CN114483417B - Water leakage defect quick identification method for guide vanes of water turbine based on voiceprint identification - Google Patents
Water leakage defect quick identification method for guide vanes of water turbine based on voiceprint identification Download PDFInfo
- Publication number
- CN114483417B CN114483417B CN202210022804.6A CN202210022804A CN114483417B CN 114483417 B CN114483417 B CN 114483417B CN 202210022804 A CN202210022804 A CN 202210022804A CN 114483417 B CN114483417 B CN 114483417B
- Authority
- CN
- China
- Prior art keywords
- water leakage
- neural network
- network model
- voiceprint
- voiceprint information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- Examining Or Testing Airtightness (AREA)
Abstract
A quick recognition method for water leakage defects of guide vanes of a water turbine based on voiceprint recognition comprises the following steps: setting a plurality of voiceprint acquisition points, wherein the average value of the voiceprint information is x, and the water leakage amount corresponding to x is y; sample statistics were performed: taking N samples, wherein one water leakage amount yn corresponds to one voiceprint information average value xn, and N represents a non-zero natural number; collecting voiceprint information by using an information collecting and processing module, denoising the voiceprint information of the N samples, and obtaining voiceprint information of the N combined lattices; the voiceprint information of the N combination lattices is sent to an RNN neural network model for training; and obtaining the water leakage quantity grade relation corresponding to the multiple x value ranges. The method can continuously and rapidly identify the water leakage defect of the guide vane of the water turbine, greatly improve the efficiency, meet a plurality of application scenes with high time efficiency requirements of a hydropower plant, make reference for on-site operation mode adjustment, help the production site reasonably arrange the operation mode, and improve the reliability and stability of the unit.
Description
Technical Field
The invention belongs to the technical field of water turbine guide vane water leakage detection, and particularly relates to a method for rapidly identifying water turbine guide vane water leakage defects based on voiceprint identification.
Background
The sealing performance of the guide vane of the water turbine has great significance for safe and stable operation of the water turbine generator set, and is also an important index for evaluating the performance of the water turbine. If the water leakage defect exists in the guide vane of the water turbine, the guide vane seal of the unit can be gradually destroyed, if the water leakage amount is large, the gap cavitation is possibly generated at the water leakage position of the guide vane, and if the water leakage amount is serious, the guide vane parent metal at the position can be destroyed, so that the water leakage gap is further increased; if the water leakage amount is increased to a certain degree, the peristaltic phenomenon of the unit can occur, and the extreme situation can lead to the serious situation that the unit cannot be stopped normally. At present, the measuring method of the water leakage quantity of the movable guide vane of the water turbine mainly adopts the measuring methods such as an air vent method, an inclined shaft method, a throttle plate method and the like.
The prior method has the characteristics, for example, patent document CN201710305226.6 discloses a measuring method and a measuring device for rated guide vane water leakage of a water turbine, patent document CN201810053307.6 discloses a detecting method and a detecting system for guide vane water leakage of a water turbine generator set, and patent document CN201910437559.3 discloses a guide vane water leakage detecting system, but the common defects are that the testing procedure is relatively complex, the falling gate measuring time is relatively long, and the method is difficult to be applied to some scenes with high timeliness requirements. The characteristics of the traditional method enable the water leakage condition of the guide vane of the water turbine to be monitored often in a special period of time, the on-site maintenance period is easy to influence during maintenance, and after some net-related tests, especially after a load shedding test, how to quickly and accurately judge the water leakage condition of the guide vane is significant for evaluating maintenance quality and arranging a subsequent operation mode; in the daily operation process of the unit, if the influence of water leakage of the guide vanes on the unit is judged according to the current traditional measurement method, the operation mode of the unit may need to be changed, an overhaul test application is submitted, special time is arranged for measurement, and the modes have certain influence on the power generation amount and the reliability index of the unit, and even the system peak period or the power failure period cannot be completed. Therefore, the method for quickly identifying the water leakage defect condition of the guide vane of the hydroelectric generating set has strong practical guiding significance on safe and stable operation of the set and improvement of the reliability index of the set.
Disclosure of Invention
In view of the technical problems existing in the background art, the rapid identification method for the water leakage defect of the guide vane of the water turbine based on voiceprint identification can continuously and rapidly identify the water leakage defect of the guide vane of the water turbine, greatly improve the efficiency, meet a plurality of application scenes with high time efficiency requirements of a hydropower plant, reference on-site operation mode adjustment, help the reasonable arrangement of operation modes on production sites and improve the reliability and stability of units.
In order to solve the technical problems, the invention adopts the following technical scheme:
a method for rapidly identifying water leakage defects of guide vanes of a water turbine based on voiceprint identification comprises the following steps:
s1: setting a plurality of voiceprint acquisition points, wherein the voiceprint acquisition points are used for acquiring voiceprint information near a guide vane of the water turbine, and the voiceprint information comprises sound decibel size;
s2: the average value of the voiceprint information is x, and the water leakage amount corresponding to x is y;
s3: sample statistics were performed: taking N samples, wherein one water leakage amount yn corresponds to one voiceprint information average value xn, and N represents a non-zero natural number;
s4: collecting voiceprint information by using an information collecting and processing module, denoising the voiceprint information of the N samples, and obtaining voiceprint information of the N combined lattices;
s5: the voiceprint information of the N combination lattices is sent to an RNN neural network model for training;
s6: and obtaining the water leakage quantity grade relation corresponding to the multiple x value ranges.
Preferably, in step S5, the method for training the RNN neural network model is as follows:
the RNN neural network model comprises an input layer x, a hidden layer h and an output layer o, wherein the hidden layer h is provided with a circulation operation, and meanwhile, linear relation parameters U, W, V of the RNN neural network model at all moments are shared; where U represents the weight of the sample input at the moment, W represents the weight of the input, and V represents the weight of the sample output;
the calculation formulas of the hidden state, the output layer state and the final predicted output in the RNN neural network model are as follows:
h t ＝f(Ux t +Wh t-1 +b)
o t ＝Vh t +c
y t ＝g(o t )
wherein x is t Representing the input of training samples at sequence index t, and accordingly x t-1 And x t+1 Representing the input of training samples at sequence index numbers t-1 and t+1;
h t represents the hidden state of the model at the sequence index t, h t From x t And h t-1 Determining together;
o t represents the output of the model at the sequence index t, o t The hidden state h is only present by the RNN neural network model t Determining;
f and g are both activation functions, b and c represent bias values, and U, W and V are both parameters;
and training the N groups of qualified voiceprint information for N times according to the RNN neural network model.
Preferably, the RNN neural network model updates the gradient using a back propagation algorithm over time, the partial derivative of V, W, U is calculated as follows:
the above formula is used to calculate the partial derivative of V, W, U.
Preferably, the voiceprint information is a value of sound decibel size, after the voiceprint information of the N combination lattice is sent to the RNN neural network model for training, the corresponding relation between the water leakage y and the sound decibel is obtained, and the water leakage y is divided into four grades according to the corresponding relation, namely, the y comprises four grades of no water leakage, tiny water leakage, small water leakage and large water leakage; the x values corresponding to the four classes are as follows:
A. x is less than or equal to 82, the judgment result is that water is not leaked, and the quantitative index of the RNN neural network model is y=0;
B. 82 < x is less than or equal to 87.3, and the judgment result is that the water leakage is tiny, and at the moment, the quantitative index of the RNN neural network model is more than 0 and less than or equal to y and less than 3;
C. 87.3 < x is less than or equal to 88.9, and the judgment result is a small amount of water leakage, and at the moment, the quantitative index of the RNN neural network model is 3 < y is less than or equal to 6;
D. and x is more than 88.9, and the judgment result is that a large amount of water is leaked, and the quantitative index of the RNN neural network model is y more than 6.
The following beneficial effects can be achieved in this patent:
1. the invention relates to a method for building a water turbine fault recognition model based on voiceprint recognition, which is characterized in that a proper algorithm model can be built only by a large number of training experiments, and aims to build the water leakage quantity of different water leakage defects of a water turbine, and different fault categories are obtained through model training, so that the water leakage defects and the fault degree of the water turbine can be recognized more quickly and accurately, and the method is divided into four stages: the method has the advantages that no water leakage, tiny water leakage, small water leakage and large water leakage are realized, targeted measures are timely taken, the water leakage prevention effect is optimal, the normal operation of the unit is not influenced by the tiny water leakage, the observation is enhanced when the small water leakage is realized, the measures are needed to be taken as soon as possible when the large water leakage is realized, the defect expansion and the misjudgment are avoided, the operation of the unit is influenced, and the operation mode and the maintenance plan of the unit are reasonably arranged and adjusted.
2. By adopting the voiceprint recognition method, the water leakage defect of the guide vane of the water turbine can be continuously and rapidly recognized, the efficiency is greatly improved, the method can meet the application scenes with high time efficiency requirements of a hydropower plant, such as after maintenance test, during the large-scale period of a flood season unit and the like, and can be used for making reference for on-site operation mode adjustment, helping the reasonable arrangement of operation modes on production sites and improving the reliability and stability of the unit.
3. The data processing method adopts an RNN neural network model which is good at processing sequence data, and has the advantages that the network structure of the RNN neural network model can process the correlation between time sequence data, the RNN neural network model comprises an input layer x, a hidden layer h and an output layer o, the hidden layer h has a cyclic operation, and meanwhile, the linear relation parameters U, W, V of the RNN neural network model at all moments are shared, so that the parameter training quantity is greatly reduced.
Drawings
The invention is further illustrated by the following examples in conjunction with the accompanying drawings:
FIG. 1 is a flow chart of a method for quickly identifying water leakage defects of guide vanes of a water turbine by voiceprint identification;
FIG. 2 is a schematic diagram of a network architecture of an RNN neural network model according to the present invention;
FIG. 3 is a training flow chart of the RNN neural network model of the present invention;
FIG. 4 is a hidden layer structure of a long-term memory network according to the present invention;
FIG. 5 is a diagram of the voiceprint recognition system of the present invention.
Detailed Description
The preferable scheme is as shown in fig. 1 to 5, and the method for rapidly identifying the water leakage defect of the guide vane of the water turbine based on voiceprint identification comprises the following steps:
s1: setting a plurality of voiceprint acquisition points, wherein the voiceprint acquisition points are used for acquiring voiceprint information near a guide vane of the water turbine, and the voiceprint information comprises sound decibel size;
s2: the average value of the voiceprint information is x, and the water leakage amount corresponding to x is y;
s3: sample statistics were performed: taking N samples, wherein one water leakage amount yn corresponds to one voiceprint information average value xn, and N represents a non-zero natural number;
s4: collecting voiceprint information by using an information collecting and processing module, denoising the voiceprint information of the N samples, and obtaining voiceprint information of the N combined lattices; the information acquisition processing module is selected from SMART SENSOR AR824 (digital display program noise meter) for measuring sound decibel. The information acquisition and processing module comprises five parts, namely a voiceprint data acquisition module, a voiceprint preprocessing module, a characteristic extraction module, an identification module and a result output module. Voiceprint recognition is largely divided into three threads: sound collection subsystem, main thread module and discernment submodule. The sound collection sub-thread is mainly responsible for calling hardware to collect data, and then triggering the sub-function to return sound to the main thread. The main thread is responsible for the operating logic of the entire software program and the invocation of various tool functions. The recognition output module is mainly responsible for extracting sound features, feature matching and recognition results.
S5: the voiceprint information of the N combination lattices is sent to an RNN neural network model for training;
in step S5, the method for training the RNN neural network model includes:
the RNN neural network model comprises an input layer x, a hidden layer h and an output layer o, wherein the hidden layer h is provided with a circulation operation, and meanwhile, linear relation parameters U, W, V of the RNN neural network model at all moments are shared; where U represents the weight of the sample input at the moment, W represents the weight of the input, and V represents the weight of the sample output; as shown in fig. 2, the RNN neural network model expansion structure diagram is shown, from which the dependency relationship between hidden layers can be more clearly seen by the RNN neural network model through the weight W.
The calculation formulas of the hidden state, the output layer state and the final predicted output in the RNN neural network model are as follows:
h t ＝f(Ux t +Wh t-1 +b)
o t ＝Vh t +c
y t ＝g(o t )
wherein x is t Representing the input of training samples at sequence index t, and accordingly x t-1 And x t+1 Representing the input of training samples at sequence index numbers t-1 and t+1;
h t representative ofHidden state of model at sequence index t, h t From x t And h t-1 Determining together;
o t represents the output of the model at the sequence index t, o t The hidden state h is only present by the RNN neural network model t Determining;
f and g are both activation functions, b and c represent bias values, and U, W and V are both parameters; f may be tanh, sigmoid, relu, etc., g is typically softmax.
And training the N groups of qualified voiceprint information for N times according to the RNN neural network model.
The RNN neural network model updates the gradient using a back-propagation algorithm over time (back-propagation through time), and the partial derivative of V, W, U is calculated as follows:
the formula is used for calculating the partial derivative of V, W, U, and the newly added sign in the formula is the sign calculated by the RNN neural network model.
The RNN neural network model in the invention adopts a long-short-period memory network (English abbreviated LSTM) with the following reasons: the RNN neural network model has excellent effect of processing time series data, but the problem that the RNN neural network model is poor in long-term dependence caused by gradient disappearance and gradient explosion, namely the RNN neural network model only has short-term memory, and the long-term memory network is a variant structure of the RNN neural network model designed for solving a plurality of problems existing in the RNN neural network model. On the aspect of gradient disappearance, the long-term memory network has a special memory storage mode. For those "memories" with a large gradient, the long-short term memory network will not immediately clear it, but will leave it partly, which can overcome the problem of gradient disappearance to some extent. In the aspect of gradient explosion, the long-term and short-term memory network can judge whether the calculated gradient exceeds a set threshold value, and reset the exceeding part to the threshold value, so that the gradient is ensured to be controllable.
Long and short term memory networks are a special type of RNN neural network model that avoids long term dependency problems by redesigning the hidden layer, making remembering long term information a model feature. In the standard RNN neural network model, the hidden layer module has only one simple structure, namely an activation function. And the RNN neural network model receives and activates all information at the t-1 moment and the current t moment and then transmits the information to the t+1 moment. The long-term and short-term memory network has three gate control (gates) structures in the hidden layer structure, namely a forgetting gate (forget gate), an input gate (input gate) and an output gate (output gate), and a hidden State called a Cell State (Cell State) is added, as shown in fig. 4.
A new state information, namely cell state, which never appears in RNNs is added to the long-short-term memory network. The cell state is the core of long-term dependence of the long-term and short-term memory network, and penetrates through all hidden layers of the whole long-term and short-term memory network model, and only a small amount of linear interaction exists in each hidden layer, so that the information flow on the hidden layers is not easy to change, and the function of preserving long-term memory is really achieved.
The gating algorithm is also the key of LSTM for ensuring long-term dependence, and the gating structure can not only reserve long-distance effective information, but also selectively forget invalid information, so that the LSTM has the capability of controlling internal information accumulation. Each gating structure in the long-short term memory network has special action objects and action modes. For example, in text translation, correct association and translation are performed for different behaviors of different subjects, so that the prior subjects are required to be forgotten correctly by forgetting the gate to control the cell state, then new subjects are added to the cell state by matching with the input gate, and finally the current main-meaning relation is adjusted through the output gate. The cell state is interacted in the hidden layer for three times, wherein the first time is forgetting door control forgetting; the second time is the input gate control input; and thirdly, calling the cell state and the output of the output gate to jointly generate the hidden layer state at the current t moment. The basic principle of voiceprint recognition in the invention is that the personal characteristics carried along in the voice signal of the speaker are extracted through the understanding capability of a computer and then matched with training templates in a database according to a certain criterion, and the identity of the speaker is identified or confirmed. The whole process consists of front-end processing, feature extraction, model training, pattern matching and the like, and is shown in fig. 5.
Aiming at the problem of fault identification under the condition of short-time voiceprint and aliasing noise of the application scene of the water turbine guide vane water leakage defect, the invention provides a fault based on a voiceprint identification algorithm
S6: and obtaining the water leakage quantity grade relation corresponding to the multiple x value ranges.
The voiceprint information is a sound decibel value, and the leaking amount y is divided into four grades, namely, the y comprises four grades of no water leakage, tiny water leakage, small water leakage and large water leakage; the x values corresponding to the four classes are as follows:
A. x is less than or equal to 82, the judgment result is that water is not leaked, and the quantitative index of the RNN neural network model is y=0;
B. 82 < x is less than or equal to 87.3, and the judgment result is that the water leakage is tiny, and at the moment, the quantitative index of the RNN neural network model is more than 0 and less than or equal to y and less than 3;
C. 87.3 < x is less than or equal to 88.9, and the judgment result is a small amount of water leakage, and at the moment, the quantitative index of the RNN neural network model is 3 < y is less than or equal to 6;
D. and x is more than 88.9, and the judgment result is that a large amount of water is leaked, and the quantitative index of the RNN neural network model is y more than 6.
As shown in FIG. 3, the algorithm of the method is based on an end-to-end deep voiceprint model, and can realize the aims of increasing the inter-class distance and reducing the intra-class distance between features, thereby improving the discernability of the voiceprint recognition model under the influence of short-time voice and aliasing noise, and being less in parameter quantity and more suitable for being deployed on embedded hardware. The voiceprint recognition is based on a deep learning algorithm, comprises two stages of data input and data output, an algorithm model is built by collecting a large amount of guide vane water leakage fault data, the algorithm model is continuously trained and optimal parameters are found in a sound modeling stage, and after the algorithm model is determined, test set data are imported into the built model to obtain a recognition result.
The above embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and the scope of the present invention should be defined by the claims, including the equivalents of the technical features in the claims. I.e., equivalent replacement modifications within the scope of this invention are also within the scope of the invention.
Claims (1)
1. A method for rapidly identifying water leakage defects of guide vanes of a water turbine based on voiceprint identification is characterized by comprising the following steps:
s1: setting a plurality of voiceprint acquisition points, wherein the voiceprint acquisition points are used for acquiring voiceprint information near a guide vane of the water turbine, and the voiceprint information comprises sound decibel size;
s2: the average value of the voiceprint information is x, and the water leakage amount corresponding to x is y;
s3: sample statistics were performed: taking N samples, wherein one water leakage amount yn corresponds to one voiceprint information average value xn, and N represents a non-zero natural number;
s4: collecting voiceprint information by using an information collecting and processing module, denoising the voiceprint information of the N samples, and obtaining voiceprint information of the N combined lattices;
s5: the voiceprint information of the N combination lattices is sent to an RNN neural network model for training;
s6: obtaining the water leakage quantity grade relation corresponding to a plurality of x value ranges;
in step S5, the method for training the RNN neural network model includes:
the RNN neural network model comprises an input layer x, a hidden layer h and an output layer o, wherein the hidden layer h is provided with a circulation operation, and meanwhile, linear relation parameters U, W, V of the RNN neural network model at all moments are shared; where U represents the weight of the sample input at the moment, W represents the weight of the input, and V represents the weight of the sample output;
the calculation formulas of the hidden state, the output layer state and the final predicted output in the RNN neural network model are as follows:
h t ＝f(Ux t +Wh t-1 +b)
o t ＝Vh t +c
y t ＝g(o t )
wherein x is t Representing the input of training samples at sequence index t, and accordingly x t-1 And x t+1 Representing the input of training samples at sequence index numbers t-1 and t+1;
h t represents the hidden state of the model at the sequence index t, h t From x t And h t-1 Determining together;
o t represents the output of the model at the sequence index t, o t The hidden state h is only present by the RNN neural network model t Determining;
f and g are both activation functions, b and c represent bias values, and U, W and V are both parameters;
training N groups of qualified voiceprint information for N times according to the RNN neural network model;
the RNN neural network model updates the gradient using a back propagation algorithm over time, the partial derivative of V, W, U is calculated as follows:
the above formula is used to calculate the partial derivative of V, W, U;
the voiceprint information is a value of sound decibel, after the voiceprint information of the N combination lattices is sent to the RNN neural network model for training, the corresponding relation between the water leakage y and the sound decibel is obtained, and the water leakage y is divided into four grades according to the corresponding relation, namely, the y comprises four grades of no water leakage, tiny water leakage, small water leakage and large water leakage; the x values corresponding to the four classes are as follows:
A. x is less than or equal to 82, the judgment result is that water is not leaked, and the quantitative index of the RNN neural network model is y=0;
B. 82 < x is less than or equal to 87.3, and the judgment result is that the water leakage is tiny, and at the moment, the quantitative index of the RNN neural network model is more than 0 and less than or equal to y and less than 3;
C. 87.3 < x is less than or equal to 88.9, and the judgment result is a small amount of water leakage, and at the moment, the quantitative index of the RNN neural network model is 3 < y is less than or equal to 6;
D. and x is more than 88.9, and the judgment result is that a large amount of water is leaked, and the quantitative index of the RNN neural network model is y more than 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210022804.6A CN114483417B (en) | 2022-01-10 | 2022-01-10 | Water leakage defect quick identification method for guide vanes of water turbine based on voiceprint identification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210022804.6A CN114483417B (en) | 2022-01-10 | 2022-01-10 | Water leakage defect quick identification method for guide vanes of water turbine based on voiceprint identification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114483417A CN114483417A (en) | 2022-05-13 |
| CN114483417B true CN114483417B (en) | 2023-06-16 |
Family
ID=81510075
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210022804.6A Active CN114483417B (en) | 2022-01-10 | 2022-01-10 | Water leakage defect quick identification method for guide vanes of water turbine based on voiceprint identification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114483417B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109579220A (en) * | 2018-10-15 | 2019-04-05 | 平安科技（深圳）有限公司 | Air-conditioning system fault detection method, device, electronic equipment |
| CN111306008A (en) * | 2019-12-31 | 2020-06-19 | 远景智能国际私人投资有限公司 | Fan blade detection method, device, equipment and storage medium |
| CN111325095A (en) * | 2020-01-19 | 2020-06-23 | 西安科技大学 | Intelligent equipment health state detection method and system based on sound wave signals |
| CN111400961A (en) * | 2020-02-17 | 2020-07-10 | 通标标准技术服务有限公司 | Wind generating set blade fault judgment method and device |
| CN112686110A (en) * | 2020-12-23 | 2021-04-20 | 武汉理工大学 | Method, device and equipment for diagnosing faults of centrifugal pump and storage medium |
| 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 |
| CN113298134A (en) * | 2021-05-20 | 2021-08-24 | 华中科技大学 | BPNN-based remote non-contact health monitoring system and method for fan blade |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7313550B2 (en) * | 2002-03-27 | 2007-12-25 | Council Of Scientific & Industrial Research | Performance of artificial neural network models in the presence of instrumental noise and measurement errors |
-
2022
- 2022-01-10 CN CN202210022804.6A patent/CN114483417B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109579220A (en) * | 2018-10-15 | 2019-04-05 | 平安科技（深圳）有限公司 | Air-conditioning system fault detection method, device, electronic equipment |
| CN111306008A (en) * | 2019-12-31 | 2020-06-19 | 远景智能国际私人投资有限公司 | Fan blade detection method, device, equipment and storage medium |
| CN111325095A (en) * | 2020-01-19 | 2020-06-23 | 西安科技大学 | Intelligent equipment health state detection method and system based on sound wave signals |
| CN111400961A (en) * | 2020-02-17 | 2020-07-10 | 通标标准技术服务有限公司 | Wind generating set blade fault judgment method and device |
| CN112686110A (en) * | 2020-12-23 | 2021-04-20 | 武汉理工大学 | Method, device and equipment for diagnosing faults of centrifugal pump and storage medium |
| 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 |
| CN113298134A (en) * | 2021-05-20 | 2021-08-24 | 华中科技大学 | BPNN-based remote non-contact health monitoring system and method for fan blade |
Non-Patent Citations (1)
| Title |
|---|
| 基于LMD-PE与神经网络的刀具故障诊断方法;杨瑞元等;计算机测量与控制;第29卷(第6期);第25-28页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114483417A (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112149316B (en) | Aero-engine residual life prediction method based on improved CNN model | |
| CN111237134B (en) | Offshore double-fed wind driven generator fault diagnosis method based on GRA-LSTM-stacking model | |
| CN108363382B (en) | A kind of complex equipment fault diagnosis method and system | |
| CN111426950B (en) | Wind driven generator fault diagnosis method of multi-scale space-time convolution depth belief network | |
| CN109102032A (en) | A kind of pumping plant unit diagnostic method based on depth forest and oneself coding | |
| CN110362048A (en) | Blower critical component state monitoring method and device, storage medium and terminal | |
| CN111612651A (en) | Abnormal electric quantity data detection method based on long-term and short-term memory network | |
| CN109447441B (en) | Transient stability risk assessment method considering uncertainty of new energy unit | |
| CN112330165B (en) | Power grid transient stability evaluation method and system based on feature separation type neural network | |
| CN101976313A (en) | Frequent subgraph mining based abnormal intrusion detection method | |
| CN110595778B (en) | Wind turbine generator bearing fault diagnosis method based on MMF and IGRA | |
| CN109617526A (en) | A method of photovoltaic power generation array fault diagnosis and classification based on wavelet multiresolution analysis and SVM | |
| CN109255333A (en) | A kind of large-scale wind electricity unit rolling bearing fault Hybrid approaches of diagnosis | |
| CN114048688A (en) | Method for predicting service life of bearing of wind power generator | |
| Li et al. | Intelligent fault diagnosis of aeroengine sensors using improved pattern gradient spectrum entropy | |
| CN107036808A (en) | Gearbox of wind turbine combined failure diagnostic method based on SVMs probability Estimation | |
| CN114483417B (en) | Water leakage defect quick identification method for guide vanes of water turbine based on voiceprint identification | |
| CN116793666A (en) | Wind turbine generator system gearbox fault diagnosis method based on LSTM-MLP-LSGAN model | |
| Tao et al. | Wind turbine blade icing diagnosis using B-SMOTE-Bi-GRU and RFE combined with icing mechanism | |
| CN116244596A (en) | Industrial time sequence data anomaly detection method based on TCN and attention mechanism | |
| CN116106198A (en) | Dam seepage flow prediction method and device based on depth forest | |
| CN114046816B (en) | Sensor signal fault diagnosis method based on lightweight gradient lifting decision tree | |
| CN115965177A (en) | Improved autoregressive error compensation wind power prediction method based on attention mechanism | |
| CN115296298A (en) | Wind power plant power prediction method | |
| Zhenhao et al. | Prediction of wind power ramp events based on deep neural network |
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 |