New energy generation power prediction analysis method and system influenced by multiple factors
By analyzing historical data and neural network models, identifying key factors, preprocessing and training, the problem of increasing difficulty in predicting power generation of new energy is solved, improving prediction accuracy and stability, and is suitable for complex power generation rules.
Patent Information
- Application Number
- CN202410868264.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-01
- Publication Date
- 2024-11-12
AI Technical Summary
The difficulty of predicting power generation power in new energy has increased, and traditional methods have large average absolute errors, so the expected prediction effect cannot be achieved.
The new energy power generation power prediction and analysis method is adopted with a variety of factors, including obtaining historical power generation power data, analyzing the impact factors using neural networks, establishing a prediction database, preprocessing and training through interpolation, correlation analysis and neuron activation functions, and output Final prediction results.
It improves prediction accuracy and model stability, identifies key factors, avoids the influence of redundant information, has strong nonlinear modeling capabilities, is suitable for complex power generation rules, optimizes power system operation and new energy utilization.
Smart Images

Figure CN118944043A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of renewable energy power generation prediction, and in particular to a method and system for predicting and analyzing renewable energy power generation influenced by multiple factors. Background Art
[0002] The power generation of renewable energy is affected by many factors, including weather conditions (such as wind speed, solar radiation intensity, temperature, humidity, etc.), geographical location, seasonal changes, equipment status, etc. Traditional statistical methods may be unable to cope with such high-dimensional, nonlinear data. In contrast, artificial neural networks (ANNs) can automatically extract complex features from input data and establish a mapping relationship from input (such as historical power data and meteorological parameters) to output (future power generation forecast) through nonlinear transformation of multi-layer structures. Common neural network models include but are not limited to BP neural networks, recursive neural networks (RNNs), long short-term memory networks (LSTMs), convolutional neural networks (CNNs), and their combined variants. BP neural network prediction model: In early applications, neural network models based on error back propagation (BP) were widely used in wind power and photovoltaic power generation prediction. This model minimizes the prediction error by adjusting the network weights and can handle multivariate inputs, but it has high requirements on the quality and quantity of training data and is prone to fall into local optimality. LSTM network model: With the development of deep learning, long short-term memory networks (LSTMs) have been favored for their superior performance in time series prediction. LSTM solves the gradient vanishing and gradient exploding problems in RNN through the gating mechanism. It is suitable for capturing long-term dependencies and is particularly suitable for short-term and ultra-short-term predictions of wind power and photovoltaic power. CNN-LSTM hybrid model: Considering that meteorological data such as wind speed and radiation intensity have spatial and temporal correlations, a model combining convolutional neural network (CNN) and LSTM is proposed. CNN is used to extract the spatial features of input data, and LSTM is used to process time series information. The combination of the two improves the prediction accuracy. Ensemble learning method: In order to further improve the prediction performance, some studies adopt an ensemble learning strategy to combine the prediction results of multiple neural network models (such as different structures or different initialization parameters) and obtain the final prediction value by voting or weighted averaging to improve the generalization ability of the model. In summary, the prediction technology of renewable energy power generation based on neural networks continues to evolve. By combining different types of neural network structures, ensemble learning strategies and advanced optimization algorithms, it continues to promote the improvement of prediction accuracy to meet the growing demand for accurate prediction in the power system.
[0003] In recent years, the renewable energy power generation industry has been developing under the support of technology and national policies. The number of power stations has been increasing, and the installed capacity has been expanding, which has increased the difficulty of power generation prediction to a certain extent. Due to the late start of domestic research on this aspect, the existing prediction theory is not mature enough, and the prediction technology level is still far behind that of developed countries. Although the power generation prediction problem has received attention and attention in the research field in recent years, relevant scholars and experts have carried out a series of studies and proposed some prediction methods and ideas, but most of the new energy photovoltaic power stations still use traditional prediction methods. There is a large average absolute error in the actual prediction, and the expected prediction effect cannot be achieved. We continue to explore a new set of prediction methods. For this reason, this research project is proposed to provide a reference for the prediction of renewable energy power generation. Summary of the invention
[0004] In view of the above-mentioned problems, the present invention is proposed.
[0005] Therefore, the technical problem solved by the present invention is: the present invention solves the problem that the number of power stations is increasing and the installed capacity is also expanding, which increases the difficulty of power generation prediction to a certain extent. At present, most new energy photovoltaic power stations still use traditional prediction methods, and there is a large average absolute error in actual prediction, and the expected prediction effect cannot be achieved.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0007] In a first aspect, the present invention provides a method for predicting and analyzing the power generation of renewable energy affected by multiple factors, including:
[0008] Acquire historical power generation, analyze the historical power generation by using a neural network, obtain influencing factors of the historical power generation, and establish a power generation prediction database based on the relationship between the historical power generation and the influencing factors;
[0009] Preprocessing the original data in the power generation prediction database by using an interpolation method to obtain a first power generation prediction database;
[0010] Using correlation analysis to calculate the correlation of the variable factors of the first power generation prediction database, and selecting the variable factors with non-zero correlation as prediction data;
[0011] A prediction model is established according to a neural network, the prediction data is normalized to obtain first prediction data, the first prediction data is input into the prediction model, the first prediction data is trained using a neuron activation function, a prediction value of the generated power is output, the prediction value is denormalized, and a final prediction result is obtained.
[0012] As a preferred solution of the method for predicting and analyzing the power generation of new energy affected by multiple factors described in the present invention, the power generation prediction database includes:
[0013] The power generation is taken as the target variable and the influencing factors are taken as independent variables to construct a decision tree model;
[0014] According to the importance index of the influencing factors output by the decision tree model, the influence of the influencing factors on the power generation is judged;
[0015] The first judgment includes: if the importance index value of the influencing factor is greater than a first threshold, the influencing factor is included in the power generation prediction database; if the importance index of the influencing factor is less than the first threshold, the influencing factor is eliminated;
[0016] A second judgment is performed on whether the importance index value is greater than the first threshold value. If the importance index value of the impact factor is greater than the second threshold value, the impact factor is a key impact factor.
[0017] As a preferred solution of the method for predicting and analyzing the power generation of renewable energy affected by multiple factors described in the present invention, the preprocessing using the interpolation method includes:
[0018] Traverse the electric power prediction database, mark the location information of all missing values, and for each missing value location, find the valid data points before and after;
[0019] Set the time interval threshold. If the time interval between adjacent valid data points is less than the first time interval threshold, the zero-order difference quotient of the previous and next valid data points is calculated as the average value of the data in the time series interval divided by the value of the next moment of the missing data minus the value of the previous moment of the missing data, and the missing part is replaced by the mean value.
[0020] If the time interval between adjacent valid data points is greater than the first time interval threshold, the missing value is filled twice using linear interpolation, valid data points before and after the missing position are found, and the interpolation value of the missing value is calculated as the value of the next moment minus the value of the previous moment plus the value of the previous moment multiplied by the missing time minus the previous moment divided by the next moment minus the previous moment, and the result of linear interpolation is used to fill the missing value;
[0021] The original data minus the minimum value of the original data is divided by the maximum value of the original data minus the minimum value to obtain normalized data, thereby obtaining a first power generation prediction database.
[0022] As a preferred solution of the method for predicting and analyzing the power generation of renewable energy affected by multiple factors described in the present invention, the method of calculating the correlation of the variable factors of the first power generation prediction database by using correlation analysis includes:
[0023] Performing data conversion on the original data of the first power generation prediction database, subtracting the normalized data from the average value of the original data and dividing it by the variance of the original data to obtain the data after mean normalization processing;
[0024] Conduct correlation analysis on the normalized data;
[0025] If the correlation coefficient is zero, the variable factor is not included in the predictor cohort; if the correlation coefficient is positive or negative, the variable factor is included in the predictor cohort.
[0026] As a preferred solution of the method for predicting and analyzing the power generation of renewable energy affected by multiple factors described in the present invention, the prediction model established according to the neural network includes:
[0027] Calculate the correlation between the input variable and the target variable, set the correlation coefficient threshold, and if the correlation coefficient between the input variable and the target variable is greater than the correlation coefficient threshold, then include the variable in the prediction model input;
[0028] The number of neurons in the hidden layer is the number of samples in the training set divided by the number of neurons in the input layer plus the number of output neurons multiplied by the value of any variable.
[0029] As a preferred solution of the method for predicting and analyzing the power generation of renewable energy affected by multiple factors described in the present invention, the training of the first prediction data using a neuron activation function includes:
[0030] The first prediction data is input from the input layer, data arrangement and mapping processing are performed in the input layer, the first prediction data is input to the next functional layer, and the first prediction data is trained using a neuron activation function in the hidden layer;
[0031] Multiply each variable in the input layer by the corresponding weight to get the weighted sum;
[0032] The weighted sum is added to the bias value to get the total input of the neuron;
[0033] The total input is transformed nonlinearly through the activation function to obtain the activation value of the neuron.
[0034] As a preferred solution of the method for predicting and analyzing the power generation of renewable energy affected by multiple factors described in the present invention, the predicted value is denormalized including:
[0035] The maximum value of the output variable minus the minimum value of the output variable multiplied by the predicted output plus the minimum value of the output variable is the output variable, which is the final prediction result.
[0036] In a second aspect, the present invention provides a system for predicting and analyzing the power generation of renewable energy affected by multiple factors, including:
[0037] A database establishment module is used to obtain historical power generation, analyze the historical power generation using a neural network, obtain the influencing factors of the historical power generation, and establish a power generation prediction database based on the relationship between the historical power generation and the influencing factors;
[0038] A preprocessing module, used for preprocessing the original data in the power generation prediction database by using an interpolation method to obtain a first power generation prediction database;
[0039] A correlation analysis module, used to perform correlation calculation on the variable factors of the first power generation prediction database by using correlation analysis, and select the variable factors with non-zero correlation as prediction data;
[0040] The training module is used to establish a prediction model based on a neural network, normalize the prediction data to obtain first prediction data, input the first prediction data into the prediction model, train the first prediction data using a neuron activation function, output a prediction value of the generated power, denormalize the prediction value, and obtain a final prediction result.
[0041] In a third aspect, the present invention provides a computing device, comprising:
[0042] Memory and processor;
[0043] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the new energy power generation power prediction and analysis method affected by the multiple factors are implemented.
[0044] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of a method for predicting and analyzing the power generation of new energy influenced by the multiple factors.
[0045] Beneficial effects of the present invention: The method and system for predicting and analyzing renewable energy power generation affected by multiple factors proposed in the present invention make full use of historical data, identify key factors affecting power generation, adopt normalization and denormalization to improve the stability of model training and the interpretability of prediction results, select important variables through correlation analysis, avoid the influence of redundant information on model performance, use neural network models with strong nonlinear modeling capabilities, and can capture complex power generation laws. The renewable energy power generation prediction technology based on neural networks has shown obvious advantages in improving prediction accuracy, optimizing power system operation and promoting efficient utilization of renewable energy due to its high flexibility, adaptability and strong modeling capabilities for complex relationships. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0047] Figure 1 The present invention provides an overall flow chart of the method for predicting and analyzing the power generation of renewable energy influenced by multiple factors. DETAILED DESCRIPTION
[0048] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0049] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0050] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0051] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0052] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0053] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0054] Example 1
[0055] Reference Figure 1 , which is an embodiment of the present invention, provides a method for predicting and analyzing the power generation of renewable energy affected by multiple factors, including:
[0056] S100: Obtain historical power generation, analyze the historical power generation using a neural network, obtain influencing factors of the historical power generation, and establish a power generation prediction database based on the relationship between the historical power generation and the influencing factors;
[0057] In the embodiment of the present application, the power generation is used as the target variable and the influencing factor is used as the independent variable to construct a decision tree model;
[0058] According to the importance index of the influencing factors output by the decision tree model, the influence of the influencing factors on the power generation is judged;
[0059] The first judgment includes: if the importance index value of the influencing factor is greater than a first threshold, the influencing factor is included in the power generation prediction database; if the importance index of the influencing factor is less than the first threshold, the influencing factor is eliminated;
[0060] A second judgment is performed on whether the importance index value is greater than the first threshold value. If the importance index value of the impact factor is greater than the second threshold value, the impact factor is a key impact factor.
[0061] Specifically, based on the historical power generation and its influencing factors, the data include but are not limited to temperature, humidity, wind speed, sunshine time, etc., and the influencing factors are used as independent variables. The data set of historical power generation is divided into a training set and a test set. The training set data is used to build a decision tree model, train the decision tree, learn the patterns in the data, and use the test set to evaluate the accuracy and generalization ability of the model. The first threshold is determined as the screening image factor, and the second threshold is determined to identify the key image factors to build a power generation prediction database. The determination of the first threshold and the second threshold is determined according to the actual application scenario.
[0062] Exemplarily, historical power generation and three influencing factors including temperature, humidity and wind are obtained, the data set is divided, a decision tree is constructed, the model is trained and evaluated, and the model outputs feature importance index values including temperature: 0.4, humidity: 0.2, and wind: 0.3; the first threshold is set to 0.25, and the second threshold is set to 0.35, and a first judgment is made on the temperature importance index value. If both the temperature importance index value and the wind importance index value are greater than the first threshold, they are included in the power generation prediction database; if the humidity importance index value is less than the first threshold, they are eliminated; a second judgment is made on the temperature importance index value and the wind importance index value greater than the first threshold. If the temperature importance index value is greater than the second threshold, it is a key influencing factor, and if the wind importance index value is less than the second threshold, it is a location key influencing factor; temperature and wind are included in the power generation prediction database, wherein temperature is a key influencing factor to construct a power generation prediction database.
[0063] It should be noted that when using neural network technology for power prediction, the neural network has a certain learning ability. It mainly analyzes historical power generation data to explore the relationship between power generation and influencing factors. Therefore, neural network prediction analysis is carried out and a power generation prediction database is established. The database saves recent photovoltaic power generation data. By identifying key influencing factors, the power generation can be predicted more accurately, thereby improving the accuracy of the prediction model. Eliminating unimportant influencing factors can simplify the data set, reduce the complexity of the model, and avoid overfitting.
[0064] S200: preprocessing the original data in the power generation prediction database by using an interpolation method to obtain a first power generation prediction database;
[0065] In the embodiment of the present application, the electric power prediction database is traversed, the location information of all missing values is marked, and for each missing value location, the previous and next valid data points are searched;
[0066] Set the time interval threshold. If the time interval between adjacent valid data points is less than the first time interval threshold, the zero-order difference quotient of the previous and next valid data points is calculated as the average value of the data in the time series interval divided by the value of the next moment of the missing data minus the value of the previous moment of the missing data, and the missing part is replaced by the mean value.
[0067] If the time interval between adjacent valid data points is greater than the first time interval threshold, the missing value is filled twice using linear interpolation, valid data points before and after the missing position are found, and the interpolation value of the missing value is calculated as the value of the next moment minus the value of the previous moment plus the value of the previous moment multiplied by the missing time minus the previous moment divided by the next moment minus the previous moment, and the result of linear interpolation is used to fill the missing value;
[0068] The original data minus the minimum value of the original data is divided by the maximum value of the original data minus the minimum value to obtain normalized data, thereby obtaining a first power generation prediction database.
[0069] Specifically, the first time interval threshold is used to determine whether to use the mean of the zero-order difference quotient for filling or linear interpolation filling. The setting of the first time interval threshold is determined according to the application scenario. A data set whose zero-order difference quotient is greater than zero is defined as a missing data set, and the missing part of the data set is replaced by the mean, which is expressed as:
[0070]
[0071] Among them, s(x,y) is the zero-order difference quotient of the data set within a time series interval, x and y represent the t-1 time data and t+1 value of the missing data at time t, respectively, and h(y)-h(x) is the average value of the time series interval data;
[0072] The data is normalized as follows:
[0073]
[0074] Among them, w is the original data in the database, k is the normalized data, and e max and e min are the maximum and minimum values of the original data.
[0075] Exemplarily, assume that the data of 2024-06-02 is missing, and the valid data points before and after are 100MW of 2024-06-01 and 120MW of 2024-06-03 respectively. Set the first time interval threshold to 1 day. Since the time interval is 1 day, which is less than the threshold, the zero-order difference quotient is used for filling. The calculated zero-order difference quotient is 20, and the missing value is 400. The original data range is 100 to 120. Normalization is performed to obtain a normalized missing value of 3. The database is updated, and the electric power value of 2024-06-02 is updated to the normalized 3 to obtain the first power generation power prediction database.
[0076] S300: Calculating the correlation of variable factors in the first power generation prediction database by using correlation analysis, and selecting variable factors with non-zero correlation as prediction data;
[0077] In the embodiment of the present application, data conversion is performed on the original data of the first power generation prediction database, and the data after mean normalization processing is obtained by subtracting the normalized data from the average value of the original data and dividing it by the variance of the original data;
[0078] Conduct correlation analysis on the normalized data;
[0079] If the correlation coefficient is zero, the variable factor is not included in the predictor queue; if the correlation coefficient is positive or negative, the variable factor is included in the predictor queue.
[0080] Specifically, there are many types of variable factors in the first power generation prediction database. If all variable factors are predicted and analyzed, prediction errors are likely to occur and the amount of calculation is increased. Therefore, it is necessary to select appropriate variable factors from the database as prediction factors. The new energy power generation data in the database have a certain correlation. The higher the correlation, the greater the prediction value of power generation. Therefore, power generation prediction factors are selected through correlation analysis. Before data correlation analysis, the data is converted and expressed as:
[0081]
[0082] Among them, g is the data after mean normalization, u is the average value of the original data, and ε is the variance of the original data set.
[0083] The normalized processing is subjected to correlation analysis to determine the linear relationship between two or more correlated variable elements, which is calculated using the Person correlation coefficient method and expressed as:
[0084]
[0085] Among them, X i and Y i are the values of the ith observation X and Y, respectively. and are the means of X and Y respectively, and M is the total number of observations; the value range of the correlation coefficient is between -1 and 1, where 1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no linear correlation. The variable factor data with non-zero correlation is selected as the predicted data.
[0086] Specifically, the present invention preferably uses person correlation analysis, and may also use principal component analysis, factor analysis, and cluster analysis.
[0087] It should be noted that principal component analysis cannot capture nonlinear relationships and is prone to losing important information. Factor analysis is not applicable to all data sets. Cluster analysis relies on domain knowledge and is sensitive to selection. Person correlation analysis is chosen instead of these methods because it is more suitable for identifying and quantifying linear relationships between variables, which is necessary for selecting variables that are highly correlated with power generation prediction. New energy power generation data has a high degree of nonlinear relationships and multivariate dependence. The Person correlation coefficient method is suitable for evaluating the linear relationship between two variables, which is crucial for identifying the key factors affecting power generation. The predictive factors selected by Person correlation analysis can improve the comprehensiveness and accuracy of the prediction and can effectively identify variables that are valuable for power generation prediction.
[0088] S400: establishing a prediction model according to a neural network, normalizing the prediction data to obtain first prediction data, inputting the first prediction data into the prediction model, training the first prediction data using a neuron activation function, outputting a prediction value of the generated power, and denormalizing the prediction value to obtain a final prediction result;
[0089] In the embodiment of the present application, the correlation between the input variable and the target variable is calculated, and a correlation coefficient threshold is set. If the correlation coefficient between the input variable and the target variable is greater than the correlation coefficient threshold, the variable is included in the prediction model input;
[0090] The number of neurons in the hidden layer is the number of samples in the training set divided by the number of neurons in the input layer plus the number of output neurons multiplied by the value of any variable.
[0091] Specifically, a prediction model is established using neural network technology, and the model is used to deeply mine the correlation data between variable factors. The model is mainly divided into three parts: hidden layer, input layer and output layer. The number of neurons is determined as:
[0092]
[0093] Among them, N i is the number of neurons in the input layer, N0 is the number of neurons in the output layer, and N s is the number of samples in the training set, and α is an arbitrary variable value, usually 2 to 10.
[0094] The correlation coefficient threshold is adjusted according to the specific conditions of the data set;
[0095] Exemplarily, the target variable is the power generation of renewable energy, the input variables are solar radiation intensity, wind speed, temperature and humidity, the number of training set samples is 1000 samples, the number of input layer neurons is 4, the number of output layer neurons is 1, and the value of any variable is 5; collect historical data of 1000 samples, including the solar radiation intensity, wind speed, temperature and humidity of each sample, and the corresponding power generation, set the correlation coefficient threshold to 0.5, if the correlation coefficients of solar radiation intensity and wind speed are 0.6 and 0.55 respectively, and the correlation coefficients of temperature and humidity are 0.4 and 0.45 respectively, and the correlation coefficients of solar radiation intensity and wind speed are greater than the correlation coefficient threshold, select solar radiation intensity and wind speed as the input variables of the model;
[0096] The number of neurons in the hidden layer was determined, and the numerical values were substituted to obtain 1666 neurons in the hidden layer. A neural network model was constructed based on an input layer with 2 neurons corresponding to the solar radiation intensity and wind speed, a hidden layer with 1666 neurons, and an output layer with 1 neuron corresponding to the predicted power generation.
[0097] In the embodiment of the present application, the first prediction data is input from the input layer, data arrangement and mapping processing are performed in the input layer, the first prediction data is input to the next functional layer, and the first prediction data is trained using a neuron activation function in the hidden layer;
[0098] Multiply each variable in the input layer by the corresponding weight to get the weighted sum;
[0099] The weighted sum is added to the bias value to get the total input of the neuron;
[0100] The total input is transformed nonlinearly through the activation function to obtain the activation value of the neuron;
[0101] The maximum value of the output variable minus the minimum value of the output variable multiplied by the predicted output plus the minimum value of the output variable is the output variable, which is the final prediction result.
[0102] Specifically, the data is input from the input layer, sorted and mapped in the input layer, and then input to the next functional layer. The data is trained using the neuron activation function in the hidden layer as follows:
[0103]
[0104] Among them, A is the activation value, d(·) is the activation function, W is the weight coefficient, R is the input and output variables of the input layer, and μ is the offset value.
[0105] The output variable is input to the output layer, and the output variable value is denormalized. In order to facilitate the calculation of the prediction model, the data corresponding to the variable factors of all parameter predictions in the model are normalized values. Therefore, the predicted value calculated above is dimensionless data. Therefore, the denormalization of the data in the output layer is expressed as:
[0106] V=A(max-min)+min
[0107] Among them, max and min are the maximum and minimum values of the output variable;
[0108] By inverse normalization, the predicted value A is converted into dimensional data, thereby completing the power generation prediction.
[0109] It should be noted that correlation analysis and setting correlation coefficient thresholds can help improve prediction accuracy. The method for determining the number of hidden layer neurons provides a method for dynamically adjusting the complexity of the hidden layer according to the size of the training set and the number of neurons in the input and output layers, which helps to avoid overfitting or underfitting of the model. By selecting input variables that are highly correlated with the target variables, the generalization ability of the model is enhanced and can better adapt to new or unknown data.
[0110] The above is a schematic scheme of the method for predicting and analyzing the power generation of renewable energy under the influence of multiple factors of this embodiment. It should be noted that the technical scheme of the device for predicting and analyzing the power generation of renewable energy under the influence of multiple factors and the technical scheme of the method for predicting and analyzing the power generation of renewable energy under the influence of multiple factors mentioned above belong to the same concept. The details of the technical scheme of the device for predicting and analyzing the power generation of renewable energy under the influence of multiple factors in this embodiment that are not described in detail can all be referred to the description of the technical scheme of the method for predicting and analyzing the power generation of renewable energy under the influence of multiple factors mentioned above.
[0111] The device for predicting and analyzing the power generation of renewable energy affected by various factors in this embodiment includes:
[0112] A database establishment module is used to obtain historical power generation, analyze the historical power generation using a neural network, obtain the influencing factors of the historical power generation, and establish a power generation prediction database based on the relationship between the historical power generation and the influencing factors;
[0113] A preprocessing module, used for preprocessing the original data in the power generation prediction database by using an interpolation method to obtain a first power generation prediction database;
[0114] A correlation analysis module, used to perform correlation calculation on the variable factors of the first power generation prediction database by using correlation analysis, and select the variable factors with non-zero correlation as prediction data;
[0115] The training module is used to establish a prediction model based on a neural network, normalize the prediction data to obtain first prediction data, input the first prediction data into the prediction model, train the first prediction data using a neuron activation function, output a prediction value of the generated power, denormalize the prediction value, and obtain a final prediction result.
[0116] This embodiment also provides a computing device suitable for predicting and analyzing the power generation of renewable energy affected by various factors, including:
[0117] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the new energy power generation power prediction and analysis method affected by multiple factors as proposed in the above embodiment.
[0118] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for predicting and analyzing the power generation of new energy influenced by multiple factors as proposed in the above embodiment is implemented.
[0119] The storage medium proposed in this embodiment and the new energy power generation power prediction and analysis method proposed in the above embodiment that realizes the influence of multiple factors belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.
[0120] Through the above description of the implementation method, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccessMemory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.
[0121] Example 2
[0122] Referring to Table 1, which is an embodiment of the present invention, a method for predicting and analyzing the power generation of renewable energy affected by multiple factors is provided to verify and illustrate the technical effects used in the method.
[0123] A new energy power station is taken as the experimental object. The installed capacity of the power station is 150MW. The historical power generation data of the power station in the past year is selected, and the data is 20,000 groups. The power generation of the power station is predicted by the method of the present invention, and two traditional methods are selected for comparison. The two traditional methods are based on convolutional neural network and cluster analysis, which are represented by traditional method 1 and traditional method 2 below.
[0124] According to the actual situation of the photovoltaic power station, a neural network model is established. The data is repaired and normalized according to the above process. The prediction factors are determined to be solar irradiance, weather, ambient temperature and relative humidity. The prediction data is iterated to calculate the power generation prediction value. The mean absolute error can reflect the accuracy of the prediction results of the new energy power generation, so it is used as the accuracy evaluation index of the prediction method, expressed as:
[0125]
[0126] Among them, F is the mean absolute error of the power generation prediction result, H is the number of test data samples, S is the output value of the neural network model, K is the actual power generation value of the power station, and the mean absolute error value ranges from 0 to 1. The larger the value, the lower the prediction accuracy. The experiment uses the number of test data samples as a variable, and the specific data are shown in Table 1.
[0127] Table 1 Comparison of prediction accuracy between traditional method 1, traditional method 2 and the present invention
[0128] Data sample number / group The method of the present invention / % Traditional method 1# / % Traditional method 2# / % 1000 0.01 0.35 0.31 2000 0.03 0.48 0.37 3000 0.04 0.52 0.46 4000 0.05 0.59 0.52 5000 0.07 0.64 0.62 6000 0.08 0.68 0.74 7000 0.09 0.69 0.78
[0129] As can be seen from the above table, the average absolute error of the method of the present invention under all sample sizes is significantly lower than that of the two traditional methods, indicating that the method of the present invention has obvious advantages in prediction accuracy;
[0130] As the number of samples increases, the mean absolute error of the proposed method increases slightly, but the increase is very small, from 0.01 for 1000 samples to 0.09 for 7000 samples. This is related to the data noise caused by the increase in data volume, but the overall impact is not significant;
[0131] In contrast, the mean absolute errors of traditional methods 1 and 2 increase significantly with the increase in the number of samples, especially traditional method 2, which increases from 0.31 for 1000 samples to 0.78 for 7000 samples, which indicates that these two traditional methods are less robust to data noise or outliers, and the prediction accuracy decreases faster with the increase in sample size.
[0132] Specifically, when there are 1000 samples, the average absolute error of the proposed method is 0.01%, which is much lower than 35% of traditional method 1 and 31% of traditional method 2;
[0133] When the number of samples increases to 2000, the average absolute error of the proposed method is 0.03%, while the errors of traditional methods 1 and 2 increase to 48% and 37%, respectively;
[0134] As the number of samples continues to increase, the mean absolute error of the method of the present invention increases to 0.09% when there are 7000 samples, but is still much lower than 69% and 78% of the two traditional methods.
[0135] The method of the present invention exhibits high prediction accuracy and good robustness under different sample numbers, while the prediction accuracy of the traditional method decreases significantly with the increase in the number of samples, showing poor robustness, which verifies the effectiveness and superiority of the method of the present invention in renewable energy power generation power prediction.
[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for predicting and analyzing the power generation of renewable energy affected by multiple factors, characterized in that: include: Acquire historical power generation, analyze the historical power generation by using a neural network, obtain influencing factors of the historical power generation, and establish a power generation prediction database based on the relationship between the historical power generation and the influencing factors; Preprocessing the original data in the power generation prediction database by using an interpolation method to obtain a first power generation prediction database; Using correlation analysis to calculate the correlation of the variable factors of the first power generation prediction database, and selecting the variable factors with non-zero correlation as prediction data; A prediction model is established according to a neural network, the prediction data is normalized to obtain first prediction data, the first prediction data is input into the prediction model, the first prediction data is trained using a neuron activation function, a prediction value of the generated power is output, the prediction value is denormalized, and a final prediction result is obtained.
2. The method for predicting and analyzing the power generation of renewable energy affected by multiple factors as claimed in claim 1, characterized in that: The power generation prediction database includes: The power generation is taken as the target variable and the influencing factors are taken as independent variables to construct a decision tree model; According to the importance index of the influencing factors output by the decision tree model, the influence of the influencing factors on the power generation is judged; The first judgment includes: if the importance index value of the influencing factor is greater than a first threshold, the influencing factor is included in the power generation prediction database; if the importance index of the influencing factor is less than the first threshold, the influencing factor is eliminated; A second judgment is performed on whether the importance index value is greater than the first threshold value. If the importance index value of the impact factor is greater than the second threshold value, the impact factor is a key impact factor.
3. The method for predicting and analyzing the power generation of renewable energy affected by multiple factors as claimed in claim 2, characterized in that: Preprocessing using interpolation methods includes: Traverse the electric power prediction database, mark the location information of all missing values, and for each missing value location, find the valid data points before and after; Set the time interval threshold. If the time interval between adjacent valid data points is less than the first time interval threshold, the zero-order difference quotient of the previous and next valid data points is calculated as the average value of the data in the time series interval divided by the value of the next moment of the missing data minus the value of the previous moment of the missing data, and the missing part is replaced by the mean value. If the time interval between adjacent valid data points is greater than the first time interval threshold, the missing value is filled twice using linear interpolation, valid data points before and after the missing position are found, and the interpolation value of the missing value is calculated as the value of the next moment minus the value of the previous moment plus the value of the previous moment multiplied by the missing time minus the previous moment divided by the next moment minus the previous moment, and the result of linear interpolation is used to fill the missing value; The original data minus the minimum value of the original data is divided by the maximum value of the original data minus the minimum value to obtain normalized data, thereby obtaining a first power generation prediction database.
4. The method for predicting and analyzing the power generation of new energy affected by multiple factors as claimed in claim 2 or 3, characterized in that: Calculating the correlation of the variable factors of the first power generation prediction database by using correlation analysis includes: Performing data conversion on the original data of the first power generation prediction database, subtracting the normalized data from the average value of the original data and dividing it by the variance of the original data to obtain the data after mean normalization processing; Conduct correlation analysis on the normalized data; If the correlation coefficient is zero, the variable factor is not included in the predictor cohort; if the correlation coefficient is positive or negative, the variable factor is included in the predictor cohort.
5. The method for predicting and analyzing the power generation of renewable energy affected by multiple factors as claimed in claim 4, characterized in that: Building a predictive model based on a neural network includes: Calculate the correlation between the input variable and the target variable, set the correlation coefficient threshold, and if the correlation coefficient between the input variable and the target variable is greater than the correlation coefficient threshold, then include the variable in the prediction model input; The number of neurons in the hidden layer is the number of samples in the training set divided by the number of neurons in the input layer plus the number of output neurons multiplied by the value of any variable.
6. The method for predicting and analyzing the power generation of renewable energy affected by multiple factors as claimed in claim 5, characterized in that: Training the first prediction data using the neuron activation function includes: The first prediction data is input from the input layer, data arrangement and mapping processing are performed in the input layer, the first prediction data is input to the next functional layer, and the first prediction data is trained using a neuron activation function in the hidden layer; Multiply each variable in the input layer by the corresponding weight to get the weighted sum; The weighted sum is added to the bias value to get the total input of the neuron; The total input is transformed nonlinearly through the activation function to obtain the activation value of the neuron.
7. The method for predicting and analyzing the power generation of renewable energy affected by multiple factors as claimed in claim 6, characterized in that: Denormalizing the predicted value includes: The maximum value of the output variable minus the minimum value of the output variable multiplied by the predicted output plus the minimum value of the output variable is the output variable, which is the final prediction result.
8. A system for predicting and analyzing the power generation of renewable energy under the influence of multiple factors, characterized in that: include: A database establishment module is used to obtain historical power generation, analyze the historical power generation using a neural network, obtain the influencing factors of the historical power generation, and establish a power generation prediction database based on the relationship between the historical power generation and the influencing factors; A preprocessing module, used for preprocessing the original data in the power generation prediction database by using an interpolation method to obtain a first power generation prediction database; A correlation analysis module, used to perform correlation calculation on the variable factors of the first power generation prediction database by using correlation analysis, and select the variable factors with non-zero correlation as prediction data; The training module is used to establish a prediction model based on a neural network, normalize the prediction data to obtain first prediction data, input the first prediction data into the prediction model, train the first prediction data using a neuron activation function, output a prediction value of the generated power, denormalize the prediction value, and obtain a final prediction result.
9. An electronic device, characterized in that: The device comprises: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.
10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Virtual power plant gateway data processing method and system based on edge computing
CN119150705A
Virtual power plant gateway data processing method and system based on edge computing
CN119150705B
New energy power generation prediction method and platform based on acquired data
CN119275838A
