CN104268436A

CN104268436A - Trend fitting based wind power intraday fluctuation continuous period recognition method and system

Info

Publication number: CN104268436A
Application number: CN201410557019.6A
Authority: CN
Inventors: 王现勋; 梅亚东; 孔艳君; 杨璐; 魏翔; 徐雨妮
Original assignee: Wuhan University WHU
Current assignee: Wuhan University WHU
Priority date: 2014-10-20
Filing date: 2014-10-20
Publication date: 2015-01-07
Anticipated expiration: 2034-10-20
Also published as: CN104268436B

Abstract

The invention discloses a trend fitting based wind power intraday fluctuation continuous period recognition method and system. The trend fitting based wind power intraday fluctuation continuous period recognition method comprises performing analysis on a wind power output process line, performing subinterval length division, trend fitting and overlapped portion fitting according to characteristic parameters of different subintervals and obtaining the integral process line fitting sequence and performing fluctuation sensitivity fitting, fluctuation recognition and continuous period recognition. According to the trend fitting based wind power intraday fluctuation continuous period recognition method and system, the period of the wind power output process line fluctuation is automatically extracted and accordingly a new judgment and recognition mode is provided, the result is simple and clear, and the implementation is convenient and easy. Compared with the prior art, the new way is provided for the period level of wind power output process line fluctuation recognition, the important innovation is achieved, the judgment of the wind power output stability analysis is facilitated, the wind power operation cost can be reduced, the wind power application efficiency can be improved, and the important actual application significance is brought to the wind power operation management and the wind power compensation adjustment.

Description

Wind power intraday fluctuation continuous time interval identification method and system based on trend fitting

Technical Field

The invention relates to the field of wind power operation stability analysis, in particular to a wind power intraday fluctuation time period identification method and system based on trend fitting.

Background

The wind power operation stability analysis research is mainly used for providing decision support for wind power operation management and wind power compensation adjustment. The improvement of the precision of the wind power operation stability analysis has great significance for improving the utilization of wind energy, saving energy and reducing emission. The wind power fluctuation time interval is mainly characterized in that the fluctuation degree of the wind power output in the time interval is large. At present, the method for depicting the fluctuation degree of wind power output at home and abroad mainly comprises the following steps: standard deviation of the output, sum of absolute values of difference of the output in adjacent time periods, first-order difference probability distribution of the output, sum of slopes of lines in the process of the output and the like. The existing method mostly focuses on judging the fluctuation degree of the output in the whole output process or a certain time period, and the fluctuation degree of the output in certain continuous time periods in the output process cannot be identified. The recognition of the degree of fluctuation of the output for successive periods of time can then be used to analyze the duration of the state of fluctuation and the redundancy required for the wind-power compensation adjustment. In view of the above, if the wind power fluctuation can be subjected to continuous period fluctuation identification, a powerful reference can be provided for wind power operation management and wind power compensation adjustment.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a wind power intraday fluctuation continuous time interval identification scheme based on trend fitting so as to judge the fluctuation of continuous time intervals.

In order to achieve the purpose, the invention provides a wind power intraday fluctuation continuous time interval identification method based on trend fitting, which comprises the following steps of:

step 1, analyzing a wind power output process line, wherein the wind power output process line is a curve obtained by taking time t as a horizontal coordinate and wind power output P as a vertical coordinate in a rectangular coordinate system according to the output process; the analysis process comprises the steps of analyzing a wind power output process line to be formed by connecting a plurality of points, setting a total of N points, respectively using the N points as control points, numbering the control points from left to right in sequence to be 1,2, …, N, and recording the coordinate of the ith control point as (t)_i,P_i) N, the vertical coordinates Pi of all control points in the wind power output process line form a sequence { P ═ 1,2_iThe total time interval length of the line in the wind power output process is 24h, and the time intervals between adjacent control points are kept consistent; step 2, inputting a subinterval characteristic parameter set { M₁,M₂,...,M_RThen initializing the current iteration times r to 1;

wherein M is_rThe R-th subinterval characteristic parameter is an integral multiple of unit, R is 1,2, …, R is M_rThe number of elements in the set; unit is the time interval between adjacent control points in the wind power output process line;

step 3, calculating the subinterval division length according to the current iteration number r as follows,

s＝2m

m = \frac{M}{unit}

wherein s is the sub-interval length, M is the sub-interval overlap length, and M is M_r；

Step 4, according to the subinterval length s, the wind power output process line is divided as follows,

dividing the part between the (K-1) × m +1 control point and the (K +1) × m +1 control point of a wind power output process line into kth subintervals, wherein K is 1, 2. If it is notThenOtherwiseint (, denotes rounding ";

dividing the part from the Kxm +1 th control point to the Nth control point left by the wind power output process line into a Km +1 th sub-interval;

the first K subintervals are composed of 2m +1 control points, and the coordinates of the control points are recorded asThe K +1 sub-interval consists of N-Kxm control points, and the coordinates of the control points are recorded in sequence

(t_{1}^{K + 1}, P_{1}^{K + 1}), (t_{2}^{K + 1}, P_{2}^{K + 1}), . . ., (t_{N - Km}^{K + 1}, P_{N - Km}^{K + 1});

Step 5, performing trend fitting on each subinterval obtained in the step 4 to obtain a corresponding fitting sequence; the coordinates of the control points of the fitting sequence of the first K subintervals are recorded as

(t_{1}^{k, f}, P_{1}^{k, f}), (t_{2}^{k, f}, P_{2}^{k . f}), . . ., (t_{2 m + 1}^{k, f}, P_{2 m + 1}^{k, f}), k = 1, 2, . . ., K,

The coordinates of the control points of the fitting sequence of the K +1 th subinterval are recorded as

(t_{1}^{K + 1, f}, P_{1}^{K + 1, f}), (t_{2}^{K + 1, f}, P_{2}^{K + 1, f}), . . ., (t_{N - Km}^{K + 1, f}, P_{N - Km}^{K + 1, f});

Step 6, the calculation of the overlap fitting weighted sequence is carried out as follows,

k overlapping parts are totally obtained among the K +1 subintervals obtained in the step 4, the fitting weighted sequence of each overlapping part is calculated by the following formula,

wherein,the ordinate, K ═ 1, 2., K, λ, of the l-th control point in the fitted weighted sequence for the K-th overlap section₁、λ₂Is a weight coefficient, l 1, 2.., m +1,is the (l + m) th control point ordinate of the fitted sequence for the kth subinterval,is the longitudinal coordinate, lambda, of the ith control point in the fitted sequence of the (k +1) th subinterval₁＝1-(l-1)/m，λ₂＝(l-1)/m；

The coordinates of the control points of the fitted weighted sequence of the k-th overlap are recorded asProcessing the K overlapped parts to obtain K fitting weighted sequences, wherein the coordinate of the last control point of the previous fitting weighted sequence is the same as the coordinate of the first control point of the next fitting weighted sequence, and connecting the K fitting weighted sequences after the weight removal end to obtain the fitting weighted sequence of the overlapped part of the whole process line, wherein the sequence comprises K multiplied by m +1 control points;

step 7, generating a line fitting sequence in the whole process as follows,

fitting the first m control points of the first subinterval fitting sequence obtained in the step 5The first part is K × m +1 control points of the fitting weighted sequence of the whole process line overlapping part obtained in the step 6, and the second part is the last N-K × m- (m +1) points of the K +1 sub-interval fitting sequence obtained in the step 5

As a third part, sequentially connecting the three parts end to obtain a line fitting sequence in the whole process;

step 8, recording the vertical coordinate of the whole process line fitting sequence obtained in the step 7 as { P_i,rIf R is less than R, R +1, returning to step 3; if R is equal to R, go to step 9;

step 9, the fluctuation sensitivity is formulated as follows,

the vertical coordinate sequence { P) of the original wind power output process line_iCombining the fitting sequences with R whole process line fitting sequences obtained in the previous iteration execution steps 3-8 to obtain an Nx (R +1) matrix A;

calculating the standard deviation of the set composed of elements contained in each row of the matrix A in turn and recording the standard deviation as sigma_iI 1, 2.., N, to get a sequence of standard differences { σ }_i}; according to sigma_iSorting the values from large to small to obtain a sorted standard deviation sequence { sigma'_i1,2, N, according to σ'_iIn sequence { sigma'_iCalculating the corresponding frequency according to the sequence ofObtaining N parameter combinations (sigma'_i，η_i) (ii) a In N number (σ'_i，η_i) Reading in combination the η closest to the predetermined parameter η_iSigma 'corresponding to'_iA value and assigned to a standard deviation threshold parameter σ;

in step 10, the wave motion recognition is performed as follows,

the sequence of standard deviations { σ } is obtained in step 9_iN, in order of increasing i value, i is 1,2Identification judgment including when sigma_iWhen > σ, the parameters corresponding to the value of i are combined (t_i,σ_i) Sequentially encoding the vector as a row vector into a matrix B;

after the identification is finished, the number of rows of the matrix B is recorded as a, the first column of the matrix B is a wind power intraday fluctuation period sequence and is recorded as { t'_ii1,2, a, the second column being the fluctuation degree of the corresponding time interval, denoted as { σ'_ii}，ii＝1,2,...,a；

Step 11, the identification of the continuous time interval is performed as follows,

according to the wind power intraday fluctuation time period sequence { t 'obtained in the step 10'_iiIdentification of continuous periods in the order of the ii value from small to large, including when t'_ii+1-t′_iiUnit, where ii 1,2, a-2, then the parameters corresponding to the value of ii are combined (t'_ii,σ′_ii) Successively programmed into matrix C as a row vector, otherwise combined with the parameter corresponding to the value ii (t'_ii,σ′_ii) Sequentially encoding the vector as a row vector into a matrix D; when t'_a-t′_a-1Combining the parameters (t'_a-1,σ′_a-1)、(t′_a,σ′_a) Coding into the C end of the matrix as a row vector in sequence, otherwise combining the parameters (t'_a-1,σ′_a-1)、(t'_a,σ'_a) Sequentially encoding the row vectors into the tail end of the matrix D;

the first column of the matrix C is a continuous time interval sequence fluctuating within a wind power day, and the second column is the fluctuation degree of a corresponding time interval; the first column of the matrix D is a discontinuous period sequence fluctuating within a wind power day, and the second column is the fluctuation degree of a corresponding period.

The invention also correspondingly provides a wind power intraday fluctuation continuous period identification system based on trend fitting, which comprises the following modules: the analysis module is used for analyzing a wind power output process line, wherein the wind power output process line is a curve obtained by taking time t as a horizontal coordinate and wind power output P as a vertical coordinate in a rectangular coordinate system according to the output process; the analysis process comprises analyzing the wind power output process line as ifThe dry point is connected and formed, a total of N points are set, the N points are respectively taken as control points and are numbered from left to right as 1,2, …, N, and the coordinate of the ith control point is marked as (t)_i,P_i) 1,2, N, the ordinate P of all control points in the wind power output process line_iComposition sequence { P_iThe total time interval length of the line in the wind power output process is 24h, and the time intervals between adjacent control points are kept consistent;

an initialization module for inputting a subinterval feature parameter set { M }₁,M₂,...,M_RThen initializing the current iteration times r to 1;

an interval length determining module, for calculating the subinterval division length according to the current iteration number r as follows,

s＝2m

m = \frac{M}{unit}

The subinterval segmentation module is used for performing the following segmentation on the wind power output process line according to the subinterval length s,

(t_{1}^{K + 1}, P_{1}^{K + 1}), (t_{2}^{K + 1}, P_{2}^{K + 1}), . . ., (t_{N - Km}^{K + 1}, P_{N - Km}^{K + 1});

The trend fitting module is used for performing trend fitting on each subinterval obtained by the subinterval segmentation module to obtain a corresponding fitting sequence; the coordinates of the control points of the fitting sequence of the first K subintervals are recorded asThe coordinates of the control points of the fitting sequence of the K +1 th subinterval are denoted as K1, 2

(t_{1}^{K + 1, f}, P_{1}^{K + 1, f}), (t_{2}^{K + 1, f}, P_{2}^{K + 1, f}), . . ., (t_{N - Km}^{K + 1, f}, P_{N - Km}^{K + 1, f});

An overlap fit weighting module for performing an overlap portion fit weighting sequence calculation as follows,

k overlapping parts are shared among K +1 subintervals obtained by the subinterval division module, the fitting weighted sequence of each overlapping part is calculated by the following formula,

wherein,the ordinate, K, of the l control point in the fitted weighting sequence for the K-th overlap portion is 1,2，λ₁、λ₂Is a weight coefficient, l 1, 2.., m +1,is the (l + m) th control point ordinate of the fitted sequence for the kth subinterval,is the longitudinal coordinate, lambda, of the ith control point in the fitted sequence of the (k +1) th subinterval₁＝1-(l-1)/m，λ₂＝(l-1)/m；

a process line fitting module for performing the whole process line fitting sequence generation as follows,

fitting the first m control points of the first subinterval fitting sequence obtained by the trend fitting moduleTaking K × m +1 control points of the fitting weighted sequence of the whole process line overlapping part obtained by the overlapping fitting weighted module as a first part, and taking the last N-K × m- (m +1) points of the K +1 sub-interval fitting sequence obtained by the trend fitting module as a second part

iterationA judging module for recording the vertical coordinate of the whole process line fitting sequence obtained by the process line fitting module as { P_i,rIf R is less than R, R is equal to R +1, and a command interval length determining module works; if R is equal to R, commanding the sensitivity module to work;

a sensitivity module for performing a fluctuation sensitivity planning as follows,

the vertical coordinate sequence { P) of the original wind power output process line_iCombining the sequence and R whole process line fitting sequences obtained by previous iteration to obtain an N (R +1) matrix A;

a fluctuation identification module for performing fluctuation identification as follows,

standard deviation sequence { sigma over sensitivity module_iAnd (e), identifying and judging the i values from small to large according to the sequence of the i values, including the step of judging when the sigma is larger_iWhen > σ, the parameters corresponding to the value of i are combined (t_i,σ_i) Sequentially encoding the vector as a row vector into a matrix B;

A period identification module for performing continuous period identification as follows,

wind power intraday fluctuation time period sequence { t 'obtained according to fluctuation identification module'_iiIdentification of continuous periods in the order of the ii value from small to large, including when t'_ii+1-t′_iiUnit, where ii 1,2, a-2, then the parameters corresponding to the value of ii are combined (t'_ii,σ′_ii) Successively programmed into matrix C as a row vector, otherwise combined with the parameter corresponding to the value ii (t'_ii,σ′_ii) Sequentially encoding the vector as a row vector into a matrix D; when t'_a-t'_a-1Combining the parameters (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Coding into the C end of the matrix as a row vector in sequence, otherwise combining the parameters (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Sequentially encoding the row vectors into the tail end of the matrix D;

The technical scheme for identifying the wind power day fluctuation time period based on trend fitting provided by the invention provides a new judgment and identification mode by automatically extracting the wind power output process line fluctuation time period, and has the advantages of simple and clear result and simple and easy implementation. Compared with the prior art, the method provides a new way for identifying the line fluctuation of the wind power output process in the continuous time interval level, is an important innovation in the technical field, is favorable for judging the stability analysis of the wind power output, reducing the wind power operation cost and improving the wind power application efficiency, and has important practical application significance for wind power operation management and wind power compensation adjustment.

Drawings

FIG. 1 is a diagram illustrating subinterval segmentation according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of the process line of each column vector in the matrix A according to the embodiment of the present invention;

fig. 3 is a diagram of a recognition result of a fluctuation time period within a wind power day according to an embodiment of the present invention.

Detailed Description

In specific implementation, the technical scheme of the invention can adopt a computer software technology to realize an automatic operation process. In order to prove the reasonability of the technical scheme, the embodiment takes the actually measured daily output process line of the wind power plant of a certain wind power base in China as an example, and the fluctuation time interval is identified. The flow of the embodiment comprises the following steps:

step 1, analyzing a wind power output process line

The wind power output process line is a curve obtained by taking time t as a horizontal coordinate and wind power output P as a vertical coordinate in a rectangular coordinate system according to the output process; the analysis process comprises the steps of analyzing a wind power output process line to be composed of a plurality of connected points, setting a total number of N points which are respectively used as control points and are numbered from left to right as 1,2, … and N (normally N is more than or equal to 96), and recording the coordinate of the ith control point as (t)_i,P_i) 1,2,., N, the same applies below. Ordinate P of all control points of wind power output process line_i(i ═ 1, 2.., N) constitutes the sequence { P_i}. The total time interval length of the line in the wind power output process is 24h, namely t_N-t₁24 h; the time interval between adjacent control points remains consistent.

Step 2, inputting a subinterval characteristic parameter set { M₁,M₂,...,M_RAnd then initializing the current iteration number r as 1.

Wherein: m_rIs a subinterval characteristic parameter, the unit of which is min (minute) and is an integral multiple of unit; unit is the length of resolution of time t in wind power output process lineI.e. the time interval between adjacent control points, unit ═ t₂-t₁In units of min; advisingThe specific value can be adjusted and set by a person skilled in the art according to the change trend of the wind power output within the day in advance, and the more violent the change is, the M is_rPreferably smaller, R is 1,2, …, R and R are M_rThe number of elements in the set is also the upper limit of the number of iterations. As shown in FIG. 2, setting M₁＝60min，M₂＝90min，M₃＝120min，M₄＝150min，M₅180min, the first row is the original wind power output process line, and the second row to the sixth row respectively correspond to M₁To M₅。

Step 3, calculating the subinterval division length according to the current iteration times r

s＝2m

m = \frac{M}{unit}

Wherein: s is the subinterval division length; m is the subinterval overlap length; m is M_rThat is, when the step is calculated 1 st time, M is M₁When the calculation of the step is carried out subsequently, M is taken out in sequence₂,M₃,...,M_R。

Step 4, segmenting the wind power output process line

And (4) performing output process line segmentation according to the current subinterval length s obtained in the step (3) executed in the iteration. And dividing the part from the (K-1) × m +1 control point to the (K +1) × m +1 control point of the output process line into kth subintervals, wherein K is 1, 2.

If it is not

Then

K = ini (\frac{N - 1}{m}) - 2,

Otherwise

K = ini (\frac{N - 1}{m}) - 1,

int (, denotes rounding ""). And dividing the part between the Kxm +1 control point and the Nth control point which are left in the force process line into K +1 sub-intervals. And m +1 points of the two adjacent subintervals are overlapped, namely the rear m +1 point of the front subinterval is overlapped with the front m +1 point of the rear subinterval. The first K subintervals are composed of 2m +1 control points, the lengths of the control points are all 2m, and the coordinates of the control points are sequentially recorded as:the K +1 sub-interval consists of N-Kxm control points, the length of the control points is N-Kxm-1, and the coordinates of the control points are sequentially recorded as:the sub-interval division diagram can be seen in fig. 1, where the sub-interval 1 includes the 1 st point to the 2m +1 st point, and the coordinates are recorded asSubinterval 2 includes m +1 th to 3m +1 th points, and coordinates are recorded asThe subinterval k comprises (k-1) m +1 th point to (k +1) m +1 th point, and the coordinates are recorded asThe subinterval k +1 includes the km +1 th point to the (k +2) m +1 th point, and the coordinates are recorded asThe subinterval K comprises (K-1) m +1 th point to (K +1) m +1 th point, and the coordinates are recorded asThe subinterval K +1 comprises Km +1 th point to Nth point, and the coordinates are recorded as

(t_{1}^{K + 1}, P_{1}^{K + 1}), (t_{2}^{K + 1}, P_{2}^{K + 1}), . . ., (t_{N - Km}^{K + 1}, P_{N - Km}^{K + 1}) .

Step 5, fitting subinterval trend

And (4) adopting a polynomial fitting method, wherein the polynomial fitting frequency is 2, and performing trend fitting on each subinterval obtained in the step (4) to obtain a corresponding fitting sequence. The coordinates of the control points of the fitting sequence of the first K subintervals are noted as:

(t_{1}^{k, f}, P_{1}^{k, f}), (t_{2}^{k, f}, P_{2}^{k . f}), . . ., (t_{2 m + 1}^{k, f}, P_{2 m + 1}^{k, f}), k = 1, 2, . . ., K,

(t_{1}^{K + 1, f}, P_{1}^{K + 1, f}), (t_{2}^{K + 1, f}, P_{2}^{K + 1, f}), . . ., (t_{N - Km}^{K + 1, f}, P_{N - Km}^{K + 1, f}) .

Step 6, calculating the fitting weighted sequence of the overlapped part

The K total overlapping parts are obtained from the K +1 subintervals obtained in the step 4, and the fitting weighted sequence of each overlapping part is calculated by the following formula:

wherein:the ordinate, K ═ 1, 2., K, λ, of the l-th control point in the fitted weighted sequence for the K-th overlap section₁、λ₂Is a weight coefficient, l 1, 2.., m +1,is the (l + m) th control point ordinate of the fitted sequence for the kth subinterval,is the ith control point ordinate in the fitted sequence of the (k +1) th subinterval. Lambda [ alpha ]₁＝1-(l-1)/m，λ₂(l-1)/m. The significance of the formula is that the weighting processing is carried out according to the influence degree of the fitting trend of the overlapping parts of the adjacent subintervals.

Let the control point coordinates of the fitted weighted sequence for the kth overlap be:and processing the K overlapped parts to obtain K fitting weighted sequences, wherein the coordinate of the last control point of the previous fitting weighted sequence is the same as the coordinate of the first control point of the next fitting weighted sequence. And carrying out the duplicate removal operation. The specific processing of the de-weighting is to reserve the coordinates of the last control point of the previous fitting weighting sequence and remove the coordinates of the first control point of the next fitting weighting sequence. After the processing, the 1 st fitting weighted sequence contains m +1 control points, and the rest fitting weighted sequences contain m control points. And connecting the K fitting weighted sequences after the duplication elimination end to obtain a fitting weighted sequence of the overlapped part of the whole process line, wherein the sequence comprises K multiplied by m +1 control points.

Step 7, generating a line fitting sequence in the whole process

As a third part, the three parts are connected end to end in sequence to obtain a whole process line fitting sequenceThe number of the control points is equal to the wind power output process line in the step 1, the control points are all N, and the abscissa of the two sequences is the same.

Step 8, circulating treatment

When the calculation of the R (R ═ 1,2, …, R) is performed up to this step, the ordinate of the whole process line fitting sequence obtained in step 7 is recorded as: { P_i,r1, 2. If R is less than R, making R equal to R +1, and returning to the step 3; if R is equal to R, go to step 9.

Step 9, drawing up fluctuation sensitivity

The vertical coordinate sequence { P) of the original wind power output process line_iAnd combining the R fitting sequences of the whole process line obtained by the previous iteration execution steps 3-8 into an N x (R +1) matrix A. Calculating the standard deviation of the set composed of elements contained in each row of the matrix A in turn and recording the standard deviation as sigma_iN, a sequence of standard deviations { σ } is available_i}. According to sigma_iSorting the values from large to small to obtain a sorted standard deviation sequence { sigma'_i1, 2. According to sigma'_iIn sequence { sigma'_iCalculating the corresponding frequency according to the sequence of the frequencyN parameter combinations (sigma ') can be obtained'_i，η_i). In specific implementation, a person skilled in the art can preset the value of the parameter η, 0<η<1, the value of eta can be generally determined according to wind power output fluctuation identification sensitivity, and the higher the sensitivity requirement is, the larger the value of eta is. In N number (σ'_i，η_i) Reading in combination the η closest to the predetermined parameter η_iSigma 'corresponding to'_iAnd is assigned to the standard deviation threshold parameter sigma.

Step 10, wave recognition

The sequence of standard deviations { σ } is obtained in step 9_iAnd (i) 1,2, N, performing identification judgment in the order of the values of i from small to large. When sigma is_iWhen > σ, the parameters corresponding to the value of i are combined (t_i,σ_i) MakingAnd sequentially compiling the row vectors into a matrix B. After the identification is finished, the number of rows of the matrix B is recorded as a. The first column of the matrix B is a wind power intraday fluctuation period sequence which is marked as { t'_iiA), the second column is the fluctuation degree of the corresponding time interval, which is marked as { σ'_ii}，(ii＝1,2,...,a)。

Step 11, continuous period identification

According to the wind power intraday fluctuation time period sequence { t 'obtained in the step 10'_iiAnd (5) carrying out continuous time interval identification according to the sequence of the ii value from small to large. When t'_ii+1-t′_iiUnit (where ii 1, 2.., a-2), then the parameters corresponding to the ii value are combined (t'_ii,σ′_ii) Successively programmed into matrix C as a row vector, otherwise combined with the parameter corresponding to the value ii (t'_ii,σ′_ii) Sequentially encoding the vector as a row vector into a matrix D; when t'_a-t'_a-1Combining the parameters (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Coding into the C end of the matrix as a row vector in sequence, otherwise combining the parameters (t'_a-1,σ'_a-1)、(t'_a,σ'_a) And the row vectors are sequentially coded into the tail end of the matrix D.

The first column of the matrix C is a continuous time interval sequence of wind power day fluctuation, and the second column is the fluctuation degree of the corresponding time interval; the first column of the matrix D is a discontinuous time period sequence fluctuating within a wind power day, and the second column is the fluctuation degree of a corresponding time period.

The process of the present invention may be implemented using MATLAB programming. For the sake of understanding the effect of the present invention, the specific process is illustrated as follows: and reading the wind power output process into MATLAB as original data, and obtaining a fluctuation time period by utilizing the flow provided by the invention. The main parameters take the values as follows: unit 15min, M_r60,90,120,150,180, η is 0.5; the results obtained were: each column process in the matrix A is shown in figure 2, a schematic diagram of the wind power intraday fluctuation identification result is shown in figure 3, and a horizontal line with the P being 50 ten thousand kW in the diagram is a fluctuation identification: "-" indicates that the corresponding period is a continuous fluctuation (matrix C), "×" indicates that the corresponding period is a discontinuous fluctuation(matrix D), fluctuation period characteristic parameter matrix C is as follows.

C parameter table of fluctuation time interval characteristic parameter matrix

From fig. 2, it can be known from comparison between the wind power output process (column 1) and the sequence processes (columns 2 to 6) based on trend fitting that the wind power output process has obvious fluctuation, the left side of the process line has obvious fluctuation of continuous time period, and the right side of the process line also has fluctuation of continuous time period. From fig. 3 and the above table, the results of this example successfully identified fluctuations of successive time periods. Therefore, the method can effectively and automatically identify the continuous time period of the wind power day fluctuation, generate the analysis result of the wind power operation stability and provide decision support for follow-up research.

The invention also correspondingly provides a wind power intraday fluctuation continuous period identification system based on trend fitting, which comprises the following modules: the analysis module is used for analyzing a wind power output process line, wherein the wind power output process line is a curve obtained by taking time t as a horizontal coordinate and wind power output P as a vertical coordinate in a rectangular coordinate system according to the output process; the analysis process comprises the steps of analyzing a wind power output process line to be formed by connecting a plurality of points, setting a total of N points, respectively using the N points as control points, numbering the control points from left to right in sequence to be 1,2, …, N, and recording the coordinate of the ith control point as (t)_i,P_i) N, the vertical coordinates Pi of all control points in the wind power output process line form a sequence { P ═ 1,2_iThe total time interval length of the line in the wind power output process is 24h, and the time intervals between adjacent control points are kept consistent;

wherein M is_rThe R-th subinterval characteristic parameter is an integral multiple of unit, R is 1,2, …, R and R areM_rThe number of elements in the set; unit is the time interval between adjacent control points in the wind power output process line;

s＝2m

m = \frac{M}{unit}

(t_{1}^{K + 1}, P_{1}^{K + 1}), (t_{2}^{K + 1}, P_{2}^{K + 1}), . . ., (t_{N - Km}^{K + 1}, P_{N - Km}^{K + 1});

The trend fitting module is used for performing trend fitting on each subinterval obtained by the subinterval segmentation module to obtain a corresponding fitting sequence; the coordinates of the control points of the fitting sequence of the first K subintervals are noted as:the coordinates of the control points of the fitting sequence of the K +1 th subinterval are denoted as K1, 2

(t_{1}^{K + 1}, P_{1}^{K + 1}), (t_{2}^{K + 1}, P_{2}^{K + 1}), . . ., (t_{N - Km}^{K + 1}, P_{N - Km}^{K + 1});

The coordinates of the control points of the fitted weighted sequence of the k-th overlap are recorded asProcessing the K overlapped parts to obtain K fitting weighted sequences, wherein the coordinate of the last control point of the previous fitting weighted sequence is the same as the coordinate of the first control point of the next fitting weighted sequence, and the K removed-of-weight sequencesThe fitting weighted sequence is connected end to obtain a fitting weighted sequence of the overlapped part of the whole process line, and the sequence comprises Kxm +1 control points;

an iteration judgment module for recording the vertical coordinate of the whole process line fitting sequence obtained by the process line fitting module as { P_i,rIf R is less than R, R is equal to R +1, and a command interval length determining module works; if R is equal to R, commanding the sensitivity module to work;

sequence of standard deviations { sigma'_iAnd (e), identifying and judging the i values from small to large according to the sequence of the i values, including the step of judging when the sigma is larger_iWhen > σ, the parameters corresponding to the value of i are combined (t_i,σ_i) Sequentially encoding the vector as a row vector into a matrix B;

In specific implementation, each module can be realized by adopting a software modularization technology, and the specific implementation corresponds to each step, which is not repeated in the invention.

It should be emphasized that the embodiments described herein are illustrative rather than restrictive, and thus the present invention is not limited to the embodiments described in the detailed description, but other embodiments derived from the technical solutions of the present invention by those skilled in the art are also within the scope of the present invention.

Claims

1. A wind power intraday fluctuation continuous period identification method based on trend fitting is characterized by comprising the following steps:

step 1, analyzing a wind power output process line, wherein the wind power output process line is a curve obtained by taking time t as a horizontal coordinate and wind power output P as a vertical coordinate in a rectangular coordinate system according to the output process; the analysis process comprises the steps of analyzing a wind power output process line to be composed of a plurality of points which are connected, setting N points in total, respectively taking the N points as control points, numbering the control points from left to right to be 1,2, …, N, and recording the coordinates of the ith control point as(t_i,P_i) 1,2, N, the ordinate P of all control points in the wind power output process line_iComposition sequence { P_iThe total time interval length of the line in the wind power output process is 24h, and the time intervals between adjacent control points are kept consistent; step 2, inputting a subinterval characteristic parameter set { M₁,M₂,...,M_RThen initializing the current iteration times r to 1;

s＝2m

m = \frac{M}{unit}

the first K sub-intervals are composed of 2m +1Control point composition, control point coordinates are recorded in sequenceThe K +1 sub-interval consists of N-Kxm control points, and the coordinates of the control points are recorded in sequence

Step 5, performing trend fitting on each subinterval obtained in the step 4 to obtain a corresponding fitting sequence; the coordinates of the control points of the fitting sequence of the first K subintervals are recorded asThe coordinates of the control points of the fitting sequence of the K +1 th subinterval are denoted as K1, 2

step 7, generating a line fitting sequence in the whole process as follows,

fitting the first subinterval to the first m control points of the sequenceAs the first part of the new sequence, the K × m +1 control points of the fitting weighted sequence of the whole process line overlapping part obtained in the step 6 are used as the second part, and the last N-K × m- (m +1) points of the K +1 sub-interval fitting sequence are used as the second partAs a third part, sequentially connecting the three parts end to obtain a line fitting sequence in the whole process;

step 9, the fluctuation sensitivity is formulated as follows,

calculating the standard deviation of the set composed of elements contained in each row of the matrix A in turn and recording the standard deviation as sigma_iI 1, 2.., N, to get a sequence of standard differences { σ }_i}; according to sigma_iSorting values from large to small to obtain a standard deviation sequence { sigma'_i1,2, N, according to σ'_iIn sequence { sigma'_iCalculating the corresponding frequency according to the sequence ofObtaining N parameter combinations (sigma'_i，η_i) (ii) a In N number (σ'_i，η_i) Reading in combination the η closest to the predetermined parameter η_iSigma 'corresponding to'_iA value and assigned to a standard deviation threshold parameter σ;

in step 10, the wave motion recognition is performed as follows,

the sequence of standard deviations { sigma'_iAnd (e), identifying and judging the i values from small to large according to the sequence of the i values, including the step of judging when the sigma is larger_iWhen > σ, the corresponding parameters are combined (t)_i,σ_i) Sequentially encoding the vector as a row vector into a matrix B;

according to the wind power intraday fluctuation time period sequence { t 'obtained in the step 10'_iiIdentification of continuous periods in the order of the ii value from small to large, including when t'_ii+1-t′_iiUnit, where ii 1,2, a-2, the respective parameters are combined (t'_ii,σ′_ii) Are successively programmed into the matrix C as row vectors, otherwise the corresponding parameters are combined (t'_ii,σ′_ii) Sequentially encoding the vector as a row vector into a matrix D; when t'_a-t'_a-1Corresponding parameters are combined (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Sequentially compiling the row vectors into the C end of the matrix, otherwise combining the corresponding parameters (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Sequentially encoding the row vectors into the tail end of the matrix D;

2. A wind power intraday fluctuation continuous period identification system based on trend fitting is characterized by comprising the following modules:

the analysis module is used for analyzing a wind power output process line, wherein the wind power output process line is a curve obtained by taking time t as a horizontal coordinate and wind power output P as a vertical coordinate in a rectangular coordinate system according to the output process; the analysis process comprises the steps of analyzing a wind power output process line to be formed by connecting a plurality of points, setting a total of N points, respectively using the N points as control points, numbering the control points from left to right in sequence to be 1,2, …, N, and recording the coordinate of the ith control point as (t)_i,P_i) 1,2, N, the ordinate P of all control points in the wind power output process line_iComposition sequence { P_iThe total time interval length of the line in the wind power output process is 24h, and the time intervals between adjacent control points are kept consistent;

initialization module for input of the sub-unitInterval feature parameter set { M₁,M₂,...,M_RThen initializing the current iteration times r to 1;

s＝2m

m = \frac{M}{unit}

the first K subintervals are composed of 2m +1 control points, and the coordinates of the control points are recorded asThe K +1 th subinterval isN-Kxm control points, the coordinates of which are recorded in sequence

(t_{1}^{K + 1, f}, P_{1}^{K + 1, f}), (t_{2}^{K + 1, f}, P_{2}^{K + 1, f}), . . ., (t_{N - Km}^{K + 1, f}, P_{N - Km}^{K + 1, f});

fitting the first subinterval to the first m control points of the sequenceAs the first part of the new sequence, the K multiplied by m +1 control points of the fitting weighted sequence of the whole process line overlapping part obtained by the overlapping fitting weighted module are used as the second part, and the last N-K multiplied by m- (m +1) points of the fitting sequence of the K +1 sub-interval are used as the second partAs a third part, sequentially connecting the three parts end to obtain a line fitting sequence in the whole process;

an iteration judgment module for recording the vertical coordinate of the whole process line fitting sequence obtained by the process line fitting module as { P_i,rIf R is less than R, R is equal to R +1, and a command interval length determining module works; if R ═ R, command sensitivityThe degree module works;

sequence of standard deviations { sigma'_iAnd (e), identifying and judging the i values from small to large according to the sequence of the i values, including the step of judging when the sigma is larger_iWhen > σ, the corresponding parameters are combined (t)_i,σ_i) Sequentially encoding the vector as a row vector into a matrix B;

wind power intraday fluctuation time period sequence { t 'obtained according to fluctuation identification module'_iiIdentification of continuous periods in the order of the ii value from small to large, including when t'_ii+1-t′_iiA-2, where ii is 1,2Combination of parameters (t'_ii,σ′_ii) Are successively programmed into the matrix C as row vectors, otherwise the corresponding parameters are combined (t'_ii,σ′_ii) Sequentially encoding the vector as a row vector into a matrix D; when t'_a-t'_a-1Corresponding parameters are combined (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Sequentially compiling the row vectors into the C end of the matrix, otherwise combining the corresponding parameters (t'_a-1,σ'_a-1)、(t'_a,σ'_a) Sequentially encoding the row vectors into the tail end of the matrix D;