CN115481791A - Water-wind power generation and power generation combined prediction method, device and equipment - Google Patents
Water-wind power generation and power generation combined prediction method, device and equipment Download PDFInfo
- Publication number
- CN115481791A CN115481791A CN202211078133.1A CN202211078133A CN115481791A CN 115481791 A CN115481791 A CN 115481791A CN 202211078133 A CN202211078133 A CN 202211078133A CN 115481791 A CN115481791 A CN 115481791A
- Authority
- CN
- China
- Prior art keywords
- power
- data
- prediction
- photovoltaic
- wind
- 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
- 238000010248 power generation Methods 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000012549 training Methods 0.000 claims abstract description 49
- 238000013136 deep learning model Methods 0.000 claims abstract description 16
- 238000011156 evaluation Methods 0.000 claims description 47
- 239000011159 matrix material Substances 0.000 claims description 39
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 35
- 230000015654 memory Effects 0.000 claims description 19
- 230000002159 abnormal effect Effects 0.000 claims description 12
- 238000003169 complementation method Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims 1
- 238000012163 sequencing technique Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 14
- 238000013528 artificial neural network Methods 0.000 description 10
- 238000004364 calculation method Methods 0.000 description 10
- 238000012360 testing method Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 230000000295 complement effect Effects 0.000 description 7
- 230000005611 electricity Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 230000002457 bidirectional effect Effects 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 3
- 150000001875 compounds Chemical class 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000005855 radiation Effects 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000001556 precipitation Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 241000284466 Antarctothoa delta Species 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000008020 evaporation Effects 0.000 description 1
- 238000001704 evaporation Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a combined prediction method, a device and equipment for water-wind power generation and power generation, wherein the method comprises the following steps: acquiring a data set, wherein the data set comprises historical meteorological data and historical hydropower, wind power and photovoltaic power data; training and checking a plurality of deep learning models according to divided seasons and weather by using a data set aiming at hydroelectric power, wind power and photovoltaic power respectively, and sequencing a preset number of deep learning models in advance with prediction accuracy as candidate models; inputting current meteorological data into a candidate model under corresponding seasons and weather conditions to obtain hydropower, wind power and photovoltaic prediction power, and forming a water-wind-light combined prediction power set; and under the condition of minimum residual load, extracting the optimal hydroelectric power, wind power and photovoltaic power from the water-wind-light combined prediction power set. According to the technical scheme provided by the invention, the accuracy of the water-wind-light combined power prediction is improved.
Description
Technical Field
The invention relates to the field of electric power energy, in particular to a combined prediction method, a combined prediction device and combined prediction equipment for water-wind power generation and photovoltaic power generation.
Background
Nowadays, low-carbon economy is vigorously developed, and hydroenergy, wind energy and photovoltaic are increasingly paid more attention to all countries in the world as environment-friendly and clean renewable energy sources. However, the inherent volatility and randomness of hydraulic energy, wind energy, and photovoltaic can pose serious challenges to the safety and stability of power systems and the quality of electric power. Therefore, by predicting the water-wind-solar power generation power, multiple energy sources of water, wind and light are aggregated to form a multi-energy complementary power generation system, and the system has great significance for reducing clean energy grid-connected impact and improving basin resource utilization rate. As meteorological factors are main factors causing the fluctuation of the water and wind power generation, the prior art refers to meteorological data to predict the power of water energy, wind energy and photovoltaic power generation, and the power prediction accuracy is improved. However, the existing power prediction methods are used for power prediction of hydropower, wind power and photovoltaic power generation independently, and the correlation influence among several energy sources of water, wind and light is not considered. Therefore, how to comprehensively consider the mutual influence among water, wind and light and carry out combined prediction on the water-wind power generation power is a problem to be considered for further improving the accuracy and the practicability of the water-electricity, wind-electricity and photovoltaic power prediction.
Disclosure of Invention
In view of this, the embodiment of the invention provides a water-wind-solar combined power prediction method, a water-wind-solar combined power prediction device and water-wind-solar combined power prediction equipment, so that the accuracy and the practicability of water-wind-solar combined power prediction are improved.
According to a first aspect, an embodiment of the present invention provides a water-wind power generation and power generation combined prediction method, where the method includes: acquiring a data set, wherein the data set comprises historical meteorological data serving as a training sample, and historical hydropower power data, historical wind power data and historical photovoltaic power data serving as labels; training a plurality of hydropower power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models by using the data set; inputting current meteorological data into each power prediction model, and predicting corresponding hydropower, wind power and photovoltaic power sets; and when water-wind-solar combined output is selected from the hydropower, wind power and photovoltaic power set, the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power under the condition of minimum residual load are selected.
Optionally, the acquiring the data set comprises: dividing the historical meteorological data and the corresponding historical hydropower power data, historical wind power data and historical photovoltaic power data according to seasons and weather; and identifying main control meteorological factors influencing hydroelectric power, wind power and photovoltaic power from the historical meteorological data according to the divided seasons and weather, and taking the main control meteorological factors as training samples.
Optionally, training a plurality of hydroelectric power prediction models, a plurality of wind power prediction models, and a plurality of photovoltaic power prediction models using the data set includes: training and checking out a plurality of deep learning models by utilizing the historical meteorological data and the historical hydropower power data to obtain a plurality of hydropower power prediction models; training and checking out a plurality of deep learning models by utilizing the historical meteorological data and the historical wind power data to obtain a plurality of wind power prediction models; training and checking out a plurality of deep learning models by utilizing the historical meteorological data and the historical photovoltaic power data to obtain a plurality of photovoltaic power prediction models; and taking the average absolute percentage error and the decision coefficient as evaluation indexes of prediction accuracy, and respectively selecting power prediction models with preset accuracy ranked in the front preset number from the plurality of hydroelectric power prediction models, the plurality of wind power prediction models and the plurality of photovoltaic power prediction models as candidate models for power prediction.
Optionally, the step of inputting the current meteorological data into each power prediction model to predict the corresponding power set of hydropower, wind power and photovoltaic includes: inputting current meteorological data into the candidate models under the season and weather conditions; and respectively forming corresponding hydropower, wind power and photovoltaic power sets by utilizing the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence output by each candidate model.
Optionally, when the combined water-wind-solar output is selected from the hydropower, wind power and photovoltaic power sets, the optimal hydropower predicted power, optimal wind power predicted power and optimal photovoltaic predicted power under the condition of minimum residual load include: respectively combining the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence which are obtained by different models, and performing summation operation to obtain a plurality of total power sequences; respectively calculating by using the total load sequence and each total power sequence to obtain a plurality of residual load standard deviations; and selecting an optimal total power sequence corresponding to the minimum residual load standard deviation, and taking the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence used in the summation calculation of the optimal total power sequence as the optimal hydropower predicted power, the optimal wind power predicted power and the optimal photovoltaic predicted power.
Optionally, when the combined water-wind-solar output is selected from the hydropower, wind power and photovoltaic power sets, the optimal hydropower predicted power, optimal wind power predicted power and optimal photovoltaic predicted power under the condition of minimum residual load include: respectively combining the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence which are obtained by different models, and performing summation operation to obtain a plurality of total power sequences; respectively calculating by using the total load sequence and each total power sequence to obtain a plurality of residual load standard deviations; respectively establishing corresponding evaluation matrixes by utilizing the standard deviation of each residual load from the angles of hydropower, wind power and photovoltaic; performing fuzzy optimization decision based on the created hydropower evaluation matrix, wind power evaluation matrix and photovoltaic evaluation matrix to determine a hydropower power prediction model, a wind power prediction model and a photovoltaic power prediction model with the maximum relative dominance from the hydropower, wind power and photovoltaic power prediction models; and taking the predicted power output by the water electric power prediction model, the wind power prediction model and the photovoltaic power prediction model with the maximum relative dominance as the optimal hydropower predicted power, the optimal wind power predicted power and the optimal photovoltaic predicted power.
Optionally, before the dividing the historical meteorological data and the corresponding historical hydroelectric power data, historical wind power data, and historical photovoltaic power data by season and weather, the method further comprises: and (3) checking abnormal values and missing values in the historical meteorological data, the historical hydropower power data, the historical wind power data and the historical photovoltaic power data, deleting the historical meteorological data missing the power data, and correcting the abnormal power data by using a K neighbor complementation method.
According to a second aspect, an embodiment of the present invention provides a water-wind power generation and power generation joint prediction apparatus, including: the data acquisition module is used for acquiring a data set, wherein the data set comprises historical meteorological data serving as a training sample, historical hydropower power data serving as a label, historical wind power data and historical photovoltaic power data; the model training module is used for training a plurality of hydropower power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models by utilizing the data set; the initial prediction module is used for inputting the current meteorological data into each power prediction model and predicting the corresponding hydropower, wind power and photovoltaic power set; and the prediction result optimizing module is used for selecting the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power under the condition of the minimum residual load when the water-wind-solar combined output is selected from the hydropower, wind power and photovoltaic power set.
According to a third aspect, an embodiment of the present invention provides a water-wind power generation and power generation combined prediction apparatus, including: a memory and a processor, the memory and the processor being communicatively coupled to each other, the memory having stored therein computer instructions, and the processor performing the method of the first aspect, or any one of the optional embodiments of the first aspect, by executing the computer instructions.
According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores computer instructions for causing a computer to thereby perform the method described in the first aspect, or any one of the optional implementation manners of the first aspect.
The technical scheme provided by the application has the following advantages:
according to the technical scheme, historical meteorological data in a historical time range and corresponding historical hydropower power data, historical wind power data and historical photovoltaic power data are obtained. And training a plurality of deep learning models for predicting the power generation power by respectively aiming at water, electricity, wind and photovoltaic by taking historical meteorological data as training samples of a machine learning model and various historical power data as labels. Then, current meteorological data (for example, meteorological data of the current day or the previous day is obtained according to weather forecast) is obtained, and then the current meteorological data is sequentially input into all the trained power prediction models, so that a plurality of hydropower prediction power sequences, a plurality of wind power prediction power sequences and a plurality of photovoltaic prediction power sequences are obtained. And then, storing the output predicted power into corresponding power sets according to the types of hydropower, wind power and photovoltaic. And then, freely combining and collocating the water and electricity predicted power sequence, the wind and electricity predicted power sequence and the photovoltaic predicted power sequence in each set. The combined predicted power is applied to load consumption (such as power values required by various load devices in an industrial park), and the optimal power combination is selected according to the condition of the minimum residual load, so that the optimal prediction model of hydropower, wind power and photovoltaic is optimal, and the multi-energy complementary power generation efficiency is highest. On one hand, the accuracy of the water-wind-light combined power prediction is improved, and on the other hand, the application value of the water-wind-light combined power prediction result in short-term scheduling planning is enhanced.
Drawings
The features and advantages of the present invention will be more clearly understood by reference to the accompanying drawings, which are illustrative and not to be construed as limiting the invention in any way, and in which:
FIG. 1 is a schematic diagram illustrating steps of a water-wind power generation and power generation combined prediction method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating final scores of candidate models for power prediction of year 2017, month 4 and day 30 in one embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating the final scores of candidate models for the 2017, 7, month, and 30-day power prediction in accordance with one embodiment of the present invention;
FIG. 4 is a schematic diagram illustrating final scores of candidate models for power prediction of year 2017, month 10 and day 30 in one embodiment of the present invention;
FIG. 5 is a schematic diagram illustrating the final scores of the candidate models for the 2017, 12-month, 31-day power prediction;
fig. 6 shows a schematic structural diagram of a water-wind power generation and power generation combined prediction device according to an embodiment of the present invention;
fig. 7 shows a schematic structural diagram of a water-wind power generation and power generation combined prediction device in an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings of the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of the present invention.
Referring to fig. 1, in one embodiment, a method for joint prediction of hydroelectric and wind-generated power includes the following steps:
step S101: acquiring a data set, wherein the data set comprises historical meteorological data serving as a training sample, and historical hydropower power data, historical wind power data and historical photovoltaic power data serving as labels.
Step S102: and training a plurality of hydropower power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models by using the data set.
Step S103: and inputting the current meteorological data into each power prediction model, and predicting the corresponding hydropower, wind power and photovoltaic power set.
Step S104: and when the water-wind-solar combined output is selected from the power set of water, electricity, wind and photovoltaic, the optimal predicted power of water and electricity, the optimal predicted power of wind and photovoltaic and the optimal predicted power of photovoltaic under the condition of the minimum residual load are obtained.
Particularly, in consideration of power prediction mainly developed around single energy sources of water, wind and light in the prior art, mutual influence among water, wind and light is not comprehensively considered, and combined prediction is carried out on water-wind-solar power generation power. Based on the above, the embodiment of the invention provides a water-wind power generation and power generation combined prediction method. Firstly, acquiring a data set, wherein the data set comprises historical meteorological data in a preset time range, such as meteorological data of the past year, and historical hydropower power data, historical wind power data and historical photovoltaic power data which are actually acquired in the time range. The historical meteorological data are training samples, and the historical hydropower power data, the historical wind power data and the historical photovoltaic power data are used as labels. Based on the data, training and verifying a plurality of deep learning models respectively aiming at the hydroelectric power, the wind power and the photovoltaic power to obtain a plurality of corresponding hydroelectric power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models. The model algorithms employed in the present embodiment include, but are not limited to, BP neural networks, LSTM neural networks, bidirectional LSTM neural networks (BI-LSTM), GRU neural networks, bidirectional GRU neural networks (BI-GRU), and Extreme Learning Machines (ELM). After the training of each power prediction model is completed, current meteorological data is collected, for example, the meteorological data of the day or the meteorological data of the previous day is collected according to weather forecast, and the warehousing flow forecast data provided by the hydrological department is collected. And then inputting the current meteorological data into each power prediction model to obtain a hydropower, wind power and photovoltaic prediction power sequence, and storing power data in the hydropower, wind power and photovoltaic power sequence into a hydropower, wind power and photovoltaic power set according to the power generation type. And then, randomly taking out one predicted power sequence from the power set of hydropower, wind power and photovoltaic respectively to carry out summation operation, and correspondingly obtaining a plurality of different total power sequences by different combination calculations. And then calculating the residual load condition corresponding to each total power sequence after the total load consumption supplied by each total power sequence is calculated, and then correspondingly obtaining the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power according to the total power under the minimum residual load condition. Therefore, the prediction models of water, electricity, wind and photovoltaic are optimal, the multi-energy complementary power generation precision is highest, and the accuracy of the combined prediction of the water, wind and electricity generation power is improved.
Specifically, in an embodiment, the step S101 specifically includes the following steps:
the method comprises the following steps: and dividing the historical meteorological data, the corresponding historical hydropower power data, the historical wind power data and the historical photovoltaic power data according to seasons and weather.
Step two: and identifying main control meteorological factors influencing hydroelectric power, wind power and photovoltaic power from historical meteorological data according to the divided seasons and weather.
Step three: and taking the main control meteorological factor as a training sample.
Specifically, in this embodiment, in order to further improve the accuracy of the model training, the training samples are first preprocessed, and the historical meteorological data obtained from the water-wind-light-energy complementary power generation system includes various meteorological factors, such as wind speed, wind direction, air pressure, precipitation, runoff, direct radiation, scattered radiation and air temperature. In addition, for hydropower, the training sample may further include hydropower station water level, ex-warehouse flow, output data, hydropower station scheduling regulations, and the like. Then, dividing historical meteorological data of water, wind and light and corresponding power data into four-season data of spring, summer, autumn and winter, and dividing photovoltaic data into three types of sunny days, cloudy days and rainy and snowy days according to the data of each season; wherein the season division adopts the solar calendar method: spring in 3-5 months, summer in 6-8 months, autumn in 9-11 months, and winter in 12-2 months. And then screening main control meteorological factors for water, wind and light participation training according to seasons and weather, specifically calculating Pearson correlation coefficients of the meteorological factors of wind power and hydropower data and the generated power of the meteorological factors in different seasons, and calculating Pearson correlation coefficients of the meteorological factors of photovoltaic power generation data and the generated power of the photovoltaic power generation data in different seasons and different weather. The pearson correlation coefficient calculation formula is as follows:
in the formula: x is the number of i Value of a certain meteorological factor in the ith time interval, y i Is the power generation power value of the i-th period,
is the average value of a certain meteorological factor,
n is the total number of time segments of the entire data set, which is the average value of the generated power.
Then, the correlation number is larger than a preset threshold value (rho) xy Not less than 0.6) as the dividing basis of the main control meteorological factor and the input factor of the prediction model.
In addition, in an embodiment, the preprocessed data can be further clustered according to the divided seasons and weather, so that the accuracy of dividing the seasons and the weather is further improved. Taking photovoltaic data as an example, the specific method comprises the following steps: carrying out photovoltaic power generation data clustering analysis based on K-means + + algorithm, and averaging the screened meteorological factorsThe value and the variance are used as input factors of a K-means + + algorithm for clustering analysis, and the contour coefficient is used for obtaining the optimal clustering result { K } of the photovoltaic power generation data of different weathers in different seasons spr ，K sum ，K aut ，K win In which K spr ，K sum ，K aut ，K win Respectively representing the best clustering results of different weather type data of spring, summer, autumn and winter).
Therefore, the power prediction model is trained in a targeted manner by utilizing the data of different seasons and different weathers to obtain model sets under different seasons and weather conditions, so that a more matched model set is selected for prediction according to the season and the weather of the current meteorological data in the subsequent prediction process, and the accuracy of power prediction is further improved.
Specifically, in an embodiment, before the first step, abnormal values and missing values existing in the historical meteorological data, the historical hydroelectric power data, the historical wind power data and the historical photovoltaic power data are also checked, the historical meteorological data missing the power data are deleted, and the abnormal power data are corrected by using a K-neighbor complementation method. The method specifically comprises the step of preprocessing abnormal values and missing values of meteorological data and power data. And (3) regarding the data continuously missing for more than 16 time periods as missing data, deleting the missing data, identifying the abnormal value by using a box plot, and processing by using a K-neighbor complementation method, thereby further improving the accuracy of the data. The calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
represents the value, x, of the K neighbor complementation method i-k Represents an abnormal value x i Sample value, x, of the k-th subsequent period i+k Indicates an abnormal value x i The sample value of the following k-th period.
Specifically, in this embodiment, the step S102 specifically includes the following steps:
step four: and training and checking out a plurality of deep learning models by using the historical meteorological data and the historical hydroelectric power data so as to obtain a plurality of hydroelectric power prediction models.
Step five: and training and checking out a plurality of deep learning models by using the historical meteorological data and the historical wind power data to obtain a plurality of wind power prediction models.
Step six: and training and checking out a plurality of deep learning models by using the historical meteorological data and the historical photovoltaic power data to obtain a plurality of photovoltaic power prediction models.
Step seven: and taking the average absolute percentage error and the decision coefficient as evaluation indexes of prediction accuracy, and respectively selecting power prediction models with preset accuracy ranked in the front preset number from the plurality of water electric power prediction models, the plurality of wind electric power prediction models and the plurality of photovoltaic power prediction models as candidate models for power prediction.
Specifically, in the present embodiment, several deep learning models, such as BP neural network, LSTM neural network, bidirectional LSTM neural network (BI-LSTM), GRU neural network, bidirectional GRU neural network (BI-GRU), and Extreme Learning Machine (ELM), are initialized in advance, and 4 to 10 models may be selected for training. The preprocessed data set is divided into a training set (60%), a verification set (20%) and a test set (20%) respectively. Training each model by using a training set, verifying and testing on the verification set and the test set, and reasonably adjusting model parameters.
Taking photovoltaic data as an example, wind power data and hydroelectric data are the same. Inputting test set data divided according to seasons and weather into each model trained by using a training set, and then calculating the average absolute percentage error MAPE and the decision coefficient R of the photovoltaic power generation predicted power output by each model 2 For evaluating the accuracy of the respective power prediction models, wherein MAPE and R 2 The calculation formula of (2) is as follows:
in the formula, MAPE k And
respectively representing the average absolute percentage error and the decision coefficient of the photovoltaic power predicted by the kth deep learning model,
represents the predicted value y of the photovoltaic power in the ith period obtained by adopting the kth deep learning model i The true value of the power of the ith time interval of the corresponding test set is represented,
and m represents the number of samples of the photovoltaic data in the test set divided according to seasons and weather.
And then, selecting a preset number of power prediction models with the highest precision (for example, 3-5 power prediction models with the highest precision ranking) according to the calculation result, thereby further improving the accuracy of model training and improving the accuracy of subsequent power prediction.
Specifically, in this embodiment, the step S103 specifically includes the following steps:
step eight: and inputting the current meteorological data into corresponding candidate models under the season and weather conditions.
Step nine: and respectively forming corresponding hydropower, wind power and photovoltaic power sets by utilizing the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence output by each candidate model.
Specifically, the power generation power is predicted according to the current weather forecast information, the weather factors screened in the second step are obtained from the current weather data, then the mean value and the variance of the day forecast data to be predicted are calculated according to the weather factors, the day is divided into seasons and weather categories corresponding to the water, wind and light data by using the calculation result, a prediction model set corresponding to the current season and weather is selected according to the corresponding seasons and weather categories, and then the candidate models with the preset number of top-ranked precision ranks are selected from the prediction model set according to the average absolute percentage error and the decision coefficient to perform power prediction. Therefore, through the steps of this embodiment, the predicted power sequences of the models with different electric powers of the water, the wind and the light in the day are calculated more accurately, for example, the predicted power sequences output by the three highest-precision water electric power prediction models, the three highest-precision wind electric power prediction models and the three highest-precision photovoltaic power prediction models are respectively:
in the set, superscripts h, w and p respectively represent hydroelectric power, wind power and photovoltaic power, subscripts 1,2 and 3 represent prediction models with the top three precision sequences selected from candidate models, and pow represents a prediction power sequence (one power sequence can be understood as a power prediction value of the model for each time interval in a period of time). To be provided with
For example, the parameter represents the power predicted value of the 1 st model in the ith time interval in the wind power prediction set, and the meanings of other parameters in the power set are the same.
Specifically, in this embodiment, the step S104 specifically includes the following steps:
step ten: and respectively combining the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence which are obtained by different models, and performing summation operation to obtain a plurality of total power sequences.
Step eleven: and calculating to obtain a plurality of residual load standard deviations by using the total load sequences and corresponding to all the total power sequences respectively.
Step twelve: and selecting an optimal total power sequence corresponding to the minimum residual load standard deviation, and taking a hydropower prediction power sequence, a wind power prediction power sequence and a photovoltaic prediction power sequence used when the optimal total power sequence is calculated by summation as the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power.
Specifically, one predicted power sequence is taken out from the hydroelectric power set, the wind power set and the photovoltaic power set, the three taken predicted power sequences are subjected to summation operation, a total power sequence is obtained, and a plurality of total power sequences can be obtained through different matching and combining conditions of a plurality of groups. And aiming at each total power sequence, calculating the difference value between the total power sequence and the total load sequence to obtain a residual load sequence, and then calculating the standard deviation of the residual load sequence. The residual load standard deviation corresponding to each total power sequence can be obtained, the minimum residual load condition can be judged based on the minimum residual standard deviation, and when a matched residual standard deviation is minimum, the predicted power sequence taken out from the power set is correspondingly used as the optimal hydropower predicted power, the optimal wind power predicted power and the optimal photovoltaic predicted power.
The residual load standard deviation was calculated as follows:
δ i,c ＝L i -SP i,c
in the formula, sl c Represents the group of model cThe standard deviation of the combined residual load, m represents the number of test lumped time segments, L i Representing the total load of the i-th period, SP i,c Representing the sum of predicted values of water, wind and photovoltaic power for the ith time interval under the combination of the model c. Delta. For the preparation of a coating i，c Representing the residual load for the i-th period under the c-th model combination,
and c represents the average residual load of all time periods under the c model combination, and the c represents different candidate model combinations of three energy sources, and the value is 1-27.
The optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power and the power prediction models with the highest precision can be determined by searching the minimum residual standard deviation in all combinations.
Specifically, in another embodiment, the step S104 specifically includes the following steps:
step thirteen: and respectively combining the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence obtained by different models, and performing summation operation to obtain a plurality of total power sequences.
Fourteen steps: and calculating to obtain a plurality of residual load standard deviations by using the total load sequences and corresponding to all the total power sequences respectively.
A fifteenth step: and respectively creating corresponding evaluation matrixes by utilizing the standard deviation of each residual load from the perspective of hydropower, wind power and photovoltaic.
Sixthly, the steps are as follows: and carrying out fuzzy optimization decision based on the created hydropower evaluation matrix, wind power evaluation matrix and photovoltaic evaluation matrix, and determining a hydropower power prediction model, a wind power prediction model and a photovoltaic power prediction model with the maximum relative dominance degree from the hydropower, wind power and photovoltaic power prediction models.
Seventeen steps: and taking the predicted power output by the water electric power prediction model, the wind power prediction model and the photovoltaic power prediction model with the maximum relative dominance as the optimal hydropower predicted power, the optimal wind power predicted power and the optimal photovoltaic predicted power.
Specifically, in this embodiment, after the residual standard deviations of the various combinations are calculated, the determination is not directly performed according to the minimum residual standard deviation, but the residual standard deviations of the various combinations are processed by a fuzzy optimal decision method, so as to select the optimal predicted power under the condition of the minimum residual load through analysis and calculation, thereby further improving the accuracy of the optimal predicted power. Specifically, corresponding evaluation matrixes are created by using the standard deviation of the residual load from the perspective of hydropower, wind power and photovoltaic.
The residual load standard deviation for the angle of interest is calculated as follows:
δ i | nn,b ＝L i -SP i | nn,b
in the formula, sl nn,b Representing the standard deviation of the residual load of the model combination, m representing the number of test lumped periods, L i Representing the total load of the i-th period, SP i | nn,b And the sum of predicted values of water, wind and solar power generation power of the ith time interval under the combination of the models is represented. Delta i | nn,b Representing the residual load for the i-th period under this model combination,
representing the average residual load for all time periods under this model combination. Mark- nn,b And the subscript nn represents the candidate model number of the current concerned angle (hydroelectric power, wind power or photovoltaic), the value can be 1,2 or 3, and the subscript b represents different candidate model combinations of other two energy sources, and the value is 1-9.
Taking hydropower, wind power and photovoltaic power prediction to respectively select 3 candidate models as an example, according to the definition of fuzzy optimal decision, the created evaluation matrix is as follows:
hydroelectric evaluation matrix:
wind power evaluation matrix:
photovoltaic evaluation matrix:
the evaluation matrix can be understood as that the evaluation matrix is respectively created by emphasizing three angles of hydropower, wind power and photovoltaic, so that a prediction power value with higher accuracy is searched. In the hydropower evaluation matrix, each row represents the standard deviation of the residual load formed by combining the same hydropower power prediction model with different wind power and different photovoltaic power prediction models; each column represents the residual load standard deviation formed by combining three different candidate hydropower power prediction models, the same wind power prediction model and the same photovoltaic power prediction model, so that three rows and nine columns of evaluation matrixes are formed. The wind power evaluation matrix and the photovoltaic evaluation matrix are in the same way. For the convenience of distinguishing, letters k, f and g represent row numbers of the hydropower evaluation matrix, the wind power evaluation matrix and the photovoltaic evaluation matrix respectively.
Then, the weight of each index in the matrix is respectively calculated by an entropy weight method:
for the hydroelectric evaluation matrix:
for the wind power evaluation matrix:
for the photovoltaic evaluation matrix:
in the formula (I), the compound is shown in the specification,
respectively representing the weights corresponding to the indexes of the b-th column in the hydropower evaluation matrix, the wind power evaluation matrix and the photovoltaic evaluation matrix,
respectively represents the information entropies corresponding to the indexes of the b-th column in the hydropower evaluation matrix, the wind power evaluation matrix and the photovoltaic evaluation matrix,
respectively represent the probability value of the kth row and the b column index of the hydropower evaluation matrix, the probability value of the fth row and the b column index of the wind power evaluation matrix and the probability value of the gth row and the b column index of the photovoltaic evaluation matrix, and the hydropower evaluation matrix comprises
Wherein k, f or g can be 1-3.
And then, calculating the relative dominance of the water, wind and light prediction models by using the evaluation matrix and the index weight, thereby screening the model with the lowest residual load which can achieve the power prediction effect. The relative membership vectors of the three candidate model sets are respectively:
wherein the parameters represent the relative dominance of the corresponding candidate model, and the specific calculation formula is as follows:
In the formula (I), the compound is shown in the specification,
respectively representing the normalization result of the indexes of the kth row and the b column in the hydropower evaluation matrix, the normalization result of the fth row and the b column in the wind power evaluation matrix and the normalization result of the gth row and the b column in the photovoltaic evaluation matrix, wherein the normalization principle is as follows:
After three relative membership vectors of hydropower, wind power and photovoltaic are obtained, an optimal prediction model is selected according to a maximum membership principle, and the optimal prediction model is specifically calculated according to the following formula:
for example: the relative membership vector of the hydropower candidate model set is
According to the formula
Calculate BM h And =2, the power prediction model with the highest accuracy in the hydropower candidate model set is the second candidate model, and the prediction result is finally used for services such as short-term scheduling planning and the like. The calculation processes of the wind power and the photovoltaic are the same, so that the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power with higher accuracy under the condition of the minimum residual load are obtained from the angles of the heavy hydropower, the heavy wind power and the heavy photovoltaic respectively through the steps.
Specifically, the following explains the water-wind power generation power joint prediction method in an actual scene embodiment:
1. meteorological data and power data are acquired. Acquiring total radiation, 70m wind speed, air temperature, precipitation, runoff, potential evaporation and wind power, photovoltaic power generation power and hydropower station output; the time resolution of all data is 1 hour, and the data sequence length is 2014-2016 years.
2. And (4) preprocessing data. And the missing-free data does not need to be processed, and abnormal data is identified by using a box diagram. Because meteorological data has strong randomness, the embodiment of the invention processes a single abnormal value by using a K-neighbor complementation method, and the K value is 2.
3. Weather and season classification. Dividing water, wind and solar power generation data into four types of data of spring, summer, autumn and winter; and the photovoltaic power generation data is further divided into a sunny day, a rainy and snowy day and a cloudy day according to weather conditions.
4. And screening master control meteorological factors participating in training in water, wind and light data. For hydroelectric prediction, the main control meteorological factors are historical runoff data; for wind power prediction, the main control meteorological factor is historical wind speed data of 70 m; for photoelectric prediction, the main control meteorological factors are historical total irradiation and air temperature data.
5. And (3) carrying out water, wind and light data clustering analysis based on a K-means + + algorithm. For spring data, photometric data are divided into 100 days in sunny days, 14 days in rainy and snowy days and 65 days in cloudy days; the wind measuring data are divided into 2 types, wherein the first type is 57 days, and the second type is 122 days; the hydraulic data were divided into 3 categories, 43 days in the first category, 48 days in the second category, and 88 days in the third category. For summer data, photometric data are divided into 108 days in a sunny day, 10 days in a rainy and snowy day and 65 days in a cloudy day; the wind measuring data are divided into 3 types, wherein the first type is 94 days, the second type is 71 days, and the third type is 18 days; the hydraulic data were divided into 3 categories, 75 days in the first category, 104 days in the second category, and 4 days in the third category. For autumn data, photometric data are divided into 66 days in sunny days, 48 days in rainy and snowy days and 104 days in cloudy days; the wind measuring data are divided into 2 types, the first type is 54 days, and the second type is 129 days; hydraulic data are divided into 2 types, the first type is 152 days, and the second type is 31 days; for winter data, photometric data are divided into 67 days in sunny days, 6 days in rainy and snowy days and 110 days in cloudy days; the wind measuring data are divided into 2 types, 159 days for the first type and 25 days for the second type; the hydraulic data were divided into three categories, 85 days in the first category, 25 days in the second category, and 74 days in the third category.
6. And building a power generation power prediction model based on a plurality of deep learning algorithms. Inputting different types of water, wind and light training set and verification set data into LSTM, BI-LSTM, GRU, BI-GRU, BP and ELM models to adjust model parameters, inputting a test set into the models, and determining a plurality of candidate models for power prediction with the top precision according to the average absolute error and the decision coefficient of the models of different types, wherein the hydroelectric results in each season are shown in tables 1-4, the wind power results in each season are shown in tables 5-8, and the photovoltaic results in each season are shown in tables 9-12.
TABLE 1 spring Water energy Power Generation Each Power prediction model Effect and candidate model
TABLE 2 prediction model effect and candidate model for each power of hydropower generation in summer
TABLE 3 autumn hydropower generation each power prediction model effect and candidate model
TABLE 4 prediction model effect and candidate model of each power of hydropower generation in winter
TABLE 5 spring wind power generation individual power prediction model effect and candidate model
TABLE 6 summer wind power generation each power prediction model effect and candidate model
TABLE 7 autumn wind power generation each power prediction model effect and candidate model
TABLE 8 winter wind power generation individual power prediction model effects and candidate models
TABLE 9 spring photovoltaic power generation each power prediction model effect and candidate model
TABLE 10 photovoltaic power generation each power prediction model effect in summer and candidate model
TABLE 11 autumn photovoltaic power generation each power prediction model effect and candidate model
TABLE 12 model effect and candidate model for forecasting each power of photovoltaic power generation in winter
7. And the self-adaptive joint prediction of the water, wind and solar power generation is realized by combining the forecast information in the day. And selecting the days to be predicted from the year 2017, 4 and 30 months, 7 and 30 months, 10 and 30 months and 12 and 31 months. The seasons and the water, wind, and light categories for the four days to be predicted are shown in table 13.
Watch 13 to-be-tested day, season and weather division watch
From tables 1 to 12, candidate models for each measurement day are searched as shown in table 14 using table 13.
TABLE 14 candidate models for days to be tested
Based on table 14, the residual load standard deviation is calculated according to the output results of the candidate models of each power generation type, and fuzzy optimal decision evaluation is performed on the residual load standard deviation, and the result of obtaining an optimal prediction model according to the relative dominance degree is as follows:
as can be seen from fig. 2, in 2017, day 4, 30: for hydroelectric data, the LSTM final score is 0.66, the GRU final score is 0.71, and the BI-GRU final score is 0.34, so the best predictive model is GRU; aiming at wind power prediction data, the GRU is finally scored to be 1, BI-GRU is finally scored to be 0.01, ELM is finally scored to be 0, and therefore the best prediction model is GRU; for the photoelectric prediction data, the final score of BI-GRU was 1, the final score of BP was 0.00094, ELM was 0, and thus the best prediction model was BI-GRU.
As can be seen from fig. 3, 30 months 7 in 2017: aiming at hydroelectric data, the LSTM has a final score of 1, the GRU has a final score of 0, and the BI-GRU has a final score of 0.80, so that the optimal prediction model is LSTM; aiming at wind power prediction data, the final score of GRU is 0.58, the final score of BI-GRU is 1, the final score of ELM is 0, and therefore the best prediction model is BI-GRU; for the photoelectric prediction data, the BI-LSTM final score was 0.01 and the GRU final score was 1, the BI-GRU final score was 0, so the best prediction model was GRU.
As can be seen from fig. 4, 10 and 30 months in 2017: aiming at hydroelectric data, the LSTM has a final score of 0.001, the GRU has a final score of 1, BI-GRU has a final score of 0, and therefore the best prediction model is GRU; aiming at wind power prediction data, the final score of EGRU is 0.41, the final score of BP is 0.02 and the final score of LM is 0.99, so that the optimal prediction model is ELM; for photoelectric prediction data, the BI-LSTM final score was 1, the BI-GRU final score was 0.20, and the best prediction model was BI-LSTM.
As can be seen from fig. 5, 12, 31 in 2017: for the hydroelectric data, the BI-LSTM final score is 0.003, the GRU final score is 0.028, the BI-GRU final score is 1, so the best prediction model is BI-GRU; aiming at wind power prediction data, the final score of LSTM is 0.46, the final score of BP is 0.15, and the final score of ELM is 0.77, so that the optimal prediction model is ELM; for photoelectric prediction data, the BI-GRU final score is 1, the BP final score is 0.15, the ELM final score is 0, and therefore the best prediction model is BI-GRU.
And finally, acquiring the predicted power output by the optimal prediction model of each day to be measured as the optimal hydropower predicted power, the optimal wind power predicted power and the optimal photovoltaic predicted power of each day to be measured, and compiling a day-ahead power generation scheduling plan according to the predicted power, the predicted power and the optimal photovoltaic predicted power, and directing the water, wind and light multi-energy complementary system to perform short-term scheduling operation.
Through the steps, according to the technical scheme provided by the application, historical meteorological data in a historical time range, and corresponding historical hydropower power data, historical wind power data and historical photovoltaic power data are obtained at first. Historical meteorological data are used as training samples of prediction models based on a deep learning algorithm, various historical power data are used as labels, and a plurality of models used for predicting power generation power are trained respectively aiming at hydropower, wind power and photovoltaics, so that a plurality of hydropower power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models are obtained. Then, current meteorological data (for example, meteorological data of the current day or the previous day is obtained according to weather forecast) is obtained, and then the current meteorological data is sequentially input into all the trained power prediction models, so that a plurality of hydropower prediction power sequences, a plurality of wind power prediction power sequences and a plurality of photovoltaic prediction power sequences are obtained. And then, storing the output predicted power sequence into corresponding power sets according to the types of hydropower, wind power and photovoltaic. And then, freely combining and collocating a hydropower predicted power sequence, a wind power predicted power sequence and a photovoltaic predicted power sequence from each set. The combined total predicted power sequence is applied to load consumption (such as power values required by various load devices in an industrial park), and then the optimal predicted power combination is selected according to the condition of the minimum residual load, so that the optimal and multi-energy complementary power generation efficiency of the hydropower, wind power and photovoltaic prediction models is the highest, and the accuracy of the combined prediction of the hydropower, wind power and photovoltaic power generation is improved.
As shown in fig. 6, the present embodiment further provides a water-wind power generation and power generation combined prediction apparatus, which includes:
the data acquisition module 101 is configured to acquire a data set, where the data set includes historical meteorological data serving as a training sample, and historical hydropower power data, historical wind power data, and historical photovoltaic power data serving as tags. For details, refer to the related description of step S101 in the above method embodiment, and details are not repeated herein.
The model training module 102 is configured to train a plurality of hydroelectric power prediction models, a plurality of wind power prediction models, and a plurality of photovoltaic power prediction models by using the data set. For details, refer to the related description of step S102 in the above method embodiment, and no further description is provided here.
And the initial prediction module 103 is used for inputting the current meteorological data into each power prediction model to predict the corresponding power set of hydropower, wind power and photovoltaic. For details, refer to the related description of step S103 in the above method embodiment, and no further description is provided here.
And the prediction result optimizing module 104 is used for selecting the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power under the condition of the minimum residual load when the water-wind-solar combined output is selected from the hydropower, wind power and photovoltaic power set. For details, refer to the related description of step S104 in the above method embodiment, and are not repeated herein.
The water-wind solar power generation combined prediction device provided by the embodiment of the invention is used for executing the water-wind solar power generation combined prediction method provided by the embodiment, the implementation manner and the principle are the same, and the detailed content refers to the related description of the method embodiment and is not repeated.
Through the cooperative cooperation of the components, the technical scheme provided by the application firstly obtains historical meteorological data in a historical time range and corresponding historical hydropower power data, historical wind power data and historical photovoltaic power data. And training a plurality of models for predicting the power generation power respectively for hydropower, wind power and photovoltaic by taking historical meteorological data as training samples of machine learning models and various historical power data as labels, thereby obtaining a plurality of hydropower power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models. Then, current meteorological data are obtained (for example, meteorological data of the day or the previous day are obtained according to weather forecast), and then the current meteorological data are sequentially input into all the trained power prediction models, so that a plurality of hydropower prediction power sequences, a plurality of wind power prediction power sequences and a plurality of photovoltaic prediction power sequences are obtained. And then, storing the output predicted power sequence into corresponding power sets according to the types of hydropower, wind power and photovoltaic. And then, freely combining and collocating a hydropower predicted power sequence, a wind power predicted power sequence and a photovoltaic predicted power sequence from each set. The combined total predicted power sequence is applied to load consumption (such as power values required by various load devices in an industrial park), and then the optimal power combination is selected according to the condition of the minimum residual load, so that the optimal prediction model of hydropower, wind power and photovoltaic has the highest multi-energy complementary power generation efficiency, and the accuracy of the combined prediction of the hydropower, wind power and photovoltaic power generation is improved.
Fig. 7 shows a water-wind power generation and power generation combined prediction device according to an embodiment of the present invention, which includes a processor 901 and a memory 902, which may be connected by a bus or by other means, and fig. 7 illustrates a connection by a bus as an example.
The memory 902, which is a non-transitory computer-readable storage medium, may be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as program instructions/modules corresponding to the methods in the above-described method embodiments. The processor 901 executes various functional applications and data processing of the processor by executing non-transitory software programs, instructions and modules stored in the memory 902, that is, implements the methods in the above-described method embodiments.
The memory 902 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created by the processor 901, and the like. Further, the memory 902 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 902 may optionally include memory located remotely from the processor 901, which may be connected to the processor 901 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
One or more modules are stored in the memory 902, which when executed by the processor 901 perform the methods in the above-described method embodiments.
The specific details of the water-wind power generation and power generation combined prediction device can be understood by referring to the corresponding related descriptions and effects in the above method embodiments, and are not described herein again.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, and the implemented program can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic Disk, an optical Disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a Flash Memory (Flash Memory), a Hard Disk (Hard Disk Drive, abbreviated as HDD), a Solid State Drive (SSD), or the like; the storage medium may also comprise a combination of memories of the kind described above.
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope defined by the appended claims.
Claims (10)
1. A water-wind power generation and power generation combined prediction method is characterized by comprising the following steps:
acquiring a data set, wherein the data set comprises historical meteorological data serving as a training sample, and historical hydropower power data, historical wind power data and historical photovoltaic power data serving as labels;
training a plurality of hydropower power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models by using the data set;
inputting current meteorological data into each power prediction model, and predicting corresponding hydropower, wind power and photovoltaic power sets;
and when the water-wind-solar combined output is selected from the water-electricity, wind-electricity and photovoltaic power set, the optimal water-electricity predicted power, the optimal wind-electricity predicted power and the optimal photovoltaic predicted power under the condition of the minimum residual load are selected.
2. The method of claim 1, wherein the acquiring the data set comprises:
dividing the historical meteorological data and corresponding historical hydropower power data, historical wind power data and historical photovoltaic power data according to seasons and weather;
according to the divided seasons and weather, main control meteorological factors influencing hydroelectric power, wind power and photovoltaic power are identified from historical meteorological data, and the main control meteorological factors are used as training samples.
3. The method of claim 1 or 2, wherein training a plurality of hydroelectric power prediction models, a plurality of wind power prediction models, and a plurality of photovoltaic power prediction models using the data set comprises:
training and checking out a plurality of deep learning models by utilizing the historical meteorological data and the historical hydropower power data to obtain a plurality of hydropower power prediction models;
training and verifying a plurality of deep learning models by using the historical meteorological data and the historical wind power data to obtain a plurality of wind power prediction models;
training and checking out a plurality of deep learning models by utilizing the historical meteorological data and the historical photovoltaic power data to obtain a plurality of photovoltaic power prediction models;
and taking the average absolute percentage error and the decision coefficient as evaluation indexes of prediction accuracy, and respectively selecting power prediction models with preset accuracy ranked in the front preset number from the plurality of hydroelectric power prediction models, the plurality of wind power prediction models and the plurality of photovoltaic power prediction models as candidate models for power prediction.
4. The method of claim 3, wherein inputting the current meteorological data into each power prediction model to predict the corresponding hydroelectric, wind and photovoltaic power set comprises:
inputting current meteorological data into the candidate models under the season and weather conditions;
and respectively forming corresponding hydropower, wind power and photovoltaic power sets by utilizing the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence output by each candidate model.
5. The method of claim 4, wherein the selecting the combined hydro-wind power output, the optimal predicted wind power output and the optimal predicted photovoltaic power output under the minimum residual load condition from the set of hydro-power, wind power and photovoltaic power comprises:
respectively combining the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence obtained by different models, and performing summation operation to obtain a plurality of total power sequences;
respectively calculating by using the total load sequence and each total power sequence to obtain a plurality of residual load standard deviations;
and selecting an optimal total power sequence corresponding to the minimum residual load standard deviation, and taking a hydropower prediction power sequence, a wind power prediction power sequence and a photovoltaic prediction power sequence used when the optimal total power sequence is calculated by summation as the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power.
6. The method of claim 4, wherein the selecting the combined hydro-wind power output, the optimal predicted wind power output and the optimal predicted photovoltaic power output under the minimum residual load condition from the set of hydro-power, wind power and photovoltaic power comprises:
respectively combining the hydropower predicted power sequence, the wind power predicted power sequence and the photovoltaic predicted power sequence which are obtained by different models, and performing summation operation to obtain a plurality of total power sequences;
respectively calculating by using the total load sequence and each total power sequence to obtain a plurality of residual load standard deviations;
respectively establishing corresponding evaluation matrixes by utilizing the standard deviation of each residual load from the angles of hydropower, wind power and photovoltaic;
carrying out fuzzy optimization decision based on the created hydropower evaluation matrix, the created wind power evaluation matrix and the created photovoltaic evaluation matrix so as to determine a water electric power prediction model, a wind power prediction model and a photovoltaic power prediction model with the maximum relative dominance from the used hydropower, wind power and photovoltaic power prediction models;
and taking the predicted power output by the water electric power prediction model, the wind power prediction model and the photovoltaic power prediction model with the maximum relative dominance as the optimal hydropower predicted power, the optimal wind power predicted power and the optimal photovoltaic predicted power.
7. The method of claim 2, wherein prior to said partitioning said historical meteorological data and corresponding historical hydroelectric power data, historical wind power data, and historical photovoltaic power data by season and weather, the method further comprises:
and (3) checking abnormal values and missing values in the historical meteorological data, the historical hydropower power data, the historical wind power data and the historical photovoltaic power data, deleting the historical meteorological data missing the power data, and correcting the abnormal power data by using a K neighbor complementation method.
8. A water-wind power generation and power generation combined prediction device, the device comprising:
the data acquisition module is used for acquiring a data set, wherein the data set comprises historical meteorological data serving as a training sample, historical hydropower power data serving as a label, historical wind power data and historical photovoltaic power data;
the model training module is used for training a plurality of hydroelectric power prediction models, a plurality of wind power prediction models and a plurality of photovoltaic power prediction models by utilizing the data set;
the initial prediction module is used for inputting the current meteorological data into each power prediction model and predicting corresponding hydropower, wind power and photovoltaic power sets;
and the prediction result optimizing module is used for selecting the optimal hydropower prediction power, the optimal wind power prediction power and the optimal photovoltaic prediction power under the condition of the minimum residual load when the water-wind-solar combined output is selected from the hydropower, wind power and photovoltaic power set.
9. A water-wind power generation and power generation combined prediction device is characterized by comprising:
a memory and a processor, the memory and the processor being communicatively coupled to each other, the memory having stored therein computer instructions, the processor performing the method of any of claims 1-7 by executing the computer instructions.
10. A computer-readable storage medium having stored thereon computer instructions for causing a computer to thereby perform the method of any one of claims 1-7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211078133.1A CN115481791A (en) | 2022-09-05 | 2022-09-05 | Water-wind power generation and power generation combined prediction method, device and equipment |
| PCT/CN2023/115461 WO2024051524A1 (en) | 2022-09-05 | 2023-08-29 | Joint prediction method and apparatus for hydraulic, wind and photovoltaic generation power |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211078133.1A CN115481791A (en) | 2022-09-05 | 2022-09-05 | Water-wind power generation and power generation combined prediction method, device and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115481791A true CN115481791A (en) | 2022-12-16 |
Family
ID=84423976
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211078133.1A Pending CN115481791A (en) | 2022-09-05 | 2022-09-05 | Water-wind power generation and power generation combined prediction method, device and equipment |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115481791A (en) |
| WO (1) | WO2024051524A1 (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116581755A (en) * | 2023-07-12 | 2023-08-11 | 长江水利委员会水文局 | Power prediction method, device, equipment and storage medium |
| CN116760028A (en) * | 2023-08-14 | 2023-09-15 | 长江三峡集团实业发展(北京)有限公司 | Wind-solar combined power prediction method, device, computer equipment and medium |
| WO2024051524A1 (en) * | 2022-09-05 | 2024-03-14 | 中国长江三峡集团有限公司 | Joint prediction method and apparatus for hydraulic, wind and photovoltaic generation power |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118054485B (en) * | 2024-04-16 | 2024-06-21 | 中成空间（深圳）智能技术有限公司 | Wind-solar energy intelligent operation and maintenance control method, system and medium based on air film system |
| CN118172047B (en) * | 2024-05-14 | 2024-08-02 | 山东大学 | Short-term wind power prediction method and system considering available installed capacity |
| CN118174331B (en) * | 2024-05-14 | 2024-08-16 | 北京国电光宇机电设备有限公司 | Coordination optimization method and system for wind power photovoltaic energy storage proportion |
| CN118501745A (en) * | 2024-05-21 | 2024-08-16 | 淮阴工学院 | Lithium battery SOC and SOH joint estimation method and system based on error correction |
| CN118396430B (en) * | 2024-06-27 | 2024-09-10 | 水利部水利水电规划设计总院 | Short-term forecasting and uncertainty analysis method and system for wind, light, water and electricity set |
| CN118469103B (en) * | 2024-07-12 | 2024-09-20 | 西安相远科技有限公司 | Energy-saving optimization method and system for industrial circulating water system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2017028983A (en) * | 2015-07-24 | 2017-02-02 | 株式会社明電舎 | Operation scheduling device for distributed energy system, and operation scheduling method |
| CN108205717A (en) * | 2017-12-30 | 2018-06-26 | 国网江苏省电力公司无锡供电公司 | A kind of photovoltaic generation power Multiple Time Scales Forecasting Methodology |
| CN109103929A (en) * | 2018-09-11 | 2018-12-28 | 燕山大学 | Based on the power distribution network economic optimization dispatching method for improving dynamic gram Li Sijin model |
| CN109740808A (en) * | 2018-12-27 | 2019-05-10 | 南瑞集团有限公司 | A kind of scene water complemental power-generation plan calculation method and system |
| CN110400056A (en) * | 2019-07-04 | 2019-11-01 | 国网四川省电力公司电力科学研究院 | Cascade hydropower based on honourable space-time characterisation Optimization Scheduling and device a few days ago |
| CN110880789A (en) * | 2019-12-06 | 2020-03-13 | 国网江苏省电力有限公司南通供电分公司 | Economic dispatching method for wind power and photovoltaic combined power generation system |
| CN111429024A (en) * | 2020-04-09 | 2020-07-17 | 云南电网有限责任公司 | Water-wind-light short-term combined scheduling method considering flexibility balance mechanism |
| CN113505938A (en) * | 2021-07-26 | 2021-10-15 | 中国电力科学研究院有限公司 | Ultra-short-term wind power combined prediction method and system |
| CN114386681A (en) * | 2021-12-28 | 2022-04-22 | 南京聚润信息科技有限公司 | Water luminous output joint prediction method |
| CN114996960A (en) * | 2022-06-16 | 2022-09-02 | 国家电网公司西南分部 | Two-stage optimized operation method and device for wind, light and water power generation system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105896578B (en) * | 2016-04-13 | 2018-12-11 | 合肥工业大学 | A kind of random production analog method for wind-solar-storage joint electricity generation system |
| KR102364176B1 (en) * | 2019-12-20 | 2022-02-18 | 한국전자기술연구원 | Comparative Evaluation Method and System of Photovoltaic Power Prediction Model |
| CN115481791A (en) * | 2022-09-05 | 2022-12-16 | 中国长江三峡集团有限公司 | Water-wind power generation and power generation combined prediction method, device and equipment |
-
2022
- 2022-09-05 CN CN202211078133.1A patent/CN115481791A/en active Pending
-
2023
- 2023-08-29 WO PCT/CN2023/115461 patent/WO2024051524A1/en unknown
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2017028983A (en) * | 2015-07-24 | 2017-02-02 | 株式会社明電舎 | Operation scheduling device for distributed energy system, and operation scheduling method |
| CN108205717A (en) * | 2017-12-30 | 2018-06-26 | 国网江苏省电力公司无锡供电公司 | A kind of photovoltaic generation power Multiple Time Scales Forecasting Methodology |
| CN109103929A (en) * | 2018-09-11 | 2018-12-28 | 燕山大学 | Based on the power distribution network economic optimization dispatching method for improving dynamic gram Li Sijin model |
| CN109740808A (en) * | 2018-12-27 | 2019-05-10 | 南瑞集团有限公司 | A kind of scene water complemental power-generation plan calculation method and system |
| CN110400056A (en) * | 2019-07-04 | 2019-11-01 | 国网四川省电力公司电力科学研究院 | Cascade hydropower based on honourable space-time characterisation Optimization Scheduling and device a few days ago |
| CN110880789A (en) * | 2019-12-06 | 2020-03-13 | 国网江苏省电力有限公司南通供电分公司 | Economic dispatching method for wind power and photovoltaic combined power generation system |
| CN111429024A (en) * | 2020-04-09 | 2020-07-17 | 云南电网有限责任公司 | Water-wind-light short-term combined scheduling method considering flexibility balance mechanism |
| CN113505938A (en) * | 2021-07-26 | 2021-10-15 | 中国电力科学研究院有限公司 | Ultra-short-term wind power combined prediction method and system |
| CN114386681A (en) * | 2021-12-28 | 2022-04-22 | 南京聚润信息科技有限公司 | Water luminous output joint prediction method |
| CN114996960A (en) * | 2022-06-16 | 2022-09-02 | 国家电网公司西南分部 | Two-stage optimized operation method and device for wind, light and water power generation system |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024051524A1 (en) * | 2022-09-05 | 2024-03-14 | 中国长江三峡集团有限公司 | Joint prediction method and apparatus for hydraulic, wind and photovoltaic generation power |
| CN116581755A (en) * | 2023-07-12 | 2023-08-11 | 长江水利委员会水文局 | Power prediction method, device, equipment and storage medium |
| CN116581755B (en) * | 2023-07-12 | 2023-09-29 | 长江水利委员会水文局 | Power prediction method, device, equipment and storage medium |
| CN116760028A (en) * | 2023-08-14 | 2023-09-15 | 长江三峡集团实业发展(北京)有限公司 | Wind-solar combined power prediction method, device, computer equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2024051524A1 (en) | 2024-03-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115481791A (en) | Water-wind power generation and power generation combined prediction method, device and equipment | |
| CN107766990B (en) | Method for predicting power generation power of photovoltaic power station | |
| CN109002915B (en) | Photovoltaic power station short-term power prediction method based on Kmeans-GRA-Elman model | |
| CN115293415A (en) | Multi-wind-farm short-term power prediction method considering time evolution and space correlation | |
| CN107194495B (en) | Photovoltaic power longitudinal prediction method based on historical data mining | |
| CN110070226A (en) | Photovoltaic power prediction technique and system based on convolutional neural networks and meta learning | |
| Melzi et al. | Hourly solar irradiance forecasting based on machine learning models | |
| CN111753893A (en) | Wind turbine generator power cluster prediction method based on clustering and deep learning | |
| CN112257941A (en) | Photovoltaic power station short-term power prediction method based on improved Bi-LSTM | |
| CN106251001A (en) | A kind of based on the photovoltaic power Forecasting Methodology improving fuzzy clustering algorithm | |
| CN106503867A (en) | A kind of genetic algorithm least square wind power forecasting method | |
| CN115271253B (en) | Method and device for constructing water-wind-solar power generation power prediction model and storage medium | |
| CN112215428B (en) | Photovoltaic power generation power prediction method and system based on error correction and fuzzy logic | |
| CN114444378A (en) | Short-term power prediction method for regional wind power cluster | |
| CN114169445A (en) | Day-ahead photovoltaic power prediction method, device and system based on CAE and GAN hybrid network | |
| CN113191918A (en) | Lunar wind-solar power generation power scene analysis method based on time sequence generation countermeasure network | |
| CN109902340A (en) | It is a kind of meter and complicated meteorology coupled characteristic multi-source-lotus combine scene generating method | |
| CN114399081A (en) | Photovoltaic power generation power prediction method based on weather classification | |
| CN115759389A (en) | Day-ahead photovoltaic power prediction method based on weather type similar day combination strategy | |
| CN116402203A (en) | Method, system and medium for predicting short-time photovoltaic power generation capacity considering weather conditions | |
| CN107403015A (en) | A kind of short-term luminous power Forecasting Methodology based on Time Series Similarity | |
| CN112149905A (en) | Photovoltaic power station short-term power prediction method based on wavelet transformation and wavelet neural network | |
| CN110852492A (en) | Photovoltaic power ultra-short-term prediction method for finding similarity based on Mahalanobis distance | |
| CN117277372A (en) | Multi-time-scale joint scheduling method and system for optical storage station and electronic equipment | |
| CN116014722A (en) | Sub-solar photovoltaic power generation prediction method and system based on seasonal decomposition and convolution network |
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 |