CN114298444A - Wind speed probability prediction method, device, equipment and storage medium - Google Patents
Wind speed probability prediction method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN114298444A CN114298444A CN202210221104.XA CN202210221104A CN114298444A CN 114298444 A CN114298444 A CN 114298444A CN 202210221104 A CN202210221104 A CN 202210221104A CN 114298444 A CN114298444 A CN 114298444A
- Authority
- CN
- China
- Prior art keywords
- wind speed
- distribution function
- collaborative
- layer
- cumulative distribution
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000005315 distribution function Methods 0.000 claims abstract description 102
- 230000001186 cumulative effect Effects 0.000 claims abstract description 80
- 238000012549 training Methods 0.000 claims abstract description 35
- 230000006870 function Effects 0.000 claims description 37
- 238000012360 testing method Methods 0.000 claims description 31
- 230000015654 memory Effects 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000011176 pooling Methods 0.000 description 7
- 230000004913 activation Effects 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 5
- 210000002569 neuron Anatomy 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 101100110035 Escherichia coli (strain K12) aslA gene Proteins 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000006386 memory function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002195 synergetic effect Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to the technical field of wind speed prediction, and discloses a wind speed probability prediction method, a wind speed probability prediction device, wind speed probability prediction equipment and a storage medium, wherein the method comprises the following steps: constructing a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse conditional cumulative distribution function network layer and a conditional cumulative distribution function network layer; performing parameter training on an inverse condition cumulative distribution function and a condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model; and inputting preset target wind speed data into the optimized collaborative prediction model for prediction operation to obtain a wind speed probability value. The method and the device solve the technical problem that in the prior art, the actual prediction result lacks reliability and accuracy due to the fact that the wind speed data with complex changes are analyzed through fixed model parameters.
Description
Technical Field
The present disclosure relates to the field of wind speed prediction technologies, and in particular, to a method, an apparatus, a device, and a storage medium for wind speed probability prediction.
Background
Wind energy is used as a renewable energy source, and has strong randomness and intermittence, so that the fluctuation of the output power of the wind generating set is strong. In this case, after the wind power is connected to the grid, a series of influences will be brought to the power system, for example, in terms of power quality, power flow distribution, power system scheduling, power market trading, and the like. Particularly, with the increase of the total installed capacity of wind power and the increase of the proportion of the grid-connected wind power capacity to the total capacity of the system, the problems of peak shaving, reactive power and voltage control and the like of a power grid can be brought, and new problems are brought to the safety, stability and normal dispatching of the power grid. The prediction of the wind speed and the wind power output power of the wind power plant by using a prediction technology is an effective way for realizing wind power management between wind power and a power grid. From the overall electricity utilization condition, the management of the wind power generation output power forecast based on the wind speed forecast not only can reduce the operation cost and the rotation reserve of a power system and improve the wind power penetration power limit, but also is beneficial to correctly making an electric energy exchange plan under the power market environment so as to fully utilize wind power resources and obtain more economic benefits and social benefits.
The prior art provides various wind speed prediction methods, and a common method directly adopts model fitting analysis such as normal distribution to a wind speed time sequence without considering the change condition of wind speed, so that a complex wind speed environment cannot be accurately expressed; and model parameters in the method adopting the off-line model prediction are unchanged, so that the prediction result lacks reliability.
Disclosure of Invention
The application provides a wind speed probability prediction method, a device, equipment and a storage medium, which are used for solving the technical problem that in the prior art, the actual prediction result lacks reliability and accuracy due to the fact that the fixed model parameters are used for analyzing complex and variable wind speed data.
In view of the above, a first aspect of the present application provides a wind speed probability prediction method based on a two-layer collaborative network, including:
constructing a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse conditional cumulative distribution function network layer and a conditional cumulative distribution function network layer;
performing parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model;
and inputting preset target wind speed data into the optimized collaborative prediction model to perform prediction operation, so as to obtain a wind speed probability value.
Preferably, the objective function of the network layer of the inverse conditional cumulative distribution function is:
wherein,
accumulating parameters of a distribution function network layer for the inverse condition,qin order to preset the place-dividing value,Xfor the data of the preset wind speed training data set,Yfor the corresponding measure of the wind speed,
in order to perform the averaging operation, the average value is calculated,
representing the inverse conditional cumulative distribution function network layer,
accumulating the output results of the distribution function network layer for the inverse condition.
Preferably, the constraint conditions of the network layer of the conditional cumulative distribution function are:
wherein,
representing the conditional cumulative distribution function network layer,
accumulating parameters of the distribution function network layer for the condition,
representing the output result of the conditional cumulative distribution function network layer,
to predict the quantile value.
Preferably, the performing parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model, and then further comprising:
testing the optimized collaborative prediction model through a preset test data set to obtain a test result;
and screening the optimized collaborative prediction model according to the test result.
The second aspect of the present application provides a wind speed probability prediction device based on a double-layer collaborative network, including:
the model building module is used for building a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse conditional cumulative distribution function network layer and a conditional cumulative distribution function network layer;
the model training module is used for carrying out parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model;
and the wind speed prediction module is used for inputting preset target wind speed data into the optimized collaborative prediction model to carry out prediction operation so as to obtain a wind speed probability value.
Preferably, the objective function of the network layer of the inverse conditional cumulative distribution function is:
wherein,
accumulating parameters of a distribution function network layer for the inverse condition,qin order to preset the place-dividing value,Xfor the data of the preset wind speed training data set,Yfor the corresponding measure of the wind speed,
to seekThe average value is taken out and the average value is obtained,
representing the inverse conditional cumulative distribution function network layer,
accumulating the output results of the distribution function network layer for the inverse condition.
Preferably, the constraint conditions of the network layer of the conditional cumulative distribution function are:
wherein,
representing the conditional cumulative distribution function network layer,
accumulating parameters of the distribution function network layer for the condition,
representing the output result of the conditional cumulative distribution function network layer,
to predict the quantile value.
Preferably, the method further comprises the following steps:
the model testing module is used for testing the optimized collaborative prediction model through a preset testing data set to obtain a testing result;
and screening the optimized collaborative prediction model according to the test result.
The third aspect of the application provides a wind speed probability prediction device based on a double-layer collaborative network, which comprises a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to execute the wind speed probability prediction method based on the two-tier collaborative network according to the first aspect according to instructions in the program code.
A fourth aspect of the present application provides a computer-readable storage medium for storing program code for executing the wind speed probability prediction method based on the two-tier collaborative network according to the first aspect.
According to the technical scheme, the embodiment of the application has the following advantages:
the application provides a wind speed probability prediction method based on a double-layer collaborative network, which comprises the following steps: constructing a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse conditional cumulative distribution function network layer and a conditional cumulative distribution function network layer; performing parameter training on an inverse condition cumulative distribution function and a condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model; and inputting preset target wind speed data into the optimized collaborative prediction model for prediction operation to obtain a wind speed probability value.
According to the wind speed probability prediction method based on the double-layer collaborative network, the constructed double-layer CNN-LSTM collaborative network model is trained through wind speed data, and model parameters are updated continuously, so that the method is more suitable for wind speed changes in different environments, and the wind speed probability value is more accurate and reliable. Therefore, the method and the device can solve the technical problem that in the prior art, the actual prediction result lacks reliability and accuracy due to the fact that the fixed model parameters are used for analyzing the wind speed data with complex changes.
Drawings
Fig. 1 is a schematic flow chart of a wind speed probability prediction method based on a double-layer collaborative network according to an embodiment of the present application;
fig. 2 is a schematic structural diagram of a wind speed probability prediction apparatus based on a two-layer collaborative network according to an embodiment of the present application;
fig. 3 is a schematic diagram of a CNN-LSTM network architecture framework provided in an embodiment of the present application;
FIG. 4 is a schematic diagram of an LSTM neuron according to an embodiment of the present application;
fig. 5 is a functional relationship diagram of an inverse conditional cumulative distribution function network layer according to an embodiment of the present application;
fig. 6 is a conceptual relationship diagram of an optimized collaborative prediction model according to an embodiment of the present application.
Detailed Description
In order to make the technical solutions of the present application better understood, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
For easy understanding, please refer to fig. 1, an embodiment of a wind speed probability prediction method based on a two-layer collaborative network provided by the present application includes:
Referring to fig. 3, a Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) are combined to obtain a comprehensive network model. The network model specifically comprises a convolutional layer (Cov 1D), a maximum pooling layer (Max Pooling), an LSTM neuron layer, a network selection layer (Droput) and a full connection layer (sense); the CNN can extract local features of wind speed data and can extract high-level features with strong representation capability in spatial dimension; and the LSTM can expand the time characteristics and process the characteristic data information with the sequence.
The convolution layer is a process of performing convolution calculation on a local area of input data, the convolution calculation is a process of traversing the whole input data by the convolution layer through a sliding convolution window, and a convolution calculation formula is as follows:
wherein,
is shown aslA first of the output layersiThe characteristics of the device are as follows,
is shown aslLayer oneiThe weight matrix of each convolution kernel is then,
is shown asl-an output result of layer 1,l>1，
is as followslLayer oneiThe offset term of each convolution kernel is,
in order to perform the convolution operation,
for the activation function, the CNN solves the non-linearity problem in the wind speed time series by a non-linear activation function, and therefore selects a rectifying linear unit (ReLU) as the activation function of the convolutional neural network. The first output layer is calculated as
Wherein
for inputting the data sequence of the first layer of convolutional layer, the first layer of convolutional layer has no output characteristics of the previous layer as the basis for calculation, so the input data sequence is directly taken for calculation, which is the convolutional expressionThrough the common calculation process of the network, detailed description is omitted.
The pooling layer adopts an up-sampling or down-sampling mode, and can effectively extract the local characteristics of the wind speed data and reduce the local characteristic dimension by alternately using the pooling layer and the convolution layer, so that the most important information of the wind speed time sequence is kept, and the quality of the input characteristics is improved. The maximum pooling is generally adopted, and the specific calculation mode is as follows:
wherein,
is the first after poolingl+1 layer ofiThe characteristic diagram is shown in the figure,
is as followsjThe number of the pooled areas is reduced,
is as followslLayer oneiElements of the feature map within the pooling range.
The Dropout layer controls the number of hidden neurons in the network, so that the fully-connected network has certain sparseness, and the synergistic effect of different characteristics is effectively reduced. Dropout is a proportional selection operation, is common in neural networks, and means that a part of neural network units are selected from the network temporarily according to a certain probability, do not participate in the training, and is equivalent to finding a lighter-weight network from the original network, so that the calculation redundancy is reduced, and the model training speed is increased.
The purpose of a full connection layer (Dense) layer is to carry out nonlinear change on the previously extracted features through a kernel function, extract the correlation among the features and finally map the correlation to an output space, wherein the Dense layer adopts ReLu and Sigmoid activation functions.
The LSTM has memory function and high efficiency for processing time series data, and is mainly composed of forgetting gate
Input gate
And output gate
The control device is used for controlling the characteristic information hidden in the wind speed. See FIG. 4 for the structure of a single LSTM neuron, in which
Representing the activation function Sigmoid or tanh function. For controlling the value output range between-1 and 1, the respective parameters in the LSTM are calculated as follows:
wherein,
、
、
、
are the weight matrices of the different network layers,
、
、
、
for the bias term in the corresponding network,
in order to activate the function(s),
、
、
are respectively astThe input layer state, the control unit state and the hidden unit state of the moment, in general
If, if
It means that the network does not start processing data, and only the input layer has a valid state, neither the control unit nor the hidden unit has a substantial state, and this invalid state is not discussed here,
the intermediate state quantity of the control unit is characterized by a characteristic numerical value.
The method is characterized in that the comprehensive description of the double-layer CNN-LSTM collaborative network model is carried out according to the functional division of the network model, the model is divided into an inverse condition cumulative distribution function network layer and a condition cumulative distribution function network layer, the inverse condition cumulative distribution function network layer calculates the characteristic value of input data through the inverse function of the condition cumulative distribution function, then the characteristic value is used as the input data of the condition cumulative distribution function network layer, the condition cumulative distribution function is adopted for calculation, and parameters in the function can be continuously optimized in the training process of the model until a reliable prediction result is obtained.
The objective function of the network layer of the inverse conditional cumulative distribution function is:
wherein,
accumulating parameters of a distribution function network layer for the inverse condition,qin order to preset the place-dividing value,Xfor the data of the preset wind speed training data set,Yfor the corresponding measure of the wind speed,
in order to perform the averaging operation, the average value is calculated,
network layer representing the inverse conditional cumulative distribution function, which may be referred to asfThe network(s) of the network(s),
accumulating the output results of the distribution function network layer for the inverse condition.
Please refer to fig. 5, if usedfRepresenting the inverse conditional cumulative distribution function network layer, it is understood that the inverse conditional cumulative distribution function network layer is guaranteed
And the output of the network layer can be expressed as
The output is used as the input of the network layer of the conditional cumulative distribution function, and the processing is continued.
The conditional cumulative distribution function network layer is represented as
Is called asgThe network, which also takes CNN-LSTM as the core, fits the condition accumulation distribution function of the wind speed time sequence, and the condition accumulation distribution function networkThe constraint conditions of the network layer are as follows:
wherein the training is optimized
Output of a network
As
The input of the network is set up,
the role of the network is to fit the value of a given place value expressed as
And
and acquiring an optimal wind speed condition cumulative distribution function.
The network layer representing the conditional cumulative function,
the parameters of the network layer of the distribution function are accumulated for the condition,
representing the output result of the conditional cumulative distribution function network layer,
to predict the quantile value.
Referring to fig. 6, a conceptual diagram of the double-layer CNN-LSTM collaborative network model may be expressed as follows, in which an inverse conditional cumulative distribution function of a wind speed time series and a collaborative loss function after the conditional cumulative distribution function are jointly fitted:
wherein,
、
respectively representgLoss function of network andfas a function of the loss of the network,
to relate to
、
And
the average value is obtained through the operation of calculating the average value,
is a quantile value distribution function, obeys a uniform distribution of (0, 1),
、
each represents a distribution function of a variable,
is composed of
A value in the case of (1), i.e.Is 1, loss function
Using a binary cross entropy function, the expression is as follows:
wherein,a，bthe wind speed actual value and the predicted value are respectively.
Will be provided with
、
And
and integrating the formulas to obtain a final cooperative loss function:
fnetwork layer andgthe network layer coordination strategy is as follows:fthe network layer fits the optimal inverse condition cumulative distribution function of the wind speed time series,gin the process of the network layer through iterative optimization,fnetwork isgNetwork provides an aid to optimization, ensuringgAnd the network acquires more comprehensive wind speed information, and in the whole cooperation process, the two loss functions are jointly optimized to learn the distribution function of the wind speed time sequence and finally obtain the interval prediction result of the wind speed time sequence.
And 102, carrying out parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model.
The preset wind speed training data set can be composed of historical data of the past year, the data needs to be converted into a form which can be input into a neural network model for processing through basic preprocessing, and the format of the following data, namely test data and target task data, needs to be unified, so that analysis is facilitated.
Further, step 102, thereafter, further includes:
testing the optimized collaborative prediction model through a preset test data set to obtain a test result;
and screening the optimized collaborative prediction model according to the test result.
The preset test data set is other data which are not input into the network training, can also be historical data, or data acquired in the existing scene, is used for testing the performance of the trained model, such as the standard information of accuracy, recall rate and the like, and can be screened as a subsequent target model when certain conditions are met.
And 103, inputting preset target wind speed data into the optimized collaborative prediction model for prediction operation to obtain a wind speed probability value.
The target wind speed data is a wind speed value obtained according to meteorological data of the current or future time, the predicted wind speed probability value is often an interval result, and the model can be suitable for various different environments and can be used for dealing with various different wind field changes.
According to the wind speed probability prediction method based on the double-layer collaborative network, the constructed double-layer CNN-LSTM collaborative network model is trained through wind speed data, and model parameters are updated continuously, so that the method is more suitable for wind speed changes in different environments, and the wind speed probability value is more accurate and reliable. Therefore, the method and the device for predicting the wind speed data can solve the technical problem that in the prior art, the actual prediction result lacks reliability and accuracy due to the fact that the fixed model parameters are used for analyzing the wind speed data with complex changes.
For easy understanding, please provide a method for evaluating an optimized collaborative prediction model for wind speed prediction, for interval prediction with a confidence level of PICP, a Prediction Interval Confidence (PICP), a prediction interval average bandwidth index (PINAW) and an interval optimization criterion based on width coverage (CWC) are introduced as performance evaluation indexes of the prediction model, and the calculation formulas of the three indexes are as follows:
wherein,
for variables with values of 0 and 1,Nto test the amount of sample data. When in use
In the prediction interval
When the temperature of the water is higher than the set temperature,
(ii) a When in use
When the number of the cells is not in the interval,
；
、
the upper limit value and the lower limit value of the prediction interval are respectively.
Wherein R is the length of the prediction interval, i.e.
。
Wherein,
and
in order to control the hyperparameter of the CWC index value, according to the empirical value,
，
，
in order to be a level of confidence,
is defined as:
the relevant parameter settings in the wind speed probability prediction model, i.e. the optimized collaborative prediction model, are shown in table 1.
TABLE 1 CNN-LSTM result parameters
At a confidence level
For example, a method such as Gaussian Process Regression (GPR) or Quantile Regression (QR) is provided for comparison analysis with the method in the embodiment of the present application, as shown in table 2, which is given by a certain 2005 yearAnd predicting the result of the PICP, PINAW and CWC index values of the result every month.
TABLE 2 comparison of results with various prediction methods
As can be seen from table 2, compared with the GPR and QR methods, the prediction probabilities of the actual values of wind speed in the method of the present embodiment both fall within the prediction interval, so that the PICP values are both 100%. Moreover, the quality of the prediction interval of the method in the embodiment is better than that of the GPR and QR methods according to the index values of PINAW and CWC.
For easy understanding, please refer to fig. 2, the present application provides an embodiment of a wind speed probability prediction apparatus based on a two-layer collaborative network, including:
the model building module 201 is used for building a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse condition cumulative distribution function network layer and a condition cumulative distribution function network layer;
the model training module 202 is used for performing parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model;
and the wind speed prediction module 203 is used for inputting preset target wind speed data into the optimized collaborative prediction model for prediction operation to obtain a wind speed probability value.
Further, the objective function of the network layer of the inverse conditional cumulative distribution function is:
wherein,
accumulating parameters of a distribution function network layer for the inverse condition,qfor presetting the positionThe value of the one or more of the one,Xfor the data of the preset wind speed training data set,Yfor the corresponding measure of the wind speed,
in order to perform the averaging operation, the average value is calculated,
representing the inverse conditional cumulative distribution function network layer,
accumulating the output results of the distribution function network layer for the inverse condition.
Further, the constraint conditions of the network layer of the conditional cumulative distribution function are as follows:
wherein,
representing the conditional cumulative distribution function network layer,
accumulating parameters of the distribution function network layer for the condition,
representing the output result of the conditional cumulative distribution function network layer,
to predict the quantile value.
Further, still include:
the model testing module 204 is used for testing the optimized collaborative prediction model through a preset testing data set to obtain a testing result;
and screening the optimized collaborative prediction model according to the test result.
It can be understood that the optimized collaborative prediction model in the present embodiment is the optimized collaborative prediction model adopted in the above method embodiment.
The application also provides wind speed probability prediction equipment based on the double-layer cooperative network, which comprises a processor and a memory;
the memory is used for storing the program codes and transmitting the program codes to the processor;
the processor is used for executing the wind speed probability prediction method based on the double-layer collaborative network in the method embodiment according to the instructions in the program code.
The present application further provides a computer-readable storage medium for storing a program code for executing the wind speed probability prediction method based on the two-tier collaborative network in the above method embodiment.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
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.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for executing all or part of the steps of the method described in the embodiments of the present application through a computer device (which may be a personal computer, a server, or a network device). And the aforementioned storage medium includes: a U disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (10)
1. A wind speed probability prediction method is characterized by comprising the following steps:
constructing a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse conditional cumulative distribution function network layer and a conditional cumulative distribution function network layer;
performing parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model;
and inputting preset target wind speed data into the optimized collaborative prediction model to perform prediction operation, so as to obtain a wind speed probability value.
2. The wind speed probability prediction method of claim 1, wherein the objective function of the network layer of the inverse conditional cumulative distribution function is:
wherein,
accumulating parameters of a distribution function network layer for the inverse condition,qin order to preset the place-dividing value,Xfor the data of the preset wind speed training data set,Yfor the corresponding measure of the wind speed,
in order to perform the averaging operation, the average value is calculated,
representing the inverse conditional cumulative distribution function network layer,
accumulating the output results of the distribution function network layer for the inverse condition.
3. The wind speed probability prediction method according to claim 2, characterized in that the constraint condition of the conditional cumulative distribution function network layer is:
wherein,
representing the conditional cumulative distribution function network layer,
accumulating parameters of the distribution function network layer for the condition,
representing the output result of the conditional cumulative distribution function network layer,
to predict the quantile value.
4. The wind speed probability prediction method according to claim 1, wherein the inverse conditional cumulative distribution function and the conditional cumulative distribution function in the double-layer CNN-LSTM collaborative network model are subjected to parameter training through a preset wind speed training data set to obtain an optimized collaborative prediction model, and then further comprising:
testing the optimized collaborative prediction model through a preset test data set to obtain a test result;
and screening the optimized collaborative prediction model according to the test result.
5. A wind speed probability prediction device, comprising:
the model building module is used for building a double-layer CNN-LSTM collaborative network model, wherein a loss function of the double-layer CNN-LSTM collaborative network model is a collaborative loss function, and the double-layer CNN-LSTM collaborative network model comprises an inverse conditional cumulative distribution function network layer and a conditional cumulative distribution function network layer;
the model training module is used for carrying out parameter training on the inverse condition cumulative distribution function and the condition cumulative distribution function in the double-layer CNN-LSTM collaborative network model through a preset wind speed training data set to obtain an optimized collaborative prediction model;
and the wind speed prediction module is used for inputting preset target wind speed data into the optimized collaborative prediction model to carry out prediction operation so as to obtain a wind speed probability value.
6. The wind speed probability prediction device of claim 5, wherein the objective function of the network layer of the inverse conditional cumulative distribution function is:
wherein,
accumulating parameters of a distribution function network layer for the inverse condition,qin order to preset the place-dividing value,Xfor the data of the preset wind speed training data set,Yfor the corresponding measure of the wind speed,
in order to perform the averaging operation, the average value is calculated,
representing the inverse conditional cumulative distribution function network layer,
accumulating the output results of the distribution function network layer for the inverse condition.
7. The wind speed probability prediction device of claim 6, wherein the constraint condition of the network layer of the conditional cumulative distribution function is:
wherein,
representing the conditional cumulative distribution function network layer,
accumulating parameters of the distribution function network layer for the condition,
representing the output result of the conditional cumulative distribution function network layer,
to predict the quantile value.
8. The wind speed probability prediction device of claim 5, further comprising:
the model testing module is used for testing the optimized collaborative prediction model through a preset testing data set to obtain a testing result;
and screening the optimized collaborative prediction model according to the test result.
9. A wind speed probability prediction device, characterized in that the device comprises a processor and a memory;
the memory is used for storing program codes and transmitting the program codes to the processor;
the processor is configured to perform the wind speed probability prediction method of any of claims 1-4 according to instructions in the program code.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium is configured to store program code for performing the wind speed probability prediction method of any of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210221104.XA CN114298444B (en) | 2022-03-09 | 2022-03-09 | Wind speed probability prediction method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210221104.XA CN114298444B (en) | 2022-03-09 | 2022-03-09 | Wind speed probability prediction method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114298444A true CN114298444A (en) | 2022-04-08 |
| CN114298444B CN114298444B (en) | 2022-05-27 |
Family
ID=80978718
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210221104.XA Active CN114298444B (en) | 2022-03-09 | 2022-03-09 | Wind speed probability prediction method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114298444B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108471137A (en) * | 2018-04-17 | 2018-08-31 | 国电南京自动化股份有限公司 | Wind speed power probability mapping method in a kind of wind power prediction |
| CN110378504A (en) * | 2019-04-12 | 2019-10-25 | 东南大学 | A kind of photovoltaic power generation climbing probability of happening prediction technique based on higher-dimension Copula technology |
| CN110648014A (en) * | 2019-08-28 | 2020-01-03 | 山东大学 | Regional wind power prediction method and system based on space-time quantile regression |
| CN111882057A (en) * | 2020-06-19 | 2020-11-03 | 苏州浪潮智能科技有限公司 | Multi-stage deep learning method oriented to space-time sequence data modeling and application |
| CN112232577A (en) * | 2020-10-23 | 2021-01-15 | 浙江八达电子仪表有限公司 | Power load probability prediction system and method for multi-core intelligent meter |
| CN112949201A (en) * | 2021-03-17 | 2021-06-11 | 华翔翔能科技股份有限公司 | Wind speed prediction method and device, electronic equipment and storage medium |
| CN113642225A (en) * | 2021-05-24 | 2021-11-12 | 国网新疆电力有限公司经济技术研究院 | CNN-LSTM short-term wind power prediction method based on attention mechanism |
| CN114004152A (en) * | 2021-10-29 | 2022-02-01 | 河海大学 | Multi-wind-field wind speed space-time prediction method based on graph convolution and recurrent neural network |
-
2022
- 2022-03-09 CN CN202210221104.XA patent/CN114298444B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108471137A (en) * | 2018-04-17 | 2018-08-31 | 国电南京自动化股份有限公司 | Wind speed power probability mapping method in a kind of wind power prediction |
| CN110378504A (en) * | 2019-04-12 | 2019-10-25 | 东南大学 | A kind of photovoltaic power generation climbing probability of happening prediction technique based on higher-dimension Copula technology |
| CN110648014A (en) * | 2019-08-28 | 2020-01-03 | 山东大学 | Regional wind power prediction method and system based on space-time quantile regression |
| CN111882057A (en) * | 2020-06-19 | 2020-11-03 | 苏州浪潮智能科技有限公司 | Multi-stage deep learning method oriented to space-time sequence data modeling and application |
| CN112232577A (en) * | 2020-10-23 | 2021-01-15 | 浙江八达电子仪表有限公司 | Power load probability prediction system and method for multi-core intelligent meter |
| CN112949201A (en) * | 2021-03-17 | 2021-06-11 | 华翔翔能科技股份有限公司 | Wind speed prediction method and device, electronic equipment and storage medium |
| CN113642225A (en) * | 2021-05-24 | 2021-11-12 | 国网新疆电力有限公司经济技术研究院 | CNN-LSTM short-term wind power prediction method based on attention mechanism |
| CN114004152A (en) * | 2021-10-29 | 2022-02-01 | 河海大学 | Multi-wind-field wind speed space-time prediction method based on graph convolution and recurrent neural network |
Non-Patent Citations (6)
| Title |
|---|
| FIRUZ AHAMED NAHID等: ""Very Short Term Wind Speed Forecasting Using Convolutional Long Short Term Memory Recurrent Neural Network"", 《2020 INTERNATIONAL CONFERENCE AND UTILITY EXHIBITION ON ENERGY, ENVIRONMENT AND CLIMATE CHANGE (ICUE)》 * |
| JIAMING LIU等: ""Deep Learning Based Visualized Wind Speed Matrix Forecasting Model for Wind Power Forecasting"", 《2020 IEEE 3RD STUDENT CONFERENCE ON ELECTRICAL MACHINES AND SYSTEMS (SCEMS)》 * |
| MALTELEHNA等: ""Forecasting day-ahead electricity prices: A comparison of time series and neural network models taking external regressors into account"", 《ENERGY ECONOMICS》 * |
| 刘升伟等: ""基于改进高斯过程回归的短期负荷概率区间预测方法"", 《电力系统保护与控制》 * |
| 雷若冰等: ""基于相关性分析的风电场群风速分布预测方法"", 《电力自动化设备》 * |
| 黎静华等: ""可再生能源电力不确定性预测方法综述"", 《高电压技术》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114298444B (en) | 2022-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ke et al. | Short-term electrical load forecasting method based on stacked auto-encoding and GRU neural network | |
| Gao et al. | Interpretable deep learning model for building energy consumption prediction based on attention mechanism | |
| CN114358213B (en) | Error ablation processing method, system and medium for nonlinear time series data prediction | |
| CN116523148B (en) | Distribution network distribution transformer overload early warning method, device and equipment | |
| CN113807589B (en) | Rolling optimization method and device for energy storage of operators based on model prediction | |
| CN113011680A (en) | Power load prediction method and system | |
| CN111697560B (en) | Method and system for predicting load of power system based on LSTM | |
| CN115238948A (en) | Method and device for predicting power generation capacity of small hydropower station | |
| CN117113202A (en) | Power loop energy consumption detection method and equipment based on joint error stacking model | |
| CN115481788A (en) | Load prediction method and system for phase change energy storage system | |
| Bajatović et al. | Application of predictive models for natural gas needs-current state and future trends review | |
| Joudaki | Long-term wind speed and power forecasting based on lstm: A comprehensive study | |
| CN114611757A (en) | Electric power system short-term load prediction method based on genetic algorithm and improved depth residual error network | |
| Hu et al. | Probabilistic net load forecasting based on transformer network and Gaussian process-enabled residual modeling learning method | |
| Jovova et al. | Air pollution forecasting using CNN-LSTM deep learning model | |
| CN114298444B (en) | Wind speed probability prediction method, device, equipment and storage medium | |
| Shang et al. | Loss of life estimation of distribution transformers considering corrupted AMI data recovery and field verification | |
| Ali et al. | Deep learning networks for short term load forecasting | |
| CN117134315A (en) | Distribution transformer load prediction method and device based on BERT algorithm | |
| CN116365512A (en) | Short-term bus load prediction method based on feature selection and combined prediction model | |
| CN115952928A (en) | Short-term power load prediction method, device, equipment and storage medium | |
| CN115759343A (en) | E-LSTM-based user electric quantity prediction method and device | |
| CN116090604A (en) | Training method, prediction method and device for photovoltaic power model in future and short term | |
| CN115618995A (en) | Short-term campus water demand prediction method based on SSA-ConvLSTM-LSTM mixed model | |
| Chen et al. | A dual attention-based CNN-GRU model for short-term electric load forecasting |
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 |