CN113470694A - Remote listening monitoring method, device and system for hydraulic turbine set - Google Patents
Remote listening monitoring method, device and system for hydraulic turbine set Download PDFInfo
- Publication number
- CN113470694A CN113470694A CN202110447167.2A CN202110447167A CN113470694A CN 113470694 A CN113470694 A CN 113470694A CN 202110447167 A CN202110447167 A CN 202110447167A CN 113470694 A CN113470694 A CN 113470694A
- Authority
- CN
- China
- Prior art keywords
- audio signal
- target
- original audio
- hydraulic turbine
- characteristic value
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 title claims abstract description 49
- 230000005236 sound signal Effects 0.000 claims abstract description 104
- 238000004458 analytical method Methods 0.000 claims abstract description 34
- 238000001514 detection method Methods 0.000 claims abstract description 21
- 230000009467 reduction Effects 0.000 claims abstract description 19
- 238000012806 monitoring device Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims description 33
- 230000006870 function Effects 0.000 claims description 24
- 239000013598 vector Substances 0.000 claims description 24
- 239000011159 matrix material Substances 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 17
- 230000014509 gene expression Effects 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 5
- 238000009472 formulation Methods 0.000 claims 1
- 239000000203 mixture Substances 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000007726 management method Methods 0.000 description 6
- 238000003745 diagnosis Methods 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 238000007405 data analysis Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000000354 decomposition reaction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000010183 spectrum analysis Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000013523 data management Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000011664 signaling Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000010977 unit operation Methods 0.000 description 2
- 108010076504 Protein Sorting Signals Proteins 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03B—MACHINES OR ENGINES FOR LIQUIDS
- F03B11/00—Parts or details not provided for in, or of interest apart from, the preceding groups, e.g. wear-protection couplings, between turbine and generator
- F03B11/008—Measuring or testing arrangements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/20—Hydro energy
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention provides a remote listening monitoring method, a remote listening monitoring device and a remote listening monitoring system for a hydraulic turbine unit, wherein the method comprises the following steps: acquiring an original audio signal with preset duration; carrying out data noise reduction on the original audio signal to obtain a target audio signal; extracting the characteristics of the target audio signal to obtain a target characteristic value; inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal; therefore, the invention solves the problem that the prior art can not monitor or judge the sound characteristics of the hydraulic turbine set during operation, and utilizes the audio analysis technology to carry out on-line analysis and processing on the field sound so as to realize remote listening of the hydraulic turbine set and meet the monitoring requirement on the running state of the hydraulic turbine set.
Description
Technical Field
The invention relates to the technical field of audio processing, in particular to a remote listening monitoring method, a remote listening monitoring device and a remote listening monitoring system for a hydraulic turbine set.
Background
At present, the monitoring and fault diagnosis technology aiming at the running state of the hydraulic turbine set mainly obtains each index data of the unit in running in real time through sensors such as current, voltage, vibration and oscillation, air gap and the like deployed on site; on the basis, the state of the equipment is analyzed in real time through a data processing and analyzing algorithm so as to ensure the reliability of the operation of the unit; therefore, in the aspect of monitoring the state of the hydroelectric generating set, the main technical means is to collect field data by arranging various sensors and cameras on the hydroelectric generating set and auxiliary units thereof by relying on a computer monitoring system and a hydroelectric generating set state monitoring system so as to realize remote measurement, remote signaling, remote regulation, remote control and remote vision.
However, for a system monitoring staff, the monitoring method in the prior art can only judge the unit operation state from a visual angle through the change of various index data, and cannot monitor and judge the sound characteristics of the unit during operation, and particularly cannot provide a more intuitive auxiliary judgment mode such as on-site listening for the monitoring staff when the unit is in a critical edge of a fault and a non-fault, and cannot meet the monitoring requirement on the water turbine unit operation state.
Disclosure of Invention
Aiming at the defects in the prior art, the remote hearing monitoring method, the remote hearing monitoring device and the remote hearing monitoring system for the hydraulic turbine set solve the problem that the prior art cannot monitor or judge the sound characteristics of the hydraulic turbine set during operation, and utilize an audio analysis technology to analyze and process the field sound on line so as to realize remote hearing of the hydraulic turbine set and meet the monitoring requirement on the running state of the hydraulic turbine set.
In a first aspect, the present invention provides a remote listening monitoring method for a hydraulic turbine set, including: acquiring an original audio signal with preset duration; carrying out data noise reduction on the original audio signal to obtain a target audio signal; extracting the characteristics of the target audio signal to obtain a target characteristic value; and inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal.
Optionally, performing data noise reduction on the original audio signal to obtain a target audio signal, including: acquiring a first eigenmode component corresponding to the original audio signal according to the original audio signal; acquiring a residual signal according to the first eigenmode component; obtaining all eigenmode components corresponding to the original audio signal according to the residual signal; and summing all eigenmode components to obtain the target audio signal.
Optionally, acquiring, according to the original audio signal, a first eigenmode component corresponding to the original audio signal includes: fitting the maximum value point and the minimum value point of the original audio signal to obtain an upper envelope line and a lower envelope line of the original audio signal; calculating the residual component of the current signal according to the upper envelope line and the lower envelope line; judging whether the residual component meets a preset constraint condition or not; when the residual component meets the preset constraint condition, taking the residual component as the first eigenmode component; and when the residual component does not meet the preset constraint condition, taking the residual component of the current signal as an original audio signal of the next fitting.
Optionally, performing feature extraction on the target audio signal to obtain a target feature value, including: performing time domain analysis on the target audio signal to obtain a time domain characteristic value; carrying out frequency domain analysis on the target audio signal to obtain a frequency domain characteristic value; and taking the time domain characteristic value and the frequency domain characteristic value as the target characteristic value.
Optionally, before the obtaining of the original audio signal of the preset duration, the method further includes: determining the number of hidden layers of the training model and the number of network nodes of each layer according to the input and output data; initializing model parameters in a training model, wherein the model parameters comprise a weight matrix and an offset vector; inputting the training data into the training model for iterative training, and calculating gradient values of all hidden layers; adjusting the weight matrix and the offset vector according to the gradient value of each hidden layer; and if the current iteration times reach the preset times, taking the adjusted weight matrix and the offset vector as target parameters of the training model to obtain the target fault model.
Optionally, inputting the training data into the training model for iterative training, and calculating gradient values of the hidden layers, including: calculating the output data of each hidden layer and the output data of the output layer according to the weight matrix, the offset vector and the activation function; acquiring a loss function according to the output data of each hidden layer and the output data of the output layer; and calculating the gradient value of each hidden layer according to the loss function.
Optionally, the formula expression of the weight matrix is: wl＝Wl-ηδl(Hl-1)T(ii) a The formula expression of the offset vector is as follows: bl＝bl-ηδl(ii) a Wherein η represents a learning rate, Wl、bl、δlRespectively representing the weight matrix, the offset vector and the gradient value of the hidden layer of the l-th layer.
Optionally, after performing feature extraction on the target audio signal to obtain a target feature value, the method further includes: comparing the target characteristic value with a preset threshold value, and judging whether the target characteristic value is greater than the preset threshold value; and outputting alarm information when the target characteristic value is larger than the preset threshold value.
In a second aspect, the present invention provides a remote listening monitoring device for a hydraulic turbine, the monitoring device comprising: the acquisition module is used for acquiring an original audio signal with preset duration; the noise reduction module is used for carrying out data noise reduction on the original audio signal to obtain a target audio signal; the extraction module is used for extracting the characteristics of the target audio signal to obtain a target characteristic value; and the checking module is used for inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal.
In a third aspect, the invention provides a remote listening monitoring system for a hydraulic turbine set, wherein the monitoring system comprises the monitoring device.
Compared with the prior art, the invention has the following beneficial effects:
according to the invention, the detection result of the original audio signal is obtained by carrying out noise reduction, spectrum analysis, feature extraction and fault detection on the acquired original audio signal, so that remote listening of the hydraulic turbine unit equipment is realized, the auxiliary judgment of on-site listening is provided for monitoring personnel, the functions of on-line monitoring, performance analysis, trend analysis, abnormity early warning and the like on the main equipment of the hydraulic turbine unit can be realized according to the detection result of the audio signal, the problem that the sound feature of the hydraulic turbine unit during operation cannot be monitored or judged in the prior art is solved, and the monitoring requirement on the operation state of the hydraulic turbine unit is met.
Drawings
Fig. 1 is a schematic flow chart of a remote listening monitoring method for a hydraulic turbine set according to an embodiment of the present invention;
fig. 2 is a schematic flowchart illustrating step S102 in fig. 1 according to an embodiment of the present invention;
fig. 3 is a schematic flow chart of another remote-listening monitoring method for a hydraulic turbine set according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a remote listening monitoring system for a hydraulic turbine set according to an embodiment of the present invention;
fig. 5 is a schematic working diagram of a remote-listening monitoring system for a hydraulic turbine set according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Fig. 1 is a schematic flow chart of a remote listening monitoring method for a hydraulic turbine set according to an embodiment of the present invention; as shown in fig. 1, the remote listening monitoring method for the hydraulic turbine set specifically includes the following steps:
step S101, obtaining an original audio signal with preset duration.
In this embodiment, pickup equipment is deployed in a hydraulic turbine set workshop and near the set to collect audio data of the set in operation under different working conditions, and the data equipment collected by the pickup equipment is uniformly stored and managed by an audio collection server, wherein the original audio signal comprises audio data with preset duration, the preset duration can be 100 milliseconds or 1 second, and the audio data can be modified according to an actual operation environment.
And S102, performing data noise reduction on the original audio signal to obtain a target audio signal.
Specifically, as shown in fig. 2, performing data noise reduction on the original audio signal to obtain a target audio signal specifically includes the following steps:
step S201, acquiring a first eigenmode component corresponding to the original audio signal according to the original audio signal;
step S202, obtaining a residual signal according to the first eigenmode component;
step S203, obtaining all eigen mode components corresponding to the original audio signal according to the residual signal;
step S204, summing all eigenmode components to obtain the target audio signal.
Further, acquiring a first eigenmode component corresponding to the original audio signal according to the original audio signal includes: fitting the maximum value point and the minimum value point of the original audio signal to obtain an upper envelope line and a lower envelope line of the original audio signal; calculating the residual component of the current signal according to the upper envelope line and the lower envelope line; judging whether the residual component meets a preset constraint condition or not; when the residual component meets the preset constraint condition, taking the residual component as the first eigenmode component; and when the residual component does not meet the preset constraint condition, taking the residual component of the current signal as an original audio signal of the next fitting.
It should be noted that, the invention uses an Empirical Mode Decomposition (EMD) algorithm to perform noise reduction processing on the audio signal, and decomposes the signal from the scale feature of the signal sequence itself to obtain fluctuation components representing different time scales implicit in the original load data, i.e. eigen Mode components (IMF), so as to weaken the non-stationary characteristic of the original signal; the residual items after decomposition represent the basic trend of the load sequence.
If the original audio signal is x (t), and t is a time interval with a preset duration, the decomposition steps of the EMD are summarized as follows:
the method comprises the following steps: respectively fitting the local maximum value point and the local minimum value point of x (t) by utilizing a cubic spline difference method to obtain an upper envelope line Us(t) and the upper and lower envelope lines Vs(t)。
Step two: calculate Us(t) and VsEnvelope mean m of (t)1And the remaining component h of the current signal1Namely:
h1＝x(t)-m1
step three: if h1If IMF component constraint condition is satisfied, then h1Is the first IMF component, i.e., the first eigenmode component; otherwise, for h1Repeating the first step to the second step until h1Satisfy the constraint condition, note imf1＝h1。
Step four: computing a residual signal r1＝x(t)-h1And repeating the first step to the third step until n IMF components are decomposed.
Step five: after decomposing n IMF components, the residual signal residual of the original signal is: r isn＝rn-1-imfn；
Accordingly, the original load sequence and the IMF component and the signal residual satisfy the following conditions:
as such, the target audio signal is:
and step S103, performing feature extraction on the target audio signal to obtain a target feature value.
Specifically, the feature extraction of the target audio signal to obtain a target feature value includes: performing time domain analysis on the target audio signal to obtain a time domain characteristic value; carrying out frequency domain analysis on the target audio signal to obtain a frequency domain characteristic value; and taking the time domain characteristic value and the frequency domain characteristic value as the target characteristic value.
It should be noted that, time domain analysis is performed on the acquired audio signal, feature information is extracted from the variation trend of each parameter along with time, and the sound symptoms of the corresponding fault category are represented, and the time domain feature value of the target audio signal can be obtained by any method in the following embodiments;
single peak method: representing the unimodal maximum of the signal, which is expressed as x if n is the corresponding number of framesp，
xp＝max(|x′n(t)|)
The root amplitude method: the signal frequency is reflected as a whole, and if n is the corresponding frame number, the square root amplitude is represented as xr，
Zero crossing rate method: the number of zero-crossing points of the signal, N is the length of the frame, N is the corresponding number of frames, and the zero-crossing rate is represented as Zn；
Short-time energy method: embodying signals at different timesN is the length of the frame, N is the corresponding frame number, and the short-time energy is expressed as En；
Short-time average amplitude difference method: the periodic characteristics can be better observed by using the short-time average amplitude difference, and if x is the signal after windowing and truncation, the short-time average amplitude difference is defined as rn(k)；
It should be further explained that the frequency domain analysis is performed on the collected audio signal, the characteristic information is extracted from the frequency domain, and the adopted frequency domain analysis means includes fourier transform and discrete fourier transform.
Fourier transform: the fourier transform of the aperiodic continuous-time signal x (t) can be expressed as follows:
discrete Fourier transform: finite length discrete signal x (t), N is 0, 1, …, N-1, then discrete fourier of audio signal
The transform is represented as follows:
wherein the content of the first and second substances,
and step S104, inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal.
It should be noted that, in this embodiment, when the abnormal sound feature is found, the system automatically pushes the warning information. System users can select appointed sound measuring points to inquire historical alarm information; defining fault rules as data-driven based diagnostic alarms: for deep fault identification and diagnosis, inputting the time sequence characteristics and the frequency domain characteristics into a neural network model, and training the model; and (3) collecting audio data on line, inputting the audio data into a network model for identification after feature extraction, and giving an alarm by the system when the identification result belongs to a specific fault type.
Compared with the prior art, the invention has the following beneficial effects:
according to the invention, the detection result of the original audio signal is obtained by carrying out noise reduction, spectrum analysis, feature extraction and fault detection on the acquired original audio signal, so that remote listening of the hydraulic turbine unit equipment is realized, the auxiliary judgment of on-site listening is provided for monitoring personnel, the functions of on-line monitoring, performance analysis, trend analysis, abnormity early warning and the like on the main equipment of the hydraulic turbine unit can be realized according to the detection result of the audio signal, the problem that the sound feature of the hydraulic turbine unit during operation cannot be monitored or judged in the prior art is solved, and the monitoring requirement on the operation state of the hydraulic turbine unit is met.
Fig. 3 is a schematic flow chart of another remote-listening monitoring method for a hydraulic turbine set according to an embodiment of the present invention; as shown in fig. 3, before the obtaining of the original audio signal with the preset duration, the method provided by the present invention further includes the following steps:
step S301, initializing model parameters in the training model, wherein the model parameters comprise a weight matrix and an offset vector;
step S302, initializing model parameters in the training model, wherein the model parameters comprise a weight matrix and an offset vector;
step S303, inputting the training data into the training model for iterative training, and calculating gradient values of all hidden layers;
step S304, adjusting the weight matrix and the offset vector according to the gradient value of each hidden layer;
step S305, if the current iteration frequency reaches the preset frequency, the adjusted weight matrix and the offset vector are used as target parameters of the training model, and the target fault model is obtained.
Further, inputting the training data into the training model for iterative training, and calculating gradient values of all hidden layers, including: calculating the output data of each hidden layer and the output data of the output layer according to the weight matrix, the offset vector and the activation function; acquiring a loss function according to the output data of each hidden layer and the output data of the output layer; and calculating the gradient value of each hidden layer according to the loss function.
In this embodiment, the formula expression of the weight matrix is: wl＝Wl-ηδl(Hl-1)T(ii) a The formula expression of the offset vector is as follows: bl＝bl-ηδl(ii) a Wherein η represents a learning rate, Wl、bl、δlRespectively representing the weight matrix, the offset vector and the gradient value of the hidden layer of the l-th layer.
It should be noted that, on the basis of audio analysis, the system provides training and testing tools based on models, provides algorithms such as neural networks, and the like, and a user can customize model parameters and perform deep training and identification on audio feature vectors, and the training process of the target fault model is as follows:
the method comprises the following steps: determining the hidden layer number L of the network model and the network node number p of each layer according to training input and output datal(L ═ 1,2, …, L), weight W between hidden and output layers is initialized randomlylAnd an offset vector bl(L ═ 2,3, …, L), learning rate η, iteration threshold ε, neuron activation function f (·).
Step two: computing hidden layer output and output Yl；
Hl＝f(zl)＝f(WlYt-1+bl)l＝2,3,…,L
Typical activation functions include a ReLU function and a tanh function, and the expressions are as follows:
f(x)＝max(0,x)
f(x)＝(ex-e-x)/(ex+e-x)
step three: calculating the gradient delta of the output layer from the loss functionLThe loss function and the gradient calculation expression are respectively:
step four: calculating the gradient delta of each hidden layerlThe expression of (a) is:
wherein L is L-1, L-2, …, 2.
Step five: according to the network output and the expected value, adjusting the weight matrix W of each hidden layer and output layerlAnd an offset vector blWeight matrix WlAnd an offset vector blThe formula expressions of (a) are respectively:
Wl＝Wl-ηδl(Hl-1)T
bl＝bl-ηδll＝2,3,…,L
step six: judging whether the training process is finished or not according to the error threshold epsilon or the upper limit of the iteration times; if not, go to step two.
In another embodiment of the present invention, after performing feature extraction on the target audio signal to obtain a target feature value, the method further includes: comparing the target characteristic value with a preset threshold value, and judging whether the target characteristic value is greater than the preset threshold value; and outputting alarm information when the target characteristic value is larger than the preset threshold value.
Specifically, the present embodiment is beneficial to the parameter out-of-limit alarm of the threshold, the present invention extracts the time domain and frequency domain features of the audio signal by using a time domain analysis method and a frequency domain analysis method, and when the feature value is greater than the set threshold, the system sends out alarm information.
In an embodiment of the present invention, the present invention provides a remote listening monitoring device for a hydraulic turbine, where the monitoring device includes: the acquisition module is used for acquiring an original audio signal with preset duration; the noise reduction module is used for carrying out data noise reduction on the original audio signal to obtain a target audio signal; the extraction module is used for extracting the characteristics of the target audio signal to obtain a target characteristic value; and the checking module is used for inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal.
In an embodiment of the present invention, the present invention provides a remote listening monitoring system for a hydraulic turbine set, where the detection system includes the monitoring device described above.
Fig. 4 is a schematic structural diagram of a remote listening monitoring system for a hydraulic turbine set according to an embodiment of the present invention; the overall architecture of the remote listening monitoring system for the hydraulic turbine set provided by the embodiment comprises a data source, infrastructure and an application platform, which are shown in fig. 4. Wherein, 1) data source: the system comprises various front-end pickup equipment which is arranged in a hydraulic turbine set workshop and near the set, and is used for acquiring audio signals of the equipment under different operating conditions; 2) infrastructure: and providing a data acquisition server, a data analysis and calculation server and a data display server. The data acquisition server provides a uniform data acquisition interface and converges audio signals of all the pickup equipment at the front end; the data analysis computing service provides an intelligent algorithm for processing and analyzing the audio signals, and the intelligent algorithm comprises analysis components such as signal noise reduction, spectrum analysis, feature extraction, a neural network model and the like; and the data display server faces the user service end, provides uniform access and visual presentation of the system and constructs an open data application development platform. 3) An application platform: under the support of software and hardware of an infrastructure layer, software functions of data storage management, data acquisition management, signal processing, data modeling, unified access, visual presentation and the like are realized.
The system mainly comprises front-end audio signal acquisition, signal analysis and processing and system management in the design process, and the system process is shown in figure 5:
(1) data acquisition: sound pickup equipment is arranged in a hydraulic turbine set workshop and near the set to collect audio data of the set in operation under different working conditions, and the data equipment collected by the sound pickup equipment is uniformly stored and managed by an audio collection server.
(2) And (3) data analysis: a user calls an audio analysis platform, audio data collected by different units and different pickup devices are checked, models such as typical device audio noise reduction, feature extraction, fault diagnosis models and deterioration analysis models are screened and established according to the device operation and maintenance requirements, and potential fault modes are mined. Meanwhile, the field personnel can play history and online real-time collected sound through the sound playing function of the platform, so that remote listening to the equipment is realized.
(3) And (3) system management: the visual operation platform is provided for users, and functions of online monitoring, performance analysis, trend analysis, abnormity early warning and the like of main equipment of the hydroelectric generating set are realized through a remote listening and audio online analysis assembly.
The remote listening monitoring system for the hydraulic turbine set comprises four functional modules, namely a data audio data acquisition module, an audio data management module, an audio data analysis module and a fault mode diagnosis module.
(1) Data acquisition: the system collects the field sound data of the unit equipment under different working conditions in real time through the data collection server, and stores the field sound data in a warehouse in real time, so that complete and detailed field audio data are provided for operation and maintenance personnel for analysis and use under the condition of normal or fault of the unit.
(2) Data management: the audio data of the platform adopts a compression storage technology, the storage interval is uninterrupted storage, all original data in the last 3 months and the original data compressed in the last 5 years are met, and all historical information of the unit is guaranteed not to be lost to the maximum extent. The storage capacity monitoring function is provided, when the capacity is insufficient, the system automatically sends out alarm information, and meanwhile, system information is pushed to a manager. The database has an automatic retrieval function, and a user can quickly obtain audio data meeting conditions by selecting a unit, pickup equipment and a time period.
(3) Audio playing: the system plays the historical and real-time online collected sound signals according to the sound measuring points and the time periods. Meanwhile, the sound in the selected frequency spectrum range is played from the angle of the frequency spectrogram, a remote listening function is provided for system management personnel, and the operation condition of the unit is judged in an auxiliary manner.
(4) Audio analysis: the platform provides noise reduction, a time domain feature extraction algorithm and a frequency domain feature extraction algorithm of the audio signal, enhances voice by eliminating noise and signal distortion, converts the corresponding voice signal from a time domain to a frequency domain, and provides an input feature vector for subsequent analysis and fault mode identification. And performing visual presentation on the key indexes by using a spectrogram and a trend graph as the result of the audio analysis.
The embodiment provides a hydraulic turbine set remote listening monitoring system, which provides a mature analysis method for fault diagnosis of the hydraulic turbine set from the aspect of audio analysis, and meanwhile, the system can play real sound under the operation state of the on-site unit in real time, provides a remote listening function, helps operation and maintenance managers to assist in judging the operation state of the unit from the aspect of audio analysis, and expands the functions of remote measurement, remote signaling, remote regulation, remote control and remote viewing of the existing unit monitoring. Meanwhile, fault indexing and fault early warning analysis are adopted for the audio data, so that the health management capability of the unit is improved, and the operation reliability of the unit is improved.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It is noted that, in this document, relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "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 phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. A remote hearing monitoring method for a hydraulic turbine unit is characterized by comprising the following steps:
acquiring an original audio signal with preset duration;
carrying out data noise reduction on the original audio signal to obtain a target audio signal;
extracting the characteristics of the target audio signal to obtain a target characteristic value;
and inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal.
2. The remote hearing monitoring method for a hydraulic turbine set according to claim 1, wherein the step of performing data noise reduction on the original audio signal to obtain a target audio signal comprises:
acquiring a first eigenmode component corresponding to the original audio signal according to the original audio signal;
acquiring a residual signal according to the first eigenmode component;
obtaining all eigenmode components corresponding to the original audio signal according to the residual signal;
and summing all eigenmode components to obtain the target audio signal.
3. The remote hearing monitoring method for the hydraulic turbine set according to claim 2, wherein the obtaining a first eigenmode component corresponding to the original audio signal according to the original audio signal comprises:
fitting the maximum value point and the minimum value point of the original audio signal to obtain an upper envelope line and a lower envelope line of the original audio signal;
calculating the residual component of the current signal according to the upper envelope line and the lower envelope line;
judging whether the residual component meets a preset constraint condition or not;
when the residual component meets the preset constraint condition, taking the residual component as the first eigenmode component;
and when the residual component does not meet the preset constraint condition, taking the residual component of the current signal as an original audio signal of the next fitting.
4. The remote hearing monitoring method for the hydraulic turbine set according to claim 1, wherein the step of performing feature extraction on the target audio signal to obtain a target feature value comprises:
performing time domain analysis on the target audio signal to obtain a time domain characteristic value;
carrying out frequency domain analysis on the target audio signal to obtain a frequency domain characteristic value;
and taking the time domain characteristic value and the frequency domain characteristic value as the target characteristic value.
5. The method of claim 1, wherein prior to said obtaining the original audio signal for the predetermined duration, the method further comprises:
determining the number of hidden layers of the training model and the number of network nodes of each layer according to the input and output data;
initializing model parameters in a training model, wherein the model parameters comprise a weight matrix and an offset vector;
inputting the training data into the training model for iterative training, and calculating gradient values of all hidden layers;
adjusting the weight matrix and the offset vector according to the gradient value of each hidden layer;
and if the current iteration times reach the preset times, taking the adjusted weight matrix and the offset vector as target parameters of the training model to obtain the target fault model.
6. The remote hearing monitoring method for a hydraulic turbine set according to claim 5, wherein inputting the training data into the training model for iterative training to calculate gradient values of hidden layers comprises:
calculating the output data of each hidden layer and the output data of the output layer according to the weight matrix, the offset vector and the activation function;
acquiring a loss function according to the output data of each hidden layer and the output data of the output layer;
and calculating the gradient value of each hidden layer according to the loss function.
7. The remote hearing monitoring method for a hydro-turbine unit according to claim 6,
formulation of the weight matrixThe formula is as follows: wl＝Wl-ηδl(Hl-1)T；
The formula expression of the offset vector is as follows: bl＝bl-ηδl；
Wherein η represents a learning rate, Wl、bl、δlRespectively representing the weight matrix, the offset vector and the gradient value of the hidden layer of the l-th layer.
8. The method for remote monitoring of a hydraulic turbine set according to claim 7, wherein after the target audio signal is subjected to feature extraction to obtain a target feature value, the method further comprises:
comparing the target characteristic value with a preset threshold value, and judging whether the target characteristic value is greater than the preset threshold value;
and outputting alarm information when the target characteristic value is larger than the preset threshold value.
9. A remote listening monitoring device for a hydraulic turbine unit, the monitoring device comprising:
the acquisition module is used for acquiring an original audio signal with preset duration;
the noise reduction module is used for carrying out data noise reduction on the original audio signal to obtain a target audio signal;
the extraction module is used for extracting the characteristics of the target audio signal to obtain a target characteristic value;
and the checking module is used for inputting the target characteristic value into a target fault model for fault detection to obtain a detection result of the original audio signal.
10. A hydraulic turbine remote listening monitoring system comprising a monitoring device according to claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110447167.2A CN113470694A (en) | 2021-04-25 | 2021-04-25 | Remote listening monitoring method, device and system for hydraulic turbine set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110447167.2A CN113470694A (en) | 2021-04-25 | 2021-04-25 | Remote listening monitoring method, device and system for hydraulic turbine set |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113470694A true CN113470694A (en) | 2021-10-01 |
Family
ID=77870559
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110447167.2A Pending CN113470694A (en) | 2021-04-25 | 2021-04-25 | Remote listening monitoring method, device and system for hydraulic turbine set |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113470694A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113931781A (en) * | 2021-10-12 | 2022-01-14 | 西安热工研究院有限公司 | Noise fault source-based hydroelectric generating set abnormal working condition protection system |
| CN116403605A (en) * | 2023-06-08 | 2023-07-07 | 宁德时代新能源科技股份有限公司 | Equipment fault prediction method, stacker fault prediction method and related devices |
| CN116994609A (en) * | 2023-09-28 | 2023-11-03 | 苏州芯合半导体材料有限公司 | Data analysis method and system applied to intelligent production line |
| CN117116291A (en) * | 2023-08-22 | 2023-11-24 | 昆明理工大学 | Sound signal processing method of sand-containing water flow impulse turbine |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163427A (en) * | 2010-12-20 | 2011-08-24 | 北京邮电大学 | Method for detecting audio exceptional event based on environmental model |
| CN102270270A (en) * | 2011-04-28 | 2011-12-07 | 东北大学 | Remote medical auscultation and consultation system |
| CN106024015A (en) * | 2016-06-14 | 2016-10-12 | 上海航动科技有限公司 | Call center agent monitoring method and system |
| CN107305774A (en) * | 2016-04-22 | 2017-10-31 | 腾讯科技（深圳）有限公司 | Speech detection method and device |
| CN109357749A (en) * | 2018-09-04 | 2019-02-19 | 南京理工大学 | A kind of power equipment audio signal analysis method based on DNN algorithm |
| CN109472288A (en) * | 2018-10-08 | 2019-03-15 | 华中科技大学 | A kind of pump-storage generator vibration hybrid feature extraction and classification method |
| CN111337277A (en) * | 2020-02-21 | 2020-06-26 | 云知声智能科技股份有限公司 | Household appliance fault determination method and device based on voice recognition |
| CN111444988A (en) * | 2020-05-11 | 2020-07-24 | 北华大学 | Rolling bearing fault diagnosis system |
| CN111783531A (en) * | 2020-05-27 | 2020-10-16 | 福建亿华源能源管理有限公司 | Water turbine set fault diagnosis method based on SDAE-IELM |
| CN112017639A (en) * | 2020-09-10 | 2020-12-01 | 歌尔科技有限公司 | Voice signal detection method, terminal device and storage medium |
| CN112193959A (en) * | 2020-09-25 | 2021-01-08 | 浙江新再灵科技股份有限公司 | Method and system for detecting abnormal sound of elevator |
| CN112700793A (en) * | 2020-12-24 | 2021-04-23 | 国网福建省电力有限公司 | Method and system for identifying fault collision of water turbine |
-
2021
- 2021-04-25 CN CN202110447167.2A patent/CN113470694A/en active Pending
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163427A (en) * | 2010-12-20 | 2011-08-24 | 北京邮电大学 | Method for detecting audio exceptional event based on environmental model |
| CN102270270A (en) * | 2011-04-28 | 2011-12-07 | 东北大学 | Remote medical auscultation and consultation system |
| CN107305774A (en) * | 2016-04-22 | 2017-10-31 | 腾讯科技（深圳）有限公司 | Speech detection method and device |
| CN106024015A (en) * | 2016-06-14 | 2016-10-12 | 上海航动科技有限公司 | Call center agent monitoring method and system |
| CN109357749A (en) * | 2018-09-04 | 2019-02-19 | 南京理工大学 | A kind of power equipment audio signal analysis method based on DNN algorithm |
| CN109472288A (en) * | 2018-10-08 | 2019-03-15 | 华中科技大学 | A kind of pump-storage generator vibration hybrid feature extraction and classification method |
| CN111337277A (en) * | 2020-02-21 | 2020-06-26 | 云知声智能科技股份有限公司 | Household appliance fault determination method and device based on voice recognition |
| CN111444988A (en) * | 2020-05-11 | 2020-07-24 | 北华大学 | Rolling bearing fault diagnosis system |
| CN111783531A (en) * | 2020-05-27 | 2020-10-16 | 福建亿华源能源管理有限公司 | Water turbine set fault diagnosis method based on SDAE-IELM |
| CN112017639A (en) * | 2020-09-10 | 2020-12-01 | 歌尔科技有限公司 | Voice signal detection method, terminal device and storage medium |
| CN112193959A (en) * | 2020-09-25 | 2021-01-08 | 浙江新再灵科技股份有限公司 | Method and system for detecting abnormal sound of elevator |
| CN112700793A (en) * | 2020-12-24 | 2021-04-23 | 国网福建省电力有限公司 | Method and system for identifying fault collision of water turbine |
Non-Patent Citations (1)
| Title |
|---|
| 程军圣: """基于EMD和分形维数的转子系统故障诊断""", 《中国机械工程》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113931781A (en) * | 2021-10-12 | 2022-01-14 | 西安热工研究院有限公司 | Noise fault source-based hydroelectric generating set abnormal working condition protection system |
| CN116403605A (en) * | 2023-06-08 | 2023-07-07 | 宁德时代新能源科技股份有限公司 | Equipment fault prediction method, stacker fault prediction method and related devices |
| CN116403605B (en) * | 2023-06-08 | 2024-06-07 | 宁德时代新能源科技股份有限公司 | Stacker fault prediction method and related device |
| CN117116291A (en) * | 2023-08-22 | 2023-11-24 | 昆明理工大学 | Sound signal processing method of sand-containing water flow impulse turbine |
| CN116994609A (en) * | 2023-09-28 | 2023-11-03 | 苏州芯合半导体材料有限公司 | Data analysis method and system applied to intelligent production line |
| CN116994609B (en) * | 2023-09-28 | 2023-12-01 | 苏州芯合半导体材料有限公司 | Data analysis method and system applied to intelligent production line |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113470694A (en) | Remote listening monitoring method, device and system for hydraulic turbine set | |
| CN111638028B (en) | High-voltage parallel reactor mechanical state evaluation method based on vibration characteristics | |
| CN111325095B (en) | Intelligent detection method and system for equipment health state based on acoustic wave signals | |
| US11069342B2 (en) | Method for training voice data set, computer device, and computer-readable storage medium | |
| CN109697207A (en) | The abnormality monitoring method and system of time series data | |
| CN110763997A (en) | Early fault early warning method for synchronous motor stator | |
| CN112838946B (en) | Method for constructing intelligent sensing and early warning model based on communication network faults | |
| CN110688617B (en) | Fan vibration abnormity detection method and device | |
| CN116226646B (en) | Method, system, equipment and medium for predicting health state and residual life of bearing | |
| KR102149495B1 (en) | Optimization apparatus for training conditions of environmental prediction model and operating thereof | |
| CN112115643B (en) | Smart train service life non-invasive prediction method | |
| CN116610998A (en) | Switch cabinet fault diagnosis method and system based on multi-mode data fusion | |
| CN112836937A (en) | Flood disaster loss evaluation method based on entropy weight and BP neural network technology | |
| CN111931849A (en) | Hydroelectric generating set operation data trend early warning method | |
| CN117371207A (en) | Extra-high voltage converter valve state evaluation method, medium and system | |
| CN115587309A (en) | Method, device and equipment for extracting key features of short-circuit resistance of transformer | |
| Hu et al. | A hybrid model for predicting the degradation trend of hydropower units based on deep learning | |
| CN116741148A (en) | Voice recognition system based on digital twinning | |
| CN115376526A (en) | Power equipment fault detection method and system based on voiceprint recognition | |
| Jarwar et al. | NEAT: A resilient deep representational learning for fault detection using acoustic signals in IIoT environment | |
| CN114708885A (en) | Fan fault early warning method based on sound signals | |
| CN113988398A (en) | Wind turbine generator power prediction method and device, electronic equipment and storage medium | |
| CN117407675A (en) | Lightning arrester leakage current prediction method based on multi-variable reconstruction combined dynamic weight | |
| CN116840743A (en) | Power transformer fault processing method and device, electronic equipment and storage medium | |
| CN114157023B (en) | Distribution transformer early warning information acquisition method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211001 |