CN112347655B - Wind power plant theoretical power calculation method based on unit operation performance evaluation - Google Patents
Wind power plant theoretical power calculation method based on unit operation performance evaluation Download PDFInfo
- Publication number
- CN112347655B CN112347655B CN202011290260.9A CN202011290260A CN112347655B CN 112347655 B CN112347655 B CN 112347655B CN 202011290260 A CN202011290260 A CN 202011290260A CN 112347655 B CN112347655 B CN 112347655B
- Authority
- CN
- China
- Prior art keywords
- data
- power
- time
- wind turbine
- output
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/06—Wind turbines or wind farms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Wind Motors (AREA)
Abstract
A wind power plant theoretical power calculation method based on unit operation performance evaluation relates to the technical field of new energy power generation, and comprises the following steps: in the data identification stage, abnormal data caused by faults or interference in links such as data acquisition or communication and the like are identified; a data restoration stage, namely reconstructing the abnormal data of the new energy operation data in three categories according to the characteristics of the abnormal data identified in the data identification stage; grouping non-sample board machines and identifying each group of output proportionality coefficients, and dynamically identifying the output proportionality coefficient of each group; and in the theoretical power calculation stage of the wind power plant, calculating the theoretical power of the wind power plant according to the calculation formula of the theoretical generating power within the formula statistic time. The invention has the beneficial effects that: the method is easy to operate, simple in modeling and capable of accurately identifying the output proportionality coefficient between each group of non-sample computers and the sample computers, and therefore the calculation accuracy of the theoretical power of the wind power plant is effectively improved.
Description
Technical Field
The invention relates to the technical field of new energy power generation, in particular to a wind power plant theoretical power calculation method based on unit operation performance evaluation.
Background
With the gradual maturity of wind power generation technology, the attention of wind power renewable energy is greatly improved, however, due to the large-scale rapid unordered operation of wind power plants and the relative lag of the construction of a power grid framework, the wind power receiving capacity of a power grid is limited, and the phenomenon of wind abandonment sometimes occurs. Therefore, the research on theoretical generating power of the wind power plant is necessary, the wind power abandoning amount of the wind power plant can be scientifically and accurately evaluated, the contradiction between the power grid and the plant is relieved, the evaluation of the actual output level of the wind power plant by a dispatching department is facilitated, and a reference is provided for scientific statistics of the power limiting electric quantity of the wind power plant.
At present, a sample board computer method is mainly adopted for calculating theoretical power of a wind power plant, namely, single-machine equipment with good performance, stable operation and relatively few faults is selected as the sample board computer, and the theoretical power of the wind power plant is calculated through actual power of the single-machine equipment.
Disclosure of Invention
The invention provides a wind power plant theoretical power calculation method based on unit operation performance evaluation, which aims to overcome the defects of the prior art and solves the problem that the theoretical power calculated by the original sampling machine method has a large error due to the uncertainty of the output proportionality coefficients of the sampling machine and a non-sampling machine caused by factors such as geographic position, arrangement mode and weather.
The invention provides a wind power plant theoretical power calculation method based on unit operation performance evaluation, which specifically comprises the following steps: the method comprises a data identification stage, a data restoration stage, a non-sample board machine grouping and force proportion coefficient identification stage and a theoretical power calculation stage of the wind power plant.
And (3) a data identification phase. The method comprises the following steps of identifying the abnormality caused by the fault or interference in the links of data acquisition or communication and the like:
step 1: and identifying the null value appearing at a certain time point or time period in the time sequence power data of all the new energy wind turbine generators according to the null value identification criterion, and removing the abnormal data.
Step 2: and identifying the out-of-limit values exceeding the reasonable range in the time sequence power data of all the new energy wind turbine generators according to the out-of-limit value identification criterion, and removing the abnormal data.
And 3, step 3: and identifying data values which are not refreshed continuously for a period of time in the time sequence power data of all the new energy wind turbine generators according to a non-refreshing value identification criterion, and removing the abnormal data.
And 4, step 4: and identifying distortion data generated in the time sequence power data of all the new energy wind turbine generators according to a distortion data identification criterion, and removing the abnormal data.
And (5) a data repairing stage. Aiming at the characteristics of the abnormal data identified in the steps 1-4, the abnormal data reconstruction is realized on the new energy running data in three categories including single-point abnormality, multi-point abnormality and continuous abnormality, and the steps are as follows:
and 5: and filling the single-point abnormal data in the time sequence power data of the new energy wind turbine generator by adopting a constant/mean value method.
And 6: for multipoint abnormal data which is not more than 4h in time sequence power data of the new energy wind turbine generator, reconstructing the abnormal data by adopting a reconstruction method based on ARMA model prediction, and establishing average credibility gamma p Calculating a forward predicted value of the weighted reconstruction value at the moment k
And reverse predicted value
Thereby improving the reliability of the reconstruction result at the time k; and regarding the multipoint exception exceeding 4h as continuous exception data, and the data reconstruction process goes to step 7.
And 7: for continuous abnormal data in time sequence power data of the new energy wind turbine generator, firstly finding out the continuous abnormal data in the time sequence power data of the new energy wind turbine generator W by adopting a reconstruction method based on sequence time delay correlation y Wind turbine generator system W with maximum time delay correlation x (x =1, 2.. An, N, N are the number of new energy wind turbine generators), and W is utilized x Force curve of (2) to W y And performing data reconstruction on the continuous abnormal operation data.
Grouping non-sample board machines and identifying each group of force proportion coefficients. In the stage, after the non-sample board machines of the same type in the new energy power station are grouped based on the unit operation performance, the output proportionality coefficient of each group is dynamically identified, so that the relative output condition of the non-sample board machines is more accurately mastered, and the method comprises the following steps:
and 8: through the basic output performance characteristic vector W of the ith non-sample board computer in the wind power plant i And the average characteristic vector of the basic output performance of the sample board machine
Pearson's correlation coefficient R is Dividing the interval unit by taking 0.1 as the interval unit, and dividing the non-sample board machines of each interval unit into one group, marking as the g group, wherein the number of the non-sample board machines is
The total number of the groups is 20.
And step 9: calculating the actual output average value sequence of the g group of non-sample plate machines of the k model fan in the statistical time
Sequence of actual output average values of sample board machine in statistical time
Power difference rate of
Obtaining a proportionality coefficient of the g group of non-sample plate machines relative to the sample plate machine output as
Thus, the output proportionality coefficients of each group are dynamically identified.
And (3) calculating theoretical power of the wind power plant.
Step 10: calculating to obtain the output proportionality coefficient of each group of non-sample computers relative to the sample computer in the k model fan of the wind power plant
Number of non-sample plate machines
And the average value of the actual output of the sample plate machine
And calculating the theoretical power of the wind power plant according to a calculation formula of the theoretical generating power within formula statistic time.
The invention has the beneficial effects that: in consideration of data abnormity caused by faults or interference of the new energy wind turbine generator in links such as data acquisition or communication, different types of abnormal data identification criteria are established to identify and remove abnormal data of the time sequence output data of the new energy wind turbine generator; considering that the integrity of the data is damaged and the usability of the data is influenced after the abnormal data are removed, different methods are adopted to reconstruct single-point, multi-point and continuous abnormal data respectively, so that errors caused by subsequent theoretical power calculation of the new energy power station are effectively reduced; in consideration of the fact that the output proportionality coefficient between the original sample board computer method and the non-sample board computer cannot be accurately calculated when the theoretical power is calculated, compared with a grouping method considering terrain and wake effect, the non-sample board computer grouping method based on wind turbine generator running performance evaluation has the advantages of easiness in operation, simplicity in modeling and capability of accurately identifying the output proportionality coefficient between each group of non-sample board computers and the sample board computers, and therefore the calculation accuracy of the theoretical power of the wind power plant is effectively improved.
Drawings
FIG. 1 is a block diagram of a theoretical power calculation process of a wind farm according to the present invention.
Detailed Description
Embodiment 1, as shown in fig. 1, the invention provides a method for calculating theoretical power of a wind farm based on unit operation performance evaluation, which includes four stages, a data identification stage, a data restoration stage, a non-sample board machine grouping and each group of output proportionality coefficient identification stage, and a theoretical power calculation stage of the wind farm; the method comprises the following steps:
step 1: according to the vacancy value identification criterion, identifying vacancy values appearing at a certain time point or time period in the time sequence power data of all the new energy wind turbine generators, and removing abnormal data;
and 2, step: according to the out-of-limit value identification criterion, identifying out-of-limit values which exceed a reasonable range in the time sequence power data of all the new energy wind turbine generators, and removing abnormal data;
and 3, step 3: according to the non-refreshing value identification criterion, identifying data values which are not refreshed for a period of time in the time sequence power data of all the new energy wind turbine generators, and removing the abnormal data;
and 4, step 4: identifying distortion data generated in time sequence power data of all new energy wind turbine generators according to distortion data identification criteria, and eliminating abnormal data;
and 5: filling single-point abnormal data in the time sequence power data of the new energy wind turbine generator by adopting a constant/mean value method;
step 6: for multipoint abnormal data which is not more than 4h in time sequence power data of the new energy wind turbine generator, reconstructing the abnormal data by adopting a reconstruction method based on ARMA model prediction, and establishing average credibility gamma p Calculating a forward predicted value of the weighted reconstruction value at the moment k
And reverse prediction
Thereby improving the reliability of the reconstruction result at the time k; regarding the multipoint abnormality exceeding 4h as continuous abnormal data, and transferring the data reconstruction process to step 7;
and 7: for continuous abnormal data in time sequence power data of the new energy wind turbine generator, firstly, a reconstruction method based on sequence time delay correlation is adopted to find out the continuous abnormal data in the time sequence power data of the new energy wind turbine generator y Wind turbine generator system W with maximum time delay correlation x (x =1, 2.. And N, N is the number of new energy wind turbines), and W is utilized x Force curve of (d) to W y Carrying out data reconstruction on the continuous abnormal operation data;
and step 8: through the basic output performance characteristic vector W of the ith non-sample board computer in the wind power plant i And the average characteristic vector of the basic output performance of the sample plate machine
Pearson's correlation coefficient R is Dividing by taking 0.1 as an interval unit, and dividing the non-sample machines of each interval unit into a group which is marked as a g group, wherein the number of the non-sample machines is
20 groups in total;
and step 9: calculating the actual output average value sequence of the g group of non-sample plate machines of the k model fan in the statistical time
Sequence of actual output average values of sample board machine in statistical time
Power difference rate of
Obtaining the proportional coefficient of the g group of non-sample plate machines relative to the sample plate machine outputIs composed of
Dynamically identifying the output proportionality coefficient of each group;
step 10: calculating to obtain the output proportionality coefficient of each group of non-sample computers relative to the sample computer in the k model fan of the wind power plant
Number of non-sample plate machines
And the average value of the actual output of the sample plate machine
And calculating the theoretical power of the wind power plant according to a calculation formula of the theoretical generating power within formula statistic time.
The theoretical power improvement calculation of the wind power plant specifically comprises the following steps:
time sequence output data X = { X) of all new energy wind turbine generators 1 ,x 2 ,...,x n And (6) identifying and reconstructing abnormal data.
The criteria for identifying the different types of anomalous data are as follows:
(1) Identification of an empty value
If x is present i Satisfies the following conditions:
then x is judged i There is a loss of data.
(2) Identification of a threshold value
If x is present i If the following relation is satisfied, x is determined i Out of the normal range of values.
x i ＞x h |x i ＜x l (2)
In the formula, x h ,x l Upper and lower limits of the normal numerical range, respectively.
(3) Identification of non-refreshed values
If the data is not refreshed for a period of time, the data is identified as not refreshed except for the first data point.
(4) Identification of distorted data
The powers of adjacent wind turbines generally have similar trends, so based on the spatial correlation of the output of the adjacent wind turbines, N power data X of the adjacent wind turbines having spatial correlation with the wind turbine A are utilized M ＝{x M1 ,x M2 ,...,x Mn } (M =1,2, \ 8230;, N), power data X for observing wind turbines A ＝{x A1 ,x A2 ,...,x An Checking is carried out, and abnormal data are identified. The method comprises the following specific steps: normalizing the power based on the formula (3) to obtain power normalization data X of the observation wind turbine generator A * ＝{x A1 * ,x A2 * ,...,x An * And power normalization data X of adjacent wind turbine generators M * ＝{x M1 * ,x M2 * ,...,x Mn * And the difference between the two at each sampling point is equation (4).
e Ai-Mi ＝|x Ai * -x Mi * | (4)
The power difference value and the average value of the wind turbine generators A and M are respectively as follows:
E A-M ＝{e A1-M1 ,e A2-M2 ,…,e An-Mn } (5)
because the abnormal conditions of the power data obtained by different wind turbines and different time periods are different, in order to obtain the identification result with higher reliability, the eliminated data is deleted and not refreshedAnd calculating the average value and standard deviation of the output power difference of the observation wind turbine generator and the adjacent wind turbine generator after normalization of the data out-of-limit data set to be identified. Because the correctness of the output data of the adjacent wind turbines cannot be determined in advance, the method ensures that
Determining abnormal data according to the criterion of Lewy-one, i.e. when
Then x is considered to be i Is the distorted data.
The reconstruction method of different abnormal data comprises the following steps:
(1) Single point exception
For the single-point abnormal data, the filling method can be adopted to carry out engineering reconstruction on the abnormal data, namely an empirical constant is selected or the abnormal data is replaced according to a certain rule, and the process is simple and time-saving and has considerable accuracy.
1) Constant filling: and based on manual experience, filling the single-point abnormal data by adopting the same specified constant value.
2) Mean value filling: the single-point abnormal data can be filled by adopting the average value of the data sequence or the average value of the data before and after the abnormal data.
(2) Multipoint exception
For multipoint abnormity, due to the fact that complete actual operation data exist on two sides of the abnormal data, the abnormal data can be reconstructed by a reconstruction method based on ARMA model prediction according to the actual operation data on the two sides.
1) Assumed time period t m ,t n ]Operating data based on intact new energy unit for missing data period
And (5) performing m-step prediction from two sides by adopting an ARMA model.
P t ＝β 1 P t-1 +β 2 P t-2 +…+β p P t-p +Z t (7)
Z t ＝ε t +λ 1 ε t-1 +λ 2 ε t-2 +…+λ 1 ε t-1 (8)
In the formula, beta i Is an autoregressive parameter; lambda [ alpha ] i Is a moving average parameter; epsilon i Is the prediction error.
2) For the prediction result of the p-th step, if the prediction error is less than the error threshold epsilon 0 And if so, the prediction result is considered to be credible. Determining the average reliability gamma of the prediction of the p-th step of the data sequence by predicting a large amount of perfect operation data p ：
In the formula, count (·) is the frequency statistics; e p Predicting error for the p step; n is the total number of times.
3) Calculating a forward predicted value of the weighted reconstruction value at the moment k
Has a prediction time length of k-t m +1, acceptance rate of
And reverse predict the value
Is t n -k +1, acceptability of
Thus, the final reconstruction result at time k is:
the reconstruction method based on the self-output rule considers the prediction effect of the ARMA, and the reconstruction time length is generally limited within 4 h.
(3) Continuous abnormality
As the time scale of the abnormal data increases, the accuracy of the reconstruction method based on the ARMA model prediction will be significantly reduced. Because a certain correlation exists among various new energy operation data, for continuous abnormal data, a reconstruction method based on sequence time delay correlation can be adopted to reconstruct the abnormal data.
1) Delay correlation
Suppose there are two time series X = { X = { [ X ] 1 ,x 2 ,…,x n And Y = { Y = 1 ,y 2 ,…,y n The calculation formula of the correlation coefficient R (l) of Y relative to the delay time l of X is:
in the formula (I), the compound is shown in the specification,
the maximum value of l is n/2, and when l is changed from 0 to n/2, R (0), R (1), \8230; R (n/2) is obtained.
The delay corresponding to the maximum value is (maximum delay correlation point), if (is correlation threshold), then X and Y have delay correlation. (three-point prediction exploration method can be adopted, the 1 st exploration point (determined by trigonometry) is arranged from the position where the maximum time delay related point is most likely to appear, and other exploration points are respectively arranged at the left side and the right side of the exploration point in a geometric progressive mode.
2) Reconstruction based on output delay correlation of new energy wind turbine generator
Assume missing data period is [ t ] m ,t n ]W is the wind turbine generator to be reconstructed 0 The rest wind turbine generators are W 1 ,W 2 8230and its preparing process. Taking each wind turbine generator at time interval [ t m -t 0 ,t m ]Internal force data. First calculate W 0 And W 1 The maximum delay correlation point of the output force in the time interval and the corresponding delay correlation coefficient if R (l) 1 ')＝max(R(l 1 '),R(l 2 '), \ 8230), then W 0 And W 1 The corresponding maximum delay correlation point is l 1 ', the corresponding delay correlation coefficient is R (l) 1 ') to a test; then respectively calculating W according to the calculated values 0 And W 2 ,W 3 8230, maximum delay correlation point of the force output in the time interval and corresponding delay correlation coefficient; if there is W 0 And W 1 When the delay correlation coefficient is maximum, W is used 1 Force curve of (2) to W 0 And reconstructing the historical operating data. I.e. W 0 And W 1 The force relationships of (d) can be fit using linear regression as:
in the formula (I), the compound is shown in the specification,
and
are respectively W 0 And W 1 At the t th m And t m +l 1 ' output value at time; a and b are coefficients to be determined, and can be obtained by a least square method.
Grouping of non-sample board machines in a wind power plant, dynamic identification of the output proportional coefficients of all groups and calculation of theoretical power are as follows:
(1) Non-sample board machine grouping based on unit operation performance evaluation
Firstly, establishing a basic power generation performance index (non-limited power condition) of the wind turbine generator set as follows:
a) Maximum power of the moon w 1
Maximum power of the moon w 1 (kW) is the actual power P of the wind turbine generator 30 days before the observation day r ＝{p 1 ,p 2 ,…,p n The maximum value in, i.e.:
w 1 ＝max{p i }, i＝1,2,…,n (13)
in the formula, n is the number of sample sampling points.
b) Monthly average power generationPower w 2
In the formula, P r,j,t To observe the actual power (kW) of the jth sample on the jth day 30 days before the day.
c) Annual rate of change w of average generated power 3
The annual change rate of the average generating power of the unit reflects the trend of the generating level of the unit changing along with time:
in the formula, P r,av,y Observing the average actual generated power (kW) for the unit in the first 365 days before the day; p r ' ,av,y Observing the average actual generated power (kW) for the unit in the second 365 days before the day; p' r,j,t Observing the actual power (kW) of the jth sampling point on the jth day in the first 365 days before the day; p r ' , ' j,t To observe the actual power (kW) of the jth sampling point on the jth day in the 365 th day; m is the number of sampling points per day.
d) Number of full hair 4
Number of full hair 4 (h) In order to observe the ratio of the actual power generation amount of the wind turbine generator to the rated power 30 days before the day, the method can be used for comparing power generation systems with different installed capacities, and the calculation formula is as follows:
in the formula, P n For observing the rated power (kW) of the wind turbine 30 days before the day, E r To observe the actual power generation 30 days before the day.
Establishing a wind turbine generator basic output performance characteristic vector W = { W ] within statistical time by using a wind turbine generator basic power generation performance index 1 ,w 2 ,w 3 ,w 4 }. The basic output performance characteristic vector of the ith non-sample board machine of the k model fan in the statistical time is set as
The average characteristic vector of the basic output performance of the k model sample plate machine in the statistical time is
Wherein M is k The total number of the full wind power fields of the k model sample machines;
the characteristic vector of the basic output performance of the mth sample board machine of the k model fan in the statistical time is obtained.
The basic output performance characteristic vector of the ith non-sample plate machine in the statistical time
And the average characteristic vector of the basic output performance of the sample plate machine
The Pearson correlation coefficient of (a) is:
in the formula (I), the compound is shown in the specification,
let 0.1 be an interval unit pair R is Is divided, usually R is E [0.1 (g-11), 0.1 (g-10) ], g being an integer, and g e [1,20 ]]. Dividing the non-sample plate machines of each interval unit into a group, and setting the number of the g-th group of non-sample plate machines as
(the length of the interval unit may be reduced in order to improve the calculation accuracy).
(2) Dynamic identification of groups of output proportionality coefficients
And setting the actual force average value sequence of the g group of non-sample plate machines in the statistical time as follows:
in the formula (I), the compound is shown in the specification,
and (4) actual output sequence of the ith group of non-sample board machines of the k model fan in the statistical time.
Setting the actual output average value sequence of a k model fan sample machine in the statistical time as follows:
in the formula (I), the compound is shown in the specification,
and (4) obtaining an actual output sequence of the ith sample board machine under the k model within the statistical time.
Then within the statistical time of the time period,
and
the average of the power difference rates is:
in the formula, n is the number of sampling points in the statistical time.
Thus, the proportionality coefficient of the g-th set of non-sample plates with respect to the plate-out force of the sample plates may be taken as
In an ideal situation, when the theoretical output of all the units in the wind power plant is equal,
(3) The theoretical power of the wind farm is calculated as follows:
and respectively calculating theoretical generating power of each group of non-sample machines by using the actual output of the sample machines, and finally obtaining a calculation formula of the theoretical generating power of the new energy power station within the statistical time:
in the formula (I), the compound is shown in the specification,
calculating the improved theoretical power of all the units of the new energy power station model k within the time;
for counting theoretical work of new energy power station in timeRate; k is the model number of the new energy power station unit.
Claims (4)
1. A wind power plant theoretical power calculation method based on unit operation performance evaluation is characterized by comprising the following steps: the method comprises four stages, namely a data identification stage, a data restoration stage, a non-sample board machine grouping and output proportional coefficient identification stage and a theoretical power calculation stage of the wind farm;
in the data identification stage, abnormal data caused by faults or interference in the data acquisition or communication link is identified, and the steps are as follows:
step 1: according to the vacancy value identification criterion, identifying vacancy values appearing at a certain time point or time period in time sequence power data of all new energy wind turbine generators, and removing abnormal data;
and 2, step: according to the out-of-limit value identification criterion, identifying out-of-limit values which exceed a reasonable range in the time sequence power data of all the new energy wind turbine generators, and removing abnormal data;
and step 3: according to the non-refreshing value identification criterion, identifying data values which are not refreshed for a period of time in the time sequence power data of all the new energy wind turbine generators, and removing the abnormal data;
and 4, step 4: identifying distortion data generated in time sequence power data of all new energy wind turbine generators according to a distortion data identification criterion, and removing abnormal data;
and in the data restoration stage, aiming at the characteristics of the abnormal data identified in the steps 1-4, the abnormal data is reconstructed by three categories including single-point abnormality, multipoint abnormality and continuous abnormality, and the steps are as follows:
and 5: filling single-point abnormal data in the time sequence power data of the new energy wind turbine generator by adopting a constant/mean value method;
and 6: for multipoint abnormal data which is not more than 4h in time sequence power data of the new energy wind turbine generator, reconstructing the abnormal data by adopting a reconstruction method based on ARMA model prediction, and establishing average credibility gamma p Calculating a weighted reconstruction valueForward predicted value at time k
And reverse predicted value
Thereby improving the reliability of the reconstruction result at the time k; regarding the multipoint abnormality exceeding 4h as continuous abnormal data, and transferring the data reconstruction process to step 7;
and 7: for continuous abnormal data in time sequence power data of the new energy wind turbine generator, firstly finding out the continuous abnormal data in the time sequence power data of the new energy wind turbine generator W by adopting a reconstruction method based on sequence time delay correlation y Wind turbine generator system W with maximum time delay correlation x X =1, 2.. And N, N is the number of new energy wind turbine generators, and W is utilized x Force curve of (d) to W y Carrying out data reconstruction on the continuous abnormal operation data;
grouping non-sample board machines and identifying each group of output proportionality coefficients, after grouping the non-sample board machines of the same type in the new energy power station based on the unit operation performance, dynamically identifying the output proportionality coefficient of each group, thereby more accurately mastering the relative output condition of the non-sample board machines, and the method comprises the following steps:
and 8: through the basic output performance characteristic vector W of the ith non-sample board computer in the wind power plant i And the average characteristic vector of the basic output performance of the sample plate machine
Pearson's correlation coefficient R is Dividing the interval unit by taking 0.1 as the interval unit, and dividing the non-sample board machines of each interval unit into one group, marking as the g group, wherein the number of the non-sample board machines is
20 groups in total;
and step 9: calculating the actual output average value sequence of the g group of non-sample plate machines of the k model fan in the statistical time
Actual output average value sequence of sample plate machine in statistical time
Power difference rate of
Obtaining a proportionality coefficient of the g group of non-sample plate machines relative to the sample plate machine
Dynamically identifying the output proportionality coefficient of each group according to the output proportionality coefficient;
the theoretical power calculation stage of the wind power plant comprises the following steps:
step 10: calculating to obtain the output proportionality coefficient of each group of non-sample computers relative to the sample computer in the k model fan of the wind power plant
Number of non-sample plate machines
And the average value of the actual output of the sample plate machine
Calculating theoretical power of the wind power plant according to a calculation formula of the theoretical generating power within formula statistic time; the non-sample board computers are grouped, and firstly, under the condition of non-limited power, basic power generation performance indexes of the wind turbine generator are established;
maximum power of the moon w 1 (kW) is the actual power P of the wind power generation unit within 30 days before the observation day r ＝{p 1 ,p 2 ,…,p n The maximum of, i.e.:
w 1 ＝max{p i },i＝1,2,…,n (13)
in the formula, n is the number of sample sampling points;
monthly average generated power w 2
In the formula, P r,j,t Observing actual power (kW) of the jth sampling point on the jth day in 30 days before the day;
annual rate of change w of average generated power 3 The annual change rate of the average generating power of the unit reflects the trend of the generating level of the unit changing along with time:
in the formula, P r,av,y Observing the average actual generated power (kW) for the unit in the first 365 days before the day; p' r,av,y Observing the average actual generated power (kW) for the unit in the second 365 days before the day; p' r,j,t Observing the actual power (kW) of the jth sampling point on the jth day in the first 365 days before the day; p ″) r,j,t To observe the actual power (kW) of the jth sampling point on the jth day in the 365 th day; m is the number of sampling points per day;
number of full hair 4 (h) In order to observe the ratio of the actual power generation amount to the rated power of the wind generating set in 30 days before the day, the method can be used for comparing power generation systems with different installed capacities, and the calculation formula is as follows:
in the formula, P n For observing the rated power (kW) of the wind generating set within 30 days before the day, E r To observe the actual power generation within 30 days before the day;
establishing a characteristic vector W = { W } of the basic output performance of the wind turbine generator within statistical time by using the basic generation performance index of the wind turbine generator 1 ,w 2 ,w 3 ,w 4 And setting a basic output performance characteristic vector of the ith non-sample board machine of the k-type fan in the statistical time as
The average characteristic vector of the basic output performance of the k model sample plate machine in the statistical time is
Wherein M is k The total number of the full wind power fields of the k model sample machines;
the characteristic vector of the basic output performance of the mth sample board machine of the k model fan in the statistical time is obtained;
the basic output performance characteristic vector of the ith non-sample plate machine in the statistical time
And the average characteristic vector of the basic output performance of the sample board machine
The Pearson correlation coefficient of (a) is:
in the formula (I), the compound is shown in the specification,
let 0.1 be an interval unit pair R is Is divided, usually R is E [0.1 (g-11), 0.1 (g-10)), g is an integer, and g e [1,20 ]]Dividing the non-sample board machines of each interval unit into a group, and setting the number of the g-th group of non-sample board machines as
And dynamically identifying the force proportion coefficients of all groups, and setting the actual force output average value sequence of the g group of non-sample plate machines in the statistical time as follows:
in the formula (I), the compound is shown in the specification,
the actual output sequence of the ith group of non-sample plate machines of the k model fan in the statistical time is obtained;
setting the actual output average value sequence of the k model fan sample plate machine in the statistical time as follows:
in the formula (I), the compound is shown in the specification,
the actual output sequence of the ith sample board machine under the k model within the statistical time is obtained;
then within the statistical time of the time period,
and
the average of the power difference rates is:
wherein n is the number of sampling points in the statistical time.
2. The wind power plant theoretical power calculation method based on unit operation performance evaluation according to claim 1, characterized in that: the abnormal data identification specifically comprises:
identification of the vacancy value: let X = { X 1 ,x 2 ,...,x n The power is the actual power of the wind turbine generator in the statistical time, if x exists i Satisfies the following conditions:
then x is judged i There is a loss of data;
identification of the threshold value: if x is present i If the following relation is satisfied, x is determined i Out of normal numerical ranges:
x i ＞x h |x i ＜x 1 (2)
in the formula, x h ,x 1 Upper limit and lower limit of the normal numerical range respectively;
identification of non-refreshed values: if the data is not refreshed continuously for a period of time, identifying the data as not refreshed except the first data point;
identification of distorted data: the power of adjacent wind turbines usually has similar trend, so based on the spatial correlation of the output of adjacent wind turbines, the power data X of N adjacent wind turbines having spatial correlation with the wind turbine A are utilized M ＝{x M1 ,x M2 ,...,x Mn M =1,2, \ 8230;, N, power data X for observing wind turbine generator A ＝{x A1 ,x A2 ,...,x An The check is carried out, and the check is carried out,identifying abnormal data as follows:
normalizing the power based on the formula (3) to obtain power normalization data X of the observation wind turbine A * ＝{x A1 * ,x A2 * ,...,x An * Power normalization data X of adjacent wind turbines M * ＝{x M1 * ,x M2 * ,...,x Mn * The difference value of the two at each sampling point is shown as formula;
e Ai-Mi ＝|x Ai * -x Mi * | (4)
the power difference value and the average value of the wind turbine generator A and the wind turbine generator M are respectively as follows:
E A-M ＝{e A1-M1 ,e A2-M2 ,…,e An-Mn } (5)
3. the wind power plant theoretical power calculation method based on unit operation performance evaluation according to claim 1, characterized in that: the abnormal data reconstruction specifically comprises the following steps:
single point exception: for the single-point abnormal data, a filling method can be adopted to carry out engineering reconstruction on the abnormal data, namely an experience constant is selected or the abnormal data is replaced and filled according to a certain rule; constant filling: based on manual experience, filling single-point abnormal data by adopting the same specified constant value; mean value filling: filling single-point abnormal data by adopting the average value of the data sequence or the average values of the data before and after the abnormal data;
multipoint exception: for multipoint abnormity, due to the fact that complete actual operation data exist on two sides of the abnormal data, the abnormal data can be reconstructed by a reconstruction method based on ARMA model prediction according to the actual operation data on the two sides;
first assume a time period t m ,t n ]For the period of missing data, based on the intact wind turbine generator operating data
Predicting from two sides by adopting an ARMA model;
P t ＝β 1 P t-1 +β 2 P t-2 +…+β p P t-p +Z t (7)
Z t ＝ε t +λ 1 ε t-1 +λ 2 ε t-2 +…+λ 1 ε t-1 (8)
in the formula, beta i Is an autoregressive parameter; lambda i Is a moving average parameter; epsilon i Is the prediction error;
secondly, for the prediction result of the p step, if the prediction error is less than an error threshold value epsilon 0 If so, the prediction result is considered to be credible; determining an average confidence gamma for a prediction of a data sequence by predicting a large amount of good operating data p ：
In the formula, count () is the frequency statistics; e p Predicting an error for the data sequence; n is the total number of times;
finally, calculating a forward predicted value of the weighted reconstruction value at the moment k
Has a prediction time length of k-t m +1, acceptance rate of
And reverse predict the value
Is t n -k +1, acceptability of
Thus, the final reconstruction result at time k is:
continuous abnormality: with the increase of the time scale of the abnormal data, the accuracy of the reconstruction method based on ARMA model prediction is obviously reduced, and as the running data of the wind turbine generator has certain correlation, the reconstruction method based on the sequence time delay correlation can be adopted to reconstruct the abnormal data for the continuous abnormal data;
delay dependence, assuming that there are two time series X = { X = { [ X ] 1 ,x 2 ,…,x n And Y = { Y = 1 ,y 2 ,…,y n The calculation formula of the correlation coefficient R (l) of Y relative to the delay time l of X is:
in the formula (I), the compound is shown in the specification,
the maximum value of l is n/2, when l is changed from 0 to n/2, R (0), R (1), \ 8230;, R (n/2) are obtained;
based on reconstruction of output time delay correlation of new energy wind turbine generator, supposing that the time period of missing data is [ t ] m ,t n ]The wind turbine generator to be reconstructed is W 0 W for the rest of the wind turbines 1 ,W 2 8230, the wind turbine generator sets are selected in the time period t m -t 0 ,t m ]Internal output data, first calculate W 0 And W 1 At maximum of output force during the periodDelay correlation points and corresponding delay correlation coefficients if R (l) 1 ')＝max(R(l 1 '),R(l 2 '), \8230;) then W 0 And W 1 The corresponding maximum delay correlation point is l 1 ', the corresponding delay correlation coefficient is R (l) 1 ') to a host; then respectively calculating W according to the calculated values 0 And W 2 ,W 3 8230that the maximum delay correlation point of the force output in the time interval and the corresponding delay correlation coefficient; if there is W 0 And W 1 When the delay correlation coefficient is maximum, W is used 1 Force curve of (d) to W 0 Is reconstructed, i.e. W 0 And W 1 The force relationship of (c) can be fit using linear regression as:
in the formula (I), the compound is shown in the specification,
and
are respectively W 0 And W 1 At the t th m And t m +l 1 ' output value at time; a and b are coefficients to be determined, and can be obtained by a least square method.
4. The wind power plant theoretical power calculation method based on unit operation performance evaluation according to claim 1, characterized in that: the theoretical power of the wind farm is calculated as follows:
and respectively calculating theoretical generating power of each group of non-sample machines by using the actual output of the sample machines, and finally obtaining a calculation formula of the theoretical generating power of the new energy power station within the statistical time:
in the formula (I), the compound is shown in the specification,
calculating the improved theoretical power of all the units of the new energy power station model k within the time;
calculating theoretical power of the new energy power station within the time; k is the model quantity of new forms of energy power station unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011290260.9A CN112347655B (en) | 2020-11-17 | 2020-11-17 | Wind power plant theoretical power calculation method based on unit operation performance evaluation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011290260.9A CN112347655B (en) | 2020-11-17 | 2020-11-17 | Wind power plant theoretical power calculation method based on unit operation performance evaluation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112347655A CN112347655A (en) | 2021-02-09 |
| CN112347655B true CN112347655B (en) | 2023-03-17 |
Family
ID=74364096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011290260.9A Active CN112347655B (en) | 2020-11-17 | 2020-11-17 | Wind power plant theoretical power calculation method based on unit operation performance evaluation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112347655B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115983663B (en) * | 2022-11-04 | 2023-09-12 | 广西电网有限责任公司 | New energy-based power generation analysis method and related equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105989236A (en) * | 2015-03-03 | 2016-10-05 | 南京理工大学 | Spatial correlation-based method for predicting wind power in typhoon passing condition |
| CN107944622A (en) * | 2017-11-21 | 2018-04-20 | 华北电力大学 | Wind power forecasting method based on continuous time cluster |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104112236B (en) * | 2014-05-29 | 2018-04-27 | 国家电网公司 | The computational methods of wind power plant generated output |
| CN104124685A (en) * | 2014-07-28 | 2014-10-29 | 国家电网公司 | Sample fan method based wind power plant theoretical power computing method |
| CN105930933B (en) * | 2016-04-26 | 2019-05-31 | 华北电力科学研究院有限责任公司 | Wind power plant theoretical power curve determines method and device |
| CN107330183B (en) * | 2017-06-29 | 2020-04-17 | 华北电力大学 | Wind power utilization rate calculation method based on operation data |
| CN109960778B (en) * | 2017-12-26 | 2023-06-27 | 北京金风慧能技术有限公司 | Method and device for calculating theoretical power of wind power plant |
-
2020
- 2020-11-17 CN CN202011290260.9A patent/CN112347655B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105989236A (en) * | 2015-03-03 | 2016-10-05 | 南京理工大学 | Spatial correlation-based method for predicting wind power in typhoon passing condition |
| CN107944622A (en) * | 2017-11-21 | 2018-04-20 | 华北电力大学 | Wind power forecasting method based on continuous time cluster |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112347655A (en) | 2021-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109308571B (en) | Distribution line variable relation detection method | |
| CN109597967B (en) | Power system distribution substation load data anomaly detection and repair method | |
| CN111680820B (en) | Distributed photovoltaic power station fault diagnosis method and device | |
| CN108400595B (en) | Voltage sag random estimation method considering new energy output correlation | |
| CN113126019B (en) | Remote estimation method, system, terminal and storage medium for error of intelligent ammeter | |
| CN106682763B (en) | Power load optimization prediction method for large amount of sample data | |
| CN111669123B (en) | Method and device for fault diagnosis of photovoltaic string | |
| CN111861023A (en) | Statistical-based hybrid wind power prediction method and device | |
| CN109657913B (en) | Transmission and distribution network joint risk assessment method considering distributed power supply | |
| CN111709569A (en) | Method and device for predicting and correcting output power of wind power plant | |
| CN111313403A (en) | Low-voltage power distribution system network topology identification method based on Markov random field | |
| CN112070121A (en) | Intelligent electric meter data filling method based on variational self-encoder | |
| CN112347655B (en) | Wind power plant theoretical power calculation method based on unit operation performance evaluation | |
| CN111654392A (en) | Low-voltage distribution network topology identification method and system based on mutual information | |
| CN109447512B (en) | Large power grid reliability assessment method based on uniform design | |
| CN113866552B (en) | Medium voltage distribution network user electricity consumption abnormality diagnosis method based on machine learning | |
| CN111221811A (en) | Low-voltage distribution network line parameter estimation method based on centralized meter reading system | |
| CN107204616B (en) | Power system random state estimation method based on self-adaptive sparse pseudo-spectral method | |
| CN113505909A (en) | Error compensation method for short-term wind power trend prediction | |
| CN112182499A (en) | Low-voltage distribution network topological structure identification method based on time sequence electric quantity data | |
| CN108665376B (en) | Nuclear density estimation method for determining maximum value of cellular load in space load prediction | |
| CN116307838A (en) | Accurate equivalence method, system and medium for power distribution network in power grid reliability evaluation | |
| CN113627655B (en) | Method and device for simulating and predicting pre-disaster fault scene of power distribution network | |
| US20240019468A1 (en) | High-resolution electrical measurement data processing | |
| CN113836125A (en) | Distribution transformer data cleaning and repairing method and system for distribution internet of things edge side |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |