CN115859148A - Fan blade vibration alarm method and device - Google Patents
Fan blade vibration alarm method and device Download PDFInfo
- Publication number
- CN115859148A CN115859148A CN202211482721.1A CN202211482721A CN115859148A CN 115859148 A CN115859148 A CN 115859148A CN 202211482721 A CN202211482721 A CN 202211482721A CN 115859148 A CN115859148 A CN 115859148A
- Authority
- CN
- China
- Prior art keywords
- layer
- alarm
- inputting
- vibration
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention provides a fan blade vibration alarm method, which comprises the following steps: acquiring a time domain curve of fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold; training a CLA model by adopting a time domain curve of fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value to obtain trained CLA model training; and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator. Based on the method, the logic relation between the input parameters and the target parameters established by training the CLA model becomes more accurate and reliable, and the number of false alarms can be reduced.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a fan blade vibration alarm method and device.
Background
The fan blade belongs to a key constituent part in the wind turbine generator and plays a role in energy conversion. In the running process of the fan, the safety and the reliability of the blades must be ensured, the conversion efficiency of absorbing wind energy can be effectively improved, and the beneficial influence is brought to the whole wind turbine set. Through improving the blade performance, also can guarantee the steady operation of fan. However, the working environment of the blade is complex, and the blade is subject to the effects of centrifugal force, aerodynamic force, thermal stress, bending stress and the like, and even has the phenomena of rain, snow, frost erosion, lightning accumulation, damage and the like. Based on the method, the vibration of the fan blade is considered to be detected and analyzed, the fatigue degree of the fan blade is accurately and quickly judged, and the method is an effective means for maintaining the safe operation of the fan and improving the wind energy utilization rate.
The existing unit is different from the actual running condition of a wind field after a fan is assembled based on wind tunnel test data before the wind tunnel test data is produced by a fan blade manufacturer, a model based on threshold value alarming cannot judge the complex condition, and the condition of false alarming often exists. In addition, because the wind farm adopts different fan models and different blade manufacturers, algorithm updating optimization cannot be carried out according to the actual operation condition of the wind farm by adopting different threshold value alarming, and therefore the accuracy of an alarming model is reduced. Therefore, a fan blade vibration alarm method is urgently needed, the accuracy of fault classification can be improved, and capture of the generator fault of the wind turbine generator and identification of different fault types are achieved.
Disclosure of Invention
The present invention is directed to solving, at least to some extent, one of the technical problems in the related art.
Therefore, the first objective of the present invention is to provide a method for alarming fan blade vibration, so as to capture the generator fault of a wind turbine and identify different fault types, thereby reducing the number of false alarms.
The second purpose of the invention is to provide a fan blade vibration alarm device.
A third object of the invention is to propose a computer device.
A fourth object of the invention is to propose a non-transitory computer-readable storage medium.
In order to achieve the above object, an embodiment of a first aspect of the present invention provides a fan blade vibration warning method, including:
acquiring a time domain curve of fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold;
training a CLA model by adopting a time domain curve of the fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold value to obtain the trained CLA model training;
and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator.
Optionally, in an embodiment of the present invention, the CLA model includes an input layer, a CNN layer, an LSTM layer, an attention layer, and an output layer, and the method includes training a pre-constructed hybrid model according to a frequency domain curve of the fan blade vibration, an alarm measured value of an amplitude index of natural frequency, and an alarm threshold value to obtain a trained hybrid model, and further includes:
inputting a time domain curve of the fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value into the input layer, and obtaining input vectors corresponding to the training samples through the input layer;
inputting the input vector into the CNN layer, and performing feature extraction on the input vector to screen out a target feature vector;
acquiring the target characteristic vector, and inputting the target characteristic vector into the LSTM layer to extract the characteristic relation corresponding to the target characteristic vector;
inputting the feature relationships corresponding to the target feature vectors into the attention layer, and screening the feature relationships according to attention weight parameter values of the feature relationships in the attention layer to obtain output vectors;
and inputting the output vector to the output layer, and outputting a vibration early warning result.
Optionally, in an embodiment of the present invention, the CNN layer includes a convolution layer and a discard drop layer, and the inputting the input vector into the CNN layer and performing feature extraction on the input vector to filter out a target feature vector includes:
inputting the input vector into the convolutional layer to obtain a plurality of characteristic vectors of the input vector extracted by the convolutional layer;
inputting a plurality of feature vectors of the input vector into the drop layer to screen out a target feature vector from the plurality of features.
