CN113392972B - Photovoltaic short-term power prediction model training method, prediction method and device - Google Patents
Photovoltaic short-term power prediction model training method, prediction method and device Download PDFInfo
- Publication number
- CN113392972B CN113392972B CN202110700398.XA CN202110700398A CN113392972B CN 113392972 B CN113392972 B CN 113392972B CN 202110700398 A CN202110700398 A CN 202110700398A CN 113392972 B CN113392972 B CN 113392972B
- Authority
- CN
- China
- Prior art keywords
- sparrow
- power prediction
- term power
- prediction model
- photovoltaic
- 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
Images
Classifications
-
- 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
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a photovoltaic short-term power prediction model training method, a photovoltaic short-term power prediction model prediction method and a photovoltaic short-term power prediction model prediction device, wherein training samples and testing samples are generated through historical environmental data and historical photovoltaic power generation data, and extreme learning machines corresponding to sparrow individuals are trained through the training samples; calculating the fitness value of each sparrow individual according to a power prediction value obtained by inputting the test sample into each initial photovoltaic short-term power prediction model and historical photovoltaic power generation power data of the test sample; the finder and the follower are selected according to the fitness value, the position of the sparrow population is updated, the position of the current generation of the sparrow population is output after the first-stage optimization meets the preset condition, the second-stage parameter optimization is carried out, the global optimal position determined from the positions of all the generation of the sparrow population is finally restored to the network parameter of the extreme learning machine, the final photovoltaic short-term power prediction model is obtained, and the technical problem that in the prior art, the local optimization is easy to fall into, and the prediction accuracy is not high is solved.
Description
Technical Field
The application relates to the technical field of photovoltaic power generation, in particular to a training method, a prediction method and a prediction device of a photovoltaic short-term power prediction model.
Background
In recent years, with the rapid development of society, people have increasingly increased demand for energy, but due to the shortage of conventional energy sources such as fossil, the energy crisis is becoming serious, and the rapid consumption of fossil fuels also causes serious environmental pollution. Renewable energy is inexhaustible, environment-friendly and is a more optimal choice for power generation. Solar energy is currently considered one of the most important renewable energy sources, thus creating photovoltaic power generation systems. However, the output power of the photovoltaic power generation system has intermittency and fluctuation, and the balance problem of random power generation and random power utilization of users brings serious influence to the stable operation of the power system. Therefore, the research on the high-precision photovoltaic power generation short-term power prediction technology has very important guiding significance for the scheduling of the power system.
Since the output characteristics of the photovoltaic panel are affected by external factors, the Maximum Power Point (MPP) of the photovoltaic panel will vary depending on these external factors. In the current research field, a direct prediction manner for short-term power prediction by referring to historical data of a photovoltaic power generation system is a hot research. However, the existing prediction method is easy to fall into local optimum, so that the prediction accuracy is not high.
Disclosure of Invention
The application provides a photovoltaic short-term power prediction model training method, a photovoltaic short-term power prediction model prediction method and a photovoltaic short-term power prediction model prediction device, which are used for solving the technical problem that in the prior art, the prediction accuracy is low due to the fact that local optimization is easy to fall into.
In view of the above, a first aspect of the present application provides a photovoltaic short-term power prediction method, including:
s1, generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data;
s2, after the network structure of the extreme learning machine is determined, the positions of the sparrow population and the first iteration times t are determined1And a second number of iterations t2Initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a group of network parameters of the extreme learning machine;
s3, training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through the training samples to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual;
s4, calculating the adaptability value of each sparrow individual according to a power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample;
s5, judging the first iteration times t1Whether a first preset iteration frequency is reached or not, if not, the sparrow individual with the largest fitness value is used as a finder, the rest sparrow individuals are used as followers, the position of the current-generation sparrow population is updated based on the positions of the finder and the followers, the position of the next-generation sparrow population is obtained, and t is set1＝t1+1, and returning to the step S3, if yes, outputting the position of the current generation sparrow population, and executing the step S6;
s6, judging the second iteration times t2Whether the second preset iteration frequency is reached or not, if not, updating the position of the current generation of sparrow population according to the global optimal position to obtain the position of the next generation of sparrow population, and setting t2＝t2And +1, returning to the step S3, and if so, restoring the global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the positions of all generations of sparrows through the fitness value.
Optionally, the calculating the fitness value of each sparrow individual according to the power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and the historical photovoltaic power generation power data of the test sample includes:
inputting the test sample into the initial photovoltaic short-term power prediction models corresponding to the sparrows to obtain power prediction values of the test sample in the initial photovoltaic short-term power prediction models;
calculating the ratio of the power prediction value of the test sample in each initial photovoltaic short-term power prediction model to the historical photovoltaic power generation power data of the test sample to obtain the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model;
and calculating the average value of the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model to obtain the fitness value of the sparrow individual corresponding to each initial photovoltaic short-term power prediction model.
Optionally, the updating the location of the contemporary sparrow population based on the locations of the finder and the follower includes:
updating the position of the finder according to the position of the finder and the first preset iteration times;
sorting the positions of the followers in a descending order according to the fitness value, and updating the positions of the followers in a preset number in the front after sorting according to the position of the finder of the next iteration;
and determining a global worst position from the positions of all generations of sparrows according to the fitness value, and updating the positions of the rest followers according to the global worst position and the number of sparrows.
Optionally, the updating the position of the present generation of sparrow population according to the global optimal position to obtain the position of the next generation of sparrow population includes:
according to the global optimal position, updating the position of the current generation of sparrow population through a preset position updating formula to obtain the position of the next generation of sparrow population, wherein the preset position updating formula is as follows:
in the formula (I), the compound is shown in the specification,
is the second iteration number t2The ith sparrow individual in the contemporary sparrow population at the position of the jth dimension,
the position of the ith sparrow individual in the next generation sparrow population in the jth dimension, K is a preset coefficient, theta is a constant, XbestAnd u and v are preset parameters and are determined by a normal distribution curve for the global optimal position.
Optionally, the historical environmental data includes historical environmental temperature data, historical illumination intensity data, and historical environmental humidity data.
The second aspect of the present application provides a photovoltaic short-term power prediction method, including:
inputting historical environment data of a first preset time period before the current time into a final photovoltaic short-term power prediction model to perform photovoltaic short-term power prediction, and obtaining a photovoltaic short-term power prediction value after a second preset time period after the current time;
the final photovoltaic short-term power prediction model is obtained by training through any one of the photovoltaic short-term power prediction model training methods in the first aspect.
The third aspect of the present application provides a photovoltaic short-term power prediction model training device, including:
the generating unit is used for generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data;
an initialization unit for determining the position of the sparrow population and the first iteration number t after the network structure of the extreme learning machine is determined1And a second number of iterations t2Initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a set of network parameters of the extreme learning machine;
the training unit is used for training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through the training samples to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual;
the calculating unit is used for calculating the adaptability value of each sparrow individual according to a power predicted value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample;
a first judging unit for judging the first iteration time t1Whether a first preset iteration number is reached or not, if not, the sparrow individual with the maximum fitness value is used as a finder, the rest sparrow individuals are used as followers, and the positions of the finder and the followers are used for updating the current-generation sparrowsThe position of the population, the position of the next generation sparrow population is obtained, and t is set1＝t1+1, triggering the training unit, if yes, outputting the position of the current sparrow population, and triggering a second judgment unit;
the second judging unit is used for judging a second iteration time t2Whether the second preset iteration frequency is reached or not, if not, updating the position of the current generation of sparrow population according to the global optimal position to obtain the position of the next generation of sparrow population, and setting t2＝t2And +1, triggering the training unit, and if so, restoring a global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the positions of all generations of sparrows through the fitness value.
Optionally, the computing unit is specifically configured to:
inputting the test sample into the initial photovoltaic short-term power prediction models corresponding to the sparrow individuals to obtain power prediction values of the test sample in the initial photovoltaic short-term power prediction models;
calculating the ratio of the power predicted value of the test sample in each initial photovoltaic short-term power prediction model to the historical photovoltaic power generation power data of the test sample to obtain the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model;
and calculating the average value of the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model to obtain the fitness value of the sparrow individual corresponding to each initial photovoltaic short-term power prediction model.
Optionally, the historical environmental data includes historical environmental temperature data, historical illumination intensity data, and historical environmental humidity data.
The fourth aspect of the present application provides a photovoltaic short-term power prediction apparatus, including:
the prediction unit is used for inputting historical environmental data of a first preset time period before the current moment into the final photovoltaic short-term power prediction model to perform photovoltaic short-term power prediction, and obtaining a photovoltaic short-term power prediction value after a second preset time period after the current moment;
wherein the final photovoltaic short-term power prediction model is obtained by training through the photovoltaic short-term power prediction model training method of any one of claims 1 to 5.
According to the technical scheme, the method has the following advantages:
the application provides a photovoltaic short-term power prediction method, which comprises the following steps: s1, generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data; s2, after the network structure of the extreme learning machine is determined, the positions of the sparrow population and the first iteration times t are determined1And a second number of iterations t2Initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a group of network parameters of the extreme learning machine; s3, training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through the training samples to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual; s4, calculating the adaptability value of each sparrow individual according to a power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample; s5, judging the first iteration times t1Whether the first preset iteration frequency is reached or not, if not, the sparrow individual with the maximum fitness value is used as a finder, the rest sparrow individuals are used as followers, the position of the current-generation sparrow population is updated based on the positions of the finder and the followers, the position of the next-generation sparrow population is obtained, and t is set1＝t1+1, returning to the step S3, if yes, outputting the position of the current sparrow population, and executing the step S6; s6, judging the second iteration times t2Whether the second preset iteration frequency is reached or not, if not, updating the position of the current generation of sparrow population according to the global optimal position to obtain the position of the next generation of sparrow population, and setting t2＝t2+1, and returning to step S3, if so, restoring the global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the network parameters by the fitness valueThe positions of all generations of sparrow populations are determined.
According to the method, after the extreme learning machine is trained through a training sample, the fitness value of each sparrow individual is calculated through a testing sample, so that a finder and a follower are determined, the positions of sparrow populations are updated aiming at the finder and the follower, a stronger global search range is provided for the finder, the follower not only has the ability of learning towards the finder, but also has the ability of jumping out of local optimum search, and convergence to local optimum can be avoided; and after the optimization process of the first stage is completed, the second stage of optimization is entered, network parameter optimization is further performed, the possibility of finding the optimal position is improved, and the prediction accuracy of the final photovoltaic short-term power prediction model is further improved, so that the technical problem that the prediction accuracy is not high due to the fact that the photovoltaic short-term power prediction model is easy to fall into local optimization in the prior art is solved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.
Fig. 1 is a schematic flowchart of a photovoltaic short-term power prediction model training method according to an embodiment of the present disclosure;
fig. 2 is a schematic structural diagram of an extreme learning machine according to an embodiment of the present application.
Detailed Description
The application provides a photovoltaic short-term power prediction model training method, a photovoltaic short-term power prediction model prediction method and a photovoltaic short-term power prediction model prediction device, which are used for solving the technical problem that in the prior art, the prediction accuracy is not high due to the fact that local optimization is easy to fall into.
In order to make those skilled in the art better understand the technical solutions of the present application, 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 the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.
For ease of understanding, referring to fig. 1, an embodiment of a photovoltaic short-term power prediction method provided herein includes:
s1, generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data.
In the embodiment of the application, the historical environment data comprises historical environment temperature data, historical illumination intensity data and historical environment humidity data, the historical environment data before a certain moment is obtained and used as three neuron variables of an input layer of the extreme learning machine, and the historical photovoltaic power generation data after a preset time period (for example, after 5 minutes) after the moment is used as a single neuron variable of an output layer of the extreme learning machine, so that a data set is constructed. Then, the data set is divided according to a certain proportion to obtain a training sample and a testing sample.
It is understood that the training samples and the testing samples may be divided after the data in the data set is normalized.
S2, after the network structure of the extreme learning machine is determined, the positions of the sparrow population and the first iteration times t are determined1And a second number of iterations t2And initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a group of network parameters of the extreme learning machine.
After the network structure of the extreme learning machine is determined, network parameters are initialized, wherein the network parameters comprise the input layer neuron number n, the output layer neuron number m and the hidden layer neuron number L of the extreme learning machine.
After determining the network structure of the extreme learning machine, the first iteration number t can be determined1And a second number of iterations t2Is initialized to 0, and the first iteration count, and the second iteration count are set to,The maximum iteration times of the second iteration times are obtained to obtain a first preset iteration time itermax1And a second predetermined number of iterations itermax2And the number of the sparrow populations can be set, and the dimension d of the variable is optimized and is equal to the sum of the number of all elements in the hidden layer input weight matrix and the hidden layer offset matrix.
Initializing the positions of the sparrow population to obtain the initial positions of all the sparrow individuals, namely giving N different hidden layer input weights and offset network parameter combinations, wherein the position of each sparrow individual in the sparrow population corresponds to one group of hidden layer input weights and offset network parameters of the extreme learning machine, and different positions correspond to different network parameters. The initial position of the sparrow individual in the embodiment of the application follows the principle of random generation.
And S3, training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through the training samples to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual.
And calling the training samples to train the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual. It can be understood that the network parameters of the extreme learning machine corresponding to each sparrow individual are positions corresponding to each sparrow individual, the extreme learning machines corresponding to different sparrow individuals have the same network structure and different network parameters, and the network parameters of the extreme learning machine corresponding to each sparrow individual in the first generation sparrow population are initial positions corresponding to each sparrow individual.
When the extreme learning machine is trained through the training sample, the training sample is utilized to solve the special solution beta of the weight between the hidden layer and the output layer according to the method for solving the least square special solution of the linear system, and the detailed process is as follows:
as shown in FIG. 2, the extreme learning machine is a novel single-layer feedforward neural network model, and the structure of the extreme learning machine only comprises 1 hidden layer. The extreme learning machine is characterized in that: only the number of nodes of the hidden layer of the network needs to be set, and the connection weight of the input layer and the hidden layer and the threshold value of the hidden layer can be randomly set; the input weight of the network and the bias of a hidden layer neuron do not need to be adjusted in the training process; the connection weight beta between the hidden layer and the output layer does not need to be adjusted iteratively, and is determined once by an equation solving formula. Therefore, the extreme learning machine has the advantages of high learning speed and good generalization performance.
Suppose that N training samples are (X)j,Tj) Wherein X isj＝[xj1,xj2,…,xjn]T,Tj＝[tj1,tj2,…,tjm]TX is historical environment data comprising historical environment temperature data, historical illumination intensity data and historical environment humidity data, and t is historical photovoltaic power generation power data, namely a true value; an extreme learning machine for L hidden layer neurons can be represented as:
in the formula ojJ (x) is the j (th) actual output value output by the extreme learning machine, g (x) is the activation function, Wi＝[wi1,wi2,…,win]TIs the weight value between the ith neuron of the input layer and the hidden layer, biIs the bias value, beta, of the ith neuron in the hidden layeriThe weight value of the ith neuron of the hidden layer and the output layer.
The training purpose of the extreme learning machine is: so that the error between the actual output value and the true value approaches zero, and the error is expressed as:
and exist of betai、WiAnd biSo that:
the matrix form can be expressed as:
T＝Hβ；
where H is the output of the hidden layer node, β is the output weight of the hidden layer and the output layer, and T is the desired output, i.e.:
according to the theory of generalized inverse, solve specially for beta*Can be expressed as:
β*＝H+T；
in the formula, H+Is the Moore-Penrose generalized inverse of the output matrix H of the hidden layer.
According to the training process, the extreme learning machine completes the training process without iterative adjustment, so that the extreme learning machine has the characteristic of high learning speed.
And S4, calculating the adaptability value of each sparrow individual according to a power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample.
Although the conventional extreme learning method has better generalization performance and extremely fast training speed, any given hidden layer bias and hidden layer weight may cause some prediction results not to reach the optimal point. Therefore, in order to obtain a higher-quality prediction result, the embodiment of the application further optimizes the network parameters of the extreme learning machine.
Before introducing an optimization algorithm for optimizing the network parameters of the extreme learning machine, an evaluation index for measuring the predictive performance needs to be formulated, and an optimized objective function is determined. In this embodiment, after the training of the photovoltaic short-term power prediction model is completed, the prediction performance of the discrete test sample measurement model is called, and an evaluation index of "prediction efficiency μ" is provided to evaluate the accuracy of the photovoltaic short-term power prediction method.
Specifically, the test sample is input into an initial photovoltaic short-term power prediction model corresponding to each sparrow individual, and a power prediction value of the test sample in each initial photovoltaic short-term power prediction model is obtained; calculating the ratio of the power predicted value of the test sample in each initial photovoltaic short-term power prediction model to the historical photovoltaic power generation power data of the test sample to obtain the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model; and calculating the average value of the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model to obtain the fitness value of the sparrow individual corresponding to each initial photovoltaic short-term power prediction model.
For a single test sample, the prediction efficiency is defined as follows:
in the formula, PpreAnd inputting the power predicted value of the initial photovoltaic short-term power prediction model for the test sample, and the true power value of the test sample, namely the historical photovoltaic power generation power data of the test sample.
For a test sample consisting of a plurality of sample data, the total prediction efficiency mu is adopted as:
in the formula, muk(k＝1,2,…,Ntest) For the prediction efficiency, N, corresponding to the kth test sampletestThe total number of samples tested.
Because the network parameters of the extreme learning machine corresponding to each sparrow individual are different, the initial photovoltaic short-term power prediction models obtained after the extreme learning machine corresponding to each sparrow individual is trained by adopting the same training sample are different, and the power prediction values output by the initial photovoltaic short-term power prediction models corresponding to the sparrow individuals may be different after the test sample is input into the initial photovoltaic short-term power prediction models corresponding to the sparrow individuals. The method comprises the steps of obtaining the total prediction efficiency of the initial photovoltaic short-term power prediction model corresponding to each sparrow individual by calculating the average value of the ratio of the power prediction value of the initial photovoltaic short-term power prediction model corresponding to each sparrow individual of a test sample to historical photovoltaic power generation power data, and taking the total prediction efficiency of each initial photovoltaic short-term power prediction model as the fitness value of the corresponding sparrow individual.
S5, judging the first iteration times t1Whether the number of first preset iteration times is reached or not, if not, the sparrow individual with the largest fitness value is used as a finder, the rest sparrow individuals are used as followers, the position of the current generation sparrow population is updated based on the positions of the finder and the followers, the position of the next generation sparrow population is obtained, and t is set1＝t1+1, and returning to step S3, if yes, outputting the position of the present generation sparrow population, and executing step S6.
After the fitness value of each sparrow individual in the contemporary sparrow population is obtained through calculation, the first stage of network parameter optimization is entered. At a first number of iterations t1Does not reach the first preset iteration number itermax1And updating the position of the current-generation sparrow population based on the positions of the finder and the follower to obtain the position of the next-generation sparrow population.
Specifically, the position of the finder is updated according to the position of the finder and the first preset iteration number; sorting the positions of the followers in a descending order according to the fitness value, and updating the positions of the followers in the preset number in the front after sorting according to the position of the finder of the next iteration; and determining a global worst position from the positions of all generations of sparrow populations according to the fitness value, and updating the positions of the remaining followers according to the global worst position and the number of sparrow individuals.
Wherein, the location updating formula of the finder is as follows:
in the formula, t1To optimize the current number of iterations of the first stage, itermax1To optimize the first preset number of iterations of the first stage, α ∈ (0, 1)]Is a random number, and is a random number,
is the first iteration number t1The position of the ith sparrow individual in dimension j, here the position of the finder,
setting the numerator in the above formula to-1 for the updated discoverer location can ensure that the update rule of each location does not deviate too much.
The location update of the follower is divided into the following two cases:
after the positions of the followers are sorted in a descending order according to the fitness value, the updating formula of the followers with the fitness values in the first half is as follows:
in the formula (I), the compound is shown in the specification,
for the location of the finder in the next iteration,
for optimizing the number t of first iterations of the first stage1The ith sparrow individual of (1) at the position of the jth dimension, here the follower,
the updated position of the ith sparrow individual in the jth dimension is gamma epsilon-1, 1]Is a random number.
The updated formula for followers with fitness values ranked in the second half is:
in the formula, XworstThe global worst position, namely the position of the sparrow individual with the minimum fitness value in all generations of sparrow populations, and sigma is a random number which follows normal distribution with the mean value of 0 and the variance of 1.
After the positions of the finder and the follower are updated through the steps, the position updating of the sparrow population of the current generation is finished, the position of the sparrow population of the next generation is obtained, and t is set1＝t1+1, and returning to step S3 to enter the next iterative training.
When the first iteration time t1Reach the first preset iteration number itermax1Then, the position of the present generation of the sparrow population is output, and the process proceeds to step S6.
The embodiment of the application provides an improved sparrow search algorithm with excellent optimizing performance, so that the optimal network parameter combination is found more possibly, the efficient and rapid optimization of the network parameters of the extreme learning machine is realized, and the high-precision prediction of the short-term power of a photovoltaic power generation system is realized; in the design of the improved sparrow search algorithm, the mutual-assistance relationship between the discoverer and the follower greatly improves the overall optimizing effect of the population, and the discoverer with the highest fitness value can guide the foraging direction for other individuals; besides, followers with low fitness values can also search undeveloped optimizing areas, and the design greatly reduces the possibility that the model converges to local optimization; in the optimization process of the improved sparrow search algorithm, for followers with higher fitness values, as the followers perceive that the finder has found better food, the followers can immediately leave the current position to go to positions around the finder for deep exploration, so that the local search capability of the algorithm is greatly enhanced, and the algorithm is helped to accelerate convergence while obtaining a global optimal solution.
S6, judging the second iteration numberNumber t2Whether the second preset iteration frequency is reached or not, if not, updating the position of the current generation of sparrow population according to the global optimal position to obtain the position of the next generation of sparrow population, and setting t2＝t2+1, and returning to the step S3, if so, restoring the global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the positions of all generation sparrow populations through fitness values.
And after the position of the contemporary sparrow population is output in the first optimizing stage, the second optimizing stage is carried out. When the second iteration number t is judged2Does not reach the second preset iteration number itermax2And updating the position of the current generation of sparrow population through a preset position updating formula according to the global optimal position to obtain the position of the next generation of sparrow population, wherein the preset position updating formula is as follows:
in the formula (I), the compound is shown in the specification,
for optimizing the second iteration number t of the second stage2The ith sparrow individual in the contemporary sparrow population is at the position of the jth dimension, wherein, the second iteration number t2The position of the contemporary sparrow population when =0 is the position of the contemporary sparrow population output in the first stage of optimizing;
the position of the ith sparrow individual in the next generation sparrow population in the jth dimension is K, which is a preset coefficient and can be set according to actual conditions; θ is a constant, preferably 1.5; xbestThe position of the sparrow individual with the largest fitness value in all generations of sparrow populations (including the current generation of sparrow populations) is the global optimal position; u and v are preset parameters and are determined by a normal distribution curve, namely:
parameter sigmau、σvCan be defined as:
σv＝1；
where Γ (·) is the integral gamma function.
When the second iteration number t is judged2Reaches a second preset iteration number itermax2And then, restoring the global optimal position to the network parameters of the extreme learning machine to obtain the final photovoltaic short-term power prediction model. The embodiment of the application provides two stages of optimization processes, after the improved sparrow algorithm is used for optimization, the sparrow population enters the second stage of optimization to enhance the algorithm optimization randomness, so that the optimization efficiency is ensured, and the possibility of finding the optimal position is improved. The two-stage optimization method has stronger global search capability, is more likely to find the optimal network parameter combination, has good convergence performance and better local search capability, and ensures that the model obtains the global optimal solution and also accelerates the convergence.
In the embodiment of the application, after the extreme learning machine is trained through the training sample, the fitness value of each sparrow individual is calculated by adopting the testing sample, so that a finder and a follower are determined, the positions of sparrow populations are updated aiming at the finder and the follower, a stronger global search range is provided for the finder, the follower not only has the ability of learning towards the finder, but also has the search ability of jumping out of local optimum, and the convergence to local optimum can be avoided; and after the optimization process of the first stage is completed, the second stage of optimization is entered, network parameter optimization is further performed, the possibility of finding the optimal position is improved, and the prediction accuracy of the final photovoltaic short-term power prediction model is further improved, so that the technical problem that the prediction accuracy is not high due to the fact that the photovoltaic short-term power prediction model is easy to fall into local optimization in the prior art is solved.
The above is an embodiment of the photovoltaic short-term power prediction model training method provided by the present application, and the following is an embodiment of the photovoltaic short-term power prediction method provided by the present application.
The photovoltaic short-term power prediction method provided by the embodiment of the application comprises the following steps:
s7, inputting historical environment data of a first preset time period before the current time into a final photovoltaic short-term power prediction model to perform photovoltaic short-term power prediction, and obtaining a photovoltaic short-term power prediction value after a second preset time period after the current time; and finally, the photovoltaic short-term power prediction model is obtained by training through the photovoltaic short-term power prediction model training method in the method embodiment.
During actual photovoltaic short-term power prediction, historical environmental data of a first preset time period before the current moment are input into a final photovoltaic short-term power prediction model for power prediction, and a photovoltaic short-term power prediction value after a second preset time period (for example, after 5 minutes) after the current moment is obtained.
In the embodiment of the application, after the extreme learning machine is trained through the training sample, the fitness value of each sparrow individual is calculated by adopting the testing sample, so that a finder and a follower are determined, the positions of sparrow populations are updated aiming at the finder and the follower, a stronger global search range is provided for the finder, the follower not only has the ability of learning towards the finder, but also has the search ability of jumping out of local optimum, and the convergence to local optimum can be avoided; and after the optimization process of the first stage is completed, the second stage of optimization is entered, network parameter optimization is further performed, the possibility of finding the optimal position is improved, and the prediction accuracy of the final photovoltaic short-term power prediction model is further improved, so that the technical problem that the prediction accuracy is not high due to the fact that the photovoltaic short-term power prediction model is easy to fall into local optimization in the prior art is solved.
The photovoltaic short-term power prediction model training device provided by the application is as follows.
The embodiment of the application provides a photovoltaic short-term power prediction model training device, including:
the generating unit is used for generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data;
an initialization unit for determining the position of the sparrow population and the first iteration number t after the network structure of the extreme learning machine is determined1And a second number of iterations t2Initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a group of network parameters of the extreme learning machine;
the training unit is used for training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through a training sample to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual;
the calculating unit is used for calculating the adaptability value of each sparrow individual according to a power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample;
a first judging unit for judging the first iteration time t1Whether the first preset iteration frequency is reached or not, if not, the sparrow individual with the maximum fitness value is used as a finder, the rest sparrow individuals are used as followers, the position of the current-generation sparrow population is updated based on the positions of the finder and the followers, the position of the next-generation sparrow population is obtained, and t is set1＝t1+1, and trigger the training unit, if yes, output the position of the present sparrow population, and trigger the second judging unit;
a second judging unit for judging a second iteration number t2Whether the second preset iteration times are reached or not is judged, if not, the position of the current generation of sparrow population is updated according to the global optimal position to obtain the position of the next generation of sparrow population, and t is set2＝t2+1, triggering a training unit, if so, restoring the global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the positions of all generations of sparrows through fitness valuesAnd (4) obtaining.
As a further improvement, the computing unit is specifically configured to:
inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual to obtain a power prediction value of the test sample in each initial photovoltaic short-term power prediction model;
calculating the ratio of the power predicted value of the test sample in each initial photovoltaic short-term power prediction model to the historical photovoltaic power generation power data of the test sample to obtain the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model;
and calculating the average value of the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model to obtain the fitness value of the sparrow individual corresponding to each initial photovoltaic short-term power prediction model.
As a further refinement, the historical environmental data includes historical ambient temperature data, historical light intensity data, and historical ambient humidity data.
In the embodiment of the application, after the extreme learning machine is trained through the training sample, the fitness value of each sparrow individual is calculated by adopting the testing sample, so that a finder and a follower are determined, the positions of sparrow populations are updated aiming at the finder and the follower, a stronger global search range is provided for the finder, the follower not only has the ability of learning towards the finder, but also has the search ability of jumping out of local optimum, and the convergence to local optimum can be avoided; and after the optimization process of the first stage is completed, the second stage of optimization is entered, network parameter optimization is further performed, the possibility of finding the optimal position is improved, and the prediction accuracy of the final photovoltaic short-term power prediction model is further improved, so that the technical problem that the prediction accuracy is not high due to the fact that the photovoltaic short-term power prediction model is easy to fall into local optimization in the prior art is solved.
The embodiment of the present application further provides a photovoltaic short-term power prediction apparatus, including:
the prediction unit is used for inputting historical environment data of a first preset time period before the current time into the final photovoltaic short-term power prediction model to perform photovoltaic short-term power prediction to obtain a photovoltaic short-term power prediction value after a second preset time period after the current time;
and finally, the photovoltaic short-term power prediction model is obtained by training through the photovoltaic short-term power prediction model training method in the method embodiment.
It can be clearly understood by those skilled in the art that, for convenience and simplicity of description, the specific working processes of the above-described apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The terms "first," "second," "third," "fourth," and the like in the description of the application and the above-described figures, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be understood that in the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" for describing an association relationship of associated objects, indicating that there may be three relationships, e.g., "a and/or B" may indicate: only A, only B and both A and B are present, wherein A and B may be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b, c may be single or plural.
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.
In addition, functional units in the embodiments of the present application 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.
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.
The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit 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 (9)
1. A photovoltaic short-term power prediction model training method is characterized by comprising the following steps:
s1, generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data;
s2, after the network structure of the extreme learning machine is determined, the positions of the sparrow population and the first iteration times t are determined1And a second number of iterations t2Initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a set of network parameters of the extreme learning machine;
s3, training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through the training samples to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual;
s4, calculating the adaptability value of each sparrow individual according to a power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample;
s5, judging the first iteration times t1Whether the first preset iteration times are reached or not is judged, and if not, the sparrow individual with the largest fitness value is used as the sparrow individualA finder and the rest sparrow individuals are taken as followers, the position of the present generation sparrow population is updated based on the positions of the finder and the followers to obtain the position of the next generation sparrow population, and t is set1＝t1+1, returning to the step S3, if yes, outputting the position of the current sparrow population, and executing the step S6;
s6, judging the second iteration times t2Whether the second preset iteration frequency is reached or not, if not, updating the position of the current generation of sparrow population through a preset position updating formula according to the global optimal position to obtain the position of the next generation of sparrow population, wherein the preset position updating formula is as follows:
in the formula (I), the compound is shown in the specification,
is the second iteration number t2The ith sparrow individual in the contemporary sparrow population at the position of the jth dimension,
the position of the ith sparrow individual in the next generation sparrow population in the jth dimension, K is a preset coefficient, theta is a constant, XbestFor global optimum position, u and v are preset parameters, determined by normal distribution curve, and t is set2＝t2And +1, returning to the step S3, and if so, restoring the global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the positions of all generations of sparrows through the fitness value.
2. The training method of the photovoltaic short-term power prediction model according to claim 1, wherein the calculating of the fitness value of each sparrow individual according to the power prediction value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample comprises:
inputting the test sample into the initial photovoltaic short-term power prediction models corresponding to the sparrows to obtain power prediction values of the test sample in the initial photovoltaic short-term power prediction models;
calculating the ratio of the power prediction value of the test sample in each initial photovoltaic short-term power prediction model to the historical photovoltaic power generation power data of the test sample to obtain the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model;
and calculating the average value of the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model to obtain the fitness value of the sparrow individual corresponding to each initial photovoltaic short-term power prediction model.
3. The photovoltaic short-term power prediction model training method according to claim 1, wherein the updating the location of the contemporary sparrow population based on the locations of the finder and the follower comprises:
updating the position of the finder according to the position of the finder and the first preset iteration number;
sorting the positions of the followers in a descending order according to the fitness value, and updating the positions of the sequenced followers in front of a preset number according to the position of the finder of the next iteration;
and determining a global worst position from the positions of all generations of sparrow populations according to the fitness value, and updating the positions of the remaining followers according to the global worst position and the number of sparrow individuals.
4. The photovoltaic short-term power prediction model training method of claim 1, wherein the historical environmental data comprises historical ambient temperature data, historical light intensity data, and historical ambient humidity data.
5. A photovoltaic short-term power prediction method is characterized by comprising the following steps:
inputting historical environmental data of a first preset time period before the current moment into a final photovoltaic short-term power prediction model to perform photovoltaic short-term power prediction, and obtaining a photovoltaic short-term power prediction value after a second preset time period after the current moment;
wherein the final photovoltaic short-term power prediction model is obtained by training through the photovoltaic short-term power prediction model training method of any one of claims 1 to 4.
6. A photovoltaic short-term power prediction model training device is characterized by comprising:
the generating unit is used for generating a training sample and a testing sample according to the acquired historical environmental data and historical photovoltaic power generation data;
an initialization unit for determining the position of the sparrow population and the first iteration number t after the network structure of the extreme learning machine is determined1And a second number of iterations t2Initializing, wherein the position of each sparrow individual in the sparrow population corresponds to a group of network parameters of the extreme learning machine;
the training unit is used for training the extreme learning machine corresponding to each sparrow individual in the contemporary sparrow population through the training samples to obtain an initial photovoltaic short-term power prediction model corresponding to each sparrow individual;
the calculating unit is used for calculating the adaptability value of each sparrow individual according to a power predicted value obtained by inputting the test sample into the initial photovoltaic short-term power prediction model corresponding to each sparrow individual and historical photovoltaic power generation power data of the test sample;
a first judging unit for judging the first iteration time t1Whether a first preset iteration frequency is reached or not, if not, the sparrow individual with the largest fitness value is used as a finder, the rest sparrow individuals are used as followers, the position of the current generation sparrow population is updated based on the positions of the finder and the followers to obtain the position of the next generation sparrow population, and t is set1＝t1+1, and triggerIf so, the training unit outputs the position of the contemporary sparrow population, triggering a second judgment unit;
the second judging unit is used for judging a second iteration time t2Whether the second preset iteration frequency is reached or not, if not, updating the position of the current generation of sparrow population through a preset position updating formula according to the global optimal position to obtain the position of the next generation of sparrow population, wherein the preset position updating formula is as follows:
in the formula (I), the compound is shown in the specification,
is the second iteration number t2The ith sparrow individual in the contemporary sparrow population at the position of the jth dimension,
the position of the ith sparrow individual in the next generation sparrow population in the jth dimension, K is a preset coefficient, theta is a constant, XbestFor global optimum position, u and v are preset parameters, determined by normal distribution curve, and t is set2＝t2And +1, triggering the training unit, and if so, restoring the global optimal position to the network parameters of the extreme learning machine to obtain a final photovoltaic short-term power prediction model, wherein the global optimal position is determined from the positions of all generation sparrow populations through the fitness value.
7. The photovoltaic short-term power prediction model training device of claim 6, wherein the computing unit is specifically configured to:
inputting the test sample into the initial photovoltaic short-term power prediction models corresponding to the sparrow individuals to obtain power prediction values of the test sample in the initial photovoltaic short-term power prediction models;
calculating the ratio of the power prediction value of the test sample in each initial photovoltaic short-term power prediction model to the historical photovoltaic power generation power data of the test sample to obtain the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model;
calculating the average value of the prediction efficiency of the test sample in each initial photovoltaic short-term power prediction model to obtain the fitness value of the sparrow individual corresponding to each initial photovoltaic short-term power prediction model.
8. The photovoltaic short-term power prediction model training device of claim 6, wherein the historical environmental data comprises historical ambient temperature data, historical light intensity data, and historical ambient humidity data.
9. A photovoltaic short-term power prediction apparatus, comprising:
the prediction unit is used for inputting historical environment data of a first preset time period before the current time into the final photovoltaic short-term power prediction model to perform photovoltaic short-term power prediction to obtain a photovoltaic short-term power prediction value after a second preset time period after the current time;
wherein the final photovoltaic short-term power prediction model is obtained by training through the photovoltaic short-term power prediction model training method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110700398.XA CN113392972B (en) | 2021-06-23 | 2021-06-23 | Photovoltaic short-term power prediction model training method, prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110700398.XA CN113392972B (en) | 2021-06-23 | 2021-06-23 | Photovoltaic short-term power prediction model training method, prediction method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113392972A CN113392972A (en) | 2021-09-14 |
| CN113392972B true CN113392972B (en) | 2022-11-01 |
Family
ID=77623637
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110700398.XA Active CN113392972B (en) | 2021-06-23 | 2021-06-23 | Photovoltaic short-term power prediction model training method, prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113392972B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115238860B (en) * | 2022-06-22 | 2024-01-23 | 中国石油天然气集团有限公司 | Leakage pressure prediction model generation method and device |
| CN116834003B (en) * | 2023-06-29 | 2024-01-02 | 广州市创博机电设备安装有限公司 | Intelligent installation method and system of photovoltaic module |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109902874A (en) * | 2019-02-28 | 2019-06-18 | 武汉大学 | A kind of micro-capacitance sensor photovoltaic power generation short term prediction method based on deep learning |
| CN111950564A (en) * | 2020-07-23 | 2020-11-17 | 江苏大学 | Pork freshness detection and classification method based on SSA-ELM algorithm |
| CN112329934A (en) * | 2020-11-17 | 2021-02-05 | 江苏科技大学 | RBF neural network optimization algorithm based on improved sparrow search algorithm |
| CN112989693A (en) * | 2021-03-02 | 2021-06-18 | 上海电机学院 | Wind power prediction method based on SSA-GRU-MSAR |
| CN112884238A (en) * | 2021-03-12 | 2021-06-01 | 国网冀北电力有限公司电力科学研究院 | Photovoltaic power generation power prediction method and device |
-
2021
- 2021-06-23 CN CN202110700398.XA patent/CN113392972B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113392972A (en) | 2021-09-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113392972B (en) | Photovoltaic short-term power prediction model training method, prediction method and device | |
| CN109978283B (en) | Photovoltaic power generation power prediction method based on branch evolution neural network | |
| CN108053077A (en) | A kind of short-term wind speed forecasting method and system based on two type T-S fuzzy models of section | |
| CN112149883A (en) | Photovoltaic power prediction method based on FWA-BP neural network | |
| CN114282646B (en) | Optical power prediction method and system based on two-stage feature extraction and BiLSTM improvement | |
| CN113468817A (en) | Ultra-short-term wind power prediction method based on IGOA (optimized El-electric field model) | |
| CN114511132A (en) | Photovoltaic output short-term prediction method and prediction system | |
| CN111355633A (en) | Mobile phone internet traffic prediction method in competition venue based on PSO-DELM algorithm | |
| CN109657147A (en) | Microblogging abnormal user detection method based on firefly and weighting extreme learning machine | |
| CN114118596A (en) | Photovoltaic power generation capacity prediction method and device | |
| CN114298377A (en) | Photovoltaic power generation prediction method based on improved extreme learning machine | |
| CN116341717A (en) | Wind speed prediction method based on error compensation | |
| CN115470987A (en) | Short-term photovoltaic power generation prediction method based on improved long-term and short-term memory neural network | |
| CN112116171A (en) | Novel photovoltaic power generation power prediction method based on neural network | |
| Guo et al. | Research on short-term forecasting method of photovoltaic power generation based on clustering SO-GRU method | |
| CN112381271B (en) | Distributed multi-objective optimization acceleration method for rapidly resisting deep belief network | |
| CN113435595A (en) | Two-stage optimization method for extreme learning machine network parameters based on natural evolution strategy | |
| CN117458480A (en) | Photovoltaic power generation power short-term prediction method and system based on improved LOF | |
| CN117374941A (en) | Photovoltaic power generation power prediction method based on neural network | |
| CN116681154A (en) | Photovoltaic power calculation method based on EMD-AO-DELM | |
| CN116822331A (en) | Multi-energy fusion wind-solar power generation prediction method | |
| CN110059871A (en) | Photovoltaic power generation power prediction method | |
| CN115238952A (en) | Bi-LSTM-Attention short-term power load prediction method | |
| CN116415177A (en) | Classifier parameter identification method based on extreme learning machine | |
| CN115579858A (en) | Photovoltaic power generation power prediction method based on time-sharing gating circulation unit |
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 |