CN112949950A - Cluster wind power mapping prediction method based on multivariate space-time correlation matrix - Google Patents
Cluster wind power mapping prediction method based on multivariate space-time correlation matrix Download PDFInfo
- Publication number
- CN112949950A CN112949950A CN202110473926.2A CN202110473926A CN112949950A CN 112949950 A CN112949950 A CN 112949950A CN 202110473926 A CN202110473926 A CN 202110473926A CN 112949950 A CN112949950 A CN 112949950A
- Authority
- CN
- China
- Prior art keywords
- wind speed
- correlation matrix
- time
- space
- time correlation
- 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
- 239000011159 matrix material Substances 0.000 title claims abstract description 149
- 238000013507 mapping Methods 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000013528 artificial neural network Methods 0.000 claims abstract description 34
- 230000015654 memory Effects 0.000 claims abstract description 32
- 238000012821 model calculation Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 230000002441 reversible effect Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000011161 development Methods 0.000 description 4
- 230000002123 temporal effect Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000007787 long-term memory Effects 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000005457 optimization 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention discloses a cluster wind power mapping prediction method based on a multivariate space-time correlation matrix, which comprises the following steps: acquiring wind speed data and cluster wind power data of a region to be predicted, wherein the wind speed data and the cluster wind power data comprise time parameters and space parameters; forming a wind speed multivariate space-time correlation matrix according to the space parameters in the wind speed data; establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network; performing model calculation on the wind speed multivariate space-time correlation matrix, and obtaining a predicted wind speed multivariate space-time correlation matrix by changing time parameters; establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on a convolution long and short time memory neural network; and inputting the predicted wind speed multivariate space-time correlation matrix into a mapping model to obtain the predicted cluster wind power. Compared with the prior art, the method and the device have the advantages that the time information and the space information of the wind speed are fully utilized, and the prediction precision is greatly improved.
Description
Technical Field
The invention relates to the technical field of cluster wind power prediction, in particular to a cluster wind power mapping prediction method based on a multivariate space-time correlation matrix.
Background
With the proposal of sustainable development strategy in recent years, wind energy has attracted wide attention as a high-quality and clean energy source, and the construction of large-scale wind power plants enables users to use clean and renewable electric energy, but also brings some problems to the electric power system. Due to the randomness and uncertainty of wind power output, certain influence is caused to the stability of a power system after grid connection. Because the output of a single fan is small, the influence on the power grid is small, and the research on the power grid is not meaningful. The output of a large-scale fan cluster is large, and the influence on the stability of the power system is also large, so that accurate wind power prediction is very important for the safe and stable operation of the power system. However, the cluster wind power is mainly influenced by the wind speed, so a high-precision wind speed prediction model is an important premise for obtaining a high-precision cluster wind power prediction result. After obtaining the wind speed result, it is also an important link to convert the wind speed result into wind power.
The wind speed prediction means that the predicted value of the wind speed condition in a future period of time in a certain area is obtained by calculation and derivation by using the known wind speed information. The application range of wind speed prediction is very wide, and the wind speed prediction method can be applied to the fields of meteorological detection, disaster early warning, wind energy utilization and the like. Especially in the field of electric power application, wind energy has received wide attention from countries around the world as a clean, pollution-free new energy source. However, due to the problems of mismatch between wind energy resource distribution and power load, insufficient power grid absorption capacity and the like, a plurality of phenomena of wind abandoning and electricity limiting appear. The aggravation of the wind abandoning phenomenon not only causes immeasurable economic loss, but also greatly weakens the market competitiveness of wind power. The reliable wind power prediction is beneficial to the power dispatching department to adjust the overall dispatching plan, configure the reasonable output of the wind generating set and save the conventional energy for power generation. Meanwhile, in the electric power market, the accuracy of wind power prediction is also a key factor for reducing the power generation cost and maintaining the competitiveness. Therefore, improvement of wind speed and power prediction methods of wind power plants becomes a research focus of wind power development, wherein wind speed prediction and mapping of wind speed-wind power are basic work and are also key links.
However, the existing wind speed prediction methods are models for directly fitting the wind speed sequence and the influencing factors, and the characteristics of the wind speed sequence are not deeply excavated. Particularly in the aspect of utilization of wind speed information, in the prior art, spatial information and temporal information are often split and used for analyzing the wind speed, and a model which does not fully combine the spatial information and the temporal information has obvious defects, so that the prediction accuracy is low. The existing mapping method of wind speed-wind power is to correspond the wind speed and the wind power through a wind power conversion curve of an ideal fan. However, the existing method does not consider the problem of actual delivery or installation of the wind turbine, so that the output of the existing wind turbine cannot meet an ideal conversion curve, or the condition of wind abandoning and electricity limiting occurs, and therefore the existing mapping method of wind speed and wind power is low in precision.
Disclosure of Invention
The invention mainly aims to provide a cluster wind power mapping prediction method based on a multivariate time-space correlation matrix, and aims to solve the problem of poor prediction effect caused by insufficient combination of time domain information and space domain information in the conventional cluster wind power prediction method.
In order to achieve the purpose, the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix provided by the invention comprises the following steps:
acquiring wind speed data and cluster wind power data of a region to be predicted, wherein the wind speed data and the cluster wind power data comprise time parameters and space parameters;
forming a wind speed multivariate space-time correlation matrix according to the space parameters in the wind speed data;
establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network;
performing model calculation on the wind speed multivariate space-time correlation matrix, and obtaining a predicted wind speed multivariate space-time correlation matrix by changing the time parameter;
establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on a convolution long and short time memory neural network;
and inputting the predicted wind speed multivariate space-time correlation matrix into the mapping model to obtain the cluster wind power.
Preferably, the step of acquiring wind speed data of the area to be predicted including the temporal parameter and the spatial parameter includes:
downloading meteorological data of different longitudes and latitudes at different moments in a specified time period in a database;
and extracting the wind speed data of the area to be predicted from the meteorological data according to the longitude and latitude of the area to be predicted.
Preferably, the step of forming a wind speed multivariate space-time correlation matrix according to the spatial parameters in the wind speed data comprises:
and forming the wind speed multi-element space-time correlation matrix at different moments in a specified time period by taking the longitude in the wind speed data as an abscissa and the latitude in the wind speed data as an ordinate.
Preferably, the step of establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network includes:
forming a wind speed multi-element space-time correlation matrix sequence by the wind speed multi-element space-time correlation matrix at different time by utilizing python software;
carrying out normalization processing on the wind speed multivariate space-time correlation matrix, and obtaining wind speed data between 0 and 1 by utilizing a MinMaxScaler () function;
dividing the wind speed multivariate space-time correlation matrix sequence into a training set, a verification set and a test set according to the ratio of 6:3: 1;
and establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network.
Preferably, the step of performing model calculation on the wind speed multivariate space-time correlation matrix and obtaining a predicted wind speed multivariate space-time correlation matrix by changing the time parameter comprises:
and inputting the previous 10 wind speed multivariate space-time correlation matrixes at the current moment to obtain the wind speed multivariate space-time correlation matrix at the next moment.
Preferably, the step of establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long-time and short-time memory neural network includes:
reading the wind speed multivariate space-time correlation matrix and the cluster wind power data;
setting first model parameters;
and building a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long-time memory neural network according to the first model parameter.
Preferably, the step of inputting the predicted wind speed multivariate space-time correlation matrix into the mapping model to obtain the cluster wind power includes:
loading the trained wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution duration memory neural network by using a load _ model () function in python software;
inputting the predicted wind speed multivariate space-time correlation matrix in the wind speed multivariate space-time correlation matrix prediction model based on the convolution long and short time memory neural network into the wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long and short time memory neural network to obtain cluster wind power at a corresponding moment;
and performing reverse normalization processing on the cluster wind power to obtain the final predicted cluster wind power.
Preferably, the time resolution of downloading weather data in the database at different times within a specified time period with different longitudes and latitudes is 15 minutes.
Preferably, before the step of establishing the wind speed multivariate space-time correlation matrix prediction model based on the convolution long-and-short-term memory neural network, the method further comprises the following steps: and displaying the wind speed multivariate space-time correlation matrix in a gray scale map form.
The wind speed multivariate spatiotemporal correlation matrix is preferably processed into the form of a grey scale map using the imread function cv2.imread () in the python software cv2 module.