Optionally, in an embodiment of the present invention, after the inputting the output vector to the output layer and outputting the vibration warning result, the method further includes:
and acquiring the vibration early warning result, and performing anti-normalization processing on the vibration alarm result.
Optionally, in an embodiment of the present invention, after the obtaining the vibration alarm result and performing inverse normalization processing on the vibration alarm result, the method further includes:
the vibration alarm result is subjected to inverse normalization to obtain an actual vibration alarm result:
wherein the content of the first and second substances,
for bearing alarm data predicted by the mixing network before de-normalization processing, based on the comparison result>
For the anti-normalized blade vibration alarm data, y min 、y max The minimum value and the maximum value in the historical output data before normalization processing are respectively.
In order to achieve the above object, a second embodiment of the present invention provides a fan blade vibration warning device, including:
the first acquisition module is used for acquiring a time domain curve of fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold;
the training module is used for training a CLA model by adopting a time domain curve of the fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold value so as to obtain the trained CLA model training;
and the second acquisition module is used for acquiring the actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain the vibration early warning result of the blades of the wind turbine generator.
Optionally, in an embodiment of the present invention, the training module is further configured to:
inputting a time domain curve of the fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value into the input layer, and obtaining input vectors corresponding to the training samples through the input layer;
inputting the input vector into the convolutional neural network layer, and performing feature extraction on the input vector to screen out a target feature vector;
acquiring the target characteristic vectors, and inputting the target characteristic vectors into the long-short term memory network layer to extract the characteristic relations corresponding to the target characteristic vectors;
inputting the feature relations corresponding to the target feature vectors into the attention layer, and screening the feature relations according to the attention weight parameter values of the feature relations in the attention layer to obtain output vectors;
and inputting the output vector to the output layer, and outputting a vibration alarm result.
Optionally, in an embodiment of the present invention, the apparatus is further configured to:
inputting the input vector into the convolutional layer to obtain a plurality of characteristic vectors of the input vector extracted by the convolutional layer;
inputting a plurality of feature vectors of the input vector to the discarding layer to screen a target feature vector from the plurality of features.
In summary, according to the method and the device for alarming fan blade vibration provided by the invention, a time domain curve of fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold are obtained, and then a CLA model is trained by using the time domain curve of fan blade vibration, the acceleration actual measurement alarm parameter and the acceleration alarm threshold, so as to obtain a trained CLA model; and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator. Based on the method, the logic relation between the input parameters and the target parameters established according to the CLA model is more accurate and reliable, the capture of the generator faults of the wind turbine generator and the recognition of different fault types can be realized, and therefore the times of false alarm can be reduced.
To achieve the above object, a third embodiment of the present invention provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the computer device implements the method according to the first embodiment of the present invention.
To achieve the above object, a fourth aspect of the present invention provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method according to the first aspect of the present invention.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
fig. 1 is a schematic flow chart of a fan blade vibration alarm method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a neuron in a drop layer network according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a hybrid model according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a fan blade vibration alarm device according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
A fan blade vibration warning method and apparatus according to an embodiment of the present invention will be described with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of a fan blade vibration alarm method according to an embodiment of the present invention.
Step S1: and acquiring a time domain curve of the fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold.
In the embodiment of the present invention, the measured acceleration alarm parameter may include: and alarming the effective value of the vibration acceleration, the peak value of the vibration acceleration, the kurtosis index of the vibration acceleration, the skewness index of the vibration acceleration, the margin index of the vibration acceleration and the effective value of the envelope of the vibration acceleration.
And, in an embodiment of the present invention, the acceleration alarm threshold may include: the method comprises the steps of a vibration acceleration effective value threshold, a vibration acceleration peak value threshold, a vibration acceleration kurtosis index threshold, a vibration acceleration skewness index threshold, a vibration acceleration margin index threshold and a vibration acceleration envelope effective value alarm threshold.
Step S2: and training the CLA (CNN + LSTM + attention) model according to a time domain curve of fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value to obtain the trained CLA model.
In an embodiment of the present invention, the CLA model includes an input layer, a Convolutional Neural Network (CNN) layer, a Long Short-term-memory Network (LSTM) layer, an Attention (Attention) layer, and an output layer.
In the embodiment of the invention, the CLA model is trained according to the time domain curve of the fan blade vibration, the acceleration measured alarm parameter and the acceleration alarm threshold value to obtain the trained CLA model, and the method further comprises the following steps:
step S201, inputting a time domain curve of fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold value into an input layer, and obtaining input vectors corresponding to training samples through the input layer.
Step S202, inputting the input vector into the CNN layer, and performing feature extraction on the input vector to screen out a target feature vector.
In the embodiment of the present invention, after obtaining the time domain curve of the fan blade vibration, the training samples of the acceleration measured alarm parameter and the acceleration alarm threshold, the training samples are input into the LCA model through the input layer to obtain the input vector converted from the training samples, for example, if the length of the training samples input in batch is m, the input vector can be used
Representing the input vector.
In the embodiment of the present invention, in order to accurately screen out the target feature vector, the CNN layer includes a convolution layer and a discard dropout layer, the input vector is input to the CNN layer, and feature extraction is performed on the input vector, so as to screen out the target feature vector, which has an implementation manner that: inputting an input vector into the convolutional layer, acquiring a plurality of feature vectors of the input vector extracted by the convolutional layer, and inputting the plurality of feature vectors of the input vector into the dropout layer so as to screen out a target feature vector from the plurality of features.
For example, in the embodiment of the present invention, when the data dimension of the solar power generation amount is 1 dimension, the convolution layer is selected as a one-dimensional convolution, and then the size of the convolution kernel is 3, and the RELU activation function is used in combination, so as to obtain a plurality of feature vectors of the input vector.
For example, in the embodiment of the present invention, in the case that the dropout layer is set to 0.2, half of the hidden neurons in the network of the dropout layer may be temporarily randomly deleted, and the input and output neurons remain unchanged. It should be noted that the circles in fig. 2 represent non-deleted neurons, and the circles with crosses represent deleted neurons, as shown in fig. 2. And then, the input neurons are propagated forwards through the modified network, the obtained loss result is propagated backwards through the modified network, and after a small batch of training samples are executed, the parameters (w, b) corresponding to the neurons which are not deleted are updated according to a random gradient descent method.
It should be noted that, after obtaining the updated corresponding parameters (w, b), in order to avoid the problem of overfitting of the trained prediction model, in some embodiments, this process may be repeated continuously: and recovering deleted neurons, wherein the deleted neurons are kept as they are at the moment, the non-deleted neurons are updated, a half-size subset is randomly selected from the hidden neurons to be temporarily deleted, parameters of the deleted neurons are backed up, and for a small batch of training samples, the parameters (w, b) are updated according to a random gradient descent method after being propagated to the previous direction and then propagated to the reverse direction to be lost so as to solve the problem of overfitting of different networks.
Where w is the parameter weight in the neural network and b is the bias in the neural network.
In other embodiments, in the case that the dropout layer is set to 0.2, if the number of neurons is n, there may be 0.2n neurons that will be deleted, where one embodiment of deleting neurons is to change the activation function value of a neuron in the network to 0 with a probability p. If the output vector length is i, the target feature vector is H c ＝[h c1 …h c1 …] T Wherein neurons of dropout compute activation functions in the networkNumerical value
One way of calculating is:
r j (l) ～Benoulli(p)
wherein, the Bernoulli function generates a probability vector r, that is, a vector of 0 and 1 is randomly generated.
Step S203, obtaining the target characteristic vector, inputting the target characteristic vector into the LSTM layer, so as to extract the characteristic relation corresponding to the target characteristic vector.
In some embodiments, the obtained target feature vector is input into an LSTM layer, the vibration alarm behavior characteristics of the fan blade are learned through the LSTM layer and a bidirectional Long-Term-Short Term Memory (bilTM) layer structure, and if the length of the first output vector is j, the target output vector of the LSTM layer is H L ＝[h L1 …h Lε …h L ] J Calculating H L One way of calculating is:
H l ＝max(dropout(L))+b r
wherein, LSTM layer needs to be connected to dropout layer and Max pooling Maxpooling layer, max is maximum function in the Max pooling layer, br is bias of the pooling layer, L is output of the LSTM layer, W is output of the LSTM layer, and the maximum pooling layer is a function of maximum value in the Max pooling layer C ,b c Respectively the weight and the bias of the LSTM layer.
Step S204, inputting the feature relations corresponding to the target feature vectors into the Attention layer, and screening the feature relations according to the Attention weight parameter values of the feature relations in the Attention layer to obtain output vectors.
And step S205, inputting the output vector to an output layer, and outputting a vibration alarm result.
In the embodiment of the invention, after the output vector is input into the output layer, the output layer obtains the vibration early warning result of the wind turbine blade through the full-connection layer, and if the predicted compensation of the output layer is n, the predicted value of the vibration early warning result of the wind turbine blade is n
An exemplary way to calculate Y is:
Y＝f(W r ·s+b r )
wherein, W r As output layer weights, b r For output layer biasing, f is the fully-connected layer activation function.
And, in the embodiment of the present invention, after inputting the output vector to the output layer and outputting the vibration alarm result, the method further includes:
and acquiring a vibration alarm result, and performing inverse normalization processing on the vibration alarm result.
Further, in the embodiment of the present invention, after obtaining the vibration alarm result and performing inverse normalization processing on the vibration alarm result, the method further includes:
the actual vibration alarm result can be obtained through inverse normalization of the vibration alarm result:
wherein the content of the first and second substances,
for pre-measuring via a hybrid networkTo bearing alarm data before inverse normalization processing, y is inverse
Normalized blade vibration alarm data, y min 、y max The minimum value and the maximum value in the historical output data before normalization processing are respectively.
Based on the above embodiments, fig. 3 is a schematic structural diagram of a CLA model according to an embodiment of the present invention. As shown in fig. 3, the CLA model includes an input layer, a CNN layer, an LSTM layer, an attention layer, and an output layer, and one embodiment of calculating the fan blade vibration alarm through the CLA model is as follows: inputting the collected training samples into an input layer, converting the training samples into input vectors, performing feature extraction on the input training samples input into a CNN layer to generate target feature vectors, inputting the target feature vectors into an LSTM layer, predicting a predicted value of fan blade vibration alarm by learning the rule in the target feature vectors extracted by a convolutional neural network layer through the LSTM layer and an attention layer, and outputting the predicted value through an output layer.
And step S3: and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator.
In summary, in the fan blade vibration alarm method provided by the invention, the time domain curve of fan blade vibration, the actually measured acceleration alarm parameter and the acceleration alarm threshold are obtained first, and then the time domain curve of fan blade vibration, the actually measured acceleration alarm parameter and the acceleration alarm threshold are adopted to train the CLA model, so as to obtain the trained CLA model training; and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator. Based on the method, the logic relation between the input parameters and the target parameters established according to the CLA model is more accurate and reliable, the capture of the generator faults of the wind turbine generator and the recognition of different fault types can be realized, and therefore the times of false alarm can be reduced.
Fig. 4 is a schematic structural diagram of a fan blade vibration alarm device according to an embodiment of the present invention.
As shown in fig. 4, the fan blade vibration warning device includes:
the first obtaining module 100 is configured to obtain a time domain curve of fan blade vibration, an acceleration measured alarm parameter, and an acceleration alarm threshold;
the training module 200 is used for training the CLA model by adopting a time domain curve of fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold value to obtain trained CLA model training;
the second obtaining module 300 is configured to obtain an actually measured time domain signal of the wind turbine generator, input the actually measured time domain signal of the wind turbine generator to the trained CLA model, and obtain a vibration early warning result of the wind turbine generator blade.
In an embodiment of the present invention, the training module 200 is further configured to:
inputting a time domain curve of fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value into an input layer, and obtaining input vectors corresponding to training samples through the input layer;
inputting the input vector into a CNN layer, and performing feature extraction on the input vector to screen out a target feature vector;
acquiring target characteristic vectors, and inputting the target characteristic vectors into an LSTM network layer to extract characteristic relations corresponding to the target characteristic vectors;
inputting the feature relations corresponding to the target feature vectors into an attention layer, and screening the feature relations according to attention weight parameter values of the feature relations in the attention layer to obtain output vectors;
and inputting the output vector to an output layer and outputting a vibration alarm result.
And, in the embodiment of the present invention, the fan blade vibration alarm device is further configured to:
inputting the input vector into the convolutional layer, and acquiring a plurality of characteristic vectors of the input vector extracted by the convolutional layer;
and inputting a plurality of feature vectors of the input vector into the drop layer to screen out a target feature vector from the plurality of features.
It should be noted that the foregoing explanation of the embodiment of the fan blade vibration alarm method is also applicable to the device of the embodiment, and reference may be made to the related description of the above embodiment, which is not repeated herein.
In summary, in the fan blade vibration alarm device provided by the invention, the time domain curve of fan blade vibration, the acceleration measured alarm parameter and the acceleration alarm threshold are obtained first, and then the time domain curve of fan blade vibration, the acceleration measured alarm parameter and the acceleration alarm threshold are adopted to train the CLA model, so as to obtain trained CLA model training; and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator. Based on the method, the logic relation between the input parameters and the target parameters established according to the CLA model is more accurate and reliable, the capture of the generator faults of the wind turbine generator and the recognition of different fault types can be realized, and therefore the times of false alarm can be reduced.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are well known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.
In addition, functional units in the embodiments of the present invention may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may also be stored in a computer readable storage medium.
The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc. Although embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are exemplary and not to be construed as limiting the present invention, and that changes, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Claims (10)
1. A fan blade vibration alarm method is characterized by comprising the following steps:
acquiring a time domain curve of fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold;
training a CLA model by adopting a time domain curve of the fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold value to obtain the trained CLA model training;
and acquiring an actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain a vibration early warning result of the blades of the wind turbine generator.
2. The alarm method of claim 1, wherein the CLA model comprises an input layer, a CNN layer, an LSTM layer, an attention layer and an output layer, and the training of the pre-constructed hybrid model according to the frequency domain curve of the fan blade vibration, the measured value of the natural frequency amplitude index alarm and the alarm threshold value results in a trained hybrid model, further comprising:
inputting a time domain curve of the fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value into the input layer, and obtaining input vectors corresponding to the training samples through the input layer;
inputting the input vector into the CNN layer, and performing feature extraction on the input vector to screen out a target feature vector;
acquiring the target characteristic vector, and inputting the target characteristic vector into the LSTM layer to extract the characteristic relation corresponding to the target characteristic vector;
inputting the feature relations corresponding to the target feature vectors into the attention layer, and screening the feature relations according to the attention weight parameter values of the feature relations in the attention layer to obtain output vectors;
and inputting the output vector into the output layer, and outputting a vibration early warning result.
3. The alarm method of claim 2, wherein the CNN layer includes a convolutional layer and a discard drop layer, and the inputting the input vector into the CNN layer and performing feature extraction on the input vector to screen out a target feature vector comprises:
inputting the input vector into the convolutional layer to obtain a plurality of characteristic vectors of the input vector extracted by the convolutional layer;
inputting a plurality of feature vectors of the input vector into the drop layer to screen out a target feature vector from the plurality of features.
4. The warning method as claimed in claim 2, further comprising, after inputting the output vector to the output layer and outputting a vibration warning result:
and acquiring the vibration early warning result, and performing anti-normalization processing on the vibration early warning result.
5. The alarm method of claim 4, wherein after said obtaining the vibration alarm result and performing denormalization on the vibration alarm result, further comprising:
the vibration alarm result is subjected to inverse normalization to obtain an actual vibration alarm result:
wherein the content of the first and second substances,
for bearing alarm data obtained by the hybrid network prediction before denormalization processing, y is inverse
And after normalization processing, the blade vibration alarm data ymin and ymax are respectively the minimum value and the maximum value in historical output data before normalization processing.
6. A fan blade vibration warning device, characterized by, includes:
the first acquisition module is used for acquiring a time domain curve of fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold;
the training module is used for training a CLA model by adopting a time domain curve of the fan blade vibration, an acceleration actual measurement alarm parameter and an acceleration alarm threshold value so as to obtain the trained CLA model training;
and the second acquisition module is used for acquiring the actually measured time domain signal of the wind turbine generator, and inputting the actually measured time domain signal of the wind turbine generator into the trained CLA model to obtain the vibration early warning result of the blades of the wind turbine generator.
7. The warning device of claim 6, wherein the training module is further configured to:
inputting a time domain curve of the fan blade vibration, an acceleration measured alarm parameter and an acceleration alarm threshold value into the input layer, and obtaining input vectors corresponding to the training samples through the input layer;
inputting the input vector into the CNN layer, and performing feature extraction on the input vector to screen out a target feature vector;
acquiring the target feature vector, and inputting the target feature vector into the LSTM network layer to extract the feature relationship corresponding to each target feature vector;
inputting the feature relations corresponding to the target feature vectors into the attention layer, and screening the feature relations according to the attention weight parameter values of the feature relations in the attention layer to obtain output vectors;
and inputting the output vector to the output layer, and outputting a vibration alarm result.
8. The warning device of claim 7, wherein the device is further configured to:
inputting the input vector into the convolutional layer to obtain a plurality of characteristic vectors of the input vector extracted by the convolutional layer;
inputting a plurality of feature vectors of the input vector into the drop layer to screen out a target feature vector from the plurality of features.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of claims 1-5 when executing the computer program.
10. A non-transitory computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, implements the method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211482721.1A CN115859148A (en) | 2022-11-24 | 2022-11-24 | Fan blade vibration alarm method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211482721.1A CN115859148A (en) | 2022-11-24 | 2022-11-24 | Fan blade vibration alarm method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115859148A true CN115859148A (en) | 2023-03-28 |
Family
ID=85665884
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211482721.1A Pending CN115859148A (en) | 2022-11-24 | 2022-11-24 | Fan blade vibration alarm method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115859148A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910570A (en) * | 2023-09-13 | 2023-10-20 | 华能新能源股份有限公司山西分公司 | Wind turbine generator system fault monitoring and early warning method and system based on big data |
| CN118167569A (en) * | 2024-05-09 | 2024-06-11 | 浙江华东测绘与工程安全技术有限公司 | Wind turbine generator blade abnormality detection method and device based on vibration |
-
2022
- 2022-11-24 CN CN202211482721.1A patent/CN115859148A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910570A (en) * | 2023-09-13 | 2023-10-20 | 华能新能源股份有限公司山西分公司 | Wind turbine generator system fault monitoring and early warning method and system based on big data |
| CN116910570B (en) * | 2023-09-13 | 2023-12-15 | 华能新能源股份有限公司山西分公司 | Wind turbine generator system fault monitoring and early warning method and system based on big data |
| CN118167569A (en) * | 2024-05-09 | 2024-06-11 | 浙江华东测绘与工程安全技术有限公司 | Wind turbine generator blade abnormality detection method and device based on vibration |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115859148A (en) | Fan blade vibration alarm method and device | |
| Joshuva et al. | An insight on VMD for diagnosing wind turbine blade faults using C4. 5 as feature selection and discriminating through multilayer perceptron | |
| CN106779200A (en) | Based on the Wind turbines trend prediction method for carrying out similarity in the historical data | |
| CN111400961B (en) | Wind generating set blade fault judging method and device | |
| Joshuva et al. | A comparative study of Bayes classifiers for blade fault diagnosis in wind turbines through vibration signals | |
| CN113822418A (en) | Wind power plant power prediction method, system, device and storage medium | |
| CN114565006A (en) | Wind driven generator blade damage detection method and system based on deep learning | |
| CN115456047A (en) | Offshore wind turbine structure state monitoring system and method based on artificial intelligence | |
| Joshuva et al. | Comparative study on tree classifiers for application to condition monitoring of wind turbine blade through histogram features using vibration signals: a data-mining approach | |
| CN108847679B (en) | Wind generating set and subsynchronous oscillation identification method, device and system for wind generating set | |
| Chen et al. | Learning deep representation for blades icing fault detection of wind turbines | |
| Joshuva et al. | A study of various blade fault conditions on a wind turbine using vibration signals through histogram features | |
| Afrasiabi et al. | Two-stage deep learning-based wind turbine condition monitoring using SCADA data | |
| CN115712860A (en) | Fan blade vibration alarm method and device | |
| CN115878992A (en) | Monitoring method and monitoring system for comprehensive pipe rack power supply system | |
| CN112855458B (en) | Anemometer fault diagnosis method, system and medium for wind generating set | |
| CN116400244B (en) | Abnormality detection method and device for energy storage battery | |
| CN115791180A (en) | Alarm method and device for fan bearing vibration | |
| CN117150705A (en) | Marine wind farm short-term wind speed prediction method based on characteristic deconstruction | |
| CN115081530A (en) | Fan blade icing detection method, system, medium and equipment | |
| Joshuva et al. | Logistic model tree classifier for condition monitoring of wind turbine blades | |
| Arockia Dhanraj et al. | Appraising machine learning classifiers for discriminating rotor condition in 50W–12V operational wind turbine for maximizing wind energy production through feature extraction and selection process | |
| CN108825452B (en) | Method and device for determining blade icing of wind generating set | |
| Hübner et al. | Wind Turbine Rotor aerodynamic imbalance detection using CNN | |
| CN117390967B (en) | Offshore wind power short-term power prediction method, device, equipment and storage medium |
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 |