CN115345387B - Wind field wind speed prediction method and device and storage medium - Google Patents
Wind field wind speed prediction method and device and storage medium Download PDFInfo
- Publication number
- CN115345387B CN115345387B CN202211272558.6A CN202211272558A CN115345387B CN 115345387 B CN115345387 B CN 115345387B CN 202211272558 A CN202211272558 A CN 202211272558A CN 115345387 B CN115345387 B CN 115345387B
- Authority
- CN
- China
- Prior art keywords
- data
- wind speed
- meteorological element
- wind
- predicted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000013528 artificial neural network Methods 0.000 claims abstract description 53
- 238000000926 separation method Methods 0.000 claims abstract description 30
- 230000006870 function Effects 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 11
- 125000004122 cyclic group Chemical group 0.000 claims description 10
- 238000010438 heat treatment Methods 0.000 claims description 4
- 238000013075 data extraction Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000000053 physical method Methods 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000001413 cellular effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis 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
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
- G06F17/13—Differential equations
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
- G06N3/08—Learning methods
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Algebra (AREA)
- Tourism & Hospitality (AREA)
- Databases & Information Systems (AREA)
- Marketing (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Water Supply & Treatment (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method, a device and a storage medium for predicting wind speed of a wind field, wherein the method comprises the following steps: acquiring historical meteorological element data of a wind field, and separating data meeting physical constraints from the historical meteorological element data; inputting data meeting physical constraints into a partial differential neural network module, extracting multi-order partial derivatives of each meteorological element variable, and predicting the wind speed of a wind field at the next time step to obtain a first predicted wind speed; inputting the multi-order partial derivatives of all meteorological element variables into a power source separation module, and calculating heated data and stressed data of atmospheric air mass in a wind field by combining an atmospheric motion equation set; inputting the heated data, the stress data and the historical meteorological element data into a data driving module, and predicting the wind speed of the wind field at the next time step to obtain a second predicted wind speed; a final predicted wind speed for the wind farm is determined based on the first predicted wind speed and the second predicted wind speed. The technical scheme of the invention can improve the prediction accuracy of the wind field wind speed.
Description
Technical Field
The invention relates to the technical field of meteorological prediction, in particular to a wind field wind speed prediction method, a wind field wind speed prediction device and a storage medium.
Background
Wind power is an important influence factor of power balance of a power grid, and prediction of wind power of a wind field is an effective means for relieving pressure of a power system, improving electric energy quality and improving wind power utilization efficiency. The method comprises the following steps of calculating the wind speed of a wind field, wherein the wind speed is calculated according to the wind speed, and the wind power is calculated according to the wind speed.
The wind speed prediction method mainly comprises a physical method, a traditional statistical method and an artificial intelligence method. The physical method is to forecast the wind speed in a future period of time by observing meteorological element values such as wind speed, wind direction, air pressure, air temperature and the like and combining a meteorological theory and a physical medium mechanics method. The traditional statistical method adopts time series and regression analysis numerical models to mine the trend of the wind speed time series so as to predict the wind speed in a future period of time. Common artificial intelligence prediction models comprise a support vector machine, a deep neural network and the like, and capture rules implicit in meteorological data through training and learning of a large amount of meteorological data so as to predict the wind speed in a future period of time.
However, the physical method requires accurate mathematical description of the physical characteristics of the atmosphere, and the established mathematical equation is difficult to be solved accurately, so that the prediction accuracy is low and the calculation amount is extremely large. The statistical method does not need to establish a specific mathematical model, but learns the rules from the data, has high calculation speed, needs a large amount of historical data and has lower prediction accuracy. The statistical model can effectively solve the problem of prediction delay, but the accuracy of long-term prediction is low.
Disclosure of Invention
The invention solves the problem of how to improve the prediction accuracy of the wind speed of the wind field.
In order to solve the above problems, the present invention provides a method, an apparatus and a storage medium for predicting wind speed of a wind field.
In a first aspect, the present invention provides a wind field wind speed prediction method, based on a wind field wind speed prediction model, where the wind field wind speed prediction model includes a partial differential neural network module, a power source separation module, and a data driving module, and the wind field wind speed prediction method includes:
acquiring historical meteorological element data of a wind field, and separating data meeting physical constraints from the historical meteorological element data, wherein the historical meteorological element data comprise historical data of a plurality of meteorological element variables of multiple isobaric surfaces;
inputting the data meeting the physical constraint into the partial differential neural network module, extracting multi-order partial derivatives of each meteorological element variable, and predicting the wind speed of the wind field at the next time step to obtain a first predicted wind speed;
inputting the multi-order partial derivatives of the meteorological element variables into the power source separation module, and calculating the heated data and the stressed data of the atmospheric air mass in the wind field by combining an atmospheric motion equation set;
inputting the heated data, the stress data and the historical meteorological element data into the data driving module, and predicting the wind speed of the wind field at the next time step to obtain a second predicted wind speed;
determining a final predicted wind speed for the wind farm based on the first predicted wind speed and the second predicted wind speed.
Optionally, the partial differential neural network module includes a plurality of neural networks, each neural network includes a plurality of convolution kernels respectively performing independent convolution, the number of the neural networks corresponds to the number of the meteorological element variables, and each convolution kernel is used for simulating a partial differential operator in a taylor expansion of a meteorological element variable function.
Optionally, the inputting the data satisfying the physical constraint into a partial differential neural network module, and the extracting the multiple-order partial derivatives of the meteorological element variables includes:
under the constraint of a convolution kernel, extracting multiple-order partial derivatives of each meteorological element variable from the data meeting the physical constraint by adopting the convolution kernel, and combining all the multiple-order partial derivatives of the meteorological element variables by using a Taylor expansion formula to obtain the first predicted wind speed.
Optionally, the data satisfying physical constraints is the historical meteorological element data conforming to an explicit equation of the system of atmospheric motion equations.
Optionally, the power source separation module includes a calculation sub-module for simulating a prediction equation of an atmospheric motion equation set, the inputting the multiple-order partial derivatives of the variables of the meteorological elements into the power source separation module, and the calculating the heated data and the stressed data of the atmospheric air mass in the wind field by combining the atmospheric motion equation set includes:
and inputting the multi-order partial derivatives of the meteorological element variables into the calculation submodule, and calculating the heated data and the stressed data of the atmospheric air mass in the wind field through a prediction equation of an atmospheric motion equation set.
Optionally, the data driving module includes a multilayer convolution cyclic neural network, and the inputting the heated data, the stressed data, and the historical meteorological element data into the data driving module to predict the wind speed of the wind field at the next time step, and obtaining a second predicted wind speed includes:
and splicing the heated data, the stress data and the historical meteorological element data in a channel dimension, projecting the spliced data to a hidden space through a multilayer convolution cyclic neural network, and predicting the wind speed of the wind field at the next time step by combining the hidden information of the data driving module at the previous time step to obtain a second predicted wind speed.
Optionally, the determining a final predicted wind speed for the wind farm from the first predicted wind speed and the second predicted wind speed comprises:
adding the first predicted wind speed to the second predicted wind speed to determine the final predicted wind speed for the wind farm.
In a second aspect, the present invention provides a wind farm wind speed prediction apparatus for implementing the wind farm wind speed prediction method according to any one of the first aspect, the apparatus comprising:
the device comprises an acquisition unit, a processing unit and a control unit, wherein the acquisition unit is used for acquiring historical meteorological element data of a wind field and separating data meeting physical constraints from the historical meteorological element data, and the historical meteorological element data comprises historical data of a plurality of meteorological element variables of a plurality of isobaric surfaces;
the physical part prediction unit is used for inputting the data meeting the physical constraint into a partial differential neural network module, extracting a multi-order partial derivative of each meteorological element variable, and predicting the wind speed of the wind field at the next time step to obtain a first predicted wind speed;
the data extraction unit is used for inputting the multi-order partial derivatives of the meteorological element variables into the power source separation module and calculating the heated data and the stressed data of the atmospheric air mass in the wind field by combining an atmospheric motion equation set;
the supplementary prediction unit is used for inputting the heated data, the stressed data and the historical meteorological element data into a data driving module, predicting the wind speed of the wind field at the next time step and obtaining a second predicted wind speed;
and the comprehensive prediction unit is used for determining the final predicted wind speed of the wind field according to the first predicted wind speed and the second predicted wind speed.
In a third aspect, the present invention provides an electronic device comprising a memory and a processor;
the memory for storing a computer program;
the processor is configured to, when executing the computer program, implement the wind farm wind speed prediction method according to any one of the first aspect.
In a fourth aspect, the invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a wind farm wind speed prediction method according to any one of the first aspect.
The wind field wind speed prediction method, the device and the storage medium have the advantages that: the method comprises the steps of obtaining historical meteorological element data of a wind field, separating data meeting physical constraints from the historical meteorological element data, wherein the data meeting the physical constraints are historical meteorological element data of an explicit equation of an atmospheric motion equation set, and the historical meteorological element data comprise historical data of a plurality of meteorological element variables of a plurality of isobaric surfaces, so that wind speed prediction can be conveniently carried out from the historical data of a plurality of dimensions, and accuracy of wind speed prediction is improved. And inputting the data meeting the physical constraint into a partial differential neural network module, and predicting the wind speed of the next time step in the physical constraint to obtain a first predicted wind speed. And extracting the multi-order partial derivative of each meteorological element variable, transmitting the multi-order partial derivative to the power source separation module, calculating the heated data and the stressed data of the atmospheric air mass of the wind field by the power source separation module through the atmospheric motion equation set, and introducing the physical priori knowledge of the atmospheric motion equation set, so that the prediction accuracy of the dynamic wind field wind speed can be improved. And the data driving module captures atmospheric motion dynamic information outside physical constraints according to the heated data, the stress data and the historical meteorological element data so as to predict the wind speed of the wind field in the next time step to obtain a second predicted wind speed. The first predicted wind speed obtained in the physical constraint is combined with the second predicted wind speed obtained outside the physical constraint, and the final predicted wind speed obtained by prediction from each dimension of the historical meteorological element data effectively improves the accuracy of the predicted wind speed.
Drawings
FIG. 1 is a schematic flow chart of a method for predicting wind speed in a wind farm according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a wind field wind speed prediction model according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a wind field wind speed prediction apparatus according to an embodiment of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. While certain embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and the embodiments of the present invention are illustrative only and are not intended to limit the scope of the present invention.
It should be understood that the various steps recited in the method embodiments of the present invention may be performed in a different order and/or performed in parallel. Moreover, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the invention is not limited in this respect.
The term "include" and variations thereof as used herein are open-ended, i.e., "including but not limited to". The term "based on" is "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "alternative embodiment". Relevant definitions for other terms will be given in the following description. It should be noted that the terms "first", "second", and the like in the present invention are only used for distinguishing different devices, modules or units, and are not used for limiting the order or interdependence relationship of the functions performed by the devices, modules or units.
It is noted that references to "a", "an", and "the" modifications in the present invention are intended to be illustrative rather than limiting, and that those skilled in the art will recognize that reference to "one or more" unless the context clearly dictates otherwise.
The names of messages or information exchanged between devices in the embodiments of the present invention are for illustrative purposes only, and are not intended to limit the scope of the messages or information.
As shown in fig. 1, a wind field wind speed prediction method provided in an embodiment of the present invention is based on a wind field wind speed prediction model, and the wind field wind speed prediction model includes a partial differential neural network module, a power source separation module, and a data driving module.
Specifically, as shown in fig. 2, an output end of the partial differential neural network module is connected to an input end of a power source separation module, a first output end of the power source separation module is connected to an input end of the data driving module, and a second output end of the power source separation module is connected to an output end of the data driving module.
The wind field wind speed prediction method comprises the following steps:
step S100, acquiring historical meteorological element data of a wind field, and separating data meeting physical constraints from the historical meteorological element data, wherein the historical meteorological element data comprises historical data of a plurality of meteorological element variables of a plurality of isobaric surfaces.
Specifically, historical meteorological element data of a wind field in the past period are collected, the historical meteorological element data are collected under an isopressure surface coordinate system, and the vertical direction of the three-dimensional space length of the isopressure surface coordinate system is divided by a height layer of an isopressure surface instead of an actual geographical height layer. The historical data of the next meteorological element variable on the single isobaric surface in the wind field can be represented in a grid form, the grid data of m x n can be regarded as a pixel image with the size of m x n, and the value of each pixel point in the pixel image is the numerical value of the meteorological element variable at the corresponding position. If the total number of meteorological element variables of different isobaric surfaces issThen the historical timetThe meteorological element data (namely, observation data) of (2) can be adopted
It is shown that, among others,
representsAnd (3) a set of m x n grid data of the meteorological element variables on a single isobaric surface. Thus, the prediction problem of wind field wind speed can be expressed as: weather element data at historical I moments
As input, determining the futurePWind field wind speed prediction value of each moment
。
Step S200, inputting the data meeting the physical constraint into the partial differential neural network module, extracting the multi-order partial derivative of each meteorological element variable, and predicting the wind speed of the wind field at the next time step to obtain a first predicted wind speed.
Optionally, the data satisfying the physical constraint is the historical meteorological element data conforming to an explicit equation of the atmospheric motion equation set, and the display equation is a prediction equation of the atmospheric motion equation set.
And S300, inputting the multi-order partial derivatives of the meteorological element variables into the power source separation module, and calculating the heated data and the stressed data of the atmospheric air mass in the wind field by combining an atmospheric motion equation set.
Specifically, the power source separation module can separate the external source power by means of three prediction equations in the explicit atmospheric motion equation set, namely calculating the heating data and the stress data of the atmospheric air mass in the wind field. Meanwhile, the output of the power source separation module can be physically constrained by utilizing two diagnostic equations in the atmospheric motion equation set.
And S400, inputting the heated data, the stress data and the historical meteorological element data into the data driving module, and predicting the wind speed of the wind field at the next time step to obtain a second predicted wind speed.
Specifically, the data driving module utilizes the heated data, the stress data and the historical meteorological element data to achieve supplementary prediction of wind field wind speed, can capture atmospheric motion dynamic information outside physical constraints, and further improves accuracy of wind speed prediction, and the data driving module stores high-dimensional dynamic information in a hidden space in a memory unit and transmits the high-dimensional dynamic information to the next time step.
Step S500, determining a final predicted wind speed of the wind field according to the first predicted wind speed and the second predicted wind speed.
In this embodiment, historical meteorological element data of a wind field is obtained, data meeting physical constraints are separated from the historical meteorological element data, the data meeting the physical constraints are historical meteorological element data of an explicit equation of an atmospheric motion equation set, the historical meteorological element data include historical data of a plurality of meteorological element variables of a plurality of isobaric surfaces, and wind speed prediction is performed conveniently from the historical data of a plurality of dimensions, so that accuracy of wind speed prediction is improved. And inputting the data meeting the physical constraint into a partial differential neural network module, and predicting the wind speed of the next time step in the physical constraint to obtain a first predicted wind speed. And extracting the multi-order partial derivative of each meteorological element variable, transmitting the multi-order partial derivative to the power source separation module, calculating the heated data and the stressed data of the atmospheric air mass of the wind field by the power source separation module through the atmospheric motion equation set, and introducing the physical priori knowledge of the atmospheric motion equation set, so that the prediction accuracy of the dynamic wind field wind speed can be improved. And the data driving module captures atmospheric motion dynamic information outside physical constraints according to the heated data, the stress data and the historical meteorological element data so as to predict the wind speed of the wind field in the next time step to obtain a second predicted wind speed. The first predicted wind speed obtained in the physical constraint is combined with the second predicted wind speed obtained outside the physical constraint, and the final predicted wind speed obtained by prediction from each dimension of the historical meteorological element data effectively improves the accuracy of the predicted wind speed.
Optionally, the partial differential neural network module includes a plurality of neural networks, each neural network includes a plurality of convolution kernels respectively performing independent convolution, the number of the neural networks corresponds to the number of the meteorological element variables, and each convolution kernel is used for simulating a partial differential operator in a taylor expansion of a meteorological element variable function.
Specifically, the partial differential neural network module comprises n neural networks, wherein n is consistent with the number of meteorological element variables, and each meteorological element variable corresponds to one convolutional neural network. Each neural network comprises
The convolution kernels, k values of which may preferably be 3, 5, 7, etc., may be trained in advance. The specific structure and data processing process of the partial differential neural network module are the prior art, and are not described herein again.
Optionally, the inputting the data satisfying the physical constraint into a partial differential neural network module, and the extracting the multiple-order partial derivatives of the meteorological element variables includes:
under the constraint of a convolution kernel, extracting multiple-order partial derivatives of each meteorological element variable from the data meeting the physical constraint by adopting the convolution kernel, and combining all the multiple-order partial derivatives of the meteorological element variables by using a Taylor expansion formula to obtain the first predicted wind speed.
Specifically, each meteorological element variable in the historical meteorological element data can be considered as one with respect to timetAnd coordinates
The function Z of (3) is that the convolution kernel extracts the multiple-order partial derivatives of meteorological element variables under the constraint of the convolution kernel, and then all the multiple-order partial derivatives are combined by using a Taylor expansion formula to predict the wind speed at the next time step, and the function Z can be expressed by the following formula:
wherein the content of the first and second substances,
representtThe function of the meteorological element variable at the moment,
a taylor expansion is shown, and,
representing the coordinates on the isobaric surface, Z representing a meteorological element variable function,
the multiple order partial derivatives are represented.
The convolution kernel constraint can be determined by the calculation formula of the partial differential neural network, and can be determined by the following formula:
wherein the content of the first and second substances,
a convolution kernel constraint formula is expressed and is,
a convolution kernel is represented that is a function of,
the matrix of objects is represented by a matrix of objects,
the number of the F-norm is expressed,i、jwhich represents the size of the convolution kernel,krepresents the maximum value of the convolution kernel,dindicating the number of channels.
Specific constraint formula
Is counted together for a group
A convolution kernel
Physical constraints imposed such that each size is
When the convolution kernel of (2) is applied to the data, it is close to differential
Obtained by processing the convolution kernel with a constraint matrix
And an object matrix
The specific calculation formulas of the constraint matrix and the target matrix, which are realized by calculating the loss, can be determined by a partial differential neural network, which is the prior art and is not described herein again.
In this optional embodiment, the partial differential neural network module may introduce the correlation and constraint between variables in the atmospheric motion equation set into the wind field wind speed prediction model, provide physical prior knowledge to assist in predicting the wind field wind speed, and improve the accuracy of wind field wind speed prediction.
A partial differential neural network which is essentially a convolutional neural network is connected in parallel in the wind field wind speed prediction model, although the structure of the multilayer convolutional cyclic neural network in the data driving module is not changed, the partial differential neural network is equivalent to a high-speed channel (the depth of the convolutional neural network is shallow) which is connected in parallel in the data driving module, and the problems of gradient disappearance and overfitting of the generated wind field wind speed prediction model can be further avoided.
Optionally, the power source separation module includes a calculation sub-module for simulating a prediction equation of an atmospheric motion equation set, the inputting the multiple-order partial derivatives of the variables of the meteorological elements into the power source separation module, and the calculating the heated data and the stressed data of the atmospheric air mass in the wind field by combining the atmospheric motion equation set includes:
and inputting the multi-order partial derivatives of the meteorological element variables into the calculation submodule, and calculating the heated data and the stressed data of the atmospheric air mass in the wind field through a prediction equation of an atmospheric motion equation set.
Specifically, the power source separation module receives the output of the partial differential neural network module, separates two pieces of external power source information, stress data and heated data according to a physical equation (namely an atmospheric motion equation set), and transmits the external information and the rest items in the equation set to the data driving module as auxiliary power information.
The power source separation module is mainly used for introducing physical priori knowledge of the atmospheric motion equation set into a wind field wind speed prediction model, expressing partial derivative terms and linear terms of meteorological element variables in the atmospheric motion equation set by utilizing partial differential operators provided by a partial differential neural network module, and calculating preliminary stress data and heating data of a local air mass by utilizing three prediction equations, namely two motion equations and a thermodynamic equation, in the atmospheric motion equation set. And the other two diagnostic equations can be added into the loss function of the wind field wind speed prediction model as a constraint because each term in the equation can be regarded as a known item and the reverse derivation of the unknown item is not needed. Specifically, the equation of the atmospheric motion in the coordinate system of the isobaric surface is as follows:
wherein, the first and the second end of the pipe are connected with each other,
under the coordinate system of representing the isobaric surface
The wind speed in the direction of the wind,
under the coordinate system of representing the isobaric surface
The wind speed in the direction of the wind,
represents the vertical wind speed under the coordinate system of the constant pressure surface,
the time is represented by the time of day,
the height of the potential is represented by,
which is indicative of the temperature of the gas,
in order to be subjected to a force,
in order to be heated up,
is the air pressure, and the air pressure is higher,
is a constant meteorological quantity, and is characterized in that,
representing the coordinates on the iso-pressure surface.
Equations (1), (2) and (4) are predictive equations of the atmospheric motion equation, and equations (3) and (5) are diagnostic equations of the atmospheric motion equation.
In this optional embodiment, the power source separation module calculates the heated data and the stressed data of the atmospheric air mass of the wind field by using the atmospheric motion equation set, introduces the physical priori knowledge of the atmospheric motion equation set to guide the wind speed prediction, and can improve the prediction accuracy of the dynamic wind field wind speed.
Optionally, the data driving module includes a multilayer convolution cyclic neural network, and the inputting the heated data, the stressed data, and the historical meteorological element data into the data driving module to predict the wind speed of the wind field at the next time step, and obtaining a second predicted wind speed includes:
and splicing the heated data, the stress data and the historical meteorological element data in a channel dimension, projecting the spliced data to a hidden space through a multilayer convolution cyclic neural network, and predicting the wind speed of the wind field at the next time step by combining the hidden information of the data driving module at the previous time step to obtain a second predicted wind speed.
Specifically, the heated data and the stressed data of the atmospheric air mass separated by the power source separation module are not completely collected and informed by the wind field wind speed prediction model, and are input into the data driving module together with the partial derivative term and the historical meteorological element data in the separated formula, so that dynamic capture and prediction are further carried out in the hidden space.
The data driving module has the main functions of simulating the wind speed change condition which cannot be completely expressed under physical constraint and extracting a hidden wind speed change rule from historical data. The data driving module can comprise a 3-layer convolutional neural network, sequentially receives the input historical meteorological element data of each time step and the external power source information extracted by the power source separation module, and predicts the wind speed of the wind field of the next time step by combining the hidden information of the data driving module of the previous time step.
The data driving module can be expressed by the following formula:
wherein the subscript
To represent
At the moment of time, the time of day,
which represents a convolution operation, is a function of,
representing a hamiltonian product operation,
is composed of
The input of the time of day is,
is composed of
The time of day is input into the gate control unit,
in order to forget the gating unit,
in order to output the gate control unit,
the state of the cell is the state of the cell,
in order to be in a hidden state,
for the corresponding convolution operation, i.e. weight
Is a pre-set weight of the weight,
bias terms for corresponding convolution operations, i.e.
Is a preset bias term.
In this optional embodiment, the data driving module captures the atmospheric motion dynamic information outside the physical constraint by using the powerful memory capability and the capability of extracting the dynamic information of the convolution cyclic neural network according to the heated data, the stress data and the historical meteorological element data, supplements the dynamic information which cannot be extracted under the physical constraint, so as to perform supplementary prediction on the wind speed of the wind field at the next time step, and can improve the accuracy of the wind speed obtained by final prediction.
Optionally, the determining a final predicted wind speed for the wind farm from the first predicted wind speed and the second predicted wind speed comprises:
adding the first predicted wind speed to the second predicted wind speed to determine the final predicted wind speed for the wind farm.
Specifically, the first predicted wind speed and the second predicted wind speed are combined to determine the final predicted wind speed of the wind field, and the wind speed prediction accuracy is improved.
As shown in fig. 3, another embodiment of the present invention provides a wind field wind speed prediction apparatus for implementing the wind field wind speed prediction method, where the apparatus includes:
the device comprises an acquisition unit, a calculation unit and a control unit, wherein the acquisition unit is used for acquiring historical meteorological element data of a wind field and separating data meeting physical constraints from the historical meteorological element data, and the historical meteorological element data comprises historical data of a plurality of meteorological element variables of a plurality of isobaric surfaces;
the physical part prediction unit is used for inputting the data meeting the physical constraint into a partial differential neural network module, extracting a multi-order partial derivative of each meteorological element variable, and predicting the wind field wind speed at the next time step to obtain a first predicted wind speed;
the data extraction unit is used for inputting the multi-order partial derivatives of the meteorological element variables into the power source separation module and calculating the heated data and the stressed data of the atmospheric air mass in the wind field by combining an atmospheric motion equation set;
the supplementary prediction unit is used for inputting the heated data, the stress data and the historical meteorological element data into a data driving module, predicting the wind speed of the wind field at the next time step and obtaining a second predicted wind speed;
and the comprehensive prediction unit is used for determining the final predicted wind speed of the wind field according to the first predicted wind speed and the second predicted wind speed.
The wind field wind speed prediction device of the present embodiment is used for implementing the wind field wind speed prediction method, and the advantages thereof over the prior art are the same as the advantages of the wind field wind speed prediction method over the prior art, and are not described herein again.
Another embodiment of the present invention provides an electronic device, including a memory and a processor; the memory for storing a computer program; the processor is configured to, when executing the computer program, implement the wind farm wind speed prediction method as described above.
A further embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the wind farm wind speed prediction method as described above.
An electronic device that can be a server or a client of the present invention, which is an example of a hardware device that can be applied to aspects of the present invention, will now be described. Electronic device is intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.
The electronic device includes a computing unit that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) or a computer program loaded from a storage unit into a Random Access Memory (RAM). In the RAM, various programs and data required for the operation of the device can also be stored. The computing unit, the ROM, and the RAM are connected to each other by a bus. An input/output (I/O) interface is also connected to the bus.
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 may be implemented by a computer program, which may be stored in a computer readable storage medium and executed by a computer to implement 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), or the like. In this application, the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention. In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
Although the present disclosure has been described above, the scope of the present disclosure is not limited thereto. Various changes and modifications may be effected therein by one of ordinary skill in the pertinent art without departing from the spirit and scope of the present disclosure, and these changes and modifications are intended to be within the scope of the present disclosure.
Claims (5)
1. A wind field wind speed prediction method is characterized in that based on a wind field wind speed prediction model, the wind field wind speed prediction model comprises a partial differential neural network module, a power source separation module and a data driving module, and the wind field wind speed prediction method comprises the following steps:
acquiring historical meteorological element data of a wind field, and separating data meeting physical constraints from the historical meteorological element data, wherein the historical meteorological element data comprise historical data of a plurality of meteorological element variables of multiple isobaric surfaces, and the data meeting the physical constraints are the historical meteorological element data which conform to an explicit equation of an atmospheric motion equation set;
inputting the data meeting the physical constraint into the partial differential neural network module, extracting the multi-order partial derivative of each meteorological element variable, predicting the wind speed of the wind field at the next time step, and obtaining a first predicted wind speed, wherein the method comprises the following steps: under the constraint of a convolution kernel, extracting multiple-order partial derivatives of each meteorological element variable from the data meeting the physical constraint by adopting the convolution kernel, and combining all the multiple-order partial derivatives of the meteorological element variables by using a Taylor expansion formula to obtain the first predicted wind speed; the partial differential neural network module comprises a plurality of neural networks, each neural network comprises a plurality of convolution kernels which respectively complete independent convolution, the number of the neural networks corresponds to that of the meteorological element variables, and each convolution kernel is used for simulating a partial differential operator in a Taylor expansion of a meteorological element variable function;
inputting the multi-order partial derivatives of the meteorological element variables into the power source separation module, wherein the power source separation module comprises a calculation submodule for simulating a prediction equation of an atmospheric motion equation set, and calculating heating data and stress data of atmospheric air mass in the wind field by combining the atmospheric motion equation set;
inputting the heated data, the stressed data and the historical meteorological element data into the data driving module, wherein the data driving module comprises a multilayer convolution cyclic neural network, predicts the wind speed of the wind field at the next time step, and obtains a second predicted wind speed, and the method comprises the following steps: splicing the heated data, the stress data and the historical meteorological element data in a channel dimension, projecting the spliced data to a hidden space through a multilayer convolution cyclic neural network, and predicting the wind speed of the wind field at the next time step by combining hidden information of the data driving module at the previous time step to obtain a second predicted wind speed;
determining a final predicted wind speed for the wind farm based on the first predicted wind speed and the second predicted wind speed.
2. The method of claim 1, wherein the determining a final predicted wind speed for the wind farm from the first predicted wind speed and the second predicted wind speed comprises:
adding the first predicted wind speed to the second predicted wind speed to determine the final predicted wind speed for the wind farm.
3. A wind farm wind speed prediction apparatus for implementing the wind farm wind speed prediction method according to claim 1 or 2, the apparatus comprising:
the acquiring unit is used for acquiring historical meteorological element data of a wind field and separating data meeting physical constraints from the historical meteorological element data, wherein the historical meteorological element data comprises historical data of a plurality of meteorological element variables of multiple isobaric surfaces, and the data meeting the physical constraints are the historical meteorological element data which conform to an explicit equation of an atmospheric motion equation set;
the physical part prediction unit is used for inputting the data meeting the physical constraint into a partial differential neural network module, extracting a multi-order partial derivative of each meteorological element variable, predicting the wind speed of the wind field at the next time step, and obtaining a first predicted wind speed, and comprises: under the constraint of a convolution kernel, extracting multiple-order partial derivatives of each meteorological element variable from the data meeting the physical constraint by adopting the convolution kernel, and combining all the multiple-order partial derivatives of the meteorological element variables by using a Taylor expansion formula to obtain the first predicted wind speed; the partial differential neural network module comprises a plurality of neural networks, each neural network comprises a plurality of convolution kernels which respectively complete independent convolution, the number of the neural networks corresponds to the number of the meteorological element variables, and each convolution kernel is used for simulating a partial differential operator in a Taylor expansion of a meteorological element variable function;
the data extraction unit is used for inputting the multi-order partial derivatives of all the meteorological element variables into the power source separation module, the power source separation module comprises a calculation submodule for simulating a prediction equation of an atmospheric motion equation set, and the heating data and the stress data of the atmospheric air mass in the wind field are calculated by combining the atmospheric motion equation set;
the supplementary prediction unit is used for inputting the heated data, the stress data and the historical meteorological element data into a data driving module, the data driving module comprises a multilayer convolution cyclic neural network, the wind field wind speed of the next time step is predicted, and a second predicted wind speed is obtained, and the supplementary prediction unit comprises: splicing the heated data, the stress data and the historical meteorological element data in a channel dimension, projecting the spliced data to a hidden space through a multilayer convolution cyclic neural network, and predicting the wind speed of a wind field at the next time step by combining hidden information of the data driving module at the previous time step to obtain a second predicted wind speed;
and the comprehensive prediction unit is used for determining the final predicted wind speed of the wind field according to the first predicted wind speed and the second predicted wind speed.
4. An electronic device comprising a memory and a processor;
the memory for storing a computer program;
the processor, when executing the computer program, for implementing the wind farm wind speed prediction method according to claim 1 or 2.
5. A computer-readable storage medium, characterized in that the storage medium has stored thereon a computer program which, when being executed by a processor, carries out the wind farm wind speed prediction method according to claim 1 or 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211272558.6A CN115345387B (en) | 2022-10-18 | 2022-10-18 | Wind field wind speed prediction method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211272558.6A CN115345387B (en) | 2022-10-18 | 2022-10-18 | Wind field wind speed prediction method and device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115345387A CN115345387A (en) | 2022-11-15 |
| CN115345387B true CN115345387B (en) | 2023-01-24 |
Family
ID=83957397
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211272558.6A Active CN115345387B (en) | 2022-10-18 | 2022-10-18 | Wind field wind speed prediction method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115345387B (en) |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003036333A1 (en) * | 2001-10-19 | 2003-05-01 | Mitsubishi Heavy Industries, Ltd. | Atmosphere condition prediction method |
| CN110442836B (en) * | 2019-07-23 | 2023-05-12 | 国能日新科技股份有限公司 | Wind power plant short-term wind speed prediction method and system |
| CN110543929B (en) * | 2019-08-29 | 2023-11-14 | 华北电力大学（保定） | Wind speed interval prediction method and system based on Lorenz system |
| CN113537648B (en) * | 2021-09-16 | 2021-12-24 | 国能日新科技股份有限公司 | Wind speed prediction method and device based on set data |
| CN114021830A (en) * | 2021-11-12 | 2022-02-08 | 天津大学 | Multi-time-range wind speed prediction method based on CNN-LSTM |
| CN114412721A (en) * | 2021-12-13 | 2022-04-29 | 明阳智慧能源集团股份公司 | Equivalent wind speed observation method for impeller of wind generating set |
| CN115099506A (en) * | 2022-06-30 | 2022-09-23 | 广州铁路职业技术学院（广州铁路机械学校） | Wind speed prediction method, system, electronic device and medium |
-
2022
- 2022-10-18 CN CN202211272558.6A patent/CN115345387B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115345387A (en) | 2022-11-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110969250B (en) | Neural network training method and device | |
| CN111538908B (en) | Search ranking method and device, computer equipment and storage medium | |
| CN110532996B (en) | Video classification method, information processing method and server | |
| CN112784778B (en) | Method, apparatus, device and medium for generating model and identifying age and sex | |
| CN107992938B (en) | Space-time big data prediction technique and system based on positive and negative convolutional neural networks | |
| CN110138595A (en) | Time link prediction technique, device, equipment and the medium of dynamic weighting network | |
| Ayodeji et al. | Causal augmented ConvNet: A temporal memory dilated convolution model for long-sequence time series prediction | |
| CN113538441A (en) | Image segmentation model processing method, image processing method and device | |
| CN113627093B (en) | Underwater mechanism trans-scale flow field characteristic prediction method based on improved Unet network | |
| CN113627422A (en) | Image classification method and related equipment thereof | |
| CN115841004A (en) | Strip steel hot rolling process mechanical property soft measurement method and device based on multidimensional data | |
| CN106446546B (en) | Meteorological data complementing method based on the automatic encoding and decoding algorithm of convolution | |
| Li et al. | Aero-engine exhaust gas temperature prediction based on LightGBM optimized by improved bat algorithm | |
| Liu et al. | Grey-based approach for estimating Weibull model and its application | |
| CN115345387B (en) | Wind field wind speed prediction method and device and storage medium | |
| CN116051699B (en) | Dynamic capture data processing method, device, equipment and storage medium | |
| CN115034121B (en) | Strip steel process regulation and control method based on organization performance intelligent prediction model | |
| Mekala et al. | Gesture recognition using neural networks based on HW/SW cosimulation platform | |
| Li et al. | Lightweight real-time makeup try-on in mobile browsers with tiny CNN models for facial tracking | |
| CN110415006B (en) | Advertisement click rate estimation method and device | |
| CN109166128B (en) | Non-equivalent three-dimensional point cloud segmentation method | |
| CN112784969B (en) | Convolutional neural network acceleration learning method for image feature extraction | |
| Ghiasi et al. | Combining thermodynamics-based model of the centrifugal compressors and active machine learning for enhanced industrial design optimization | |
| CN113675844B (en) | Multi-node power load prediction method | |
| Dumont et al. | A fast multipole implementation of the simplified hybrid boundary element method: application to 2D potential problems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |