CN115935261A - Group equipment non-absolute forward feedback method based on industrial Internet - Google Patents
Group equipment non-absolute forward feedback method based on industrial Internet Download PDFInfo
- Publication number
- CN115935261A CN115935261A CN202211722389.1A CN202211722389A CN115935261A CN 115935261 A CN115935261 A CN 115935261A CN 202211722389 A CN202211722389 A CN 202211722389A CN 115935261 A CN115935261 A CN 115935261A
- Authority
- CN
- China
- Prior art keywords
- vector
- eigenvector
- feature vector
- feature
- gaussian
- 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
- 238000000034 method Methods 0.000 title claims abstract description 59
- 239000013598 vector Substances 0.000 claims abstract description 386
- 230000004044 response Effects 0.000 claims abstract description 83
- 238000004804 winding Methods 0.000 claims abstract description 22
- 238000009826 distribution Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 71
- 238000010248 power generation Methods 0.000 claims description 25
- 230000008569 process Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 abstract description 8
- 238000004519 manufacturing process Methods 0.000 description 10
- 238000007792 addition Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 8
- 238000003860 storage Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 6
- 238000011161 development Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
Classifications
-
- 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
- Y02E10/70—Wind energy
- Y02E10/72—Wind turbines with rotation axis in wind direction
Landscapes
- Wind Motors (AREA)
Abstract
The application relates to the field of industrial internet, and particularly discloses a group equipment non-absolute forward feedback method and system based on the industrial internet, wherein high-dimensional associated features of blade spacing, facing direction, impeller rotating speed, turn number of accessed generating windings and highest rated value of each fan in an air-out motor group on a time sequence dimension are extracted through a time sequence encoder, and due to the fact that response relations exist among five obtained feature vectors, in order to achieve non-absolute forward feedback, response tolerance indexes among the feature vectors are further calculated to obtain a sixth feature vector and a seventh feature vector, and then the two feature vectors are fused on the basis of a Gaussian density map, so that the Gaussian density map is used as a learning target of a deep neural network, and a better probability distribution among high-dimensional features of different target scales is generated along with model learning update parameters. Therefore, the running state of the wind turbine generator can be judged, and the risk is reduced.
Description
Technical Field
The present invention relates to the field of industrial internet, and more particularly, to a group device non-absolute forward feedback method and system based on industrial internet.
Background
The essence of the industrial internet is that equipment, production lines, factories, suppliers, products and customers are tightly connected and integrated through an open and global industrial-level network platform, various element resources in industrial economy are efficiently shared, and therefore, the cost is reduced, the efficiency is improved, the manufacturing industry is helped to extend an industrial chain, and the transformation development of the manufacturing industry is promoted through an automatic and intelligent production mode. With the continuous development of industrialization and informatization, more and more information technologies are applied to the industrial field, and most of the information technologies need to be networked in the using process.
Nowadays, a neural network with a final result as a feedback guide has become one of important links of machine learning, but in the face of group devices, a feedback step in the machine learning neural network needs to solve a problem of how to select a learnable "experience" from a plurality of results, and this problem is particularly obvious in the face of a non-absolute forward conclusion, and it is difficult to smoothly guide a learning path of a machine.
A fan power generation set in clean energy is a typical group device. Along with the continuous development of industrial internet, the fan improves the generating efficiency through the whole connecting network, can "exchange" each other between the fan, when head group fan perception weather variation such as wind direction, can cooperate the adjustment blade interval and face orientation to improve the whole generated energy of unit. According to statistics, compared with a fan power generation set accessed to the industrial internet technology, the power production cost is reduced from 3 yuan per kilowatt hour to 0.3 yuan per kilowatt hour.
However, feedback training with absolute positive conclusions inside the fan pack is often accompanied by some other non-obvious losses or even risks: the fan loss is too high due to the fact that the rotating speed of the impeller is too high, and the increasing acceleration of the overall generating efficiency of the unit is rapidly reduced when the rotating speed of the impeller exceeds a rated value; in addition, under bad weather, the stress of the internal supporting structure of the fan can be increased, the material fatigue is improved, the service life of the wind turbine generator is shortened, and the power generation cost is increased in a phase-changing manner. At present, the relation between the generating efficiency and the hidden cost can be balanced to a certain degree by methods such as increasing the number of turns of a connected generating winding or improving the flexibility of the highest rated value in a high wind speed state, but how to use a non-absolute forward conclusion mode to replace a traditional training mode of absolute forward conclusion feedback is a key problem to be solved at present. Therefore, it is desirable to provide a group device non-absolute forward feedback scheme based on the industrial internet.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a group equipment non-absolute forward feedback method based on an industrial internet and a system thereof, wherein a time sequence encoder is used for extracting high-dimensional correlation characteristics of the blade spacing, the facing direction, the impeller rotating speed, the turn number of connected power generation windings and the highest rated value of each fan in an air-out motor group in a time sequence dimension, and as the obtained five characteristic vectors have a response relation, in order to realize the non-absolute forward feedback, a response tolerance index between the characteristic vectors is further calculated to obtain a sixth characteristic vector and a seventh characteristic vector, and then the two characteristic vectors are fused based on a Gaussian density map to serve as a learning target of a deep neural network through the Gaussian density map so as to generate a more optimal probability distribution between the high-dimensional characteristics of different target scales along with model learning updating parameters. Therefore, the running state of the wind turbine generator can be judged to reduce the risk.
According to one aspect of the application, an industry internet-based group device non-absolute forward feedback method is provided, which comprises the following steps:
acquiring the blade spacing, the facing orientation, the impeller rotating speed, the number of turns of a connected power generation winding and the highest rated value of each fan in the wind turbine generator;
arranging various data of each fan in the wind turbine generator into input vectors respectively, and then obtaining first to fifth characteristic vectors through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer;
calculating a first response tolerance index of the third feature vector and the first and second feature vectors to obtain a sixth feature vector, wherein the first response tolerance index is based on a correlation between a feature vector resulting from a position-wise multiplication between a difference vector between the third feature vector and the first feature vector and a position-wise reciprocal vector of the second feature vector and a feature vector resulting from a position-wise multiplication between the first feature vector and the position-wise reciprocal vector of the second feature vector;
calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector to obtain a seventh feature vector, wherein the second response tolerance index is related to a feature vector resulting from a position-wise multiplication between a difference vector between the third feature vector and the fourth feature vector and a position-wise multiplication between a reciprocal-by-position vector of the fifth feature vector and a feature vector resulting from a position-wise multiplication between the fourth feature vector and a reciprocal-by-position vector of the fifth feature vector;
constructing a gaussian density map of the sixth eigenvector and the seventh eigenvector, wherein a mean vector of the gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector;
performing Gaussian discretization processing on the Gaussian density map to obtain a classification matrix; and
and passing the classification matrix through a classifier to obtain a classification result, wherein the classification result is used for representing the current running state of the wind turbine generator.
In the above group device non-absolute forward feedback method based on the industrial internet, after arranging each item of data of each fan in the wind turbine generator as an input vector respectively, obtaining first to fifth eigenvectors through a time sequence encoder including a one-dimensional convolution layer and a full connection layer, the method includes: arranging various data of each fan in the wind turbine generator into one-dimensional input vectors respectively; performing the input vector using the fully-connected layer of the sequential encoder in the following formulaAnd full-connection coding is carried out to extract high-dimensional implicit features of feature values of all positions in the input vector, wherein the formula is as follows:wherein X is an input vector, Y is an output vector, W is a weight matrix, B is an offset vector, and B is a->Represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
In the above group device non-absolute forward feedback method based on the industrial internet, calculating a first response tolerance index of the third feature vector and the first and second feature vectors to obtain a sixth feature vector includes: calculating a first response tolerance index of the third feature vector and the first and second feature vectors with the following formula to obtain a sixth feature vector; wherein the formula is:
whereinAnd & _ indicates position-wise addition, subtraction and multiplication of the vector, respectively, and & _ 1 indicates that the position-wise inverse of the vector is taken, and I is a unit vector.
In the above group device non-absolute forward feedback method based on industrial internet, calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector to obtain a seventh feature vector includes: calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector with the following formula to obtain a seventh feature vector; wherein the formula is:
whereinAnd & -1 respectively indicate an addition, subtraction and multiplication by an on-position of the vector, an-1 indicates that an on-position of the vector is reciprocal, and I is a unit vector.
In the above non-absolute positive feedback method for group devices based on the industrial internet, constructing a gaussian density map of the sixth feature vector and the seventh feature vector includes: constructing a gaussian density map of the sixth feature vector and the seventh feature vector in the following formula;
wherein the formula is:
x represents the synthesized Gaussian vector, mu i Represents a mean value of feature values of each position of the sixth feature vector and the seventh feature vector, and σ i A variance of the feature value representing each position of the sixth feature vector and the seventh feature vector.
In the above non-absolute forward feedback method for group devices based on the industrial internet, the performing gaussian discretization on the gaussian density map to obtain a classification matrix includes: performing Gaussian discretization processing on the Gaussian distribution of each position in the Gaussian density map to reduce the dimension of the Gaussian distribution of each position in the Gaussian density map into a one-dimensional feature vector; and two-dimensionally arranging the one-dimensional feature vectors of the positions to generate the classification matrix.
In the above non-absolute forward feedback method for group devices based on the industrial internet, the classifying matrix is passed through a classifier to obtain a classification result, and the classification result is used to represent the current operating state of the wind turbine generator, and the method includes: the classifier processes the classification matrix to generate a classification result according to the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) Project (F), where Project (F) represents projecting the classification matrix as a vector, W 1 To W n As a weight matrix for each fully connected layer, B 1 To B n A bias matrix representing the layers of the fully connected layer.
According to another aspect of the present application, there is provided an industrial internet-based group device non-absolute forward feedback system, comprising: the data acquisition unit is used for acquiring the blade spacing, the facing orientation, the impeller rotating speed, the number of turns of the accessed power generation winding and the highest rated value of each fan in the wind turbine generator; the encoding unit is used for respectively arranging various data of each fan in the wind turbine generator set, which are obtained by the data obtaining unit, into input vectors and then obtaining first to fifth characteristic vectors through a time sequence encoder comprising a one-dimensional convolution layer and a full-connection layer; a first response tolerance index calculation unit configured to calculate a first response tolerance index of the third eigenvector obtained by the encoding unit and the first eigenvector obtained by the encoding unit and the second eigenvector obtained by the encoding unit to obtain a sixth eigenvector, wherein the first response tolerance index is related to an eigenvector obtained by position-wise multiplication between a difference vector between the third eigenvector and the first eigenvector and a position-wise reciprocal vector of the second eigenvector and an eigenvector obtained by position-wise multiplication between the first eigenvector and the position-wise reciprocal vector of the second eigenvector; a second response tolerance index calculation unit configured to calculate a second response tolerance index between the fourth eigenvector obtained by the encoding unit and the fifth eigenvector obtained by the encoding unit and the third eigenvector obtained by the encoding unit to obtain a seventh eigenvector, wherein the second response tolerance index is related to an eigenvector obtained by position-wise multiplication between a differential vector between the third eigenvector and the fourth eigenvector and a position-wise reciprocal vector of the fifth eigenvector and an eigenvector obtained by position-wise multiplication between the fourth eigenvector and the position-wise reciprocal vector of the fifth eigenvector; a gaussian density map constructing unit, configured to construct a gaussian density map of the sixth eigenvector obtained by the first response tolerance index calculating unit and the seventh eigenvector obtained by the second response tolerance index calculating unit, where a mean vector of the gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector; a gaussian discretization unit, configured to perform gaussian discretization processing on the gaussian density map obtained by the gaussian density map constructing unit to obtain a classification matrix; and the classification unit is used for enabling the classification matrix obtained by the Gaussian discretization unit to pass through a classifier so as to obtain a classification result, and the classification result is used for representing the current running state of the wind turbine generator.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the industrial internet based group device non-absolute forward feedback method as described above.
Compared with the prior art, the group equipment non-absolute forward feedback method and system based on the industrial internet extract high-dimensional associated features of the blade spacing, the facing direction, the impeller rotating speed, the turn number of the accessed generating winding and the highest rated value of each fan in the wind power generation unit on the time sequence dimension through the time sequence encoder, and further calculate the response tolerance index between the feature vectors to obtain a sixth feature vector and a seventh feature vector in order to realize the non-absolute forward feedback because of the response relationship among the obtained five feature vectors, and then fuse the two feature vectors based on the Gaussian density map to serve as the learning target of the deep neural network through the Gaussian density map so as to generate more optimal probability distribution among the high-dimensional features of different target scales along with the model learning updating parameters. Therefore, the running state of the wind turbine generator can be judged to reduce the risk.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 is an application scenario diagram of a group device non-absolute forward feedback method based on an industrial internet according to an embodiment of the present application;
FIG. 2 is a flow chart of a group device non-absolute forward feedback method based on industrial Internet according to an embodiment of the application;
FIG. 3 is a schematic diagram of a system architecture of a group device non-absolute forward feedback method based on industrial Internet according to an embodiment of the present application;
FIG. 4 is a block diagram of an industrial Internet based crowd device non-absolute forward feedback system according to an embodiment of the present application;
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of scenes
As described above, the nature of the industrial internet is to tightly connect and integrate equipment, production lines, factories, suppliers, products and customers through an open and global industrial-level network platform, to efficiently share various essential resources in the industrial economy, thereby reducing the cost and improving the efficiency through an automatic and intelligent production mode, helping the manufacturing industry to extend the industrial chain, and promoting the transformation development of the manufacturing industry. With the continuous development of industrialization and informatization, more and more information technologies are applied to the industrial field, and the use process of most information technologies needs networking.
Nowadays, a neural network with a final result as a feedback guide has become one of important links of machine learning, but in the face of group devices, a feedback step in the machine learning neural network needs to solve a problem of how to select a learnable "experience" from a plurality of results, and this problem is particularly obvious in the face of a non-absolute forward conclusion, and it is difficult to smoothly guide a learning path of a machine.
A fan power generation set in clean energy is a typical group device. Along with the continuous development of industrial internet, the fan improves the generating efficiency through the whole connecting network, can "exchange" each other between the fan, when head group fan perception weather variation such as wind direction, can cooperate the adjustment blade interval and face orientation to improve the whole generated energy of unit. According to statistics, compared with a fan power generation set accessed to the industrial internet technology for ten years, the power production cost is reduced from 3 yuan per kilowatt hour to 0.3 yuan per kilowatt hour.
However, feedback training with absolute positive conclusions inside the fan pack is often accompanied by some other non-obvious losses or even risks: the fan loss is too high due to the fact that the rotating speed of the impeller is too high, and the increasing acceleration of the overall generating efficiency of the unit is rapidly reduced when the rotating speed of the impeller exceeds a rated value; in addition, under bad weather, the stress of the internal supporting structure of the fan can be increased, the material fatigue is improved, the service life of the wind turbine generator is shortened, and the power generation cost is increased in a phase-changing manner. At present, the relationship between the power generation efficiency and the hidden cost can be balanced to a certain extent by methods of increasing the number of turns of the connected power generation winding or improving the flexibility of the highest rated value in a high wind speed state, but how to use a non-absolute forward conclusion mode to replace a traditional training mode of absolute forward conclusion feedback is a key problem to be solved at present. Therefore, it is desirable to provide a group device non-absolute forward feedback scheme based on the industrial internet.
Based on this, in the technical scheme of the application, if the operating state of the current wind turbine generator meets the requirement, it is necessary to start with five aspects of the blade pitch, the facing direction (the rotation angle relative to the vertical direction), the impeller rotation speed, the number of turns of the accessed power generation winding, and the highest rated value, and this is essentially a classification problem, that is, based on the data of these five aspects, a deep neural network model is used to extract the features, and a classifier is used to perform classification judgment on whether the operating state of the current wind turbine generator meets the requirement.
Specifically, in the technical scheme of the application, the blade pitch, the facing direction, the impeller rotating speed, the number of turns of the accessed generating winding and the highest rated value of each fan in the wind turbine generator are firstly obtained, each data of the fan set is respectively arranged into an input vector, and a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer is input to obtain first to fifth eigenvectors.
Since the third eigenvector has a response relationship with the first eigenvector and the second eigenvector, and the fourth eigenvector and the fifth eigenvector have a response relationship with the third eigenvector, in order to realize non-absolute forward feedback, it is desirable that the response has a certain tolerance relationship, that is, a first response tolerance index of the third eigenvector with the first eigenvector and the second eigenvector, and a second response tolerance index of the fourth eigenvector and the fifth eigenvector with the third eigenvector are calculated respectively, and expressed as:
whereinAnd & _ indicates position-wise addition, subtraction and multiplication of the vector, respectively, and & _ 1 indicates that the position-wise inverse of the vector is taken, and I is a unit vector.
Then, the sixth and seventh feature vectors v 6 And v 7 And fusing by a Gaussian density graph, and obtaining a classification result representing the current running state of the fan unit.
Based on this, the application provides a group device non-absolute forward feedback method based on the industrial internet, which includes: acquiring the blade spacing, the facing orientation, the impeller rotating speed, the number of turns of a connected power generation winding and the highest rated value of each fan in the wind turbine generator; arranging various data of each fan in the wind turbine generator into input vectors respectively, and then obtaining first to fifth characteristic vectors through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer; calculating a first response tolerance index of the third eigenvector with the first eigenvector and the second eigenvector to obtain a sixth eigenvector, wherein the first response tolerance index is related to an eigenvector resulting from a position-wise multiplication between a difference vector between the third eigenvector and the first eigenvector and a position-wise reciprocal vector of the second eigenvector and an eigenvector resulting from a position-wise multiplication between the position-wise reciprocal vectors of the first eigenvector and the second eigenvector; calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector to obtain a seventh feature vector, wherein the second response tolerance index is related to a feature vector resulting from a position-wise multiplication between a difference vector between the third feature vector and the fourth feature vector and a position-wise multiplication between a reciprocal-by-position vector of the fifth feature vector and a feature vector resulting from a position-wise multiplication between the fourth feature vector and a reciprocal-by-position vector of the fifth feature vector; constructing a Gaussian density map of the sixth eigenvector and the seventh eigenvector, wherein a mean vector of the Gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the Gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector; performing Gaussian discretization processing on the Gaussian density map to obtain a classification matrix; and enabling the classification matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for representing the current operation state of the wind turbine generator.
Fig. 1 illustrates an application scenario diagram of a group device non-absolute forward feedback method based on the industrial internet according to an embodiment of the application. As shown in fig. 1, in this application scenario, first, the blade pitch, facing orientation, impeller speed, number of turns of the incoming generation winding, and maximum rating of each fan (e.g., F1-Fn as illustrated in fig. 1) in the wind turbine are obtained by each sensor (e.g., T1-Tn as illustrated in fig. 1) provided in the wind turbine (e.g., W as illustrated in fig. 1). Then, the obtained blade pitch, facing orientation, impeller rotation speed, number of turns of the accessed power generation winding and the maximum rated value of each fan are input into a server (for example, S as illustrated in fig. 1) deployed with an industrial internet-based swarm equipment non-absolute forward feedback algorithm, wherein the server can process the obtained blade pitch, facing orientation, impeller rotation speed, number of turns of the accessed power generation winding and the maximum rated value of each fan by the industrial internet-based swarm equipment non-absolute forward feedback algorithm to generate a classification result for representing the current operating state of the wind turbine.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
FIG. 2 illustrates a flow chart of an industrial Internet based group device non-absolute forward feedback method. As shown in fig. 2, the group device non-absolute forward feedback method based on the industrial internet according to the embodiment of the application includes: s110, acquiring the blade spacing, the facing orientation, the impeller rotating speed, the turn number of a connected power generation winding and the highest rated value of each fan in the wind turbine; s120, arranging various data of each fan in the wind turbine generator into input vectors respectively, and then obtaining first to fifth characteristic vectors through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer; s130, calculating a first response tolerance index of the third feature vector and the first and second feature vectors to obtain a sixth feature vector, wherein the first response tolerance index is related to a feature vector obtained by position-wise multiplication between a difference vector between the third feature vector and the first feature vector and a reciprocal-by-position vector of the second feature vector and a feature vector obtained by position-wise multiplication between the reciprocal-by-position vectors of the first and second feature vectors; s140, calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector to obtain a seventh feature vector, wherein the second response tolerance index is related to a feature vector obtained by position-wise multiplication between a difference vector between the third feature vector and the fourth feature vector and a reciprocal-by-position vector of the fifth feature vector and a feature vector obtained by position-wise multiplication between the fourth feature vector and the reciprocal-by-position vector of the fifth feature vector; s150, constructing a Gaussian density map of the sixth eigenvector and the seventh eigenvector, wherein a mean vector of the Gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the Gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector; s160, carrying out Gaussian discretization processing on the Gaussian density map to obtain a classification matrix; and S170, enabling the classification matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for representing the current operation state of the wind turbine generator.
Fig. 3 illustrates an architecture diagram of a group device non-absolute positive feedback method based on the industrial internet according to an embodiment of the application. As shown in fig. 3, in the network architecture of the group device non-absolute forward feedback method based on the industrial internet, first, after arranging data items (e.g., P1-P5 as illustrated in fig. 3) of each wind turbine in the wind turbine into input vectors (e.g., V1-V5 as illustrated in fig. 3), respectively, a time sequence encoder (e.g., E as illustrated in fig. 3) including one-dimensional convolutional layers and fully-connected layers is passed to obtain first to fifth feature vectors (e.g., VF1-VF5 as illustrated in fig. 3); then, calculating a first response tolerance index of the third eigenvector and the first eigenvector and the second eigenvector to obtain a sixth eigenvector (e.g., VF6 as illustrated in fig. 3); then, calculating a second response tolerance index between the fourth and fifth eigenvectors and the third eigenvector to obtain a seventh eigenvector (e.g., VF7 as illustrated in fig. 3); next, constructing a gaussian density map of the sixth and seventh eigenvectors (e.g., GD as illustrated in fig. 3); then, gaussian discretizing the gaussian density map to obtain a classification matrix (e.g., MF as illustrated in fig. 3); and finally, passing the classification matrix through a classifier (e.g., a classifier as illustrated in fig. 3) to obtain a classification result, wherein the classification result is used for representing the current operation state of the wind turbine.
In step S110 and step S120, the blade pitch, the facing direction, the impeller rotation speed, the number of turns of the connected power generation winding, and the maximum rated value of each fan in the wind turbine are obtained, and each item of data of each fan in the wind turbine is respectively arranged as an input vector and then passes through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer to obtain first to fifth eigenvectors. It can be understood that the fan loss is too high due to the fact that the rotating speed of the impeller is too high, and the acceleration of the increase of the overall generating efficiency of the unit is rapidly reduced when the rated value is exceeded; in addition, under bad weather, the stress of the internal supporting structure of the fan can be increased, the material fatigue is improved, the service life of the wind turbine generator is shortened, and the power generation cost is increased in a phase-changing manner. At present, the relationship between the power generation efficiency and the hidden cost can be balanced to a certain degree by methods of increasing the number of turns of the accessed power generation winding or improving the flexibility of the highest rated value in a high wind speed state.
Therefore, in the technical solution of the present application, if it is desired to accurately determine whether the current operating state of the wind turbine generator meets the requirement, it is necessary to start with five aspects, namely, the blade pitch, the facing direction (the rotation angle relative to the vertical direction), the impeller rotation speed, the number of turns of the connected power generation winding, and the highest rated value, which is a classification problem essentially, that is, based on the data of the five aspects, a deep neural network model is used to extract the features, and a classifier is used to perform classification determination on whether the current operating state of the wind turbine generator meets the requirement.
Specifically, in the technical scheme of this application, at first through setting up blade interval, the facing orientation, the impeller rotational speed, insert the winding number of turns and the highest rated value of generating electricity of each fan in the wind turbine generator system of obtaining through each sensor in the wind turbine generator system. Then, after the data of each fan in the wind turbine generator are respectively arranged into input vectors, coding processing is carried out in a time sequence coder comprising a one-dimensional convolution layer and a full connection layer, so that high-dimensional implicit correlation characteristics of the data of each fan in the wind turbine generator are extracted, and therefore first to fifth characteristic vectors are obtained.
Specifically, in this embodiment of the present application, the process of obtaining the first to fifth eigenvectors through a time sequence encoder including a one-dimensional convolution layer and a full link layer after arranging each item of data of each fan in the wind turbine generator as an input vector respectively includes: arranging various data of each fan in the wind turbine generator into one-dimensional input vectors respectively; using a full-connection layer of the time sequence encoder to perform full-connection encoding on the input vector by using the following formula to extract high-dimensional implicit features of feature values of all positions in the input vector, wherein the formula is as follows:where X is an input vector, Y is an output vector, W is a weight matrix, B is a bias vector, and>represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
In step S130, a first response tolerance index of the third eigenvector and the first eigenvector and the second eigenvector is calculated to obtain a sixth eigenvector, wherein the first response tolerance index is related to an eigenvector obtained by position-wise multiplication between a difference vector between the third eigenvector and the first eigenvector and a position-wise reciprocal vector of the second eigenvector and an eigenvector obtained by position-wise multiplication between the position-wise reciprocal vectors of the first eigenvector and the second eigenvector. It should be appreciated that since the third eigenvector has a response relationship with the first eigenvector and the second eigenvector, a certain tolerance relationship is desirable for the response in order to achieve non-absolute forward feedback. That is, in the technical solution of the present application, the first response tolerance indexes of the third feature vector and the first feature vector and the second feature vector are calculated, respectively, so as to obtain a sixth feature vector.
Specifically, in an embodiment of the present application, the process of calculating the first response tolerance index of the third feature vector and the first and second feature vectors to obtain a sixth feature vector includes: calculating a first response tolerance index of the third feature vector and the first and second feature vectors with the following formula to obtain a sixth feature vector;
wherein the formula is:
whereinAnd & _ indicates position-wise addition, subtraction and multiplication of the vector, respectively, and & _ 1 indicates that the position-wise inverse of the vector is taken, and I is a unit vector.
In step S140, a second response tolerance index between the fourth eigenvector and the fifth eigenvector and the third eigenvector is calculated to obtain a seventh eigenvector, wherein the second response tolerance index is related to the eigenvector obtained by position-wise multiplication between the difference vector between the third eigenvector and the fourth eigenvector and the reciprocal position-wise vector of the fifth eigenvector and the eigenvector obtained by position-wise multiplication between the fourth eigenvector and the reciprocal position-wise vector of the fifth eigenvector. It should be understood that since the fourth feature vector and the fifth feature vector have a response relationship with the third feature vector, a certain tolerance relationship is expected for the response in order to realize non-absolute forward feedback. That is, in the technical solution of the present application, second response tolerance indexes between the fourth feature vector and the fifth feature vector and between the third feature vector are calculated respectively to obtain a seventh feature vector.
Specifically, in this embodiment, the process of calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector to obtain a seventh feature vector includes: calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector with the following formula to obtain a seventh feature vector;
wherein the formula is:
whereinAnd & -1 respectively indicate an addition, subtraction and multiplication by an on-position of the vector, an-1 indicates that an on-position of the vector is reciprocal, and I is a unit vector.
In step S150, a gaussian density map of the sixth eigenvector and the seventh eigenvector is constructed, a mean vector of the gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector. It should be appreciated that due to the difference in scale and format of the above data, if the two feature vectors are directly arranged in two dimensions as a feature matrix, the feature matrix will have non-aggregated feature manifold in the high-dimensional feature space, thereby reducing the accuracy of the subsequent classification, and therefore needs to be unified in scale and format. That is, specifically, in the technical solution of the present application, the sixth and seventh feature vectors v are set 6 And v 7 And fusing by using a Gaussian density map, wherein a mean vector of the Gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the Gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector.
Specifically, in the embodiment of the present application, the process of constructing the gaussian density map of the sixth feature vector and the seventh feature vector includes: constructing a gaussian density map of the sixth feature vector and the seventh feature vector with the following formula; wherein the formula is:
x represents the synthesized Gaussian vector, mu i Represents a mean value of feature values of each position of the sixth feature vector and the seventh feature vector, and σ i A variance of the feature value representing each position of the sixth feature vector and the seventh feature vector.
In step S160 and step S170, performing gaussian discretization on the gaussian density map to obtain a classification matrix, and passing the classification matrix through a classifier to obtain a classification result, where the classification result is used to indicate an operating state of the current wind turbine. That is, in the technical scheme of the application, the gaussian density map is further subjected to gaussian discretization to obtain a classification matrix, so that no information loss is generated during feature fusion and augmentation, and the classification accuracy is improved. Accordingly, in a specific example, first, the gaussian distribution of each position in the gaussian density map is subjected to gaussian discretization processing to reduce the gaussian distribution of each position in the gaussian density map into a one-dimensional feature vector; then, the one-dimensional feature vectors of the respective positions are two-dimensionally arranged to generate the classification matrix. Then, the classification matrix can be passed through a classifier to obtain a classification result representing the current operating state of the wind turbine.
Specifically, in the embodiment of the present application, the classifying matrix is passed through a classifier to obtain a classification result, where the classification result is used in a process of representing an operation state of a current wind turbine, and the process includes: the classifier processes the classification matrix to generate a classification result according to the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) L Project (F) }, where Project (F) denotes the projection of the classification matrix as a vector, W 1 To W n As a weight matrix for all connected layers of each layer, B 1 To B n Representing fully-connected layers of each layerThe matrix is biased.
In summary, the group device non-absolute forward feedback method based on the industrial internet according to the embodiment of the present application is elucidated, which extracts high-dimensional associated features of blade spacing, facing orientation, impeller rotation speed, turn number of accessed generation windings and highest rated value of each fan in the wind turbine generator set on a time sequence dimension through a time sequence encoder, and further calculates a response tolerance index between feature vectors to obtain a sixth feature vector and a seventh feature vector in order to realize non-absolute forward feedback due to response relations among five obtained feature vectors, so as to fuse the two feature vectors based on a gaussian density map to serve as a learning target of a deep neural network through the gaussian density map to generate a better probability distribution among high-dimensional features of different target scales along with model learning update parameters. Therefore, the running state of the wind turbine generator can be judged to reduce the risk.
Exemplary System
FIG. 4 illustrates a block diagram of an industrial Internet based group device non-absolute forward feedback system in accordance with an embodiment of the present application. As shown in fig. 4, the group device non-absolute forward feedback system 400 based on the industrial internet according to the embodiment of the present application includes: the data acquisition unit 410 is used for acquiring the blade spacing, the facing direction, the impeller rotating speed, the number of turns of the accessed power generation winding and the maximum rated value of each fan in the wind turbine; the encoding unit 420 is configured to arrange each item of data of each fan in the wind turbine generator, which is obtained by the data obtaining unit 410, into input vectors respectively, and then obtain first to fifth feature vectors through a time sequence encoder including a one-dimensional convolutional layer and a full connection layer; a first response tolerance index calculation unit 430, configured to calculate a first response tolerance index of the third eigenvector obtained by the encoding unit 420 and the first eigenvector obtained by the encoding unit 420 and the second eigenvector obtained by the encoding unit 420 to obtain a sixth eigenvector, where the first response tolerance index is related to an eigenvector obtained by position-wise multiplication between a difference vector between the third eigenvector and the first eigenvector and a position-wise reciprocal vector of the second eigenvector and an eigenvector obtained by position-wise multiplication between the first eigenvector and the position-wise reciprocal vector of the second eigenvector; a second response tolerance index calculation unit 440, configured to calculate a second response tolerance index between the fourth feature vector obtained by the encoding unit 420 and the fifth feature vector obtained by the encoding unit 420 and the third feature vector obtained by the encoding unit 420 to obtain a seventh feature vector, where the second response tolerance index is related to a feature vector obtained by position-wise multiplication between a difference vector between the third feature vector and the fourth feature vector and a reciprocal position-wise vector of the fifth feature vector, and a feature vector obtained by position-wise multiplication between the fourth feature vector and the reciprocal position-wise vector of the fifth feature vector; a gaussian density map constructing unit 450, configured to construct a gaussian density map of the sixth eigenvector obtained by the first response tolerance index calculating unit 430 and the seventh eigenvector obtained by the second response tolerance index calculating unit 440, wherein a mean vector of the gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector; a gaussian discretization unit 460, configured to perform gaussian discretization processing on the gaussian density map obtained by the gaussian density map constructing unit 450 to obtain a classification matrix; and a classification unit 470, configured to pass the classification matrix obtained by the gaussian discretization unit 460 through a classifier to obtain a classification result, where the classification result is used to represent an operation state of the current wind turbine.
In an example, in the above group device non-absolute positive feedback system 400 based on the industrial internet, the encoding unit 420 is further configured to: arranging various data of each fan in the wind turbine generator into one-dimensional input vectors respectively; using all of said sequential encodersThe connecting layer performs full-connection coding on the input vector by using the following formula to extract high-dimensional implicit features of feature values of all positions in the input vector, wherein the formula is as follows:wherein X is an input vector, Y is an output vector, W is a weight matrix, B is an offset vector, and B is a->Represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
In one example, in the above-mentioned group device non-absolute positive feedback system based on industrial internet 400, the first response tolerance index calculation unit 430 is further configured to: calculating a first response tolerance index of the third feature vector and the first and second feature vectors in a formula to obtain a sixth feature vector;
wherein the formula is:
whereinAnd [ ] respectively represent a position-wise addition, subtraction and multiplication of the vector, [ - ] -1 represents a position-wise reciprocal of the vector, and I is a unit vector。
In one example, in the above group device non-absolute positive feedback system based on industrial internet 400, the second response tolerance index calculating unit 440 is further configured to: calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector with the following formula to obtain a seventh feature vector;
wherein the formula is:
whereinAnd & -1 respectively indicate an addition, subtraction and multiplication by an on-position of the vector, an-1 indicates that an on-position of the vector is reciprocal, and I is a unit vector.
In one example, in the above group device non-absolute positive feedback system 400 based on the industrial internet, the gaussian density map constructing unit 450 is further configured to: constructing a gaussian density map of the sixth feature vector and the seventh feature vector in the following formula;
wherein the formula is:
x represents the synthesized Gaussian vector, mu i Represents a mean value of feature values of each position of the sixth feature vector and the seventh feature vector, and σ i A variance of the feature value representing each position of the sixth feature vector and the seventh feature vector.
In one example, in the above-mentioned swarm equipment non-absolute positive feedback system 400 based on the industrial internet, the gaussian discretization unit 460 is further configured to: performing Gaussian discretization processing on the Gaussian distribution of each position in the Gaussian density map to reduce the dimension of the Gaussian distribution of each position in the Gaussian density map into a one-dimensional feature vector; and two-dimensionally arranging the one-dimensional feature vectors of the positions to generate the classification matrix.
In one example, in the above group device non-absolute positive feedback system 400 based on the industrial internet, the classifying unit 470 is further configured to: the classifier processes the classification matrix to generate a classification result according to the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) L Project (F) }, where Project (F) denotes the projection of the classification matrix as a vector, W 1 To W n As a weight matrix for each fully connected layer, B 1 To B n A bias matrix representing the fully connected layers of each layer.
Here, it can be understood by those skilled in the art that the detailed functions and operations of the respective units and modules in the above-mentioned industrial internet based community device non-absolute forward feedback system 400 have been described in detail in the above description of the industrial internet based community device non-absolute forward feedback method with reference to fig. 1 to 3, and thus, a repetitive description thereof will be omitted.
As described above, the group device non-absolute forward feedback system 400 based on the industrial internet according to the embodiment of the present application can be implemented in various terminal devices, such as a server of the group device non-absolute forward feedback algorithm based on the industrial internet, and the like. In one example, the industrial internet based group device non-absolute positive feedback system 400 according to the embodiment of the present application can be integrated into a terminal device as one software module and/or hardware module. For example, the industrial internet-based group device non-absolute forward feedback system 400 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the industrial internet-based group device non-absolute positive feedback system 400 can also be one of many hardware modules of the terminal device.
Alternatively, in another example, the industrial internet based group device non-absolute forward feedback system 400 and the terminal device may be separate devices, and the industrial internet based group device non-absolute forward feedback system 400 may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information according to an agreed data format.
Exemplary computer program product and computer-readable storage Medium
In addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the functions of the industrial internet based group device non-absolute forward feedback method according to the various embodiments of the present application described in the "exemplary methods" section above of this specification.
The computer program product may be written with program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform the steps in the industrial internet based group device non-absolute forward feedback method described in the "exemplary methods" section above in this specification.
The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably herein. As used herein, the words "or" and "refer to, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. A group device non-absolute forward feedback method based on industrial Internet is characterized by comprising the following steps: acquiring the blade spacing, the facing orientation, the impeller rotating speed, the number of turns of a connected power generation winding and the highest rated value of each fan in the wind turbine generator; arranging various data of each fan in the wind turbine generator into input vectors respectively, and then obtaining first to fifth characteristic vectors through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer; calculating a first response tolerance index of the third eigenvector with the first eigenvector and the second eigenvector to obtain a sixth eigenvector, wherein the first response tolerance index is related to an eigenvector resulting from a position-wise multiplication between a difference vector between the third eigenvector and the first eigenvector and a position-wise reciprocal vector of the second eigenvector and an eigenvector resulting from a position-wise multiplication between the position-wise reciprocal vectors of the first eigenvector and the second eigenvector; calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector to obtain a seventh feature vector, wherein the second response tolerance index is related to a feature vector resulting from a position-wise multiplication between a difference vector between the third feature vector and the fourth feature vector and a position-wise multiplication between a reciprocal-by-position vector of the fifth feature vector and a feature vector resulting from a position-wise multiplication between the fourth feature vector and a reciprocal-by-position vector of the fifth feature vector; constructing a gaussian density map of the sixth eigenvector and the seventh eigenvector, wherein a mean vector of the gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector; performing Gaussian discretization processing on the Gaussian density map to obtain a classification matrix; and enabling the classification matrix to pass through a classifier to obtain a classification result, wherein the classification result is used for representing the current running state of the wind turbine generator.
2. The group device non-absolute forward feedback method based on the industrial internet according to claim 1, wherein after the data of each fan in the wind turbine generator are respectively arranged as input vectors, a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer is used for obtaining a first characteristic vector, a second characteristic vector and a fifth characteristic vector, and the method comprises the following steps: arranging various data of each fan in the wind turbine generator into one-dimensional input vectors respectively; using a full-connection layer of the time sequence encoder to perform full-connection encoding on the input vector by using the following formula to extract high-dimensional implicit features of feature values of all positions in the input vector, wherein the formula is as follows:wherein X is an input vector, Y is an output vector, W is a weight matrix, B is an offset vector, and B is a->Represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
3. The industrial internet-based group device non-absolute forward feedback method of claim 2, wherein calculating a first response tolerance index of the third feature vector with the first and second feature vectors to obtain a sixth feature vector comprises: calculating a first response tolerance index of the third feature vector and the first and second feature vectors in a formula to obtain a sixth feature vector; wherein the formula is:
4. The industrial internet-based group device non-absolute forward feedback method of claim 3, wherein calculating a second response tolerance index between the fourth and fifth feature vectors and the third feature vector to obtain a seventh feature vector comprises: calculating a second response tolerance index between the fourth feature vector and the fifth feature vector and the third feature vector with the following formula to obtain a seventh feature vector; wherein the formula is:
5. The industrial internet-based group device non-absolute forward feedback method of claim 4, wherein constructing a Gaussian density map of the sixth and seventh eigenvectors comprises: constructing a gaussian density map of the sixth feature vector and the seventh feature vector in the following formula;
wherein the formula is:
x represents the synthesized Gaussian vector, mu i Represents a mean value of feature values of each position of the sixth feature vector and the seventh feature vector, and σ i A variance of the feature value representing each position of the sixth feature vector and the seventh feature vector.
6. The industrial internet-based group device non-absolute forward feedback method of claim 5, wherein the gaussian discretization of the gaussian density map to obtain a classification matrix comprises: performing Gaussian discretization processing on the Gaussian distribution of each position in the Gaussian density map to reduce the dimension of the Gaussian distribution of each position in the Gaussian density map into a one-dimensional feature vector; and two-dimensionally arranging the one-dimensional feature vectors of the positions to generate the classification matrix.
7. The industrial internet-based group device non-absolute forward feedback method of claim 6The method comprises the following steps of obtaining a classification result by passing the classification matrix through a classifier, wherein the classification result is used for representing the operation state of the current wind turbine generator and comprises the following steps: the classifier processes the classification matrix to generate a classification result according to the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) L Project (F) }, where Project (F) denotes the projection of the classification matrix as a vector, W 1 To W n As a weight matrix for each fully connected layer, B 1 To B n A bias matrix representing the layers of the fully connected layer.
8. An industry internet based group device non-absolute forward feedback system, comprising: the data acquisition unit is used for acquiring the blade spacing, the facing orientation, the impeller rotating speed, the number of turns of the accessed power generation winding and the highest rated value of each fan in the wind turbine generator; the encoding unit is used for respectively arranging various data of each fan in the wind turbine generator set, which are obtained by the data obtaining unit, into input vectors and then obtaining first to fifth characteristic vectors through a time sequence encoder comprising a one-dimensional convolution layer and a full connection layer; a first response tolerance index calculation unit configured to calculate a first response tolerance index of the third eigenvector obtained by the encoding unit and the first eigenvector obtained by the encoding unit and the second eigenvector obtained by the encoding unit to obtain a sixth eigenvector, wherein the first response tolerance index is related to an eigenvector obtained by position-wise multiplication between a difference vector between the third eigenvector and the first eigenvector and a position-wise reciprocal vector of the second eigenvector and an eigenvector obtained by position-wise multiplication between the position-wise reciprocal vectors of the first eigenvector and the second eigenvector; a second response tolerance index calculation unit configured to calculate a second response tolerance index between the fourth feature vector obtained by the encoding unit and the fifth feature vector obtained by the encoding unit and the third feature vector obtained by the encoding unit to obtain a seventh feature vector, wherein the second response tolerance index is related to a feature vector obtained by position-wise multiplication between a difference vector between the third feature vector and the fourth feature vector and a reciprocal-by-position vector of the fifth feature vector, and a feature vector obtained by position-wise multiplication between the fourth feature vector and a reciprocal-by-position vector of the fifth feature vector; a gaussian density map constructing unit, configured to construct a gaussian density map of the sixth eigenvector obtained by the first response tolerance index calculating unit and the seventh eigenvector obtained by the second response tolerance index calculating unit, where a mean vector of the gaussian density map is a mean vector between the sixth eigenvector and the seventh eigenvector, and a value of each position in a covariance matrix of the gaussian density map is a variance between eigenvalues of corresponding positions in the sixth eigenvector and the seventh eigenvector; a gaussian discretization unit, configured to perform gaussian discretization on the gaussian density map obtained by the gaussian density map constructing unit to obtain a classification matrix; and the classification unit is used for enabling the classification matrix obtained by the Gaussian discretization unit to pass through a classifier so as to obtain a classification result, and the classification result is used for representing the current running state of the wind turbine generator.
9. The industrial internet-based group device non-absolute forward feedback system of claim 8, wherein the encoding unit is further configured to: arranging various data of each fan in the wind turbine generator into one-dimensional input vectors respectively; using a full-connection layer of the time sequence encoder to perform full-connection encoding on the input vector by using the following formula to extract high-dimensional implicit features of feature values of all positions in the input vector, wherein the formula is as follows:wherein X is an input vector, Y is an output vector, W is a weight matrix, B is an offset vector, and B is a->Represents a matrix multiplication; performing one-dimensional convolutional coding on the input vector by using a one-dimensional convolutional layer of the time sequence encoder according to the following formula to extract high-dimensional implicit correlation characteristics among characteristic values of all positions in the input vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
10. The industrial internet-based group device non-absolute forward feedback system of claim 8, wherein the first response tolerance index calculation unit is further configured to: calculating a first response tolerance index of the third feature vector and the first and second feature vectors with the following formula to obtain a sixth feature vector; wherein the formula is:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211722389.1A CN115935261A (en) | 2022-12-30 | 2022-12-30 | Group equipment non-absolute forward feedback method based on industrial Internet |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211722389.1A CN115935261A (en) | 2022-12-30 | 2022-12-30 | Group equipment non-absolute forward feedback method based on industrial Internet |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115935261A true CN115935261A (en) | 2023-04-07 |
Family
ID=86699170
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211722389.1A Pending CN115935261A (en) | 2022-12-30 | 2022-12-30 | Group equipment non-absolute forward feedback method based on industrial Internet |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115935261A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116321620A (en) * | 2023-05-11 | 2023-06-23 | 杭州行至云起科技有限公司 | Intelligent lighting switch control system and method thereof |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020152870A1 (en) * | 2019-01-25 | 2020-07-30 | 国立大学法人東北大学 | Signal processing method, signal processing device, and signal processing program |
WO2022017122A1 (en) * | 2020-07-24 | 2022-01-27 | Huawei Technologies Co.,Ltd. | Hybrid message-passing-algorithm-based deep learning neural network for a massive beam-forming system |
CN114580545A (en) * | 2022-03-07 | 2022-06-03 | 华北电力大学(保定) | Wind turbine generator gearbox fault early warning method based on fusion model |
CN114647198A (en) * | 2022-03-09 | 2022-06-21 | 深圳市经纬纵横科技有限公司 | Intelligent home control method and system based on Internet of things and electronic equipment |
CN115171317A (en) * | 2022-03-09 | 2022-10-11 | 深圳市经纬纵横科技有限公司 | Internet of things smart home method and system and electronic equipment |
-
2022
- 2022-12-30 CN CN202211722389.1A patent/CN115935261A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020152870A1 (en) * | 2019-01-25 | 2020-07-30 | 国立大学法人東北大学 | Signal processing method, signal processing device, and signal processing program |
WO2022017122A1 (en) * | 2020-07-24 | 2022-01-27 | Huawei Technologies Co.,Ltd. | Hybrid message-passing-algorithm-based deep learning neural network for a massive beam-forming system |
CN114580545A (en) * | 2022-03-07 | 2022-06-03 | 华北电力大学(保定) | Wind turbine generator gearbox fault early warning method based on fusion model |
CN114647198A (en) * | 2022-03-09 | 2022-06-21 | 深圳市经纬纵横科技有限公司 | Intelligent home control method and system based on Internet of things and electronic equipment |
CN115171317A (en) * | 2022-03-09 | 2022-10-11 | 深圳市经纬纵横科技有限公司 | Internet of things smart home method and system and electronic equipment |
Non-Patent Citations (3)
Title |
---|
ROYA NOURI等: "Blind image steganalysis based on reciprocal singular value curve", 2015 9TH IRANIAN CONFERENCE ON MACHINE VISION AND IMAGE PROCESSING (MVIP), pages 124 - 127 * |
刘青;张进华;黄军勤;: "基于机器视觉的钢球表面缺陷检测和分类", 轴承, no. 10, pages 44 - 48 * |
郭旭: "面向结构化数据的自动化特征工程算法与系统研究", 中国优秀硕士学位论文全文数据库信息科技辑, no. 09, pages 138 - 98 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116321620A (en) * | 2023-05-11 | 2023-06-23 | 杭州行至云起科技有限公司 | Intelligent lighting switch control system and method thereof |
CN116321620B (en) * | 2023-05-11 | 2023-08-11 | 杭州行至云起科技有限公司 | Intelligent lighting switch control system and method thereof |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111157899B (en) | Method for estimating SOC of battery based on model fusion idea | |
Philip et al. | An evolutionary computation based approach for reduced order modelling of linear systems | |
Li et al. | An effective federated learning verification strategy and its applications for fault diagnosis in industrial IOT systems | |
CN114219147A (en) | Power distribution station fault prediction method based on federal learning | |
CN115935261A (en) | Group equipment non-absolute forward feedback method based on industrial Internet | |
CN114580520A (en) | Monitoring system and monitoring method for secondary equipment of power system | |
CN113420837B (en) | Fault diagnosis method based on multi-source compressed sensing | |
Yu et al. | Forecasting a short‐term wind speed using a deep belief network combined with a local predictor | |
Habib et al. | Retracted: Forecasting model for wind power integrating least squares support vector machine, singular spectrum analysis, deep belief network, and locality‐sensitive hashing | |
CN111725802B (en) | Method for judging transient stability of alternating current-direct current hybrid power grid based on deep neural network | |
Malik et al. | Application of GEP to investigate the imbalance faults in direct‐drive wind turbine using generator current signals<? show [AQ ID= Q1]?> | |
CN115324843A (en) | Wind generating set fault diagnosis system and method based on monitoring data | |
CN116826743A (en) | Power load prediction method based on federal graph neural network | |
Wang et al. | Cluster division in wind farm through ensemble modelling | |
CN116842463A (en) | Electric automobile charging pile equipment fault diagnosis method | |
CN116432117A (en) | Motor fault classification method and device based on multi-source information fusion | |
CN116010831A (en) | Combined clustering scene reduction method and system based on potential decision result | |
Yin et al. | Power transformer fault diagnosis based on support vector machine with cross validation and genetic algorithm | |
CN114861506A (en) | Uncertainty-considered consumption reduction optimization method for induction motor | |
CN113328466B (en) | Electric vehicle charging scene generation method | |
CN112531725B (en) | Method and system for identifying parameters of static var generator | |
Wen et al. | Health-diagnosis of electromechanical system with a principal-component bayesian neural network algorithm | |
Mestha et al. | Support vector machine based fault detection in inverter‐fed electric vehicle | |
CN115765191B (en) | Electric quantity data transmission system | |
CN118013400B (en) | Motor fault diagnosis method, device, electronic 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 | ||
AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20240628 |