According to the technical scheme, time parameters and space parameters in wind speed data are obtained simultaneously, a wind speed multivariate space-time correlation matrix is obtained through the space parameters, and a wind speed image is indirectly obtained through the wind speed multivariate space-time correlation matrix in a visual mode; a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time memory neural network is established, and a predicted wind speed multivariate space-time correlation matrix is obtained by changing time parameters; establishing a mapping model of the wind speed multivariate space-time incidence matrix and the cluster wind power according to the real-time mapping relation of the wind speed multivariate space-time incidence matrix and the cluster wind power; and inputting the predicted wind speed multivariate space-time correlation matrix into a mapping model so as to obtain the predicted cluster wind power. Compared with the prior art that only the time information parameter or the space parameter is utilized independently or only the time information parameter and the space parameter are simply combined, the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix fully utilizes the time information and the space information of the wind speed, directly establishes the mapping relation between the wind speed and the wind power, and greatly improves the accuracy of cluster wind power prediction.
Drawings
FIG. 1 is a schematic flow chart of a first embodiment of a cluster wind power mapping prediction method based on a multivariate space-time correlation matrix according to the present invention;
FIG. 2 is a schematic view of a comparison of a predicted wind speed image and an actually measured wind speed image;
FIG. 3 is a schematic diagram illustrating a comparison between a wind speed prediction result of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix and a wind speed prediction effect in the prior art;
FIG. 4 is a schematic diagram showing a comparison between a cluster wind power prediction result of the cluster wind power mapping prediction method based on the multivariate time-space correlation matrix and a prediction effect in the prior art.
The objects, features and advantages of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, the invention provides a cluster wind power mapping prediction method based on a multivariate spatiotemporal correlation matrix, which includes the following steps:
step S10, acquiring wind speed data and cluster wind power data of a region to be predicted, wherein the wind speed data and the cluster wind power data comprise time parameters and space parameters;
step S20, forming a wind speed multivariate space-time correlation matrix according to the space parameters in the wind speed data;
step S30, establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network;
step S40, performing model calculation on the wind speed multivariate space-time correlation matrix, and obtaining a predicted wind speed multivariate space-time correlation matrix by changing the time parameter;
step S50, establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on a convolution long-time and short-time memory neural network;
and step S60, inputting the predicted wind speed multivariate space-time correlation matrix into the mapping model to obtain the cluster wind power.
According to the technical scheme, time parameters and space parameters in wind speed data are obtained simultaneously, a wind speed multivariate space-time correlation matrix is obtained through the space parameters, and a wind speed image is indirectly obtained through the wind speed multivariate space-time correlation matrix in a visual mode; a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time memory neural network is established, and a predicted wind speed multivariate space-time correlation matrix is obtained by changing time parameters; establishing a mapping model of the wind speed multivariate space-time incidence matrix and the cluster wind power according to the real-time mapping relation of the wind speed multivariate space-time incidence matrix and the cluster wind power; and inputting the predicted wind speed multivariate space-time correlation matrix into a mapping model so as to obtain the predicted cluster wind power. Compared with the prior art that only the time information parameter or the space parameter is utilized independently or only the time information parameter and the space parameter are simply combined, the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix fully utilizes the time information and the space information of the wind speed, directly establishes the mapping relation between the wind speed and the wind power, and greatly improves the accuracy of cluster wind power prediction.
The time parameter refers to a time period corresponding to the wind speed data, and the spatial parameter refers to a geographical area corresponding to the wind speed data. The step of acquiring the wind speed data of the area to be predicted, which includes the time parameter and the space parameter, may specifically be: and acquiring wind speed data of the area to be predicted in a specified time period. In the same time period, the wind speeds in different regions have great differences, for example, at the same time, the wind speeds in coastal regions and inland regions have great differences, so when the wind speed of a certain region to be predicted is predicted, the historical wind speed data of the same region needs to be acquired as a prediction basis, but the historical wind speed data of other regions except the region to be predicted cannot be adopted as the prediction basis, so as to ensure the accuracy of the wind speed prediction.
Please refer to table 1, which shows the comparison between the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix and the prior art (LSTM prediction method and BPNN prediction method) for predicting wind speed data of a certain area for 4 continuous hours, and it can be clearly seen from the table information that the wind speed prediction result of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention is closer to the actually measured wind speed.
TABLE 1 prediction results comparison Table
Please refer to fig. 2, which is a schematic diagram illustrating a comparison between a predicted wind speed image and an actually measured wind speed image, and it can be seen that the prediction result of the cluster wind power mapping prediction method based on the multivariate spatiotemporal correlation matrix is very similar to the actually measured result.
Please refer to fig. 3, which is a schematic diagram illustrating a comparison between a wind speed prediction result of the cluster wind power mapping prediction method based on the multivariate spatiotemporal correlation matrix and a wind speed prediction effect in the prior art, and it can be seen from the diagram that the prediction effect of the present invention is closer to a real situation compared with the prediction effect in the prior art.
Based on the first embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, and the second embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, the step of obtaining the wind speed data of the region to be predicted including the time parameter and the space parameter in the step S10 includes:
step S11, downloading meteorological data of different latitudes and longitudes at different moments in a specified time period in a database;
and step S12, extracting the wind speed data of the area to be predicted from the meteorological data according to the longitude and latitude of the area to be predicted.
The meteorological data of different longitudes and latitudes at different moments in the specified time period comprise time parameters and space parameters, and simultaneously comprise the corresponding relation between the time participation and the space parameters.
The step S10 further includes:
and step S13, constructing a wind speed database of the area to be predicted.
The database for downloading the meteorological data of different longitudes and latitudes at different moments in a specified time interval is a Cowbeniy climate data repository. And determining the specific plane position of the wind speed area to be predicted according to the longitude and latitude information, and extracting the wind speed data in the area independently. The wind speed data comprises instantaneous wind speed data and wind direction data, and the development trend of wind in the area to be predicted can be known along a time axis by combining the instantaneous wind speed and the wind direction. And storing the extracted wind speed data into a wind speed database of the area to be predicted for subsequent extraction.
In this embodiment, the spatial parameter includes a longitude and a latitude, and as a further extension of the present technical solution, the spatial parameter may further include an altitude. After the altitude is added as a space parameter, the wind speed under different altitudes in the area range to be predicted can be accurately predicted by the aid of the method, so that wind speed parameters are provided for ground operation, unmanned aerial vehicle operation or aviation and the like.
Based on the first embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, and the third embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, the step S20 includes:
and step S21, forming the wind speed multi-element space-time correlation matrix at different moments in a specified time period by taking the longitude in the wind speed data as an abscissa and the latitude in the wind speed data as an ordinate.
The longitude and latitude in the wind speed data can uniquely determine the plane position, and the wind speed multivariate space-time correlation matrix at the wind speed can be formed by combining the wind speed information of the corresponding position at a certain moment. By changing the time parameters, a plane wind speed multi-element space-time correlation matrix at different moments in a specified time period can be formed.
The step S20 further includes:
and step S22, arranging the wind speed multivariate space-time correlation matrix at the corresponding moment into a wind speed multivariate space-time correlation matrix set according to the sequence of time.
When the wind speed multi-element space-time correlation matrix at a certain moment needs to be extracted, corresponding time information only needs to be input from the wind speed multi-element space-time correlation matrix in a centralized mode.
In this embodiment, the wind speed data includes instantaneous wind speed information and wind direction information, the instantaneous wind speed information is independently selected and combined with the longitude and latitude, an instantaneous wind speed multivariate time-space correlation matrix of each coordinate point at different times can be obtained, the instantaneous wind speeds of adjacent coordinate points are connected, the instantaneous wind speed information is updated along a time axis, and a wind speed change trend of the whole area at the corresponding time can be obtained. And the wind direction information is independently selected to be combined with the longitude and latitude, so that a wind direction matrix of each coordinate point at different moments can be obtained, the directions pointed by the wind directions of the adjacent coordinate points are connected, the wind direction information is updated along the time axis, and the wind direction change trend of the whole area at the corresponding moment, namely the trend of wind, can be obtained. As a further extension of the technical scheme, longitude is used as an x-axis coordinate, latitude is used as a y-axis coordinate, and altitude is used as a z-axis coordinate, so that a three-dimensional wind speed multi-element space-time correlation matrix is constructed and formed, and the development trend of wind can be reflected in a three-dimensional space.
Based on the first embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, and the fourth embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, the step S30 includes:
step S31: forming a wind speed multi-element space-time correlation matrix sequence by the wind speed multi-element space-time correlation matrix at different time by utilizing python software;
step S32: carrying out normalization processing on the wind speed multivariate space-time correlation matrix, and obtaining wind speed data between 0 and 1 by utilizing a MinMaxScaler () function;
step S33: dividing the wind speed multivariate space-time correlation matrix sequence into a training set, a verification set and a test set according to the ratio of 6:3: 1;
step S34: and establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network.
Wherein, the step S32 is as shown in formula 1;
in the formula, s (t) and s' (t) represent the wind speed before and after normalization, respectively.
In addition, the wind speed multi-element space-time correlation matrix prediction model based on the convolution long-term memory neural network has five layers of networks, wherein the first four layers are ConvLSTM layers and are used for capturing time and space sequence information of the wind speed multi-element space-time correlation matrix; and the fifth layer is an output layer, and Conv2D is used as an output for acquiring a wind speed multi-element space-time correlation matrix at the next moment.
After step S34, the method further includes:
step S35: setting second model parameters including the size of a convolution kernel, the number of filters and the like;
step S36: taking 10 continuous wind speed multivariate space-time correlation matrixes as input, and taking 1 wind speed multivariate space-time correlation matrix at the next moment as a label to output, and training a model;
step S37: in the training process of the model, MSE is used for calculating the loss of the result, namely the precision of the model is calculated;
step S38: when the loss of the model is less than a certain threshold value, the model is saved by using a torch.save () function in python.
Based on the first embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, and the fifth embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, the step S40 includes:
and step S41, obtaining the wind speed multi-element space-time correlation matrix at the next moment by inputting the previous 10 wind speed multi-element space-time correlation matrices at the current moment.
The step S41 includes:
step S41a, loading the trained model by using a load _ model () function in python software;
step S41b, inputting the former 10 wind speed multivariate space-time correlation matrixes at the current moment into the model to obtain the wind speed multivariate space-time correlation matrix at the next moment, namely the predicted wind speed multivariate space-time correlation matrix.
After step S41, the method further includes:
step S42, performing inverse normalization processing on the predicted wind speed multivariate space-time correlation matrix to obtain a final predicted wind speed multivariate space-time correlation matrix;
and step S43, visually displaying the multivariate spatiotemporal correlation matrix of the predicted wind speed in the form of a gray scale map.
Based on the first embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, and the sixth embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, the step S50 includes:
step S51, reading the wind speed multivariate space-time correlation matrix and the cluster wind power data;
step S52, setting first model parameters;
and step S53, constructing the wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution length-time memory neural network according to the first model parameters.
The step S53 includes:
step S53a, carrying out normalization processing on the cluster wind power data, and obtaining wind power data between 0 and 1 by utilizing a MinMaxScalter () function.
Step S53 b: and dividing the wind speed multivariate space-time correlation matrix and the cluster wind power data into a training set, a verification set and a test set according to the ratio of 6:3: 1.
Step S53 c: and establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution length-time memory neural network.
The wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long-term memory neural network has five layers of networks, wherein the first two layers are ConvLSTM layers and are used for capturing time and space sequence information of the wind speed multivariate space-time correlation matrix; the third layer is an LSTM layer, and the fourth and fifth layers are full connection layers and are used for outputting corresponding wind power data.
After step S53c, the method further includes:
step S53 d: setting second model parameters including the size of a convolution kernel, the number of filters, the number of neurons and the like;
step S53 e: taking 1 wind speed multivariate space-time correlation matrix as input, and outputting cluster wind power data at corresponding moments as labels to train a model;
step S53 f: in the training process of the model, MSE is used for calculating the loss of the result, namely the precision of the model is calculated;
step S53 g: when the loss of the model is less than a certain threshold value, the model is saved by using a torch.save () function in python.
The wind speed sequence and the cluster wind power are predicted through algorithms such as deep learning, and a prediction result with high precision is obtained through training of historical wind speed and wind power data samples and optimization of parameters. Because the wind speeds of different longitude and latitude coordinates have relativity, the wind speed multi-element space-time correlation matrix sequence under different coordinates is deeply learned, the information of time parameters can be taken into account, the relativity of space parameter information can be learned, and the correlation support each other, so that the prediction precision is greatly improved.
Based on the first embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, in the seventh embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, the step S60 includes:
step S61: loading the trained wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution duration memory neural network by using a load _ model () function in python software;
step S62: inputting the predicted wind speed multivariate space-time correlation matrix in the wind speed multivariate space-time correlation matrix prediction model based on the convolution long and short time memory neural network into the wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long and short time memory neural network to obtain cluster wind power at a corresponding moment;
step S63: and performing inverse normalization processing on the cluster wind power to obtain the final predicted cluster wind power, as shown in formula 2.
s(t)＝s'(t)·max(s) (2)
Please refer to fig. 4, which is a schematic diagram comparing a wind power prediction result of the cluster wind power mapping prediction method based on the multivariate spatiotemporal correlation matrix of the present invention with a prediction effect of the prior art, and it can be seen from the diagram that, compared with the prediction effect of the prior art, the prediction effect of the present invention is closer to the real cluster wind power.
Please refer to table 2, which shows the comparison between the cluster wind power mapping prediction method based on the multivariate spatio-temporal correlation matrix and the prior art (LSTM prediction method, BPNN prediction method, ConvLSTM-LSTM method and ConvLSTM-BPNN method) for predicting wind speed data of a certain area for 4 continuous hours at the same time, and it can be clearly seen from the table information that the wind power prediction result of the cluster wind power mapping prediction method based on the multivariate spatio-temporal correlation matrix of the present invention is closer to the actually measured wind power data.
TABLE 2 prediction results comparison Table
Based on the second embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix, the time resolution of the meteorological data downloaded into the database at different moments with different longitudes and latitudes in a specified time period is 15 minutes in the eighth embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix.
The time resolution refers to the time interval between two adjacent detections when the same target is repeatedly detected. The system can provide information of dynamic changes of the ground features, can be used for monitoring the changes of the ground features, and can also provide additional information for accurate classification of some special elements.
The meteorological data at different time points can be obtained by changing the time resolution, so that the prediction accuracy is influenced. According to the accuracy requirement of the wind speed predicted value in different applications, the time resolution can be changed, so that the prediction accuracy meeting the corresponding requirement is obtained. The temporal resolution also includes 0.2 hours, 0.5 hours, 1 hour, 2 hours, 5 hours, and 10 hours.
Based on any one of the first to eighth embodiments of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, in a ninth embodiment of the cluster wind power mapping prediction method based on the multivariate space-time correlation matrix of the present invention, before step S30, the method further includes:
and step S20a, displaying the wind speed multi-element space-time correlation matrix in the form of a gray scale map.
Preferably, in the step S20a, the wind speed multivariate spatiotemporal correlation matrix is processed into a form of a gray scale map by using an imread function cv2.imread () in a python software cv2 module.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. The cluster wind power mapping prediction method based on the multivariate space-time correlation matrix is characterized by comprising the following steps of:
acquiring wind speed data and cluster wind power data of a region to be predicted, wherein the wind speed data and the cluster wind power data comprise time parameters and space parameters;
forming a wind speed multivariate space-time correlation matrix according to the space parameters in the wind speed data;
establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network;
performing model calculation on the wind speed multivariate space-time correlation matrix, and obtaining a predicted wind speed multivariate space-time correlation matrix by changing the time parameter;
establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on a convolution long and short time memory neural network;
and inputting the predicted wind speed multivariate space-time correlation matrix into the mapping model to obtain the cluster wind power.
2. The cluster wind power mapping prediction method based on the multivariate space-time correlation matrix as claimed in claim 1, wherein the step of obtaining wind speed data of the region to be predicted comprising the time parameter and the space parameter comprises:
downloading meteorological data of different longitudes and latitudes at different moments in a specified time period in a database;
and extracting the wind speed data of the area to be predicted from the meteorological data according to the longitude and latitude of the area to be predicted.
3. The multi-element spatiotemporal correlation matrix-based cluster wind power mapping prediction method according to claim 1, wherein the step of forming a wind speed multi-element spatiotemporal correlation matrix according to the spatial parameters in the wind speed data comprises:
and forming the wind speed multi-element space-time correlation matrix at different moments in a specified time period by taking the longitude in the wind speed data as an abscissa and the latitude in the wind speed data as an ordinate.
4. The cluster wind power mapping prediction method based on the multivariate spatiotemporal correlation matrix as claimed in claim 1, wherein the step of establishing a wind speed multivariate spatiotemporal correlation matrix prediction model based on a convolution long-and-short-term memory neural network comprises:
forming a wind speed multi-element space-time correlation matrix sequence by the wind speed multi-element space-time correlation matrix at different time by utilizing python software;
carrying out normalization processing on the wind speed multivariate space-time correlation matrix, and obtaining wind speed data between 0 and 1 by utilizing a MinMaxScaler () function; dividing the wind speed multivariate space-time correlation matrix sequence into a training set, a verification set and a test set according to the ratio of 6:3: 1;
and establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network.
5. The cluster wind power mapping prediction method based on the multivariate spatiotemporal correlation matrix as claimed in claim 1, wherein the step of performing model calculation on the wind speed multivariate spatiotemporal correlation matrix and obtaining the predicted wind speed multivariate spatiotemporal correlation matrix by changing the time parameters comprises:
and inputting the previous 10 wind speed multivariate space-time correlation matrixes at the current moment to obtain the wind speed multivariate space-time correlation matrix at the next moment.
6. The method for cluster wind power mapping prediction based on multivariate space-time correlation matrix as claimed in claim 1, wherein the step of establishing a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on a convolution long and short time memory neural network comprises:
reading the wind speed multivariate space-time correlation matrix and the cluster wind power data;
setting first model parameters;
and building a wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long-time memory neural network according to the first model parameter.
7. The method for cluster wind power mapping prediction based on multivariate space-time correlation matrix as claimed in claim 1, wherein the step of inputting the predicted wind speed multivariate space-time correlation matrix into the mapping model to obtain cluster wind power comprises:
loading the trained wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution duration memory neural network by using a load _ model () function in python software;
inputting the predicted wind speed multivariate space-time correlation matrix in the wind speed multivariate space-time correlation matrix prediction model based on the convolution long and short time memory neural network into the wind speed multivariate space-time correlation matrix-cluster wind power mapping model based on the convolution long and short time memory neural network to obtain cluster wind power at a corresponding moment;
and performing reverse normalization processing on the cluster wind power to obtain the final predicted cluster wind power.
8. The method for cluster wind power mapping prediction based on multivariate space-time correlation matrix as claimed in claim 2, wherein the time resolution of downloading meteorological data with different longitudes and latitudes at different times within a specified time period in the database is 15 minutes.
9. The clustered wind power mapping prediction method based on multivariate space-time correlation matrix as claimed in any one of claims 1-8, wherein before the step of establishing a wind speed multivariate space-time correlation matrix prediction model based on a convolution long-time and short-time memory neural network, the method further comprises:
and displaying the wind speed multivariate space-time correlation matrix in a gray scale map form.
10. The multi-element spatiotemporal correlation matrix-based cluster wind power mapping prediction method according to claim 9, characterized in that the wind speed multi-element spatiotemporal correlation matrix is processed into a form of a gray scale by using an imread function cv2.imread () in python software cv2 module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110473926.2A CN112949950A (en) | 2021-04-29 | 2021-04-29 | Cluster wind power mapping prediction method based on multivariate space-time correlation matrix |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110473926.2A CN112949950A (en) | 2021-04-29 | 2021-04-29 | Cluster wind power mapping prediction method based on multivariate space-time correlation matrix |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112949950A true CN112949950A (en) | 2021-06-11 |
Family
ID=76233684
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110473926.2A Pending CN112949950A (en) | 2021-04-29 | 2021-04-29 | Cluster wind power mapping prediction method based on multivariate space-time correlation matrix |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112949950A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103390116A (en) * | 2013-08-07 | 2013-11-13 | 华北电力大学（保定） | Method for predicting electricity generation power of photovoltaic power station in step-by-step way |
| CN106446494A (en) * | 2016-05-11 | 2017-02-22 | 新疆大学 | Wavelet packet-neural network-based wind/photovoltaic power prediction method |
| CN110782071A (en) * | 2019-09-25 | 2020-02-11 | 天津大学 | Method for predicting wind power by convolutional neural network based on time-space characteristic fusion |
| CN111680838A (en) * | 2020-06-08 | 2020-09-18 | 中国电力科学研究院有限公司 | Air conditioner load aggregated power prediction method and system |
| CN111784041A (en) * | 2020-06-28 | 2020-10-16 | 中国电力科学研究院有限公司 | Wind power prediction method and system based on graph convolution neural network |
-
2021
- 2021-04-29 CN CN202110473926.2A patent/CN112949950A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103390116A (en) * | 2013-08-07 | 2013-11-13 | 华北电力大学（保定） | Method for predicting electricity generation power of photovoltaic power station in step-by-step way |
| CN106446494A (en) * | 2016-05-11 | 2017-02-22 | 新疆大学 | Wavelet packet-neural network-based wind/photovoltaic power prediction method |
| CN110782071A (en) * | 2019-09-25 | 2020-02-11 | 天津大学 | Method for predicting wind power by convolutional neural network based on time-space characteristic fusion |
| CN111680838A (en) * | 2020-06-08 | 2020-09-18 | 中国电力科学研究院有限公司 | Air conditioner load aggregated power prediction method and system |
| CN111784041A (en) * | 2020-06-28 | 2020-10-16 | 中国电力科学研究院有限公司 | Wind power prediction method and system based on graph convolution neural network |
Non-Patent Citations (4)
| Title |
|---|
| CHEN, SZ: "A Wind Power Prediction Method Based on Deep Convolutional Network with Multiple Features", 《INTERNATIONAL CONFERENCE ON NEURAL INFORMATION PROCESSING》 * |
| JIAMING LIU: "Deep Learning Based Visualized Wind Speed Matrix Forecasting Model for Wind Power Forecasting", 《2020 IEEE 3RD STUDENT CONFERENCE ON ELECTRICAL MACHINES AND SYSTEMS (SCEMS)》 * |
| WENZU WU: "Probabilistic Short-term Wind Power Forecasting Based on Deep Neural Networks", 《2016 INTERNATIONAL CONFERENCE ON PROBABILISTIC METHODS APPLIED TO POWER SYSTEMS (PMAPS)》 * |
| 尹瑞: "基于灵活性矩阵与优劣解距离的新能源功率预测评价方法", 《电力建设》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107247938B (en) | high-resolution remote sensing image urban building function classification method | |
| Yang et al. | Multitime-scale data-driven spatio-temporal forecast of photovoltaic generation | |
| EP2856339B1 (en) | Method and apparatus for forecasting solar radiation and solar power production using synthetic irradiance imaging | |
| CN110705727B (en) | Photovoltaic power station shadow shielding diagnosis method and system based on random forest algorithm | |
| CN112329977A (en) | Wind power prediction system for extreme scene | |
| Haupt et al. | Big data and machine learning for applied weather forecasts: Forecasting solar power for utility operations | |
| CN106228278A (en) | Photovoltaic power prognoses system | |
| CN102254239A (en) | Power grid wind damage early warning system based on micro-landform wind field distribution and typhoon superimposed effect | |
| CN110346518B (en) | Traffic emission pollution visualization early warning method and system thereof | |
| CN112906955B (en) | Tidal data fitting method based on multi-source data driving | |
| CN113505521B (en) | Urban waterlogging rapid forecasting method based on neural network-numerical simulation | |
| Alimohammadi et al. | Multi-stage algorithm for uncertainty analysis of solar power forecasting | |
| Cheng | Hybrid solar irradiance now-casting by fusing Kalman filter and regressor | |
| CN112183625A (en) | PM based on deep learning2.5High-precision time-space prediction method | |
| CN111461423B (en) | High-precision gridding air quality inference method, system, terminal equipment and storage medium | |
| CN113984969A (en) | Air quality prediction method and system based on multi-source space-time data fusion | |
| CN115358060A (en) | Efficient algorithm framework supporting high-precision prediction of new energy power generation power | |
| Weekes et al. | Low‐cost wind resource assessment for small‐scale turbine installations using site pre‐screening and short‐term wind measurements | |
| CN110059972B (en) | Daily solar radiation resource assessment method based on functional deep belief network | |
| CN115936238B (en) | Method, system, equipment and medium for predicting medium-long-term output of global wind power | |
| CN116720156A (en) | Weather element forecasting method based on graph neural network multi-mode weather data fusion | |
| CN112949950A (en) | Cluster wind power mapping prediction method based on multivariate space-time correlation matrix | |
| Wang et al. | Neural network based irradiance mapping model of solar PV power forecasting using sky image | |
| Zhou et al. | Smart grid data mining and visualization | |
| Sivakumar et al. | Machine learning approach on time series for PV-solar energy |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210611 |
|
| RJ01 | Rejection of invention patent application after publication |