CN110161119B - Wind power blade defect identification method - Google Patents
Wind power blade defect identification method Download PDFInfo
- Publication number
- CN110161119B CN110161119B CN201910493788.7A CN201910493788A CN110161119B CN 110161119 B CN110161119 B CN 110161119B CN 201910493788 A CN201910493788 A CN 201910493788A CN 110161119 B CN110161119 B CN 110161119B
- Authority
- CN
- China
- Prior art keywords
- defect
- ultrasonic detection
- wavelet packet
- wind power
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/04—Analysing solids
- G01N29/043—Analysing solids in the interior, e.g. by shear waves
-
- 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/023—Solids
Abstract
The invention discloses a wind power blade defect identification method, which is used for carrying out ultrasonic detection on a wind power blade, carrying out wavelet packet transformation on an obtained ultrasonic detection signal according to a frequency band characteristic window, inputting an obtained energy spectrum coefficient into a BP neural network as a characteristic vector, and outputting corresponding defect types by the neural network so as to realize automatic identification of different defects of the wind power blade. The defect identification method provided by the invention is effective and feasible, the automatic identification of the wind power blade defects becomes possible, and the average identification rate is up to 90%.
Description
Technical Field
The invention belongs to the technical field of intelligent detection, and particularly relates to a defect identification method for a wind power blade.
Background
The wind power blade is used as a core component of a wind generating set, has the characteristics of complex structure, various processes, special materials and the like, and inevitably has defects such as inclusion, glue deficiency, wrinkles and the like in the production process, and different defect types have different influence degrees on the rigidity and the strength of the blade, so that the defect type of the blade is very necessary to be identified in order to ensure the service life of the blade. Ultrasonic detection is widely applied to the field of composite materials such as wind power blades due to the advantages of strong penetrating power, convenience in use and the like, but is mainly used for detecting the existence and the positioning of bonding defects in the tail edge of the wind power blade at present, and for the identification of different defect modes in a wind power blade web plate and a crossbeam, the comprehensive judgment is carried out mainly depending on manual experience according to the processing technology, the material, the structure and the detection data of a detected workpiece; the method has the advantages of large workload, low efficiency and large influence of subjective factors of inspectors on the inspection level, and the reason is that for wind power blade materials, due to the effects of sound attenuation, scattering, reflection and the like of the composite material, ultrasonic signals of the blades are complex, hidden defect information is difficult to extract from a large number of complex ultrasonic signals, and the defect condition in the blades cannot be accurately evaluated.
Disclosure of Invention
In order to solve the technical problem, the invention provides a wind power blade defect identification method based on a frequency band characteristic window, which is used for automatically identifying defects in a wind power blade web or a crossbeam.
The invention provides a wind power blade defect identification method, which specifically comprises the following steps:
step a) carrying out ultrasonic detection on a sample containing defects, and selecting a db4 function with the maximum correlation with signal waveforms as a wavelet basis of wavelet packet decomposition according to a time domain diagram of ultrasonic detection signals;
b) carrying out frequency spectrum transformation on the ultrasonic detection signal obtained in the step a, and finding out a characteristic frequency band window by combining frequency spectrum analysis and wavelet packet decomposition so as to determine the optimal number m of layers of wavelet packet transformation;
c) performing m-layer db4 wavelet packet decomposition on the ultrasonic detection signal obtained in the step a, and obtaining decomposed nodes and energy spectrum coefficients thereof through the wavelet packet decomposition;
step d) constructing a BP neural network, taking the energy spectrum coefficient obtained in the step c as an input vector, taking the corresponding defect type as an output result, and training and verifying the neural network;
and e) carrying out ultrasonic detection on the wind power blade to be detected, inputting an ultrasonic detection signal into the neural network constructed in the step d, and automatically identifying the defect type in the wind power blade.
Preferably, the specific operation of step (a) is: and carrying out ultrasonic detection on the wind power blade sample containing the defect to obtain a corresponding defect ultrasonic detection signal, and observing the signal waveform in the time domain graph of the defect ultrasonic detection signal to select db4 with the maximum correlation as a wavelet basis for wavelet packet decomposition.
Preferably, the specific operation of step (b) is: b, performing frequency spectrum transformation on the defect ultrasonic detection signal provided in the step a, analyzing the frequency characteristics of the obtained frequency spectrum signal, performing k-layer wavelet packet decomposition on the frequency characteristics of the frequency spectrum signal, finding out a characteristic frequency band window (the characteristic frequency band window is a data window corresponding to different characteristic frequency ranges in data analysis software, and performing data analysis by using matlab), and determining the optimal number m of decomposed layers of the wavelet packets, so that each node after m layers of decomposition of the wavelet packets can effectively separate characteristic information in the frequency spectrum signal; wherein k is a natural number, k is not equal to 0, and m belongs to k.
Preferably, the specific operation of step (c) is: b, performing m-layer wavelet packet decomposition on the defect ultrasonic detection signal provided in the step a by using db4 wavelets, and extracting node energy spectrum coefficients corresponding to the nodes; selecting a node as a horizontal coordinate, and selecting an energy spectrum coefficient corresponding to the node as a vertical coordinate to obtain an energy spectrum coefficient graph; and distinguishing each defect type according to different node energies of different defects in the energy spectrum coefficient diagram, and verifying the correctness of the optimal decomposition layer number m.
Preferably, the specific operation of step (d) is: and constructing a BP neural network, wherein input nodes of the BP neural network are set as the number of the energy spectrum coefficients after m layers of wavelet packet decomposition, output nodes are set as the defect type number, and the mean square error of the output layer of the neural network reaches the minimum value through training and verification of the neural network.
Preferably, the specific operation of step (e) is: and d, carrying out ultrasonic detection on the wind power blade to be detected, inputting an energy spectrum coefficient obtained after carrying out m-layer db4 wavelet packet decomposition on an ultrasonic detection signal into the neural network constructed in the step d, and automatically identifying the defect type of the wind power blade through a computer.
It should be noted that when determining the optimal number of wavelet packet decomposition layers, the defect ultrasonic detection signal may have a frequency band aliasing phenomenon under a general wavelet packet analysis, and defect feature information cannot be distinguished, and the obtained wavelet packet decomposition coefficient cannot be used as a basis for distinguishing different defect types. And only when the m-th layer wavelet packet decomposition is carried out based on the frequency band characteristic window, different defect signal characteristic values have difference (characteristic frequency band separation of defect ultrasonic signals). That is, in this decomposition method, we find the characteristic parameters capable of distinguishing different defect types.
Compared with the prior art, the wind power blade defect identification method based on the frequency band characteristic window is characterized in that ultrasonic detection is carried out on the wind power blade, wavelet packet transformation is carried out on ultrasonic signals according to the frequency band characteristic window, and the obtained energy spectrum coefficient is input into a constructed BP neural network as defect frequency characteristics, so that intelligent identification of different defects in the wind power blade is achieved. And finding out a frequency band characteristic window by combining spectral analysis and wavelet packet decomposition, thereby realizing the separation of defect frequency characteristics. And extracting the spectral energy characteristics after wavelet packet decomposition, constructing a characteristic vector of a defect signal as an input vector, inputting the characteristic vector into the trained BP neural network, and outputting a corresponding defect type by the BP neural network to realize automatic identification of the wind power blade defect. The defect identification method provided by the invention is effective and feasible, avoids subjective errors of manual identification, makes automatic identification of the defects of the wind power blade possible, and has an average automatic identification rate as high as 90%.
Drawings
FIG. 1 is a defect prediction chart for a wind turbine blade;
FIG. 2a is a time domain diagram of an ultrasonic detection signal of a non-defective part of a wind turbine blade obtained in step a of the embodiment of the present invention;
FIG. 2b is a time domain diagram of an ultrasonic detection signal of a wind turbine blade inclusion defect part obtained in step a of the embodiment of the invention;
FIG. 2c is a time domain diagram of an ultrasonic detection signal of a wind turbine blade fold defect portion obtained in step a according to the embodiment of the invention;
FIG. 2d is a time domain diagram of an ultrasonic detection signal of a glue-lacking defect portion of the wind turbine blade obtained in step a according to the embodiment of the invention;
FIG. 3a is a frequency spectrum diagram of the glue deficiency defect of the wind turbine blade obtained in step b according to the embodiment of the present invention;
FIG. 3b is a lossless frequency spectrum diagram of the wind turbine blade obtained in step b according to the embodiment of the present invention;
FIG. 4a is an energy spectrum of an ultrasonic detection signal of a non-defective portion of a wind turbine blade obtained in step c according to an embodiment of the present invention;
FIG. 4b is an energy spectrum diagram of an ultrasonic detection signal of a wind turbine blade inclusion defect portion obtained in step c of the embodiment of the present invention;
FIG. 4c is an energy spectrum diagram of an ultrasonic detection signal of a wind turbine blade wrinkle defect portion obtained in step c of the embodiment of the present invention;
FIG. 4d is an energy spectrum diagram of an ultrasonic detection signal of a glue-lacking defect portion of the wind turbine blade obtained in step c of the embodiment of the present invention;
fig. 5 is a graph of the training result of the BP neural network provided in the embodiment of the present invention.
Detailed Description
The present invention will be further specifically described with reference to the drawings and examples.
Fig. 1 shows a wind turbine blade sample (glass fiber cloth is glass fiber cloth) containing defects prefabricated in this embodiment, the preset defect types are respectively wrinkle, glue deficiency and inclusion, and the specific defect parameters are shown in table 1. The wind power blade is made of glass fiber cloth and epoxy resin through compounding, and is formed through a vacuum suction injection process, wherein a layer of glass fiber is laid, then a layer of epoxy resin is laid, 37 layers are laid together, the 1 st layer and the 37 th layer are both made of glass fiber cloth, and the forming thickness of each layer is 1.2mm and 44.4 mm. Three defects of inclusion, glue deficiency and wrinkle are prefabricated in a sample, namely the wrinkle defect is formed by pre-embedding cylindrical glass fiber reinforced plastics between the 17 th layer and the 18 th layer of glass fiber cloth, and a blade is placed between the 19 th layer and the 20 th layer of glass fiber cloth to simulate the defect of inclusion and the defect of simulated glue deficiency by adhering polytetrafluoroethylene. The invention adopts matlab software to analyze data.
TABLE 1 Experimental sample parameter Table
The defect identification method provided by the invention is adopted to detect the defects of the wind power blade sample prepared by the embodiment:
step (1): and (3) carrying out ultrasonic detection on the sample by adopting an ultrasonic detector, and selecting a db4 function with the maximum correlation with the signal waveform as a wavelet basis of wavelet packet decomposition through an ultrasonic detection signal time domain diagram.
Specifically, the sampling frequency of the instrument is 10MHz, the probe adopts an Olympus R101 probe, and the frequency is 0.5 MHz. Flaw detection is carried out on the prepared sample defects in a water coupling mode to obtain defect signals, and 60 groups of ultrasonic detection signals are collected for each defect. The obtained ultrasonic detection signal is subjected to fourier transform, the time domain signal is shown in fig. 2, the ordinate of the time domain signal is the amplitude, and the abscissa of the time domain signal is the depth of the a-type reflected wave. As shown in fig. 2a, the bottom echo occurs at a 44.4mm position, i.e. corresponding to the thickness of the pre-fabricated sample. Fig. 2b, 2c, 2d show defect echoes of 21.4mm, 23.8mm, and 25.0mm at the defect preparation location, respectively, in addition to the bottom echo at the 44.4mm location, which is very similar to the location of the defect prepared by the inventor. From the waveform shown in fig. 2, it can be seen that the sample has no defects and a defect depth, but the corresponding defect type cannot be identified. Based on the characteristics of the wavelet basis functions, in combination with the time domain waveform of the ultrasound signal, the present embodiment selects the db4 function that is most similar to the detected signal as the wavelet basis of the wavelet packet decomposition.
Step (2): and (3) carrying out frequency spectrum transformation on the ultrasonic detection signal obtained in the step (1), and determining a frequency band characteristic window by combining frequency spectrum and wavelet packet decomposition, thereby determining the optimal decomposition layer number m of the wavelet packet decomposition.
Specifically, in this embodiment, 1-5 layers of wavelet packet decomposition are performed on the ultrasonic detection signals obtained in step 1, and by analyzing the frequency band feature window obtained in each layer, the layer of the frequency band feature window where the energy distribution difference of the ultrasonic signals can be seen is selected, that is, the optimal number of decomposition layers for wavelet packet decomposition. In the present embodiment, the results of the spectrum analysis of the glue-deficient sample and the nondestructive sample are taken as an example for explanation during analysis, and fig. 3 is a spectrogram of the nondestructive and glue-deficient defects of the wind turbine blade provided in the present embodiment. It can be seen from fig. 3 that the primary frequencies of both the glue-missing and the non-destructive signals are concentrated around 0.5MHZ, which coincides with a transmission frequency of 0.5MHZ for the probe. The lossless samples have troughs at 0.4MHZ and 0.5MHZ and peaks at 0.35MHZ, 0.45MHZ and 0.55MHZ, and the two troughs indicate the difference of the two spectrograms, so that the two troughs are separated in wavelet packet decomposition to find the characteristic value of the signal. In the step, after the ultrasonic signal frequency spectrum is decomposed by 4 layers of wavelet packets, 16 nodes are obtained, the frequency bandwidth of each node is 0.3125MHZ, and then the frequency of the second node is 0.3125-0.625MHZ, it can be obviously seen that the second node includes 0.4MHZ and 0.5MHZ, the two nodes are not separated, and the decomposition scale is small; after 5 layers of wavelet packet decomposition are carried out on the signal frequency spectrum, 32 wavelet packet decomposition frequency bands are obtained, each frequency band width is 0.15625MHZ, and calculation shows that 0.4MHZ is 0.3125-0.46875MHZ at the third node, and 0.5MHZ is 0.46875-0.625 at the fourth node
MHZ. Therefore, a frequency band characteristic window can be found through 5-layer wavelet packet decomposition, so that the characteristic points of the two ultrasonic signals are separated, and the difference of energy distribution of the two ultrasonic signals is reflected, so that the optimal decomposition layer number of the wavelet packet in the embodiment is 5.
And (3): performing 5-layer wavelet packet decomposition on the defect ultrasonic detection signal provided in the step 1 by using db4 wavelet to obtain 25Each node corresponds to a frequency segment of the defect signal, and the frequency segments of different defects have different characteristics and corresponding energy spectrum coefficients. The spectral coefficients of the 32 nodes are extracted, and a wavelet packet spectral coefficient histogram (two-dimensional histogram) is made, as shown in fig. 4. It can be seen from fig. 4 that most of the energy is concentrated within the first 10 nodes, the energy of the latter 22 nodes is almost 0, and the signal energy is mainly concentrated at the 3 rd, 4 th and 7 th, 8 th nodes. As is obvious from the histogram of the spectral energy, the wavelet energy spectral coefficients of different defect types are distributed differently, thereby proving the correctness of the inventor for selecting 5-layer wavelet packet decomposition to distinguish the defect types by taking the energy spectral coefficients as characteristic parameters.
Ultrasonic detection signals of three defects of glue deficiency, inclusion and wrinkle are subjected to wavelet packet transformation to extract energy spectrum coefficients, 60 groups of data are counted in 60 groups of each defect test, 40 groups of data are selected from 60 groups of data of each defect to form a training sample set, and the remaining 20 groups of data form a test sample set. Because of more data, the embodiment selects representative data in the training set to be displayed in the table, as shown in table 2. The test set selects and displays representative 9 sets of data (3-10 nodes), which are numerical features that can distinguish defect types, as shown in table 3.
TABLE 2 training sample set
TABLE 3 test sample set
And (4): the energy spectrum is a characteristic parameter which can be quantized, so that a wavelet energy spectrum coefficient obtained by calculation after wavelet packet decomposition is used as an input vector of the neural network to construct the BP neural network. When the BP neural network is constructed, the frequency spectrum energy characteristic value after 32 wavelet packet decomposition is extracted by considering that 5-layer wavelet packet decomposition is carried out on a defect ultrasonic detection signal, and the number of corresponding neural network input nodes is set to be 32. Three defects need to be automatically identified, and the output nodes of the network are set to be 3, wherein (100) represents inclusion defects, (010) represents wrinkle defects, and (001) represents glue-missing defects. Generally, a neural network has only one hidden layer. If the BP neural network comprises one hidden layer, the hyperplane division of a small sample space can be realized, but the number of samples is large in the embodiment, so that two hidden layers are selected, the network scale can be reduced, and the training precision of the neural network can be improved. The number of hidden nodes is now generally chosen by this empirical formula:
where n is the number of hidden nodes, niIs the number of nodes of the input layer, n0A is a constant of 1 to 10 for the number of output layer nodes.
According to the empirical formula, the error generated by the prediction result of the BP neural network is the minimum and the neural network structure is stable when the number of the hidden layer nodes is 8 as found by a trial and error method. Finally, the number of the first hidden nodes is set to be 8, and the number of the second hidden nodes is set to be 4. Because the input and output data after normalization are all in the range of [0,1], in order to reduce errors, S-type Sigmoid functions are adopted by both hidden layers. And training the network by adopting an algorithm with a variable learning rate, and only determining the initial learning rate. In order to ensure the stability of the system, the learning rate value is generally selected from 0.01 to 1, in this embodiment, the initial learning rate is selected to be 0.8, the set target error is 0.001, and the maximum training frequency is 5000.
The data (table 2) of the three defects extracted by wavelet packet decomposition is input into the BP neural network as a training set. The training result is shown in fig. 5, where the broken line is the training error and the dotted line is the target error (0.001), and the broken line in the BP neural network shows a steady descending trend in the 9-step round trip training process and finally falls below the target error value, thereby achieving network convergence. The trained network was input with the test sample set (table 3) for testing, and the recognition results are shown in table 4. As can be seen from Table 4, the recognition effect of the three defects is good, and the average recognition accuracy rate is over 90%.
TABLE 4 BP neural network recognition results
The method is based on the frequency band characteristic window, combines spectral analysis and wavelet packet decomposition, and applies the BP neural network to analyze the ultrasonic signals, so that the defect type of the wind power blade is automatically identified. Specifically, a frequency band characteristic window is found by combining spectral analysis and wavelet packet analysis, so that the number of wavelet decomposition layers is determined to be 5. And 5-layer wavelet packet decomposition is carried out on the wind power blade defect signal, the decomposed wavelet packet frequency spectrum energy characteristics are extracted, the characteristic vector of the defect signal is further constructed, and the characteristic vector is input into a BP neural network for training and effect inspection. Experimental results show that the defect identification method provided by the invention is effective and feasible, the automatic identification of the wind power blade defects becomes possible, and the average identification rate is up to 90%.
Claims (6)
1. A wind power blade defect identification method is characterized by comprising the following steps:
step a) carrying out ultrasonic detection on a sample containing defects, and selecting a db4 function with the maximum correlation with signal waveforms as a wavelet basis of wavelet packet decomposition according to a time domain diagram of ultrasonic detection signals;
b) carrying out frequency spectrum transformation on the ultrasonic detection signal obtained in the step a, analyzing the frequency characteristics of the obtained frequency spectrum signal, carrying out k-layer wavelet packet decomposition, selecting the layer of the frequency band characteristic window capable of showing the energy distribution difference of the ultrasonic signal by analyzing the frequency band characteristic window obtained in each layer, and finding out signal characteristic values from different wave troughs in the wavelet packet decomposition so as to determine the optimal number m of layers of the wavelet packet transformation;
c) performing m-layer db4 wavelet packet decomposition on the ultrasonic detection signal obtained in the step a to obtain a decomposed node and an energy spectrum coefficient thereof;
step d) constructing a BP neural network, taking the energy spectrum coefficient obtained in the step c as an input vector, taking the corresponding defect type as an output result, and training and verifying the neural network;
and e) carrying out ultrasonic detection on the wind power blade to be detected, inputting an ultrasonic detection signal into the neural network constructed in the step d, and automatically identifying the defect type in the wind power blade.
2. The wind turbine blade defect identification method according to claim 1, wherein the specific operation of the step (a) is as follows: and carrying out ultrasonic detection on the wind power blade sample containing the defect to obtain a corresponding defect ultrasonic detection signal, and observing the signal waveform in the time domain graph of the defect ultrasonic detection signal to select db4 with the maximum correlation as a wavelet basis for wavelet packet decomposition.
3. The wind turbine blade defect identification method according to claim 1, wherein the specific operation of the step (b) is as follows: b, performing frequency spectrum transformation on the defect ultrasonic detection signal provided in the step a, analyzing the frequency characteristics of the obtained frequency spectrum signal, performing k-layer wavelet packet decomposition on the frequency characteristics of the frequency spectrum signal, finding out a characteristic frequency band window, and determining the optimal number m of decomposed layers of the wavelet packet, so that each node after m-layer decomposition of the wavelet packet can effectively separate characteristic information in the frequency spectrum signal; wherein k is a natural number, k is not equal to 0, and m belongs to k.
4. The wind turbine blade defect identification method according to claim 1, wherein the specific operation of the step (c) is as follows: b, performing m-layer wavelet packet decomposition on the defect ultrasonic detection signal provided in the step a by using db4 wavelets, and extracting node energy spectrum coefficients corresponding to the nodes; selecting a node as a horizontal coordinate, and selecting an energy spectrum coefficient corresponding to the node as a vertical coordinate to obtain an energy spectrum coefficient graph; and distinguishing each defect type according to different node energies of different defects in the energy spectrum coefficient diagram, and verifying the correctness of the optimal decomposition layer number m.
5. The wind turbine blade defect identification method according to claim 1, wherein the specific operation of the step (d) is as follows: and constructing a BP neural network, wherein input nodes of the BP neural network are designed to be the number of energy spectrum coefficients after m layers of wavelet packet decomposition, output nodes are set to be defect types, and the mean square error of an output layer of the neural network reaches the minimum value through training and verification of the neural network.
6. The wind turbine blade defect identification method according to claim 1, wherein the specific operation of the step (e) is as follows: and d, carrying out ultrasonic detection on the wind power blade to be detected, inputting an energy spectrum coefficient obtained after carrying out m-layer db4 wavelet packet decomposition on an ultrasonic detection signal into the neural network constructed in the step d, and automatically identifying the defect type of the wind power blade through a computer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910493788.7A CN110161119B (en) | 2019-06-07 | 2019-06-07 | Wind power blade defect identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910493788.7A CN110161119B (en) | 2019-06-07 | 2019-06-07 | Wind power blade defect identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110161119A CN110161119A (en) | 2019-08-23 |
| CN110161119B true CN110161119B (en) | 2021-11-26 |
Family
ID=67627874
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910493788.7A Active CN110161119B (en) | 2019-06-07 | 2019-06-07 | Wind power blade defect identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110161119B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110702042B (en) * | 2019-10-15 | 2021-07-02 | 河海大学常州校区 | Thickness estimation method under signal aliasing condition in ultrasonic pulse echo thickness measurement |
| CN111122705B (en) * | 2019-12-26 | 2023-01-03 | 中国科学院工程热物理研究所 | Ultrasonic nondestructive testing method for wind turbine blade |
| CN111652098B (en) * | 2020-05-25 | 2022-03-01 | 四川长虹电器股份有限公司 | Product surface defect detection method and device |
| CN115166049B (en) * | 2022-09-07 | 2022-12-02 | 广东工业大学 | Laser ultrasonic real-time detection system and method based on additive manufacturing |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102628835A (en) * | 2012-04-19 | 2012-08-08 | 迪皮埃复材构件（太仓）有限公司 | System and method for detecting bonding quality of wind power generation blade web plate |
| CN103776903A (en) * | 2014-01-15 | 2014-05-07 | 南京航空航天大学 | Wind power blade delamination detection method and system |
| CN104458910A (en) * | 2014-12-16 | 2015-03-25 | 中国科学院工程热物理研究所 | Nondestructive testing method for bonding defects of shell of wind turbine blade and web plate |
| CN106092578A (en) * | 2016-07-15 | 2016-11-09 | 西安交通大学 | A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine |
| CN106289780A (en) * | 2016-09-20 | 2017-01-04 | 北京工业大学 | A kind of rolling bearing damage degree recognition methods based on Sparsogram and Sample Entropy |
| CN107228900A (en) * | 2016-03-24 | 2017-10-03 | 迪皮埃复材构件（太仓）有限公司 | A kind of blade web trailing edge quality determining method |
| CN108152376A (en) * | 2017-12-22 | 2018-06-12 | 吉林重通成飞新材料股份公司 | A kind of ultrasonic detection method of wind electricity blade |
| EP2476033B1 (en) * | 2009-09-08 | 2019-06-19 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Model-based method for monitoring the condition of rotor blades |
-
2019
- 2019-06-07 CN CN201910493788.7A patent/CN110161119B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2476033B1 (en) * | 2009-09-08 | 2019-06-19 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Model-based method for monitoring the condition of rotor blades |
| CN102628835A (en) * | 2012-04-19 | 2012-08-08 | 迪皮埃复材构件（太仓）有限公司 | System and method for detecting bonding quality of wind power generation blade web plate |
| CN103776903A (en) * | 2014-01-15 | 2014-05-07 | 南京航空航天大学 | Wind power blade delamination detection method and system |
| CN104458910A (en) * | 2014-12-16 | 2015-03-25 | 中国科学院工程热物理研究所 | Nondestructive testing method for bonding defects of shell of wind turbine blade and web plate |
| CN107228900A (en) * | 2016-03-24 | 2017-10-03 | 迪皮埃复材构件（太仓）有限公司 | A kind of blade web trailing edge quality determining method |
| CN106092578A (en) * | 2016-07-15 | 2016-11-09 | 西安交通大学 | A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine |
| CN106289780A (en) * | 2016-09-20 | 2017-01-04 | 北京工业大学 | A kind of rolling bearing damage degree recognition methods based on Sparsogram and Sample Entropy |
| CN108152376A (en) * | 2017-12-22 | 2018-06-12 | 吉林重通成飞新材料股份公司 | A kind of ultrasonic detection method of wind electricity blade |
Non-Patent Citations (2)
| Title |
|---|
| "基于小波包一BP网络的超声检测缺陷类型识别";周西峰;《解放军理工大学学报(自然科学版)》;20121031;第13卷(第5期);第521-506页 * |
| "基于小波包分解的滚动轴承故障诊断";黄中华;《湖南科技大学学报( 自然科学版)》;20080630;第 23 卷(第 2 期);第32-34页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110161119A (en) | 2019-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110161119B (en) | Wind power blade defect identification method | |
| CN110333285B (en) | Ultrasonic lamb wave defect signal identification method based on variational modal decomposition | |
| CN110286155B (en) | Damage detection method and system for multilayer composite material | |
| CN109490730A (en) | Cable discharge detection method, device, storage medium and processor | |
| CN105806270B (en) | A kind of detection method of material surface micro-crack depth | |
| CN108181378B (en) | Ultrasonic defect identification method for detecting laminated structure of mixed-layer composite material | |
| CN103245311A (en) | Ultrasonic thickness measurement device and method for multilayered wave-absorbing coatings | |
| Chapon et al. | Deconvolution of ultrasonic signals using a convolutional neural network | |
| CN102608212A (en) | Method for measuring acoustic impedance and acoustic attenuation of thin layer based on sound pressure reflection coefficient power spectrum | |
| Li et al. | A correlation filtering-based matching pursuit (CF-MP) for damage identification using Lamb waves | |
| CN109596710A (en) | The device and method of sleeve grouting defect ultrasound detection based on wavelet-packet energy | |
| CN114264724A (en) | Ultrasonic signal processing method based on CEEMDAN combined wavelet packet threshold | |
| CN2809640Y (en) | Ultrasonic online intelligent recognition system for structural defect | |
| US10261054B2 (en) | Method for the non-destructive ultrasonic testing of a part by echo analysis | |
| CN116818914A (en) | Glass and nondestructive testing method for processed product thereof | |
| Selim et al. | Wavelet transform applied to internal defect detection by means of laser ultrasound | |
| CN108195934B (en) | Ultrasonic guided wave detection frequency optimization method based on time-frequency analysis | |
| CN113533510B (en) | Rail fatigue micro-crack identification method and device | |
| CN114169373A (en) | Method for extracting GFRP internal defect characteristics based on variational modal decomposition | |
| CN113884567A (en) | Steel rail weld damage detection method and device based on ultrasonic Lamb waves | |
| CN106442744A (en) | Ultrasonic signal processing method for removing response characteristics of ultrasonic probe | |
| Kraljevski et al. | Autoencoder-based Ultrasonic NDT of Adhesive Bonds | |
| Zhu et al. | A denoising method for ultrasonic testing of rubber composites based on improved symplectic geometric mode decomposition | |
| Ling et al. | Wideband Dispersion Reversal based Corrosion Inspection Using A0 Mode Lamb Waves | |
| Yahiaoui et al. | Analysis of the Composite Materials using the Wavelet Transforms |
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 |