant characteristics is avoided, and the accuracy of characteristic selection is improved; the screened key acoustic features are used for training the prediction model, so that the input features of the model are more targeted, and the precision and accuracy of the prediction model can be improved; by defining the relationship between the acoustic features and the historical target features, the decision process of the system is more interpretable and understandable, increasing the confidence level of the user and related personnel to the system. Because the screened features are more critical, the system can respond to real-time data faster, provide faster decision and feedback, and adapt to different scenes and requirements more easily. The screened key feature set is more representative, so that the adaptability of the system to different working conditions and environmental changes can be improved, and the robustness of the system can be enhanced.
S102, mapping the screening features into a target space, extracting multi-scale envelope information corresponding to the screening features by combining time information corresponding to the screening features, and creating a noise image corresponding to the sound source information by combining a space grid corresponding to the target space;
illustratively, using screening features, acoustic features are mapped to a target space, which may be accomplished by multiple linear regression, principal Component Analysis (PCA), or other dimension reduction techniques, the mapped features should be more representative, enabling a better description of power device states. For each screening feature, decomposing it into sub-signals of different scales using wavelet transformation; and (3) applying Hilbert transformation to each scale sub-signal, extracting an analysis signal, and extracting amplitude information from the analysis signal to obtain multi-scale envelope information.
The two-dimensional or three-dimensional space grid is divided according to the dimension and the range of the target space. Each grid represents a space unit, and the granularity of the grids can be adjusted according to the requirement to determine the spatial resolution of the noise image. The multi-scale envelope information of each screening feature is mapped onto a corresponding spatial grid, and within each spatial cell, the multi-scale envelope information of the plurality of screening features is combined, which may be a simple weighted average or a more complex combination rule, and the combined values represent noise levels within the spatial cell, creating a noise image.
In an alternative embodiment of the present invention,
mapping the screening feature into a target space, extracting multi-scale envelope information corresponding to the screening feature by combining time information corresponding to the screening feature, and creating a noise image corresponding to the sound source information by combining a space grid corresponding to the target space, wherein the step of creating the noise image comprises the following steps:
according to the screening characteristics, combining time information corresponding to the screening characteristics, decomposing the time information into a plurality of time-scale sub-signals through wavelet transformation, performing Hilbert transformation on each time-scale sub-signal to obtain a plurality of analysis signals, extracting amplitude information of each analysis signal as envelope information of the current scale, and obtaining multi-scale envelope information in a weighted average mode;
dividing space grids based on the dimension and the range of the target space, wherein each space grid is used for indicating pixels of a noise image, and distributing space weights to the screening features based on the position correlation between the multi-scale envelope information and the target space;
and distributing the corresponding screening characteristics to the space grid of the target space based on the space weights of the screening characteristics, and creating a noise image corresponding to the sound source information.
Illustratively, the filtering features are combined with corresponding time information to construct a time-frequency domain data set, which is decomposed using wavelet transform to obtain sub-signals of different time scales. And performing Hilbert transformation on each time scale sub-signal to obtain an analysis signal, and extracting amplitude information from the analysis signal to serve as envelope information of the current scale. And carrying out weighted average on the envelope information extracted by each time scale to obtain multi-scale envelope information, wherein the weighted average can use fixed weight or dynamically adjust the weight based on the importance of the characteristics.
Specifically, determining the dimensions and extent of the target space, e.g., the length and width of the two-dimensional space, defines the resolution of the grid, i.e., determines the size of each spatial grid, and may be adjusted as desired. Dividing the target space into a corresponding number of small grids to form a two-dimensional (or three-dimensional) matrix, wherein each small grid represents a space unit, and dividing the two-dimensional or three-dimensional space grids according to the dimension and the range of the target space, and each grid represents a space unit. An average of the multi-scale envelope information in each spatial cell is calculated as a characteristic representation of the spatial cell. For each time scale, an average of the multi-scale envelope information in each grid within the target space is calculated. This may be achieved by taking the average of the multi-scale envelope information within each grid. For two-dimensional space, the average multiscale envelope information for each grid forms a matrix.
Each screening feature is assigned a spatial weight based on the correlation of the mean value of the multi-scale envelope information and the spatial unit location. Multiplying the multi-scale envelope information in each space unit by corresponding space weights to obtain weighted multi-scale envelope information. The weighted multiscale envelope information is mapped onto corresponding locations (pixels) of the noise image creating the noise image.
Through the steps, the combination of screening characteristics and time information, and the extraction and weighted average of multi-scale envelope information are realized. Meanwhile, according to the division of the space grids and the distribution of the space weights, the multi-scale envelope information is successfully mapped onto the noise image, and the spatial presentation of the sound source information is realized. The noise image can more intuitively represent the space-time distribution characteristic of sound, and provides a visual basis for subsequent analysis and decision.
In addition, the noise image combines the sound characteristics with the space positions, and intuitively displays the sound characteristics of different positions, so that a user can clearly know the distribution condition of sound in the space through the image, and the rapid identification of faults or abnormal areas is facilitated. The noise image contains information of a plurality of screening features, and the comprehensive influence of each feature at different spatial positions can be more comprehensively analyzed through comprehensive representation of weighted multi-scale envelope information. This allows the system to comprehensively consider various sound characteristics, improving the ability to comprehensively evaluate the status of the device. The spatialization presentation of the noise image causes the source of the abnormal sound to present a distinct feature on the image. By combining the image processing technology, the fault source in the equipment can be more easily identified and positioned, and the accuracy and precision of fault diagnosis are improved.
S103, performing pattern recognition and fault positioning on the noise image based on a preset fault positioning model, and determining the fault type and the fault position of the power equipment.
In an alternative embodiment of the present invention,
the method further includes training a fault localization model:
obtaining a failure training data set, wherein samples of the failure training data set comprise a plurality of failure sample tags;
initializing network weight and bias parameters of a fault location model to be trained, inputting a fault training data set into the fault location model to be trained, determining a model output of the fault location model and a prediction deviation of a sample label corresponding to a sample of the fault training data set;
and determining the gradient of the network weight and the bias parameter of the fault location model to be trained on the loss function through a back propagation algorithm based on the loss function of the fault location model in combination with the prediction deviation, iteratively and automatically adjusting the learning rate and updating the network weight and the bias parameter of the fault location model to be trained so as to minimize the loss value of the loss function.
In an alternative embodiment of the present invention,
constructing a loss function includes:
；
wherein,LOSSrepresenting the loss value corresponding to the loss function,i,jrespectively represent the firstiSeed failure sample tag and the firstjA seed failure sample tag is used to identify,Nrepresenting the number of samples to be taken,w i represent the firstiThe failure weight corresponding to the failure sample tag,p i 、y i respectively represent the firstiModel output and model number corresponding to seed fault sample labeliA seed failure sample tag is used to identify,r ij represent the firstiSeed failure sample tag and the firstjThe correlation weight corresponding to the failure sample tag,y j represent the firstjA failure sample tag.
Illustratively, a training data set containing a plurality of fault sample tags is obtained from historical data of an actual power device, which samples should cover various fault conditions that the device may encounter. Initializing network weight and bias parameters of a fault location model to be trained, wherein random initialization or pre-trained model parameters can be adopted; and inputting the fault training data set into a fault positioning model, and carrying out forward propagation to obtain the output of the model. And comparing the output of the model with the sample labels of the training data set, and calculating a prediction deviation, namely an error between the model output and the actual labels.
The loss function of the fault location model is defined, typically using a Mean Square Error (MSE) or cross entropy loss function. The predicted deviation is combined with the loss function to calculate a loss value. The gradient of the loss function to the model parameters is calculated using a back propagation algorithm. The learning rate is adaptively adjusted, and a learning rate decay strategy (e.g., adam optimizer) or an adaptive learning rate algorithm (e.g., adagrad, RMSprop, etc.) may be used. And updating the network weight and the bias parameters of the fault location model according to the learning rate and the gradient by using a gradient descent algorithm so as to minimize the loss value of the loss function. The process is iterated until the loss function converges or reaches a preset training round.
Through learning the characteristics and the label relation of different fault samples in the training data set, the model can map the observed sound characteristics to correct fault positions, and accurate fault positioning is realized. The trained fault positioning model can be used for rapidly deducing on real-time data, so that automatic diagnosis of the faults of the power equipment is realized, the efficiency of fault diagnosis is improved, the requirement of manual intervention is reduced, and the time and the cost are saved. Because the training data set contains various fault samples, the fault positioning model has diversity, can cope with various types of power equipment faults, and can effectively diagnose faults caused by abrasion, looseness, circuit faults or other reasons.
Fig. 2 is a schematic structural diagram of a noise fault location detection system for an electrical device according to an embodiment of the present invention, as shown in fig. 2, where the system includes:
the device comprises a first unit, a second unit and a third unit, wherein the first unit is used for acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and selecting the acoustic characteristics based on the relevance of the acoustic characteristics and pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics, wherein the acoustic characteristics comprise at least one of frequency spectrum characteristics, time domain characteristics and energy characteristics;
a second unit, configured to map the screening feature to a target space, extract multi-scale envelope information corresponding to the screening feature in combination with time information corresponding to the screening feature, and create a noise image corresponding to the sound source information in combination with a spatial grid corresponding to the target space;
and the third unit is used for carrying out pattern recognition and fault location on the noise image based on a preset fault location model and determining the fault type and the fault position of the power equipment.
In a third aspect of an embodiment of the present invention,
there is provided an electronic device including:
a processor;
a memory for storing processor-executable instructions;
wherein the processor is configured to invoke the instructions stored in the memory to perform the method described previously.
In a fourth aspect of an embodiment of the present invention,
there is provided a computer readable storage medium having stored thereon computer program instructions which, when executed by a processor, implement the method as described above.
The present invention may be a method, apparatus, system, and/or computer program product. The computer program product may include a computer readable storage medium having computer readable program instructions embodied thereon for performing various aspects of the present invention.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.
Claims (10)
1. A noise fault location detection method for an electrical device, comprising:
acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and carrying out characteristic selection on the acoustic characteristics based on the relevance of the acoustic characteristics and the pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics;
mapping the screening characteristics into a target space, extracting multi-scale envelope information corresponding to the screening characteristics by combining time information corresponding to the screening characteristics, and creating a noise image corresponding to the sound source information by combining space grids corresponding to the target space;
and carrying out pattern recognition and fault positioning on the noise image based on a preset fault positioning model, and determining the fault type and the fault position of the power equipment.
2. The method of claim 1, wherein, when the acoustic signature is a spectral signature,
extracting features of the sound source information, wherein determining acoustic features corresponding to the sound source information comprises:
converting the sound source information into a frequency domain signal through a fast Fourier transform algorithm, and determining amplitude and phase information of different frequency components from the frequency domain signal as frequency spectrum information;
and carrying out smoothing processing on the spectrum information through an average sliding window, converting the smoothed spectrum information into analysis information by combining Hilbert transformation, extracting the amplitude of the analysis information as envelope information, and taking the envelope information as spectrum characteristics.
3. The method of claim 1, wherein selecting the acoustic feature based on the correlation of the acoustic feature with a pre-acquired historical target feature to obtain a screening feature corresponding to the acoustic feature comprises:
respectively determining historical target features corresponding to all the acoustic features, and determining first relevance between all the features and the historical target features and second relevance between all the features;
and determining a sorting score of each feature based on the ratio of the first relevance to the second relevance, and taking the acoustic feature with the sorting score exceeding a preset screening threshold as a screening feature.
4. The method of claim 1, wherein mapping the screening feature into a target space, extracting multi-scale envelope information corresponding to the screening feature in combination with time information corresponding to the screening feature, and creating a noise image corresponding to the sound source information in combination with a spatial grid corresponding to the target space comprises:
according to the screening characteristics, combining time information corresponding to the screening characteristics, decomposing the time information into a plurality of time-scale sub-signals through wavelet transformation, performing Hilbert transformation on each time-scale sub-signal to obtain a plurality of analysis signals, extracting amplitude information of each analysis signal as envelope information of the current scale, and obtaining multi-scale envelope information in a weighted average mode;
dividing space grids based on the dimension and the range of the target space, wherein each space grid is used for indicating pixels of a noise image, and distributing space weights to the screening features based on the position correlation between the multi-scale envelope information and the target space;
and distributing the corresponding screening characteristics to the space grid of the target space based on the space weights of the screening characteristics, and creating a noise image corresponding to the sound source information.
5. The method of claim 1, wherein the acoustic features comprise at least one of spectral features, temporal features, and energy features.
6. The method of claim 1, further comprising training a fault localization model:
obtaining a failure training data set, wherein samples of the failure training data set comprise a plurality of failure sample tags;
initializing network weight and bias parameters of a fault location model to be trained, inputting a fault training data set into the fault location model to be trained, determining a model output of the fault location model and a prediction deviation of a sample label corresponding to a sample of the fault training data set;
and determining the gradient of the network weight and the bias parameter of the fault location model to be trained on the loss function through a back propagation algorithm based on the loss function of the fault location model in combination with the prediction deviation, iteratively and automatically adjusting the learning rate and updating the network weight and the bias parameter of the fault location model to be trained so as to minimize the loss value of the loss function.
7. The method of claim 6, wherein constructing a loss function comprises:
；
wherein,LOSSrepresenting the loss value corresponding to the loss function,i,jrespectively represent the firstiSeed failure sample tag and the firstjA seed failure sample tag is used to identify,Nrepresenting the number of samples to be taken,w i represent the firstiThe failure weight corresponding to the failure sample tag,p i 、y i respectively represent the firstiModel output and model number corresponding to seed fault sample labeliA seed failure sample tag is used to identify,r ij represent the firstiSeed failure sample tag and the firstjThe correlation weight corresponding to the failure sample tag,y j represent the firstjA failure sample tag.
8. A noise fault location detection system for an electrical device, comprising:
the device comprises a first unit, a second unit and a third unit, wherein the first unit is used for acquiring sound source information of power equipment in a target area in real time through a sensor array, extracting characteristics of the sound source information, determining acoustic characteristics corresponding to the sound source information, and selecting the acoustic characteristics based on the relevance of the acoustic characteristics and pre-acquired historical target characteristics to obtain screening characteristics corresponding to the acoustic characteristics, wherein the acoustic characteristics comprise at least one of frequency spectrum characteristics, time domain characteristics and energy characteristics;
a second unit, configured to map the screening feature to a target space, extract multi-scale envelope information corresponding to the screening feature in combination with time information corresponding to the screening feature, and create a noise image corresponding to the sound source information in combination with a spatial grid corresponding to the target space;
and the third unit is used for carrying out pattern recognition and fault location on the noise image based on a preset fault location model and determining the fault type and the fault position of the power equipment.
9. An electronic device, comprising:
a processor;
a memory for storing processor-executable instructions;
wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of claims 1 to 7.
10. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311334263.1A CN117110989B (en) | 2023-10-16 | 2023-10-16 | Noise fault positioning detection method and system for power equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311334263.1A CN117110989B (en) | 2023-10-16 | 2023-10-16 | Noise fault positioning detection method and system for power equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117110989A CN117110989A (en) | 2023-11-24 |
| CN117110989B true CN117110989B (en) | 2024-02-06 |
Family
ID=88805738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311334263.1A Active CN117110989B (en) | 2023-10-16 | 2023-10-16 | Noise fault positioning detection method and system for power equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117110989B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112666430A (en) * | 2020-12-07 | 2021-04-16 | 北京博研中能科技有限公司 | Intelligent transformer voiceprint fault detection method and system |
| CN114509266A (en) * | 2022-01-05 | 2022-05-17 | 东南大学 | Bearing health monitoring method based on fault feature fusion |
| KR20220093942A (en) * | 2020-12-28 | 2022-07-05 | 주식회사 렉터슨 | Apparatus and method for detecting fault of structure considering noise environment |
| CN115963350A (en) * | 2022-11-08 | 2023-04-14 | 国网河北省电力有限公司电力科学研究院 | Fault positioning method and device for direct-current power distribution network |
| CN116380445A (en) * | 2023-06-05 | 2023-07-04 | 吉林市特种设备检验中心(吉林市特种设备事故调查服务中心) | Equipment state diagnosis method and related device based on vibration waveform |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7602985B2 (en) * | 2005-09-14 | 2009-10-13 | University Of Massachusetts | Multi-scale enveloping spectrogram signal processing for condition monitoring and the like |
| WO2013180723A1 (en) * | 2012-05-31 | 2013-12-05 | University Of Connecticut | Methods apparatuses for defect diagnosis in a mechanical system |
| WO2017116353A1 (en) * | 2015-12-30 | 2017-07-06 | Ford Otomotiv Sanayi A. S. | Vibration and noise mapping system and method |
-
2023
- 2023-10-16 CN CN202311334263.1A patent/CN117110989B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112666430A (en) * | 2020-12-07 | 2021-04-16 | 北京博研中能科技有限公司 | Intelligent transformer voiceprint fault detection method and system |
| KR20220093942A (en) * | 2020-12-28 | 2022-07-05 | 주식회사 렉터슨 | Apparatus and method for detecting fault of structure considering noise environment |
| CN114509266A (en) * | 2022-01-05 | 2022-05-17 | 东南大学 | Bearing health monitoring method based on fault feature fusion |
| CN115963350A (en) * | 2022-11-08 | 2023-04-14 | 国网河北省电力有限公司电力科学研究院 | Fault positioning method and device for direct-current power distribution network |
| CN116380445A (en) * | 2023-06-05 | 2023-07-04 | 吉林市特种设备检验中心(吉林市特种设备事故调查服务中心) | Equipment state diagnosis method and related device based on vibration waveform |
Non-Patent Citations (7)
| Title |
|---|
| Research on Discharge Sound Recognition Based on Machine Learning and Convolutional Neural Network Training Algorithm;Yifan Shen;2021 IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA);全文 * |
| 变电站全站设备故障声学成像检测系统的研制;陈敬德;吴继健;徐友刚;周南;罗林根;;电气自动化(05);全文 * |
| 基于复小波多尺度包络谱的滚动轴承故障诊断研究;胡易平;安钢;王凯;贺亮;;煤矿机械(11);全文 * |
| 基于小波变换与SVM的复杂装备故障诊断研究;郑铌;张琳;汤煜;王君;唐晓兵;;火力与指挥控制(06);全文 * |
| 基于最优尺度小波包络流形的轴承故障诊断;许强;方健;刘方;陆思良;刘永斌;赵吉文;;机电一体化(07);全文 * |
| 声信号特征融合的变压器故障判断方法;杨光辉;北 京 理 工 大 学 学 报;全文 * |
| 大型发电厂噪声源的监测及其反演方法研究;曾鹏亚;中国硕士学位论文 信息科技辑;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117110989A (en) | 2023-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111625763A (en) | Operation risk prediction method and prediction system based on mathematical model | |
| CN112213687B (en) | Gateway electric energy meter data anomaly detection method and system based on pseudo-anomaly point identification | |
| Sun et al. | Data-driven fault diagnosis method based on second-order time-reassigned multisynchrosqueezing transform and evenly mini-batch training | |
| CN115376526A (en) | Power equipment fault detection method and system based on voiceprint recognition | |
| CN113642439B (en) | Mechanical state abnormity detection method, device and equipment for on-load tap-changer | |
| CN117371207A (en) | Extra-high voltage converter valve state evaluation method, medium and system | |
| CN116230013A (en) | Transformer fault voiceprint detection method based on x-vector | |
| CN117110989B (en) | Noise fault positioning detection method and system for power equipment | |
| Lei et al. | High voltage shunt reactor acoustic signal denoising based on the combination of VMD parameters optimized by coati optimization algorithm and wavelet threshold | |
| Chaudhuri et al. | Denoising of partial discharge signal using a hybrid framework of total variation denoising-autoencoder | |
| CN117079005A (en) | Optical cable fault monitoring method, system, device and readable storage medium | |
| CN116884432A (en) | VMD-JS divergence-based power transformer fault voiceprint diagnosis method | |
| CN117116293A (en) | Machine equipment fault diagnosis system in complex sound field environment | |
| CN116205817A (en) | Data content complexity targeted detection system | |
| CN116522080A (en) | Partial discharge signal noise reduction method | |
| CN114492146B (en) | Bolt group loosening positioning and quantitative analysis method and system based on transfer learning | |
| CN114897032A (en) | Current transformer fault diagnosis method and device based on width learning | |
| CN117313001B (en) | Mixed event decomposition and identification method for distributed optical fiber acoustic wave sensing system | |
| CN115659150B (en) | Signal processing method, device and equipment | |
| CN116482526A (en) | Analysis system for non-fault phase impedance relay | |
| CN117473275B (en) | Energy consumption detection method for data center | |
| CN117911401A (en) | Power equipment fault detection method, system, storage medium and equipment | |
| CN114034375A (en) | System and method for measuring noise of ultra-high voltage transmission line | |
| Dmitriev et al. | Digital Signal Processing Techniques Applied to Partial Discharge Monitoring and Classification | |
| CN117909805A (en) | Automatic separation method, device, equipment and storage medium for partial discharge mixed signals |
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 |