CN113935408A - Wind power generation equipment state monitoring method based on simplified nuclear principal component network - Google Patents
Wind power generation equipment state monitoring method based on simplified nuclear principal component network Download PDFInfo
- Publication number
- CN113935408A CN113935408A CN202111156322.1A CN202111156322A CN113935408A CN 113935408 A CN113935408 A CN 113935408A CN 202111156322 A CN202111156322 A CN 202111156322A CN 113935408 A CN113935408 A CN 113935408A
- Authority
- CN
- China
- Prior art keywords
- vector
- matrix
- power generation
- wind power
- generation equipment
- 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.)
- Granted
Links
- 238000010248 power generation Methods 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012544 monitoring process Methods 0.000 title claims abstract description 40
- 238000000513 principal component analysis Methods 0.000 claims abstract description 10
- 238000004458 analytical method Methods 0.000 claims abstract description 7
- 239000013598 vector Substances 0.000 claims description 99
- 239000011159 matrix material Substances 0.000 claims description 43
- 238000005070 sampling Methods 0.000 claims description 20
- 210000002569 neuron Anatomy 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000001133 acceleration Effects 0.000 claims description 5
- 238000007621 cluster analysis Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 125000004432 carbon atom Chemical group C* 0.000 claims description 2
- 238000013461 design Methods 0.000 claims description 2
- 238000005259 measurement Methods 0.000 claims description 2
- 230000007935 neutral effect Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 abstract description 12
- 238000012549 training Methods 0.000 abstract description 4
- 238000009826 distribution Methods 0.000 abstract description 2
- 238000000605 extraction Methods 0.000 abstract description 2
- 238000003064 k means clustering Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03D—WIND MOTORS
- F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Resources & Organizations (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Biology (AREA)
- Marketing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Mechanical Engineering (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Primary Health Care (AREA)
- Sustainable Energy (AREA)
- Sustainable Development (AREA)
- Probability & Statistics with Applications (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses a wind power generation equipment state monitoring method based on a simplified nuclear principal component network. Specifically, the method disclosed by the invention implements sample distribution characteristic analysis on a training data set of the wind power generation equipment, and uses edge data points and clustering center data points to construct a hidden layer of the principal component network, so that the calculated amount during online monitoring can be greatly reduced, and the application feasibility and the real-time property of the state monitoring method based on the simplified principal component network are ensured. In addition, the method respectively calculates corresponding outputs through the two output layer coefficient matrixes, and compared with the traditional kernel principal component analysis algorithm which only has one output, the diversity of feature extraction is further guaranteed.
Description
Technical Field
The invention relates to a state monitoring method for wind power generation equipment, in particular to a state monitoring method for wind power generation equipment based on a simplified nuclear principal component network.
Background
Due to the fluctuation of the price of primary energy such as petroleum and the like and the environmental protection characteristic of wind power generation, the advantage of wind power generation is more and more obvious. At present, the top five countries of the world with the wind power installed capacity are China, America, Germany, Spain and India. China is rich in wind energy resources, the wind energy storage capacity is 32 hundred million kilowatts, the developable installed capacity is about 2.5 hundred million kilowatts, and the wind energy resources live at the first place in the world. The installed capacity of wind power in China is rapidly increased in recent years, and meanwhile, the related support and the happy planning of domestic new energy are continuously provided, so that the development of domestic wind power generation is further promoted. Wind power generation equipment (namely a wind turbine) is a power machine for converting wind energy into mechanical energy, meanwhile, the main utilization form of wind energy is power generation at present, and the wind power generation is fastest growing in the industries of new energy and renewable energy.
Generally speaking, most wind power generation equipment operates in regions with bad weather, and the operating conditions are extremely poor, so that the wind power generation set has more faults, and a large number of faults cause the wind power generation set to stop operating or not reach rated output. Therefore, the running state of the wind power generation equipment is monitored in real time, the running abnormality is found in time, the maintenance of the equipment is organized rapidly, and the method has important practical significance for improving the wind power utilization rate and the wind power generation service quality. In the existing wind power generation equipment set, a plurality of sensors are usually installed in a matching manner, and data information such as the rotating speed of a generator, generated electric power, acceleration and the like are measured and fed back in real time. The data collected by these sensors provide a solid data base for implementing data-driven wind power plant condition monitoring. Under the current wind tide of intelligent manufacturing and big data, the scheme of monitoring the state of the wind power generation equipment by using the sampled data is very suitable.
However, the operating state of the wind turbine is directly influenced by the wind speed of the external environment, and changes along with the change of the wind speed. Because the intermittent characteristics, the non-linear characteristics and the time sequence variation characteristics of the wind power are not manually and accurately predictable or controllable, the operating characteristics of the wind power generator, which are directly affected by the wind speed, provide challenges for implementing the state monitoring of the data-driven wind power generation equipment. The development of the equipment fault diagnosis technology in the field of wind power equipment at present in China is in an attempt stage, large-scale wind generating sets in China develop rapidly in recent years, but the state monitoring system products of the corresponding large-scale wind generating sets are in a primary stage, the running state monitoring system products of the large-scale wind generating sets in China are immature, and autonomous real-time intelligent monitoring equipment is lacked.
In recent years, in the field of data-driven process monitoring, nuclear principal component analysis algorithms have been successfully applied to monitor the operating state of non-linear industrial processes. Although the kernel principal component analysis algorithm can realize the monitoring of the running state by mining the nonlinear characteristics of the sampled data, a key problem exists: the computational complexity of implementing online monitoring is proportional to the size of the training data set, but larger training data sets are more advantageous for implementing condition monitoring. Therefore, if the condition monitoring of the wind power generation equipment is implemented by using the kernel principal component analysis algorithm, the on-line calculation amount is a problem to be considered in order to ensure the feasibility and real-time performance of the practical application.
Disclosure of Invention
The invention aims to solve the main technical problems that: how to build a simplified principal component network model and implement real-time state monitoring on the wind power generation equipment on the basis of the model. Specifically, the method disclosed by the invention implements sample distribution characteristic analysis on a training data set of the wind power generation equipment, and uses edge data points and clustering center data points to construct a hidden layer of the principal component network, so that the calculated amount during online monitoring can be greatly reduced, and the application feasibility and the real-time property of the state monitoring method based on the simplified principal component network are ensured.
The technical scheme adopted by the method for solving the problems is as follows: a wind power generation equipment state monitoring method based on a simplified core principal component network comprises the following steps:
step (1): after data which can be measured in real time by the wind power generation equipment is determined, acquiring the data according to the inherent sampling time interval of the data acquisition system of the wind power generation equipment in the normal operation state of the wind power generation equipment; wherein, 11 data that can real-time measurement at each sampling moment are specifically in turn: wind speed, rotor speed, generator speed, mechanical torque, generated power, blade pitch angle, blade azimuth, blade root moment, top horizontal axis acceleration, top longitudinal axis acceleration and yaw error.
Step (2): n sample data vectors x with wind speed between cut-in wind speed (typically 3 meters per second) and cut-out wind speed (typically 25 meters per second)1,x2,...,xnThe composition matrix X ═ X1,x2,...,xn]And for X ∈ R11×nEach row vector is normalized to obtain a new matrixWherein the ith sample data vector xi∈R11×1The 11 data in (1) are arranged in sequence according to the sequence in step (1), i belongs to {1, 211×nRepresenting a matrix of real numbers in dimensions 11 Xn, R representing a set of real numbers, R11×1Representing a real number vector of dimension 11 x 1.
And (3): for new matrixAfter kernel principal component analysis is implemented, the kernel parameter delta and the kernel score matrix Y epsilon R are reservedD×nThe specific implementation process is shown in the steps (3.1) to (3.5).
Step (3.1): according to the formulaComputing new matricesVector of the ith columnAnd j-th column vectorThe squared distance ζ (i, j) therebetween; where j ∈ {1, 2., n }, the superscript T denoting the transpose of a matrix or vector.
Step (3.2): the kernel parameter δ is determined according to the formula (i) as follows:
step (3.3): calculating a kernel matrix K epsilon R according to the formula ∈ Rn×nRow i and column j element K (i, j):
in the above equation, e represents a natural constant, i.e., e is equal to about 2.718281828.
Step (3.4): calculating n eigenvalues λ of the kernel matrix K1≥λ2≥...≥λnAnd their respectively corresponding feature vectors alpha1,α2,...,αnThen, the lambda is added1,λ2,...,λnIn is not less than λ0The number of the characteristic values is recorded as D; wherein λ is0Representing the mean value of n characteristic values, i.e. λ0=(λ1+λ2+...+λn) N, feature vector α1,α2,...,αnAre all 1.
Step (3.5): d feature vectors alpha1,α2,...,αDForming a characteristic transformation matrix A ═ alpha1,α2,...,αD]Then, according to the formula Y ═ ATK is calculated to obtain a kernel score matrix Y epsilon RD×n。
And (4): for new matrixColumn vector ofCarrying out edge characteristic point analysis to obtain E edge point vectors xi1,ξ2,...,ξEThe specific implementation process is shown in the steps (4.1) to (4.5).
Step (4.1): the initialization i is 1.
Step (4.2): will be provided withThe neutral column vectorColumn vectors having a squared distance between them less than δ are recorded asWherein N isiTo representThe number of column vectors satisfying the condition ζ (i, j) < δ.
In the above formula, the first and second carbon atoms are,representation calculationAnd1, 2, N, is equal toi。
Step (4.4): calculating the column vector according to the formulaCorresponding edge point index gi:
In the above formula, b is in the form of {1, 2i},θbIs binary number, and the value law is as follows:
step (4.5): judging whether a condition i is less than n; if yes, after i is set to i +1, returning to the step (4.2); if not, according to the value to the edge point index g1,g2,...,gnPerforming descending arrangement, and recording the column vectors corresponding to the largest E edge point indexes as edge point vectors xi1,ξ2,...,ξE。
And (5): for new matrixColumn vector ofPerforming cluster analysis to obtain C cluster central point vectors h1,h2,...,hC。
It should be noted that the various clustering algorithms that can be used for performing the clustering analysis in the step (5) are not limited to the k-means clustering (k-means clustering) algorithm adopted in the specific embodiment.
And (6): respectively sequentially arranging the column vectorsComputing the hidden layer neuron output vector z of the simplified kernel principal component network according to the following formula (iv) as an input vector1,z2,...,zn:
In the above formula, zi∈R(E+C)×1Representing the ith hidden layer neuronThe output vector, i ∈ {1, 2., n }, R ∈(E+C)×1Represents a real number vector of (E + C) × 1 dimension,
and (7): will z1,z2,...,znMerging into a hidden layer output matrix Z ═ Z with (E + C) x n dimension1,z2,...,zn]Then according to formula B1=(ZTZ)-1ZYTCalculating to obtain a first output layer coefficient matrix B of the simplified core principal component network1。
And (8): computing matricesE + C eigenvalues η of1≥η2≥...≥ηE+CAnd their respectively corresponding feature vectors beta1,β2,...,βE+CThen, η1,η2,...,ηE+CIn is not less than eta0The number of the characteristic values of (a) is recorded as d; wherein eta is0Representing the mean of E + C characteristic values, i.e. eta0=(η1+η2+...+ηE+C) /(E + C), feature vector β1,β2,...,βE+CAre all 1.
And (9): will beta1,β2,...,βdSecond output layer series matrix B combined into simplified core principal component network2=[β1,β2,...,βd]Then, the first monitoring index vector Q is calculated according to the formula (v)1And a second monitoring index vector Q2:
In the above formula, diag { } denotes an operation of converting a matrix diagonal element in braces into a column vector,andrespectively represent Λ1=B1 TZZTB1And Λ2=B2 TZZTB2The inverse matrix of (c).
Step (10): respectively combine Q1And Q2The element of the maximum value in (b) is recorded as the upper control limit q1And q is2Thereafter, the off-line modeling process is ended and step (11) is performed.
Step (11): the method comprises the steps of utilizing a data acquisition system of the wind driven generator to acquire 11 data of the wind driven generator at the latest sampling moment and forming a sample data vector x belonging to R11×1Judging whether the first data in x is between the cut-in wind speed and the cut-out wind speed or not; if yes, executing step (12); if not, the wind driven generator is in an offline state, and the step (11) is repeated to continue to utilize the sample data vector at the latest sampling moment to implement state monitoring; wherein, 11 data in x are arranged in sequence according to the sequence in the step (1).
Step (12): subjecting the elements of each row in x to the same normalization process as in step (2), thereby obtaining a data vector
Step (13): will be provided withWhen the vector is made into an input vector, the hidden layer neuron output vector z E R is obtained by calculation according to the formula ∈ R(E+C)×1:
Step (14): according to the formulaAndcalculating to obtain a monitoring index J1And J2Then, whether the conditions are met is judged again: j. the design is a square1≤q1And J2≤q2(ii) a If yes, the wind power generation equipment operates normally at the current sampling moment, and the step (11) is returned; if not, executing step (15).
Step (15): returning to the step (11) to continue to monitor the state of the wind power generation equipment by using the sample data vector at the latest sampling moment; if the monitoring indexes of the continuous 6 latest sampling moments do not meet the judgment condition in the step (14), triggering a working abnormity alarm, and arranging personnel to maintain in time; and otherwise, the wind power generation equipment normally operates, and the step (11) is returned to continue to monitor the state of the wind power generation equipment.
The main advantages of the process of the invention compared to conventional processes are detailed below.
It can be seen from the above step (13) that the online calculation amount of the method of the present invention mainly comes from the formula (C), i.e., a vector containing E + C data is calculated. When the traditional kernel principal component analysis algorithm is implemented on-line monitoring, the traditional kernel principal component analysis algorithm needs to be aimed atA column vector containing n data is calculated as follows:
since n is equal to the total number of sample data vectors, and E and C represent the number of edge feature points and cluster center points, respectively, E + C is generally much smaller than n. Therefore, the online calculation amount of the method is greatly reduced.
In addition, corresponding outputs are respectively calculated through the two output layer coefficient matrixes, and compared with the traditional kernel principal component analysis algorithm with only one output, the diversity of feature extraction is optimized and guaranteed. Finally, the present specification will further explain the advantages of the process of the present invention by means of specific implementation procedures.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
FIG. 2 is a schematic diagram of a simplified kernel principal component network structure according to the method of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
The invention discloses a wind power generation equipment state monitoring method based on a simplified core principal component network, and the specific implementation mode of the method is described by combining with an implementation flow diagram shown in figure 1.
Step (1): after the data which can be measured in real time by the wind power generation equipment is determined, the data are collected according to a fixed sampling time interval under the normal operation state of the wind power generation equipment.
Step (2): n sample data vectors x with wind speed between 3 and 25 meters per second1,x2,...,xnThe composition matrix X ═ X1,x2,...,xn]And for X ∈ R11×nEach row vector is normalized to obtain a new matrix
And (3): for new matrixAfter kernel principal component analysis is implemented, the kernel parameter delta and the kernel score matrix Y epsilon R are reservedD×nThe specific implementation process is shown in the steps (3.1) to (3.5).
And (4): for new matrixColumn vector ofCarrying out edge characteristic point analysis to obtain E edge point vectors xi1,ξ2,...,ξEThe specific implementation process is shown in the steps (4.1) to (4.5).
And (5): for new matrixColumn vector ofPerforming cluster analysis to obtain C cluster central point vectors h1,h2,...,hC。
In this embodiment, the k-means clustering (k-means clustering) algorithm is used in the step (5). It should be noted that the implementation step (5) of the method of the present invention is not limited to the use of k-means clustering algorithm, as long as the column vector can be implementedClustering analysis algorithms that cluster into C clusters can be used.
And (6): respectively sequentially arranging the column vectorsComputing the hidden layer neuron output vector z of the simplified kernel principal component network according to the formula (iv) as an input vector1,z2,...,zn。
And (7): will z1,z2,...,znMerging into a hidden layer output matrix Z ═ Z with (E + C) x n dimension1,z2,...,zn]Then according to formula B1=(ZTZ)-1ZYTCalculating to obtain a first output layer coefficient matrix B of the simplified core principal component network1。
And (8): computing matricesE + C eigenvalues η of1≥η2≥...≥ηE+CAnd their respectively corresponding feature vectors beta1,β2,...,βE+CThen, η1,η2,...,ηE+CIn is not less than eta0The number of the characteristic values of (a) is recorded as d;
and (9): will beta1,β2,...,βdSecond output layer series matrix B combined into simplified core principal component network2=[β1,β2,...,βd]Then, according to the above-mentioned formula (v), respectively calculating first monitoring index vector Q1And a second monitoring index vector Q2。
Passing through two output layer series matrixes B in the step (7) and the step (9)1And B2A reduced core principal component network structure as shown in fig. 2 can be built. As can be seen from FIG. 2, the number of hidden layer neurons of the reduced kernel principal component network according to the method of the present invention is equal to E + C, and the output layer neurons are divided into two groups, respectively using B1And B2As their respective coefficient matrices.
Step (10): respectively combine Q1And Q2The element of the maximum value in (b) is recorded as the upper control limit q1And q is2Thereafter, the off-line modeling process is ended and step (11) is performed.
Step (11): by utilizing a data acquisition system of the wind driven generator, acquiring a sample data vector x ∈ R at the latest sampling moment11×1Judging whether the first data in x is between the cut-in wind speed and the cut-out wind speed or not; if yes, executing step (12); if not, the wind driven generator is in an offline state, and the step (11) is repeated to continue to utilize the sample data vector at the latest sampling moment to implement state monitoring; wherein, 11 data in x need to be arranged in sequence according to the sequence in the step (1).
Step (12): subjecting the elements of each row in x to the same normalization process as in step (2), thereby obtaining a data vector
Step (13): will be provided withWhen the vector is made into an input vector, the hidden layer neuron output vector z E R is obtained by calculation according to the formula |(E+C)×1。
Step (14): according to the formulaAndcalculating to obtain a monitoring index theta1And theta2Then, whether the conditions are met is judged again: theta1≤q1And theta2≤q2(ii) a If yes, the wind power generation equipment operates normally at the current sampling moment, and the step (11) is returned; if not, executing step (15).
Step (15): returning to the step (11) to continue to monitor the state of the wind power generation equipment by using the sample data vector at the latest sampling moment; if the monitoring indexes at the continuous 6 sampling moments do not meet the judgment condition in the step (14), triggering an abnormal alarm, and arranging personnel to maintain in time; and otherwise, the wind power generation equipment normally operates, and the step (11) is returned to continue to monitor the state of the wind power generation equipment.
Claims (1)
1. A wind power generation equipment state monitoring method based on a simplified core principal component network is characterized by comprising the following steps:
step (1): after data which can be measured in real time by the wind power generation equipment is determined, acquiring the data according to a fixed sampling time interval under the normal operation state of the wind power generation equipment; wherein, 11 data that can real-time measurement at each sampling moment are specifically in turn: wind speed, rotor speed, generator speed, mechanical torque, generated power, blade pitch angle, blade azimuth, blade root moment, top horizontal axis acceleration, top longitudinal axis acceleration and yaw error;
step (2): the wind speed is between the cut-in wind speed andn sample data vectors x between cut-out wind speeds1,x2,…,xnThe composition matrix X ═ X1,x2,…,xn]And for X ∈ R11×nEach row vector is normalized to obtain a new matrixWherein the ith sample data vector xi∈R11×1The 11 data in (1) are arranged in sequence, i belongs to {1, 2, …, n }, R belongs to11×nRepresenting a matrix of real numbers in dimensions 11 Xn, R representing a set of real numbers, R11×1A real number vector representing 11 × 1 dimensions;
and (3): for new matrixAfter kernel principal component analysis is implemented, the kernel parameter delta and the kernel score matrix Y epsilon R are reservedD×nThe specific implementation process is shown in the steps (3.1) to (3.5);
step (3.1): according to the formulaComputing new matricesVector of the ith columnAnd j-th column vectorThe squared distance ζ (i, j) therebetween; wherein j belongs to {1, 2, …, n }, and the upper label T represents the transpose of a matrix or a vector;
step (3.2): the kernel parameter δ is determined according to the formula (i) as follows:
step (3.3): calculating a kernel matrix K epsilon R according to the formula ∈ Rn×nRow i and column j element K (i, j):
in the above formula, e represents a natural constant;
step (3.4): calculating n eigenvalues λ of the kernel matrix K1≥λ2≥…≥λnAnd its corresponding feature vector alpha1,α2,…,αnThen, the lambda is added1,λ2,…,λnIn is not less than λ0The number of the characteristic values is recorded as D; wherein the feature vector alpha1,α2,…,αnAre all equal to 1, lambda0=(λ1+λ2+…+λn)/n;
Step (3.5): d feature vectors alpha1,α2,…,αDForming a characteristic transformation matrix A ═ alpha1,α2,…,αD]Then, according to the formula Y ═ ATK is calculated to obtain a kernel score matrix Y epsilon RD×n;
And (4): for new matrixColumn vector ofCarrying out edge characteristic point analysis to obtain E edge point vectors xi1,ξ2,…,ξEThe specific implementation process is shown in the steps (4.1) to (4.5);
step (4.1): initializing i to 1;
step (4.2): will be provided withThe neutral column vectorN with square distance less than delta betweeniEach column vector is recorded as
In the above formula, the first and second carbon atoms are,representation calculationAndb is 1, 2, …, Ni;
Step (4.4): calculating the column vector according to the formulaCorresponding edge point index gi:
In the above formula, θbThe value law is as follows:
step (4.5): judging whether i is smaller than n; if yes, after i is set to i +1, returning to the step (4.2); if not, according to the value to the edge point index g1,g2,…,gnPerforming descending arrangement, and recording the column vectors corresponding to the largest E edge point indexes as edge point vectors xi1,ξ2,…,ξE;
And (5): for new matrixColumn vector ofPerforming cluster analysis to obtain C cluster central point vectors h1,h2,…,hC;
And (6): respectively sequentially arranging the column vectorsAs an input vector, computing a hidden layer neuron output vector z of the reduced kernel principal component network according to the formula1,z2,…,zn:
In the above formula, zi∈R(E+C)×1The representation corresponds toIs the hidden layer neuron output vector of (i) is in the range of {1, 2, …, n }, R(E+C)×1Represents a real number vector of (E + C) × 1 dimension,
and (7): will z1,z2,…,znMerging into a hidden layer output matrix Z ═ Z with (E + C) x n dimension1,z2,…,zn]Then according to formula B1=(ZTZ)-1ZYTCalculating to obtain a first output layer coefficient matrix B of the simplified core principal component network1;
And (8): computing matricesE + C eigenvalues η of1≥η2≥…≥ηE+CAnd their respectively corresponding feature vectors beta1,β2,…,βE+CThen, η1,η2,…,ηE+CIn is not less than eta0The number of the characteristic values of (a) is recorded as d; wherein the feature vector beta1,β2,…,βE+CAll have a length of 1, eta0=(η1+η2+…+ηE+C)/(E+C);
And (9): will beta1,β2,…,βdSecond output layer series matrix B combined into simplified core principal component network2=[β1,β2,…,βd]Then, the first monitoring index vectors Q are calculated according to the following formula1And a second monitoring index vector Q2:
In the above formula, diag { } denotes an operation of converting a matrix diagonal element in braces into a column vector,andrespectively represent Λ1=B1 TZZTB1And Λ2=B2 TZZTB2The inverse matrix of (d);
step (10): respectively combine Q1And Q2The element of the maximum value in (b) is recorded as the upper control limit q1And q is2Then, ending the off-line modeling process and implementing the step (11);
step (11): acquiring 11 data of the wind power generation equipment at the latest sampling moment, and forming a sample data vector x ∈ R11×1Judging whether the first data in x is between the cut-in wind speed and the cut-out wind speed or not; if yes, executing step (12); if not, the wind driven generator is in an offline state, and the step (11) is repeated to continue to utilize the sample data vector at the latest sampling moment to implement state monitoring; wherein, 11 data in x need to be arranged in sequence according to the sequence in the step (1);
step (12): the same normalization process as in step (2) is applied to the data of each row in x, thereby obtaining a data vector
Step (13): will be provided withWhen an input vector is made, the hidden layer neuron output vector z epsilon R is obtained by utilizing the following formula [ + ](E+C)×1:
Step (14): according to the formulaAndcalculating to obtain a monitoring index J1And J2Then, whether the conditions are met is judged again: j. the design is a square1≤q1And J2≤q2(ii) a If yes, the wind power generation equipment operates normally at the current sampling moment, and the step (11) is returned; if not, executing the step (15);
step (15): returning to the step (11) to continue to monitor the state of the wind power generation equipment by using the sample data vector at the latest sampling moment; if the monitoring indexes of the continuous 6 latest sampling moments do not meet the judgment condition in the step (14), triggering an abnormal alarm and arranging personnel to maintain in time; and otherwise, the wind power generation equipment normally operates, and the step (11) is returned to continue to monitor the state of the wind power generation equipment.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111156322.1A CN113935408B (en) | 2021-09-22 | Wind power generation equipment state monitoring method based on simplified core principal component network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111156322.1A CN113935408B (en) | 2021-09-22 | Wind power generation equipment state monitoring method based on simplified core principal component network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113935408A true CN113935408A (en) | 2022-01-14 |
CN113935408B CN113935408B (en) | 2024-07-02 |
Family
ID=
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111914471A (en) * | 2020-06-13 | 2020-11-10 | 宁波大学 | Rectification tower fault detection method based on rapid nuclear independent component analysis |
CN111914889A (en) * | 2020-06-13 | 2020-11-10 | 宁波大学 | Rectifying tower abnormal state identification method based on brief kernel principal component analysis |
CN111914886A (en) * | 2020-06-13 | 2020-11-10 | 宁波大学 | Nonlinear chemical process monitoring method based on online brief kernel learning |
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111914471A (en) * | 2020-06-13 | 2020-11-10 | 宁波大学 | Rectification tower fault detection method based on rapid nuclear independent component analysis |
CN111914889A (en) * | 2020-06-13 | 2020-11-10 | 宁波大学 | Rectifying tower abnormal state identification method based on brief kernel principal component analysis |
CN111914886A (en) * | 2020-06-13 | 2020-11-10 | 宁波大学 | Nonlinear chemical process monitoring method based on online brief kernel learning |
Non-Patent Citations (1)
Title |
---|
黄保海;李岩;王东风;韩璞;: "基于KPCA和KFCM集成的汽轮机故障诊断", 电力自动化设备, no. 07, 10 July 2010 (2010-07-10) * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Qian et al. | Data-driven condition monitoring approaches to improving power output of wind turbines | |
CN112613554B (en) | Wind driven generator variable pitch system fault prediction method and system | |
CN105069236B (en) | Consider the broad sense load joint probability modeling method of wind power plant node space correlation | |
CN110889535B (en) | Multi-point wind speed prediction method in wind power plant based on convolution cyclic neural network | |
CN110009135B (en) | Wind power prediction method based on width learning | |
CN111091236A (en) | Multi-classification deep learning short-term wind power prediction method classified according to pitch angles | |
CN108196444A (en) | Based on the control of the variable pitch wind energy conversion system of feedback linearization sliding formwork and SCG and discrimination method | |
CN105649896B (en) | Wind turbines intelligent monitoring operation control system and its control method | |
Pang et al. | Multi kernel fusion convolutional neural network for wind turbine fault diagnosis | |
CN113935408A (en) | Wind power generation equipment state monitoring method based on simplified nuclear principal component network | |
CN113935408B (en) | Wind power generation equipment state monitoring method based on simplified core principal component network | |
CN112039051A (en) | Real-time modeling method for accessing double-fed wind driven generator into substation bus load | |
CN112270081B (en) | Wind driven generator fault detection method based on parallel Elman-NN | |
CN109611268A (en) | A kind of bilobed wheel horizontal-shaft wind turbine design optimization method | |
CN113946608B (en) | Wind generating set running state monitoring method based on online RBF neural network | |
Wenzhou et al. | Design of a large scale wind turbine generator set yaw system | |
Qian et al. | A weighted kNN fault detection based on multistep index and dynamic neighborhood scale under complex working conditions | |
CN112232427B (en) | Wind driven generator fault detection method based on distributed regression model | |
Tingting et al. | Early warning method for power station auxiliary failure considering large-scale operating conditions | |
CN110985287A (en) | Indirect rotating speed control method based on width learning | |
CN112267978B (en) | Wind driven generator fault detection method based on distributed ESN model | |
CN113822344B (en) | Method for monitoring state of front bearing of generator of wind turbine generator based on data driving | |
Peng et al. | Short-Term Prediction of Generator Blade Ice Fault Based on Multi-AN | |
CN116976650B (en) | Power grid lean management regulation and control method based on big data | |
Li | Research on New Energy Power Generation Power Prediction Method Based on Machine Learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant |