CN118232308A - Power prediction method and device based on wind power fluctuation recognition and storage medium - Google Patents
Power prediction method and device based on wind power fluctuation recognition and storage medium Download PDFInfo
- Publication number
- CN118232308A CN118232308A CN202410105791.8A CN202410105791A CN118232308A CN 118232308 A CN118232308 A CN 118232308A CN 202410105791 A CN202410105791 A CN 202410105791A CN 118232308 A CN118232308 A CN 118232308A
- Authority
- CN
- China
- Prior art keywords
- power
- fluctuation
- power generation
- wind turbine
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000003860 storage Methods 0.000 title claims abstract description 10
- 238000010248 power generation Methods 0.000 claims abstract description 228
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000004364 calculation method Methods 0.000 claims description 14
- 238000012937 correction Methods 0.000 claims description 5
- 238000007906 compression Methods 0.000 claims description 4
- 230000006835 compression Effects 0.000 claims description 4
- 238000012549 training Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000001105 regulatory effect Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000007637 random forest analysis Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention provides a power prediction method and device based on wind power fluctuation recognition and a storage medium, and belongs to the field of power generation power prediction. The method comprises the following steps: acquiring historical operation data of a wind turbine generator; preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine; calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine by using the trained power prediction model; respectively calculating a first fluctuation trend of historical theoretical power generation data of the wind turbine generator and a second fluctuation trend of historical forecast power generation data of the wind turbine generator; calculating a power numerical difference and a Bayesian probability relation between the fluctuation trends according to the wind turbine generator set historical theoretical power generation data, the wind turbine generator set historical prediction power generation data, the first fluctuation trend and the second fluctuation trend; and adjusting future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain future prediction power generation data.
Description
Technical Field
The invention relates to the field of power generation power prediction, in particular to a power prediction method based on wind power fluctuation recognition, a power prediction device based on wind power fluctuation recognition and a machine-readable storage medium.
Background
The new energy supply capability is closely related to the change of natural conditions, and the fluctuation is normal. Besides the factor research for influencing new energy sources such as wind power, solar energy and the like is carried out jointly by combining meteorological departments and geological departments, weather prejudgement is enhanced, and wind power fluctuation can be utilized for prejudgement.
Disclosure of Invention
The invention aims to provide a power prediction method, a device and a storage medium based on wind power fluctuation recognition, wherein the method utilizes wind turbine historical operation data to establish a power prediction model, utilizes the power prediction model to predict wind turbine historical prediction power generation data, carries out fluctuation recognition and analysis on wind turbine historical theory power generation data and wind turbine historical prediction power generation data, combines future prediction initial power generation data to carry out early intervention adjustment on future possible fluctuation, improves power prediction precision, establishes effective counter measures for power generation enterprises and power grid enterprises, and solves energy supply risk and provides technical support for maintaining a power grid and power supply.
To achieve the above object, a first aspect of the present invention provides a power prediction method based on wind power fluctuation identification, the method comprising:
Acquiring historical operation data of a wind turbine generator;
Preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine;
calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine by using the trained power prediction model;
Respectively calculating a first fluctuation trend of historical theoretical power generation data of the wind turbine generator and a second fluctuation trend of historical forecast power generation data of the wind turbine generator;
Calculating a power numerical difference and a Bayesian probability relation between the fluctuation trends according to the wind turbine generator set historical theoretical power generation data, the wind turbine generator set historical prediction power generation data, the first fluctuation trend and the second fluctuation trend;
and adjusting future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain future prediction power generation data.
According to the technical means, historical operation data of the wind turbine are processed into historical theoretical power generation data of the wind turbine, the trained power prediction model is adopted to predict historical predicted power generation data and future predicted initial power generation data of the wind turbine, the historical theoretical power generation data of the wind turbine and the historical predicted power generation data of the wind turbine are used as bases, volatility of the historical theoretical power generation data of the wind turbine and the future predicted initial power generation data is identified, bayesian probability relations among the fluctuation trends are calculated according to the fluctuation trends, and finally future possible fluctuation is interfered and regulated in advance by combining the future predicted initial power generation data, so that power prediction accuracy is improved, effective counter measures are formulated for power generation enterprises and power grid enterprises, energy supply risks are solved, and technical support is provided for power grid and power supply.
In the embodiment of the application, the fluctuation trend is calculated by the following steps:
Setting a power generation power fluctuation amplitude threshold value and a power generation power fluctuation width threshold value;
Calculating the upper slope and the lower slope of the power generation power variation of each time point according to the power generation power fluctuation width threshold value, the power generation power in the original data and the time corresponding to the power generation power;
Performing iterative compression according to the upper slope and the lower slope, and recording time points meeting the requirement that the upper slope is greater than or equal to the lower slope to obtain a first set;
judging the fluctuation trend of each time point in the first set according to the fluctuation amplitude threshold of the generated power and the generated power of each time point;
and restoring and filling according to the fluctuation trend of each time point in the first set to obtain the fluctuation trend of the original data.
According to the technical means, the upper slope and the lower slope of each time point can be obtained through calculation according to the fluctuation width threshold value, the power generation power and the time, fluctuation iteration is carried out according to the calculated upper slope and lower slope, a point set of fluctuation change is determined, the fluctuation trend of each time point in the point set is determined, the fluctuation trend of the original data is obtained through reduction and filling, the fluctuation trend is prevented from being judged one by one, the calculation time is shortened, and finally the accuracy of the fluctuation trend of the obtained original data is high.
In an embodiment of the present application, the upper slope is calculated by the following formula:
；
Wherein, Is the upper slope of the ith time point,/>Power generation power of (i+1) th time point position,/>The generated power of the ith time point is E is the generated power fluctuation width threshold value,/>For the (i+1) th time point,/>Is the ith time point;
The lower slope is calculated by the following formula:
；
Wherein, Is the lower slope of the ith time point.
According to the technical means, the calculated up-down slope can accurately represent the relationship between power and time.
In the embodiment of the present application, determining a fluctuation trend of each time point in the first set according to a generated power fluctuation amplitude threshold and generated power of each time point includes:
Calculating the power generation power difference value between each time point in the first set and the previous time point;
Judging the fluctuation trend of each time point according to the fluctuation amplitude threshold value of the power generation power and the power generation power difference value:
If the generated power difference value is larger than the generated power fluctuation amplitude threshold value, judging that the generated power is fluctuated upwards;
if the generated power difference value is smaller than the opposite number of the generated power fluctuation amplitude threshold value, judging that the generated power is fluctuated downwards;
And if the generated power difference value is larger than or equal to the opposite number of the generated power fluctuation amplitude threshold value and smaller than or equal to the generated power fluctuation amplitude threshold value, judging that the state is stable.
According to the technical means, the fluctuation trend of the current point position can be judged according to the generated power difference value of the adjacent time point positions and the generated power fluctuation amplitude threshold value.
In the embodiment of the application, the method for restoring and filling the fluctuation trend of the original data according to the fluctuation trend of each time point in the first set comprises the following steps:
Determining two corresponding time points in the original data according to any two adjacent time points in the first set;
Filling trend marks of two adjacent time points in the first set into the corresponding two time points in the original data;
and filling the trend marks of the later time points into the time points between the two time points in the original data to obtain the fluctuation trend of the original data.
According to the technical means, the fluctuation trend in the original data can be restored according to the fluctuation trend of the time points in the first set, the time points in the first set represent the fluctuation time points, and the time points between the two time points are consistent with the time points of the later time point, so that the time points between the two time points are filled forward with trend marks of the time points of the later time point, and the fluctuation trend of the original data can be restored accurately.
In the embodiment of the application, in the process of calculating the upper slope and the lower slope of each time point, the method further comprises the following steps:
each time a new time point is calculated, determining the maximum upper slope according to the upper slope calculated value of the current time point and the upper slope of the previous time point, and determining the upper slope of the current time point;
And determining the minimum lower slope as the lower slope of the current time point according to the lower slope calculated value of the current time point and the lower slope of the previous time point.
According to the technical means, the maximum value in the upper slope is used as the upper slope of the current time point, the minimum value in the lower slope is used as the lower slope of the current time point, the slopes can be processed to be consistent within a certain time, and the problem that robustness is poor due to the oscillation of the generated power in the running process of the fan is avoided.
In the embodiment of the application, the Bayesian probability relation between the power numerical difference and the fluctuation trend is calculated according to the wind turbine generator set historical theoretical power generation data, the wind turbine generator set historical prediction power generation data, the first fluctuation trend and the second fluctuation trend, and the Bayesian probability relation comprises the following steps:
Calculating a first probability of electric power fluctuation of the historical forecast power generation data of the wind turbine generator according to the first fluctuation trend and the second fluctuation trend under the condition that the electric power fluctuation of the historical theoretical power generation data of the wind turbine generator occurs;
calculating the prior probability of electric power fluctuation of historical theoretical power generation data of the wind turbine generator according to the first fluctuation trend statistics;
calculating a second probability of electric power fluctuation of the historical forecast power generation data of the wind turbine generator according to the second fluctuation trend statistics;
Calculating the probability of actually generating power fluctuation under the condition of numerical weather forecast fluctuation according to a Bayesian formula:
；
Wherein, The probability of actually generating power fluctuation under the condition of numerical weather forecast fluctuation is represented; /(I)Is a first probability; /(I)The prior probability of electric power fluctuation for the historical theoretical power generation data of the wind turbine generator; /(I)Is a second probability;
and calculating the power numerical value difference of the same time point according to the theoretical power in the historical theoretical power generation data of the wind turbine and the predicted power in the historical predicted power generation data of the wind turbine.
According to the technical means, based on the Bayesian probability relation, the probability of generating power fluctuation actually occurs under the condition that the current point position is in the numerical weather forecast fluctuation can be calculated, so that the initial generating data can be predicted in the future for adjustment and correction.
In the embodiment of the application, the future prediction initial power generation data is adjusted according to the Bayesian probability relation and the power numerical value difference to obtain the future prediction power generation data, which comprises the following steps:
if the Bayesian probability is larger than a preset value, judging that adjustment is needed, otherwise, not needing adjustment;
Calculating the average value of the power value differences with the same fluctuation trend as an adjustment amplitude;
And linearly superposing the amplitude of the future prediction initial power generation data and the adjustment amplitude to obtain the future prediction power generation data.
According to the technical means, the Bayesian probability indicates the probability that the current point is in the condition of fluctuation of numerical weather forecast, and the larger the numerical value is, the more easily the fluctuation is, so that the fluctuation judgment can be carried out on the future prediction initial power generation data, and when the adjustment is judged to be needed, the average value of the power numerical value differences with the same fluctuation trend is taken as the adjustment amplitude, so that the future prediction initial power generation data is corrected to be more fit with the actual power generation condition, and the accuracy of the prediction power is improved.
A second aspect of the present application provides a power prediction apparatus based on wind power fluctuation identification, the apparatus comprising:
the data acquisition unit is used for acquiring historical operation data of the wind turbine generator;
the data preprocessing unit is used for preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine;
The power generation data prediction unit is used for calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine generator set by using the trained power prediction model;
the fluctuation trend calculation unit is used for respectively calculating a first fluctuation trend of the historical theoretical power generation data of the wind turbine and a second fluctuation trend of the historical forecast power generation data of the wind turbine;
the Bayesian probability calculation unit is used for calculating the power numerical difference and the Bayesian probability relation between the fluctuation trends according to the wind turbine generator set history theoretical power generation data, the wind turbine generator set history prediction power generation data, the first fluctuation trend and the second fluctuation trend;
and the future prediction power generation data correction unit is used for adjusting the future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain the future prediction power generation data.
According to the technical means, historical operation data of the wind turbine are processed into historical theoretical power generation data of the wind turbine, the trained power prediction model is adopted to predict historical predicted power generation data and future predicted initial power generation data of the wind turbine, the historical theoretical power generation data of the wind turbine and the historical predicted power generation data of the wind turbine are used as bases, volatility of the historical theoretical power generation data of the wind turbine and the future predicted initial power generation data is identified, bayesian probability relations among the fluctuation trends are calculated according to the fluctuation trends, and finally future possible fluctuation is interfered and regulated in advance by combining the future predicted initial power generation data, so that power prediction accuracy is improved, effective counter measures are formulated for power generation enterprises and power grid enterprises, energy supply risks are solved, and technical support is provided for power grid and power supply.
A third aspect of the application provides a machine-readable storage medium having stored thereon instructions for causing a machine to perform the method of power prediction based on wind power fluctuation identification.
Through the technical scheme, the wind turbine theoretical power model is built by utilizing the historical operation data of the wind turbine. And identifying and analyzing the fluctuation of the model, combining numerical weather forecast data, establishing a probability prediction model, and performing early intervention on the possible future fluctuation. Compared with the traditional prediction model, the technology improves the model precision, makes effective counter measures for power generation enterprises and power grid enterprises, solves the energy supply risk, and provides technical support for maintaining the power grid and power supply. The technology can be applied to all wind power plants and has good adaptability.
Additional features and advantages of embodiments of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings are included to provide a further understanding of embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain, without limitation, the embodiments of the invention. In the drawings:
FIG. 1 is a flow chart of a power prediction method based on wind power fluctuation identification according to one embodiment of the present invention;
FIG. 2 is a schematic diagram of historical theoretical power generation data of a wind turbine provided in an embodiment of the invention;
FIG. 3 is a schematic diagram of a restoration result of historical theoretical power generation data of a wind turbine generator according to an embodiment of the invention;
FIG. 4 is a schematic diagram showing the comparison of actual data and predicted data fluctuation identifications for days 2023, 2, and 1-7 in an application example of the present invention;
FIG. 5 is a schematic diagram of actual data and comparison of predicted data of 2023, 9, 18-24 days in an application example of the present invention;
FIG. 6 is a block diagram of a power prediction device based on wind power fluctuation identification according to an embodiment of the present invention.
Detailed Description
The following describes specific embodiments of the present invention in detail with reference to the drawings. It should be understood that the detailed description and specific examples, while indicating and illustrating the invention, are not intended to limit the invention.
According to the method, historical operation data of the wind turbine are utilized, the historical operation data are subjected to data cleaning and restoration through the operation state of the wind turbine, and a power prediction model is built. And automatically identifying and analyzing the fluctuation of the theoretical power generation, and carrying out probability prediction by combining numerical forecast data of the same period. And combining future numerical weather forecast data, identifying and intervening possible fluctuation in the future in advance, outputting a power generation predicted value of the interference result through a power prediction model, and adding a factor of a time dimension for the prediction model. Through the continuous accumulation of historical data, the model is continuously updated and iterated, and the fluctuation characteristics can be automatically identified. The model avoids the influence of the traditional machine learning model on the power generated at the same time point by only considering multiple independent variables at a single time point, not only utilizes the advantages of the machine learning model, but also utilizes the fluctuation recognition technology to output the predicted power in the space dimension, adjusts the fluctuation of the power generated in the time dimension, and improves the prediction precision.
FIG. 1 is a flowchart of a power prediction method based on wind power fluctuation identification according to an embodiment of the present invention. As shown in fig. 1, the method includes:
S1: and acquiring historical operation data of the wind turbine.
In the embodiment of the application, the historical operation data of the wind turbine comprise wind turbine numbers, time of a fixed time period, historical operation parameters of the wind turbine, historical data of numerical weather forecast and the like.
S2: preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine.
According to different configurations of the wind turbine, the historical operation data of the wind turbine may have data without operation state identifiers of the wind turbine, and the data with the operation state identifiers of the wind turbine are preprocessed by different methods respectively, for example, the data with the operation state identifiers are used for deleting power data of faults, limited power, capacity reduction operation and high wind cut-out of the states of the wind turbine according to the operation state identifiers, and the wind speed data is reserved, so that power corresponding to wind speed can be supplemented according to adjacent power or theoretical power curves in the subsequent processing process. For the data without running state identification, a common unsupervised machine learning algorithm is adopted for clustering, such as K-means (K-means algorithm) or DBSCAN (Density clustering method), the power data excluding the algorithm is deleted, the wind speed data is reserved, and then cleaner wind turbine data is obtained.
The wind turbine generator system data obtained in this way comprises wind speed and power data, then the wind speed interval method is used for processing the data to obtain wind turbine generator system historical theoretical power generation data, and the wind turbine generator system historical theoretical power generation data can be displayed in a power-time curve. The wind speed interval method firstly needs to divide wind speed into different interval sections according to a set wind speed interval, generally, the wind speed interval is 0.5m/s, and the initial range of the wind speed interval section needs to be lower than a certain value of the cut-out wind speed, so that all wind speeds in wind turbine generator data can be enveloped. Typically, the cut-in wind speed of the wind turbine is 1m/s less than that of the wind turbine. Each interval section at least comprises a preset number of wind speed values, and for the extremely low wind speed part and the extremely high wind speed part, if the number of wind speed values is smaller than the preset number, the intervals need to be combined, and the preset number is 10. And finally, respectively taking an average value, a maximum value and a minimum value of the power data in each interval, and finally forming historical theoretical power generation data of the wind turbine according to the wind speed interval value corresponding to each interval, the average value of the power data, an upper limit value and a lower limit value, wherein the upper limit is the maximum value, the lower limit is the minimum value, and the formed historical theoretical power generation data of the wind turbine is shown in figure 2.
For example, assuming that the cut-in wind speed of the wind turbine generator is 3m/s, the start value of the wind speed interval is 2m/s, the interval is divided by 0.5m/s, the interval is 2m/s-2.5m/s,2.5m/s-3m/s, and so on, assuming that the first interval has only 2 pieces of data, the second interval has 21 pieces of data, since the first interval is less than 10 pieces of data, the first interval and the second interval need to be combined, and the combined interval has 23 pieces of data, so that the average value, the maximum value and the minimum value of the power data of the combined interval are calculated. Assuming that there are 4 data in the interval of 13.5m/s-14m/s, which is less than 10 data, at this time, the data is preferably merged back, i.e. with the interval of 14m/s-14.5m/s, until 10 data are satisfied, if there is no data more than 14m/s, at this time, the interval of 13.5m/s-14m/s is merged with the previous interval (13 m/s-13.5 m/s) to ensure that the data amount satisfies 10.
For the deleted data in the data processing process or the originally deleted data in the acquired historical operation data, firstly judging whether the power data of the previous time point and the next time point of the data are not deleted, wherein the power data are positioned in the upper limit value and the lower limit value of the same wind speed interval of the data, if so, restoring the power generation value of the deleted/deleted time point to be the average value of the power data of the previous time point and the next time point.
Otherwise, the generated power value of the deleted/missing time point is recovered to be the average value of the power data corresponding to the wind speed interval value of the deleted/missing time point.
The complete wind turbine generator set historical theoretical power generation data can be obtained through the repairing, and a complete wind turbine generator set theoretical power generation time sequence can be formed, and the result is shown in fig. 3.
The method for identifying, processing and restoring the abnormal data of the wind turbine generator can clean and restore the data with abnormal identification, truly reflects the theoretical power generation capacity of the wind turbine generator under the running condition, and lays a data foundation for the application of subsequently establishing a power prediction model.
S3: and calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine by using the trained power prediction model.
In the embodiment of the application, the power prediction model is obtained by training a regression model, such as XGBoost (eXtreme Gradient Boosting, extreme gradient lifting), lightGBM (LIGHT GRADIENT Boosting Machine), randomForest (random forest) three common Machine learning models. In the model training process, the historical theoretical power generation data of the wind turbine generator are divided into a training set, a verification set and a detection set, the training set, the verification set and the detection set are generally divided in a ratio of 7:2:1, model training is carried out by taking three models as basic models respectively in a parallel mode, and the three models after training utilize the most accurate models in the RMSE evaluation detection set, so that a power prediction model is determined.
And then taking the numerical weather forecast historical data as input, and predicting by using a power prediction model to obtain the wind turbine generator set historical prediction power generation data. And inputting the future numerical weather forecast data in a digital mode, and predicting by using a power prediction model to obtain the future prediction initial power generation data.
Wherein XGBoost is a tree model for breaking feature splitting, each new tree fits the residual error of the previous prediction, gradient descent is rapidly and accurately carried out by using second-order Taylor expansion, fitting is controlled by using a regularization term, and the method has the advantages of rapid and parallel processing of small and medium-sized data sets.
LightGBM adopts a histogram algorithm to accelerate, adopts a unilateral gradient algorithm to filter samples in training and carries out splitting based on finding one leaf with the largest splitting gain from all current leaves each time, thereby having faster training speed and higher efficiency, reducing the memory utilization rate and having better accuracy.
RandomForest is based on a tree algorithm with subsampling and attribute splitting, which has better adaptability to various problems.
S4: and respectively calculating a first fluctuation trend of the historical theoretical power generation data of the wind turbine and a second fluctuation trend of the historical forecast power generation data of the wind turbine.
In the embodiment of the application, the fluctuation trend is calculated by the following steps:
Firstly, setting a power generation power fluctuation amplitude threshold value and a power generation power fluctuation width threshold value; in one embodiment, the power fluctuation amplitude threshold value P is set to be 15% of rated power of the wind turbine, the power fluctuation width threshold value E is set to be 5% of rated power of the wind turbine, and the set of raw data is recorded as 。
Then, the upper slope and the lower slope of the power generation power variation at each time point are calculated according to the power generation power fluctuation width threshold, the power generation power in the original data and the time corresponding to the power generation power.
The upper slope is calculated by the following formula:
；
Wherein, Is the upper slope of the ith time point,/>Power generation power of (i+1) th time point position,/>The generated power of the ith time point is E is the generated power fluctuation width threshold value,/>For the (i+1) th time point,/>Is the ith time point;
The lower slope is calculated by the following formula:
；
Wherein, Is the lower slope of the ith time point.
Then, performing iterative compression according to the upper slope and the lower slope, recording time points meeting the upper slope and the lower slope to obtain a first setWherein/>For the number of recordings. For compressed collections/>The number of the points in the set is smaller than or equal to the number of the original points.
Then judging the fluctuation trend of each time point in the first set according to the fluctuation amplitude threshold of the generated power and the generated power of each time point, wherein the method specifically comprises the following steps:
Calculating the power generation power difference value between each time point in the first set and the previous time point;
Judging the fluctuation trend of each time point according to the fluctuation amplitude threshold value of the power generation power and the power generation power difference value:
If the generated power difference value is larger than the generated power fluctuation amplitude threshold value P, the upward fluctuation is judged to be +1;
if the power generation power difference value is smaller than the inverse number-P of the power generation power fluctuation amplitude threshold value, judging that the power generation power fluctuation is downwards fluctuated, and recording the power generation power difference value as-1;
If the generated power difference value is greater than or equal to the inverse number-P of the generated power fluctuation amplitude threshold value and is smaller than or equal to the generated power fluctuation amplitude threshold value P, judging that the state is stable, and recording the state as 0, wherein the trend of the initial point position in the first set is 0.
According to the technical means, the fluctuation trend of the current point position can be judged according to the generated power difference value of the adjacent time point positions and the generated power fluctuation amplitude threshold value.
Finally, restoring and filling according to the fluctuation trend of each time point in the first set to obtain the fluctuation trend of the original data, wherein the method specifically comprises the following steps:
Determining two corresponding time points in the original data according to any two adjacent time points in the first set;
Filling trend marks of two adjacent time points in the first set into the corresponding two time points in the original data;
and filling the trend marks of the later time points into the time points between the two time points in the original data to obtain the fluctuation trend of the original data.
For example, a first set ofThe time point corresponding to the two points m 1 and m 2,m1 in the model is t 1, the fluctuation trend is 1, the time point corresponding to the m 2 is t 6, the fluctuation trend is 0, and when the fluctuation trend is restored and filled, the original dataset/>' is found according to t 1 and t 6 Raw data p 1 corresponding to t 1 and raw data p 6 corresponding to t 6 in the set, filling the fluctuation trend of p 1 into 1, filling the fluctuation trend of p 6 into 0, finally filling the fluctuation trend of p 5、p4、p3、p2 into 0 according to p 6, and so on, and finally filling to obtain the fluctuation trend of all data in the raw data set.
According to the technical means, the fluctuation trend in the original data can be restored according to the fluctuation trend of the time points in the first set, the time points in the first set represent the fluctuation time points, and the time points between the two time points are consistent with the time points of the later time point, so that the time points between the two time points are filled forward with trend marks of the time points of the later time point, and the fluctuation trend of the original data can be restored accurately.
According to the technical means, the upper slope and the lower slope of each time point can be obtained through calculation according to the fluctuation width threshold value, the power generation power and the time, fluctuation iteration is carried out according to the calculated upper slope and lower slope, a point set of fluctuation change is determined, the fluctuation trend of each time point in the point set is determined, the fluctuation trend of the original data is obtained through reduction and filling, the fluctuation trend is prevented from being judged one by one, the calculation time is shortened, and finally the accuracy of the fluctuation trend of the obtained original data is high.
In the embodiment of the application, the calculation method of the first fluctuation trend of the historical theoretical power generation data of the wind turbine and the calculation method of the second fluctuation trend of the historical forecast power generation data of the wind turbine are the same, the original data adopted by the calculation of the first fluctuation trend is the historical theoretical power generation data of the wind turbine, and the original data adopted by the calculation of the second fluctuation trend is the historical forecast power generation data of the wind turbine. According to the method, the linear fitting compression algorithm is introduced without depending on selection of a time window, the most important trend of the original data is identified and reserved, and the original data is reversely filled, so that missing detection and false detection of fluctuation identification caused by local fluctuation in the process of selecting complex types of fluctuation of fixed time intervals, complex meteorological conditions or complex terrain conditions are avoided.
Table one is the waving trend data for a portion of the dot bits in one embodiment of the application, because of the 17945 th dot>/>At this time, the 17945 th point is judged as a stage point, and the set/> isrecorded; And therefore, the 17941 th point is the stage point, but the difference of the power of the two points is smaller than the threshold value, the 17941 th point to the 17945 th point are considered to be stable trends, and the stationary trends are marked as 0.
List one
In the embodiment of the application, in the process of calculating the upper slope and the lower slope of each time point, the method further comprises the following steps:
Each time a new time point is calculated, determining the maximum upper slope according to the upper slope calculated value of the current time point and the upper slope of the previous time point, wherein the upper slope calculated value of the current time point is The upper slope of the previous time point is/>Then, the upper slope of the current point location is/>。
Determining the minimum lower slope as the lower slope of the current time point according to the lower slope calculated value of the current time point and the lower slope of the previous time point, wherein the lower slope calculated value of the current time point isThe lower slope of the previous time point is/>Then, the lower slope of the current point location is/>。
According to the technical means, the maximum value in the upper slope is used as the upper slope of the current time point, the minimum value in the lower slope is used as the lower slope of the current time point, the slopes can be processed to be consistent within a certain time, and the problem that robustness is poor due to the oscillation of the generated power in the running process of the fan is avoided.
S5: and calculating the power numerical difference and the Bayesian probability relation between the fluctuation trends according to the wind turbine generator set historical theoretical power generation data, the wind turbine generator set historical prediction power generation data, the first fluctuation trend and the second fluctuation trend.
The basic principle of the bayesian probability relationship is the bayesian theorem, which states that the probability of an event can be updated by new evidence given a priori probabilities. Specifically, the bayesian formulation can be expressed by the following formula:
Wherein, Expressed at/>In case of occurrence,/>Probability of occurrence; /(I)Expressed at/>In case of occurrence,/>Probability of occurrence; /(I)Representation/>The prior probability of occurrence; /(I)Representation/>Probability of occurrence.
In the embodiment of the application, the Bayesian probability relation between the power numerical difference and the fluctuation trend is calculated according to the wind turbine generator set historical theoretical power generation data, the wind turbine generator set historical prediction power generation data, the first fluctuation trend and the second fluctuation trend, and the Bayesian probability relation comprises the following steps:
According to the first fluctuation trend and the second fluctuation trend, the first probability of the electric power fluctuation of the wind turbine generator set historical prediction power generation data under the condition that the electric power fluctuation of the wind turbine generator set historical theoretical power generation data occurs is calculated, namely at the same time point, the wind turbine generator set historical theoretical power generation data fluctuation trend is 1, the wind turbine generator set historical prediction power generation data fluctuation trend is 1, or the wind turbine generator set historical theoretical power generation data fluctuation trend is-1, and the ratio of the data quantity A1 of the wind turbine generator set historical prediction power generation data fluctuation trend is-1 to the data quantity A2 of the wind turbine generator set historical theoretical power generation data fluctuation trend is 1 or-1.
And calculating the prior probability of electric power fluctuation of the historical theoretical power generation data of the wind turbine generator according to the first fluctuation trend statistics, namely, the ratio of the quantity A2 of the electric power fluctuation trend of 1 or-1 of the historical theoretical power generation data of the wind turbine generator to the total data quantity B0 of the historical theoretical power generation data of the wind turbine generator.
Calculating a second probability of electric power fluctuation of the historical forecast power generation data of the wind turbine generator according to the second fluctuation trend statistics; the ratio of the quantity A3 of the historical forecast power generation data fluctuation trend of the wind turbine generator to the total data quantity B1 of the historical forecast power generation data of the wind turbine generator is 1 or-1.
Calculating the probability of actually generating power fluctuation under the condition of numerical weather forecast fluctuation according to a Bayesian formula:
；
Wherein, The probability of actually generating power fluctuation under the condition of numerical weather forecast fluctuation is represented; /(I)Is a first probability; /(I)The prior probability of electric power fluctuation for the historical theoretical power generation data of the wind turbine generator; /(I)Is a second probability;
and calculating the power value difference of the same time point according to the theoretical power in the historical theoretical power generation data of the wind turbine and the predicted power in the historical predicted power generation data of the wind turbine, wherein the power value difference is the difference value between the theoretical power and the predicted power of the same time point.
According to the technical means, based on the Bayesian probability relation, the probability of generating power fluctuation of the current point location under the condition of numerical weather forecast fluctuation can be calculated, so that adjustment and correction can be carried out on the initial generating data predicted in the future, and the blank of identifying time fluctuation of a machine learning model is filled by utilizing the Bayesian probability relation between the historical generating data and the historical predicting data in the time dimension, the numerical difference and the upper limit and the lower limit in the space dimension.
In one embodiment of the application, the calculated probability relationships are shown in Table II.
Watch II
S6: and adjusting future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain future prediction power generation data.
In the embodiment of the application, the future prediction initial power generation data is adjusted according to the Bayesian probability relation and the power numerical value difference to obtain the future prediction power generation data, which comprises the following steps:
if the Bayesian probability is larger than a preset value, the adjustment is judged to be needed, otherwise, the adjustment is not needed, and in general, the probability of the entering Bayesian is larger than 50%, namely, the large probability fluctuation time judgment is considered to be needed to be subjected to the intervention adjustment.
Calculating the average value of the power value differences with the same fluctuation trend as an adjustment amplitude; for example, if the current fluctuation trend of the future predicted power generation data is 1, determining the historical theoretical power generation data of the wind turbine and the time points with the same fluctuation trend of 1 in the historical predicted power generation data of the wind turbine, and calculating the average value as the adjustment amplitude according to the power value difference of the time points.
And linearly superposing the amplitude of the future prediction initial power generation data and the adjustment amplitude to obtain the future prediction power generation data.
According to the technical means, the Bayesian probability indicates the probability that the current point is in the condition of fluctuation of numerical weather forecast, and the larger the numerical value is, the more easily the fluctuation is, so that the fluctuation judgment can be carried out on the future prediction initial power generation data, and when the adjustment is judged to be needed, the average value of the power numerical value differences with the same fluctuation trend is taken as the adjustment amplitude, so that the future prediction initial power generation data is corrected to be more fit with the actual power generation condition, and the accuracy of the prediction power is improved.
According to the technical means, historical operation data of the wind turbine are processed into historical theoretical power generation data of the wind turbine, the trained power prediction model is adopted to predict historical predicted power generation data and future predicted initial power generation data of the wind turbine, the historical theoretical power generation data of the wind turbine and the historical predicted power generation data of the wind turbine are used as bases, volatility of the historical theoretical power generation data of the wind turbine and the future predicted initial power generation data is identified, bayesian probability relations among the fluctuation trends are calculated according to the fluctuation trends, and finally future possible fluctuation is interfered and regulated in advance by combining the future predicted initial power generation data, so that power prediction accuracy is improved, effective counter measures are formulated for power generation enterprises and power grid enterprises, energy supply risks are solved, and technical support is provided for power grid and power supply.
Future predicted power generation data can be predicted according to the method, and in order to verify the accuracy of the predicted result, the following formula can be used for measurement:
。
the power prediction method disclosed by the application can be used for directly predicting the power of the wind turbine generator set and can also be applied to predicting the power of a wind power plant.
In one embodiment, a certain wind farm of an inner Mongolian autonomous region is used as a test wind farm, the installed capacity of the wind farm is 49.5MW, the wind farm is influenced by hilly terrain and wind resources, gusts are large, fluctuation is large, and the short-time up-down fluctuation of the generated power is obvious. The predicted data is short-term predicted data of 0:00-23:45 of the second day of daily 06:30 prediction, the training test data is subjected to data processing, model training and optimization by using 2023, 2, 1, 8, 6 days, and is verified by using 9, 18, 9, 24, and 24 days, the verification time range is long, and the method has certain representative significance.
Training data is shown in figure four, which shows actual data from day 1 to day 7 of year 2 of 2023 compared with predicted data fluctuation identification. The left-hand coordinates are power and the right-hand coordinates are fluctuation marks, wherein 1 represents rising; 0 represents a gentle value; -1 represents a drop. As can be seen from the figures: the original prediction has stronger correlation (77%) than the actual data, but judging that the original prediction has far less fluctuation than actual power fluctuation according to the fluctuation mark distribution, and strengthening the prediction fluctuation is required. The set fluctuation recognition parameters are respectively the initial width=7.5%, Fluctuation amplitude threshold/>15%.
According to Bayesian probability relation and power value difference, future prediction initial power generation data are adjusted, the adjustment results are shown in a fifth graph and a third graph, when gust fluctuation occurs in the prediction data using the technology of the invention, the prediction fluctuation is obviously increased, and the fluctuation of power generation power can be predicted in advance, and the prediction upper limit and the prediction lower limit are given. The improvement is obvious for 19 days, 22 days and 23 days in 9 months, and the accuracy is respectively improved by 1.5%, 3.2% and 0.5%.
Watch III
A second aspect of the present application provides a power prediction apparatus based on wind power fluctuation recognition, as shown in fig. 6, the apparatus comprising:
the data acquisition unit is used for acquiring historical operation data of the wind turbine generator;
the data preprocessing unit is used for preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine;
The power generation data prediction unit is used for calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine generator set by using the trained power prediction model;
the fluctuation trend calculation unit is used for respectively calculating a first fluctuation trend of the historical theoretical power generation data of the wind turbine and a second fluctuation trend of the historical forecast power generation data of the wind turbine;
the Bayesian probability calculation unit is used for calculating the power numerical difference and the Bayesian probability relation between the fluctuation trends according to the wind turbine generator set history theoretical power generation data, the wind turbine generator set history prediction power generation data, the first fluctuation trend and the second fluctuation trend;
and the future prediction power generation data correction unit is used for adjusting the future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain the future prediction power generation data.
According to the technical means, historical operation data of the wind turbine are processed into historical theoretical power generation data of the wind turbine, the trained power prediction model is adopted to predict historical predicted power generation data and future predicted initial power generation data of the wind turbine, the historical theoretical power generation data of the wind turbine and the historical predicted power generation data of the wind turbine are used as bases, volatility of the historical theoretical power generation data of the wind turbine and the future predicted initial power generation data is identified, bayesian probability relations among the fluctuation trends are calculated according to the fluctuation trends, and finally future possible fluctuation is interfered and regulated in advance by combining the future predicted initial power generation data, so that power prediction accuracy is improved, effective counter measures are formulated for power generation enterprises and power grid enterprises, energy supply risks are solved, and technical support is provided for power grid and power supply.
A third aspect of the application provides a machine-readable storage medium having stored thereon instructions for causing a machine to perform the method of power prediction based on wind power fluctuation identification.
Those skilled in the art will appreciate that all or part of the steps in a method for implementing the above embodiments may be implemented by a program stored in a storage medium, where the program includes several instructions for causing a single-chip microcomputer, chip or processor (processor) to perform all or part of the steps in a method according to the embodiments of the invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The alternative embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the embodiments of the present invention are not limited to the specific details of the above embodiments, and various simple modifications may be made to the technical solutions of the embodiments of the present invention within the scope of the technical concept of the embodiments of the present invention, and all the simple modifications belong to the protection scope of the embodiments of the present invention. In addition, the specific features described in the above embodiments may be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the various possible combinations of embodiments of the invention are not described in detail.
In addition, any combination of the various embodiments of the present invention may be made, so long as it does not deviate from the idea of the embodiments of the present invention, and it should also be regarded as what is disclosed in the embodiments of the present invention.
Claims (10)
1. A power prediction method based on wind power fluctuation identification, the method comprising:
Acquiring historical operation data of a wind turbine generator;
Preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine;
calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine by using the trained power prediction model;
Respectively calculating a first fluctuation trend of historical theoretical power generation data of the wind turbine generator and a second fluctuation trend of historical forecast power generation data of the wind turbine generator;
Calculating a power numerical difference and a Bayesian probability relation between the fluctuation trends according to the wind turbine generator set historical theoretical power generation data, the wind turbine generator set historical prediction power generation data, the first fluctuation trend and the second fluctuation trend;
and adjusting future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain future prediction power generation data.
2. The power prediction method based on wind power fluctuation identification according to claim 1, wherein the fluctuation trend is calculated by the steps of:
Setting a power generation power fluctuation amplitude threshold value and a power generation power fluctuation width threshold value;
Calculating the upper slope and the lower slope of the power generation power variation of each time point according to the power generation power fluctuation width threshold value, the power generation power in the original data and the time corresponding to the power generation power;
Performing iterative compression according to the upper slope and the lower slope, and recording time points meeting the requirement that the upper slope is greater than or equal to the lower slope to obtain a first set;
judging the fluctuation trend of each time point in the first set according to the fluctuation amplitude threshold of the generated power and the generated power of each time point;
and restoring and filling according to the fluctuation trend of each time point in the first set to obtain the fluctuation trend of the original data.
3. The power prediction method based on wind power fluctuation identification according to claim 2, wherein the upper slope is calculated by the following formula:
；
Wherein, Is the upper slope of the ith time point,/>Power generation power of (i+1) th time point position,/>The generated power of the ith time point is E is the generated power fluctuation width threshold value,/>For the (i+1) th time point,/>Is the ith time point;
The lower slope is calculated by the following formula:
；
Wherein, Is the lower slope of the ith time point.
4. The method for predicting power based on wind power fluctuation identification according to claim 2, wherein determining a fluctuation trend of each time point in the first set according to the generated power fluctuation amplitude threshold and the generated power of each time point comprises:
Calculating the power generation power difference value between each time point in the first set and the previous time point;
Judging the fluctuation trend of each time point according to the fluctuation amplitude threshold value of the power generation power and the power generation power difference value:
If the generated power difference value is larger than the generated power fluctuation amplitude threshold value, judging that the generated power is fluctuated upwards;
if the generated power difference value is smaller than the opposite number of the generated power fluctuation amplitude threshold value, judging that the generated power is fluctuated downwards;
And if the generated power difference value is larger than or equal to the opposite number of the generated power fluctuation amplitude threshold value and smaller than or equal to the generated power fluctuation amplitude threshold value, judging that the state is stable.
5. The power prediction method based on wind power fluctuation identification according to claim 2, wherein the step of restoring and filling the fluctuation trend of the raw data according to the fluctuation trend of each time point in the first set comprises the steps of:
Determining two corresponding time points in the original data according to any two adjacent time points in the first set;
Filling trend marks of two adjacent time points in the first set into the corresponding two time points in the original data;
and filling the trend marks of the later time points into the time points between the two time points in the original data to obtain the fluctuation trend of the original data.
6. The method for predicting power based on wind power fluctuation identification of claim 2, wherein in calculating the upper slope and the lower slope of each time point, the method further comprises:
each time a new time point is calculated, determining the maximum upper slope according to the upper slope calculated value of the current time point and the upper slope of the previous time point, and determining the upper slope of the current time point;
And determining the minimum lower slope as the lower slope of the current time point according to the lower slope calculated value of the current time point and the lower slope of the previous time point.
7. The power prediction method based on wind power fluctuation recognition according to claim 1, wherein calculating the bayesian probability relationship between the power numerical difference and the fluctuation trend from the wind turbine historical theoretical power generation data, the wind turbine historical predicted power generation data, the first fluctuation trend and the second fluctuation trend comprises:
Calculating a first probability of electric power fluctuation of the historical forecast power generation data of the wind turbine generator according to the first fluctuation trend and the second fluctuation trend under the condition that the electric power fluctuation of the historical theoretical power generation data of the wind turbine generator occurs;
calculating the prior probability of electric power fluctuation of historical theoretical power generation data of the wind turbine generator according to the first fluctuation trend statistics;
calculating a second probability of electric power fluctuation of the historical forecast power generation data of the wind turbine generator according to the second fluctuation trend statistics;
Calculating the probability of actually generating power fluctuation under the condition of numerical weather forecast fluctuation according to a Bayesian formula:
；
Wherein, The probability of actually generating power fluctuation under the condition of numerical weather forecast fluctuation is represented; /(I)Is a first probability; /(I)The prior probability of electric power fluctuation for the historical theoretical power generation data of the wind turbine generator; /(I)Is a second probability;
and calculating the power numerical value difference of the same time point according to the theoretical power in the historical theoretical power generation data of the wind turbine and the predicted power in the historical predicted power generation data of the wind turbine.
8. The power prediction method based on wind power fluctuation recognition according to claim 1, wherein the adjusting future prediction initial power generation data according to bayesian probability relation and power numerical value difference to obtain future prediction power generation data comprises:
if the Bayesian probability is larger than a preset value, judging that adjustment is needed, otherwise, not needing adjustment;
Calculating the average value of the power value differences with the same fluctuation trend as an adjustment amplitude;
And linearly superposing the amplitude of the future prediction initial power generation data and the adjustment amplitude to obtain the future prediction power generation data.
9. A power prediction device based on wind power fluctuation identification, the device comprising:
the data acquisition unit is used for acquiring historical operation data of the wind turbine generator;
the data preprocessing unit is used for preprocessing historical operation data of the wind turbine to obtain historical theoretical power generation data of the wind turbine;
The power generation data prediction unit is used for calculating historical prediction power generation data and future prediction initial power generation data of the wind turbine generator set by using the trained power prediction model;
the fluctuation trend calculation unit is used for respectively calculating a first fluctuation trend of the historical theoretical power generation data of the wind turbine and a second fluctuation trend of the historical forecast power generation data of the wind turbine;
the Bayesian probability calculation unit is used for calculating the power numerical difference and the Bayesian probability relation between the fluctuation trends according to the wind turbine generator set history theoretical power generation data, the wind turbine generator set history prediction power generation data, the first fluctuation trend and the second fluctuation trend;
and the future prediction power generation data correction unit is used for adjusting the future prediction initial power generation data according to the Bayesian probability relation and the power numerical value difference to obtain the future prediction power generation data.
10. A machine-readable storage medium having instructions stored thereon for causing a machine to perform the wind power fluctuation identification-based power prediction method of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410105791.8A CN118232308A (en) | 2024-01-25 | 2024-01-25 | Power prediction method and device based on wind power fluctuation recognition and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410105791.8A CN118232308A (en) | 2024-01-25 | 2024-01-25 | Power prediction method and device based on wind power fluctuation recognition and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118232308A true CN118232308A (en) | 2024-06-21 |
Family
ID=91499857
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410105791.8A Pending CN118232308A (en) | 2024-01-25 | 2024-01-25 | Power prediction method and device based on wind power fluctuation recognition and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118232308A (en) |
-
2024
- 2024-01-25 CN CN202410105791.8A patent/CN118232308A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111091233B (en) | Short-term wind power prediction modeling method for wind power plant | |
| CN112597691B (en) | LightGBM algorithm-based fault early warning method for wind turbine generator variable pitch motor temperature sensor | |
| CN114792156B (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| CN110766200A (en) | Method for predicting generating power of wind turbine generator based on K-means mean clustering | |
| US20230258161A1 (en) | A method and an apparatus for computer-implemented prediction of power production of one or more wind turbines in a wind farm | |
| CN114202129A (en) | Wind power output prediction method, electronic device, storage medium and system | |
| CN113657662B (en) | Downscaling wind power prediction method based on data fusion | |
| CN112669168B (en) | Short-term prediction method for wind power | |
| CN111695736A (en) | Photovoltaic power generation short-term power prediction method based on multi-model fusion | |
| CN117013527A (en) | Distributed photovoltaic power generation power prediction method | |
| CN116629416A (en) | Photovoltaic power station power prediction method and device | |
| CN112598148A (en) | Fan variable pitch motor temperature fault early warning method based on collaborative expression and LightGBM algorithm | |
| CN116402203A (en) | Method, system and medium for predicting short-time photovoltaic power generation capacity considering weather conditions | |
| CN111697560B (en) | Method and system for predicting load of power system based on LSTM | |
| CN111967660B (en) | Ultra-short-term photovoltaic prediction residual error correction method based on SVR | |
| CN117374920A (en) | Ultra-short-term prediction method, device and medium considering environmental factors | |
| CN117200352A (en) | Photovoltaic power generation regulation and control method and system based on cloud edge fusion | |
| CN118232308A (en) | Power prediction method and device based on wind power fluctuation recognition and storage medium | |
| CN116484998A (en) | Distributed photovoltaic power station power prediction method and system based on meteorological similar day | |
| CN111178601B (en) | Wind turbine generator power prediction method based on meteorological data post-processing | |
| CN113780643A (en) | Photovoltaic power station short-term output prediction method based on case reasoning | |
| CN113344293A (en) | Photovoltaic power prediction method based on NCA-fusion regression tree model | |
| CN112633565A (en) | Photovoltaic power aggregation interval prediction method | |
| CN111178593A (en) | Photovoltaic system output power prediction method and device | |
| Huang et al. | Probabilistic prediction intervals of wind speed based on explainable neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |