CN115618205A - Portable voiceprint fault detection system and method - Google Patents
Portable voiceprint fault detection system and method Download PDFInfo
- Publication number
- CN115618205A CN115618205A CN202211332248.9A CN202211332248A CN115618205A CN 115618205 A CN115618205 A CN 115618205A CN 202211332248 A CN202211332248 A CN 202211332248A CN 115618205 A CN115618205 A CN 115618205A
- Authority
- CN
- China
- Prior art keywords
- voiceprint
- fault
- equipment
- monitoring analyzer
- collector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a portable voiceprint fault detection system and a method, wherein the system comprises a voiceprint monitoring analyzer, a sound pickup, a collector and an edge computing gateway; the sound pickup is arranged at a collection node of the equipment to be detected according to the equipment property, the collector is connected with the voiceprint monitoring analyzer, the sound pickup is connected with the collector, the collector transmits collected data to the edge computing gateway, and the data are sent to the voiceprint monitoring analyzer through the processing of the edge computing gateway; the edge computing gateway is connected with the voiceprint monitoring analyzer through a wireless network; a fault model database is stored in the voiceprint monitoring analyzer; the voiceprint monitoring analyzer adopts a cmfmc engine to carry out weighting dimension reduction optimization based on MFCC characteristic vectors, applies a vector quantization algorithm to identify noise signals of equipment, and judges the state and the fault type of primary equipment by combining with a fault model database for comparison. The invention directly contacts the detection equipment body, and the detection result is slightly influenced by the environment.
Description
Technical Field
The invention belongs to the technical field of voiceprint detection, and particularly relates to a portable voiceprint fault detection system and method.
Background
The on-line monitoring of the operation of the power equipment is an important technical means for ensuring reliable operation, and wide interference sources such as corona discharge, impact generated by switching action, partial discharge possibly occurring in adjacent high-voltage electrical equipment and the like exist on site; part of primary equipment is in a severe operating environment, and the continuous operation in a high-load place is a normal state. Therefore, in the long-term operation of large-scale primary electric power equipment, the aging of equipment parts and the loosening of parts can cause unplanned shutdown, resulting in serious accidents.
The traditional hidden danger identification technology of the primary power equipment is realized based on structural change or sudden change of electrical quantity. The detection based on the structural change cannot identify the tiny defects, and the detection result is greatly influenced by the environment and light rays and cannot accurately reflect the health state of the equipment; the detection technology based on sudden change of the electrical quantity needs to perform electrical model modeling on different devices, and various sensors are used for accurately and quickly detecting the electrical quantity, and the current technology cannot directly relate the hidden dangers of all the devices to the electric quantity, so that the detection technology based on sudden change of the electrical quantity has limited accuracy.
Disclosure of Invention
The invention mainly aims to overcome the defects in the prior art and provides a portable voiceprint fault detection system and a portable voiceprint fault detection method.
In order to achieve the purpose, the invention adopts the following technical scheme:
a portable voiceprint fault detection system comprises a voiceprint monitoring analyzer, a sound pickup, a collector and an edge computing gateway;
the sound pickup is arranged at a collection node of the equipment to be detected according to the equipment property, the collector is connected with the voiceprint monitoring analyzer, the sound pickup is connected with the collector, the collector transmits collected data to the edge computing gateway, and the data are sent to the voiceprint monitoring analyzer through the processing of the edge computing gateway;
the edge computing gateway is connected with the voiceprint monitoring analyzer through a wireless network;
a fault model database is stored in the voiceprint monitoring analyzer;
the voiceprint monitoring analyzer adopts a cmfmc engine to perform weighted dimensionality reduction optimization based on MFCC characteristic vectors, applies a vector quantization algorithm to identify noise signals of equipment, and judges the state and the fault type of primary equipment by combining a fault model database for comparison.
Furthermore, the sound pickup is an acoustic emission sensor or a voiceprint sensor, and the acoustic emission sensor and the voiceprint sensor adopt low-frequency narrow-band sensors.
Furthermore, when voiceprint data are collected, a trigger threshold and a filter are adopted to carry out amplitude filtering and frequency filtering, and noise influence is eliminated.
Furthermore, the voiceprint monitoring analyzer identifies voiceprint signals collected by the sound pickup by adopting a parameter analysis method and a waveform analysis method.
Further, the adapter is installed at the collection node of waiting to examine equipment through the mode of magnetism.
The invention also comprises a fault detection method based on the provided system, which comprises the following steps:
installing a pickup at a collection node of equipment to be detected according to equipment properties, and collecting voiceprint data and noise signals;
preprocessing a noise signal, extracting an MFCC characteristic vector, weighting, and reducing the dimension of the MFCC characteristic vector;
and combining a fault model database to perform 1:1, comparing, namely comparing the voiceprint with a normal voiceprint model to judge whether a fault exists; if the fault exists, the following steps are carried out: n, analyzing the fault, namely comparing the voiceprint with n models in the abnormal voiceprint model library, judging the fault type and storing the fault type information into a fault model database;
and (6) deriving a diagnosis report.
Further, the preprocessing specifically includes framing and windowing:
the framing specifically comprises:
when the noise signal is framed, in order to ensure the continuity between two adjacent frames of signals, two frames are overlapped;
total number of samples: m = sr time, i.e. total number of points of samples = sampling rate time;
number of overlaps within a single frame: overlap = wlen-inc i.e. number of overlaps = frame length-frame movement length;
considering the relation of the overlapped signal framing, the frame number after framing is expressed as:
the windowing specifically comprises the following steps:
discrete Fourier transform is performed on the framed signal, and a Hamming window is applied to each framed signal.
Further, extracting the MFCC feature vector specifically includes:
respectively solving a feature vector for the framing signals to form a feature vector group, wherein the solving process comprises Fourier transform, mel filtering, logarithmic transform and discrete cosine transform; the Mel filtering is realized by a filter bank consisting of a plurality of triangular band-pass filters;
setting the number of the filters as p, and obtaining p parameters m after filtering the signals i The calculation formula is as follows:
wherein N is the number of Fourier transform points, X (k) is the Fourier transform of the preprocessed framed signal, H i (k) Is the filter parameter, expressed as:
B(f[i+1])-B(f[i])＝B(f[i])-B(f[i-1])
wherein f [ i ] is the center frequency of the triangular filter;
m is obtained by calculation i Then taking logarithm of the obtained product, performing discrete cosine transform, and calculating to obtain C i Namely, the high-dimensional MFCC feature vector of the framing signal; the method specifically comprises the following steps:
further, the dimension reduction of the MFCC feature vector is specifically as follows:
and reducing and simplifying the obtained high-dimensional MFCC feature vector by adopting a K-L transformation algorithm, and simultaneously ensuring that the noise feature of the equipment is accurately obtained, wherein the K-L transformation calculation is as follows:
e eigenvectors are set to form a matrix G, the dimensionality of each eigenvector is h, and G is expressed as:
calculating a correlation matrix R of G:
R＝G T G/(e-1)
the eigenvalue lambda of the correlation matrix R is calculated therefrom 1 ，λ 2 ，λ 3 ，…λ h And corresponding feature vector u 1 ，u 2 ，u 3 ，…u h ；
Calculating the variance contribution rate eta i And cumulative variance contribution η (l):
further, the construction of the fault model database specifically comprises:
the voiceprint fault detection system conducts active training through a cmfmc algorithm according to existing voiceprint data, voiceprint characteristics are extracted, then labels are built for training models, and a fault model database is generated.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. according to the portable voiceprint fault detection system, the fault abnormal sound mixed in the noise is identified by using the equipment noise, and the fault type is matched according to the characteristics of the fault abnormal sound, so that the fault cause and effect tracing and the time marking are facilitated to be completed; the system is directly contacted with the equipment body to be detected, and the detection result is slightly influenced by the environment; when the fault and the hidden danger are detected, the device does not need to be deeply inserted into the device, the operation can be carried out on the surface of the device, and the device is not damaged.
2. Compare with traditional monitoring platform, the problem that power equipment exists is gone out in monitoring that can convenient and fast, uses the operational environment at different power equipment, and is portable, high-efficient more.
3. The invention can be widely applied to timely early warning of various fault states of power equipment, can be displayed on a portable voiceprint monitoring analyzer in real time, can carry out artificial intelligent deep learning on the fault phenomenon in an abnormal state, provides a change monitoring curve of the equipment for operation and inspection personnel, and can make auxiliary study and judgment for correct operation and inspection decisions.
Drawings
FIG. 1 is a schematic diagram of a system according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of an embodiment of the present invention;
FIG. 3 is a schematic diagram of a fault identification process in an embodiment;
FIG. 4 is a schematic diagram of a construction flow of a fault model database in the embodiment;
FIG. 5 is a flow chart of a method of an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the embodiments of the present invention are not limited thereto.
Examples
As shown in fig. 1 and 2, a portable voiceprint fault detection system of the present invention includes a voiceprint monitoring analyzer, a sound pickup, a collector, and an edge computing gateway; in this embodiment, the detection of a transformer fault is taken as an example.
The pickup is arranged at a collection node of the equipment to be detected in a magnetic attraction manner according to the property of the equipment, the collector is connected with the voiceprint monitoring analyzer, the pickup is connected with the collector, the collector transmits collected data to the edge computing gateway, and the data are sent to the voiceprint monitoring analyzer through the processing of the edge computing gateway;
the edge computing gateway is connected with the voiceprint monitoring analyzer through a wireless network;
a fault model database is stored in the voiceprint monitoring analyzer;
the voiceprint monitoring analyzer adopts a cmfmc engine to carry out weighting dimension reduction optimization based on MFCC characteristic vectors, applies a vector quantization algorithm to identify noise signals of equipment, and judges the state and the fault type of primary equipment by combining with a fault model database for comparison.
In this embodiment, the sound pickup is an acoustic emission sensor or a voiceprint sensor, and the acoustic emission sensor and the voiceprint sensor use low-frequency narrow-band sensors.
In this embodiment, when collecting voiceprint data, a trigger threshold and a filter are used to perform amplitude filtering and frequency filtering, and eliminate noise influence.
The voiceprint monitoring analyzer identifies the voiceprint signals collected by the sound pick-up by adopting a parameter analysis method and a waveform analysis method.
In another embodiment, a fault detection method based on the system of the above embodiment is provided, as shown in fig. 3 and 5, including the following steps:
s1, mounting a pickup at a collecting node of equipment to be detected according to the equipment properties, and collecting voiceprint data and noise signals;
s2, preprocessing the noise signal, extracting an MFCC characteristic vector, weighting, and reducing the dimension of the MFCC characteristic vector; in this embodiment, the preprocessing specifically includes framing and windowing:
the framing specifically includes:
when the noise signal is framed, in order to ensure the continuity between two adjacent frames of signals, two frames are overlapped;
total number of samples: m = sr time, i.e. total number of samples = sampling rate time;
number of overlaps within a single frame: overlap = wlen-inc i.e. number of overlaps = frame length-frame movement length;
considering the relation of the overlapped signal framing, the frame number after framing is expressed as:
the windowing specifically comprises:
discrete Fourier transform is performed on the framed signal, and a Hamming window is applied to each framed signal.
In this embodiment, extracting the MFCC feature vector specifically includes:
respectively solving a feature vector from the framing signals to form a feature vector group, wherein the solving process comprises Fourier transform, mel filtering, logarithmic transform and discrete cosine transform; the Mel filtering is realized by a filter bank consisting of a plurality of triangular band-pass filters;
setting the number of the filters as p, and obtaining p parameters m after filtering the signals i The calculation formula is as follows:
wherein N is the number of Fourier transform points, X (k) is the FFT (Fourier transform) of the preprocessed framing signal, H i (k) Is the filter parameter, expressed as:
B(f[i+1])-B(f[i])＝B(f[i])-B(f[i-l])
wherein f [ i ] is the center frequency of the triangular filter;
m is obtained by calculation i Then taking logarithm of the obtained C, performing discrete cosine transform, and calculating to obtain C i That is, the MFCC feature vector of the framing signal specifically includes:
in this embodiment, the dimension reduction of the MFCC feature vector specifically includes:
and reducing and simplifying the obtained high-dimensional MFCC feature vector by adopting a K-L transformation algorithm, and simultaneously ensuring that the noise feature of the equipment is accurately obtained, wherein the K-L transformation calculation is as follows:
e eigenvectors are arranged to form a matrix G, the dimensionality of each eigenvector is h, and G is expressed as:
calculating a correlation matrix R of G:
R＝G T G/(e-1)
the eigenvalue lambda of the correlation matrix R is calculated therefrom 1 ,λ 2 ,λ 3 ,...λ h And corresponding feature vector u 1 ,u 2 ,u 3 ,…u h ；
Calculating the variance contribution rate eta i And cumulative variance contribution η (l):
and S3, combining a fault model database, and performing 1: comparing 1, namely comparing the voiceprint with a normal voiceprint model and judging whether a fault exists; if the fault exists, the following steps are carried out: n, analyzing the fault, namely comparing the voiceprint with n models in the abnormal voiceprint model library, judging the fault type and storing the fault type information into a fault model database; as shown in fig. 4, the construction of the fault model database specifically includes:
the voiceprint fault detection system conducts active training through a cmfmc algorithm according to existing voiceprint data, voiceprint characteristics are extracted, then labels are built for training models, and a fault model database is generated.
And S4, deriving a diagnosis report.
It should also be noted that in the present specification, terms such as "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A portable voiceprint fault detection system is characterized by comprising a voiceprint monitoring analyzer, a sound pickup, a collector and an edge computing gateway;
the sound pickup is arranged at a collection node of the equipment to be detected according to the equipment property, the collector is connected with the voiceprint monitoring analyzer, the sound pickup is connected with the collector, the collector transmits collected data to the edge computing gateway, and the data are sent to the voiceprint monitoring analyzer through the processing of the edge computing gateway;
the edge computing gateway is connected with the voiceprint monitoring analyzer through a wireless network;
a fault model database is stored in the voiceprint monitoring analyzer;
the voiceprint monitoring analyzer adopts a cmfmc engine to perform weighted dimensionality reduction optimization based on MFCC characteristic vectors, applies a vector quantization algorithm to identify noise signals of equipment, and judges the state and the fault type of primary equipment by combining a fault model database for comparison.
2. The portable voiceprint failure detection system of claim 1 wherein the sound pickup is an acoustic emission sensor or a voiceprint sensor, and the acoustic emission sensor and the voiceprint sensor employ low frequency narrow band sensors.
3. The portable voiceprint failure detection system of claim 1 wherein a trigger threshold and a filter are employed to perform amplitude filtering, frequency filtering, and noise effects are eliminated when voiceprint data is collected.
4. The portable voiceprint failure detection system of claim 1 wherein the voiceprint monitoring analyzer identifies voiceprint signals collected by a microphone using a parametric analysis method and a waveform analysis method.
5. The portable voiceprint failure detection system of claim 1 wherein the pickup is mounted at the collection node of the device under inspection by means of magnetic attraction.
6. A method of fault detection based on the system of any one of claims 1 to 5, characterized in that it comprises the following steps:
installing a pickup at a collection node of equipment to be detected according to the equipment properties, and collecting voiceprint data and noise signals;
preprocessing a noise signal, extracting an MFCC characteristic vector, weighting, and reducing the dimension of the MFCC characteristic vector;
and combining a fault model database to perform 1:1, comparing, namely comparing the voiceprint with a normal voiceprint model to judge whether a fault exists; if the fault exists, the following steps are carried out: n, analyzing the fault, namely comparing the voiceprint with n models in the abnormal voiceprint model library, judging the fault type and storing the fault type information into a fault model database;
and (6) deriving a diagnosis report.
7. The method according to claim 6, wherein the preprocessing specifically comprises framing and windowing:
the framing specifically comprises:
when the noise signal is framed, in order to ensure the continuity between two adjacent frames of signals, two frames are overlapped;
total number of samples: m = sr time, i.e. total number of samples = sampling rate time;
number of overlaps within a single frame: overlap = wlwn-inc i.e. number of overlaps = frame length-frame movement length;
considering the relationship of the overlapped signal framing, the number of frames after framing is expressed as:
the windowing specifically comprises the following steps:
discrete Fourier transform is performed on the framed signal, and a Hamming window is applied to each framed signal.
8. The method of fault detection according to claim 7, wherein extracting MFCC feature vectors is specifically:
respectively solving a feature vector for the subframe signals to form a feature vector group, wherein the solving process comprises Fourier transform, mel filtering, logarithmic transform and discrete cosine transform; the Mel filtering is realized by a filter bank consisting of a plurality of triangular band-pass filters;
setting the number of the filters as p, and obtaining p parameters m after filtering the signals i The calculation formula is as follows:
wherein N is the number of Fourier transform points, X (k) is the Fourier transform of the preprocessed framing signal, H i (k) Is the filter parameter, expressed as:
B(f[i+1])-B(f[i])＝B(f[i])-B(f[i-1])
wherein f [ i ] is the center frequency of the triangular filter;
m is obtained by calculation i Then taking logarithm of the obtained product, performing discrete cosine transform, and calculating to obtain C i Namely, the high-dimensional MFCC feature vector of the framing signal is obtained; the method specifically comprises the following steps:
9. the method of claim 8, wherein the dimension reduction of the MFCC eigenvector is specifically:
and reducing and simplifying the obtained high-dimensional MFCC feature vector by adopting a K-L transformation algorithm, and simultaneously ensuring that the noise feature of the equipment is accurately obtained, wherein the K-L transformation calculation is as follows:
e eigenvectors are set to form a matrix G, the dimensionality of each eigenvector is h, and G is expressed as:
calculating a correlation matrix R of G:
R＝G T G/(e-1)
from this, the eigenvalue λ of the correlation matrix R is calculated 1 ,λ 2 ,λ 3 ,...λ h And corresponding feature vector u 1 ,u 2 ,u 3 ,…u h ；
Calculating the variance contribution rate eta i And cumulative variance contribution η (l):
10. the fault detection method according to claim 6, wherein the fault model database is constructed by:
the voiceprint fault detection system conducts active training through a cmfmc algorithm according to existing voiceprint data, voiceprint characteristics are extracted, then labels are built for training models, and a fault model database is generated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211332248.9A CN115618205A (en) | 2022-10-28 | 2022-10-28 | Portable voiceprint fault detection system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211332248.9A CN115618205A (en) | 2022-10-28 | 2022-10-28 | Portable voiceprint fault detection system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115618205A true CN115618205A (en) | 2023-01-17 |
Family
ID=84876786
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211332248.9A Pending CN115618205A (en) | 2022-10-28 | 2022-10-28 | Portable voiceprint fault detection system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115618205A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116774109A (en) * | 2023-06-26 | 2023-09-19 | 国网黑龙江省电力有限公司佳木斯供电公司 | Transformer fault identification system based on voiceprint detection information |
| CN116839883A (en) * | 2023-07-04 | 2023-10-03 | 安徽中科昊音智能科技有限公司 | Iron tower screw loosening diagnosis method and device based on voiceprint recognition |
| CN117894317A (en) * | 2024-03-14 | 2024-04-16 | 沈阳智帮电气设备有限公司 | Box-type transformer on-line monitoring method and system based on voiceprint analysis |
-
2022
- 2022-10-28 CN CN202211332248.9A patent/CN115618205A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116774109A (en) * | 2023-06-26 | 2023-09-19 | 国网黑龙江省电力有限公司佳木斯供电公司 | Transformer fault identification system based on voiceprint detection information |
| CN116774109B (en) * | 2023-06-26 | 2024-01-30 | 国网黑龙江省电力有限公司佳木斯供电公司 | Transformer fault identification system based on voiceprint detection information |
| CN116839883A (en) * | 2023-07-04 | 2023-10-03 | 安徽中科昊音智能科技有限公司 | Iron tower screw loosening diagnosis method and device based on voiceprint recognition |
| CN116839883B (en) * | 2023-07-04 | 2024-02-13 | 安徽中科昊音智能科技有限公司 | Iron tower screw loosening diagnosis method and device based on voiceprint recognition |
| CN117894317A (en) * | 2024-03-14 | 2024-04-16 | 沈阳智帮电气设备有限公司 | Box-type transformer on-line monitoring method and system based on voiceprint analysis |
| CN117894317B (en) * | 2024-03-14 | 2024-05-24 | 沈阳智帮电气设备有限公司 | Box-type transformer on-line monitoring method and system based on voiceprint analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115618205A (en) | Portable voiceprint fault detection system and method | |
| CN112201260B (en) | Transformer running state online detection method based on voiceprint recognition | |
| CN110398647B (en) | Transformer state monitoring method | |
| CN111948487A (en) | High-voltage power equipment fault diagnosis method and system based on artificial intelligence | |
| CN108169639B (en) | Method for identifying switch cabinet fault based on parallel long-time and short-time memory neural network | |
| CN107748314A (en) | Transformer Faults Analysis system based on sound wave shock detection | |
| CN112067701B (en) | Fan blade remote auscultation method based on acoustic diagnosis | |
| CN112331231B (en) | Broiler feed intake detection system based on audio technology | |
| CN112660745A (en) | Intelligent diagnosis method and system for carrier roller fault and readable storage medium | |
| CN107367676A (en) | Partial discharge indicator based on audio intelligent identification | |
| CN115932659A (en) | Transformer fault detection method based on voiceprint characteristics | |
| CN116778964A (en) | Power transformation equipment fault monitoring system and method based on voiceprint recognition | |
| CN113345399A (en) | Method for monitoring sound of machine equipment in strong noise environment | |
| CN115839763A (en) | Transformer vibration noise monitoring system | |
| CN114708885A (en) | Fan fault early warning method based on sound signals | |
| CN116705039A (en) | AI-based power equipment voiceprint monitoring system and method | |
| CN117169639B (en) | Product detection method and system for power adapter production | |
| CN115376526A (en) | Power equipment fault detection method and system based on voiceprint recognition | |
| CN114999527A (en) | Transformer anomaly detection model training and deploying method and device | |
| CN114487952A (en) | Quench detection system and method using acoustic optical fiber | |
| CN117607598A (en) | Transformer fault detection method and system based on voiceprint characteristics | |
| CN117292713A (en) | Transformer voiceprint monitoring method and system based on end Bian Yun cooperation | |
| JP2002323371A (en) | Sound diagnostic device and sound diagnostic method | |
| CN117116293A (en) | Machine equipment fault diagnosis system in complex sound field environment | |
| CN112034312A (en) | Power equipment insulation defect mode identification 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 |