Photovoltaic output prediction model construction method based on environment variable correlation analysis and application

Abstract

The invention discloses a photovoltaic output prediction model based on environmental variable correlation analysis, which is characterized in that all available weather data are used for predicting photovoltaic output, and meteorological and electric power data are analyzed by using a statistical method to determine an importance degree sequence of input variables; and finally, a photovoltaic output prediction model based on environment variable correlation analysis is formed. According to the method, the irradiance, the environment temperature, the wind speed and the relative humidity are comprehensively considered, the function relation between the meteorological data and the photovoltaic output is established, accurate prediction of the photovoltaic output is achieved, a power grid operator can be helped to better integrate photovoltaic power generation, and the stability and economical efficiency of a power grid are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic output prediction, and specifically relates to a photovoltaic output prediction model construction method and application based on environmental variable correlation analysis. Background Art

[0002] Photovoltaic power generation, as a clean and renewable energy source, has developed rapidly in recent years. However, since the output of photovoltaic power generation is affected by a variety of meteorological factors, such as irradiance, ambient temperature, wind speed and relative humidity, the uncertainty and variability of these factors lead to significant intermittency and volatility in photovoltaic output. This characteristic poses a challenge to the dispatching management and stability of the power grid, especially in the case of large-scale photovoltaic power generation connected to the grid. In order to meet this challenge, it is crucial to accurately predict the output capacity of photovoltaic power generation. This is not only a prerequisite for efficient planning of power production and reasonable allocation of resources, but also a key link to ensure the overall reliability of hybrid power systems, optimize power grid dispatching strategies, and maintain stable operation of power grids. Therefore, it is necessary to establish a correlation model between photovoltaic output and environmental variables, and improve the prediction accuracy of photovoltaic power output through scientific methods to better adapt to the dispatching and operation needs of the power system. Summary of the invention

[0003] The technical problem to be solved by the present invention is how to consider the relationship between photovoltaic output and environmental variables, and thus more accurately predict photovoltaic output.

[0004] The present invention solves the above technical problems through the following technical means:

[0005] A photovoltaic output prediction model construction method based on environmental variable correlation analysis includes the following steps:

[0006] (1) Obtain various environmental data of the photovoltaic installation site and the maximum photovoltaic power output data P max , calculate the correlation coefficient r between environmental variables and photovoltaic power generation;

[0007] (2) Select the prediction model variables based on the correlation coefficient r and create a regression analysis model between different variables; if |r is greater than the set value, use the variable itself in the regression analysis model, otherwise multiply the variable by the POA value;

[0008] (3) Analyze the mean absolute error, root mean square error, and mean absolute percentage error indicators to verify the effectiveness of the prediction model.

[0009] Furthermore, step (1) is specifically as follows:

[0010] (11) Obtain the total horizontal radiation GHI and ambient temperature T at the photovoltaic installation site a, wind speed WS and relative humidity RH and other meteorological data, and measure the temperature T of the photovoltaic device m , array plane irradiance POA and maximum photovoltaic power output data P max ;

[0011] (12) Determine P max The correlation coefficient r between the two variables x i and i The degree of correlation between them is as follows:

[0012]

[0013] Among them, x i is the maximum photovoltaic power output data P max ,y i are other environmental variables, n is the number of samples; For variable x i The mean of is the variable y i The mean of

[0014] Thus we can deduce:

[0015]

[0016] in, and n are the mean and sample size respectively, {x i ,y i} is the i-th sample point.

[0017] Furthermore, the coefficient calculation process of the regression analysis model in step (2) is as follows: For the multiple regression analysis model, the dependent variable is described as:

[0018]

[0019] The superordinate and independent variables can be described by the following formula:

[0020] y i =β0+β1x 1i +β2x 2i +β3x 3i +…+e i (4)

[0021] Among them, y is the parent variable, that is, the maximum photovoltaic power output data Pmax, x is the independent variable, that is, other environmental variables, β is the coefficient, and e is the error term;

[0022] The error term has a mean of zero, a variance that follows a normal distribution, and a constant variance across all measurements; the coefficient β is estimated by least squares regression as follows:

[0023]

[0024]

[0025] Y=(y1,y2,…,yn)T

[0026] Where X' is the transpose of X and Y' is the transpose of Y. In this case, the row equation is solved by setting the partial differentials of the squared error and L with respect to the coefficients to zero. This process yields an estimate of the regression coefficient β. The coefficient of determination is used for linear regression as follows:

[0027]

[0028] (23) For the univariate simple linear regression analysis model, the determination coefficient is defined as follows:

[0029] 1) The ratio of the variance of the error in the regression analysis model to the total variance of the dependent variable;

[0030] 2) The square of the correlation coefficient between the observed dependent variable and the regression analysis model;

[0031] The above determination coefficient definition can be expressed as:

[0032]

[0033] Where SSTO is the total sum of squares. The greater the change in , the greater the SSTO; SSE represents the error sum of squares, SSR represents the regression sum of squares, R 2 is the determination coefficient.

[0034] Furthermore, step (3) is specifically as follows:

[0035] The mean absolute error (AAD), root mean square error (RMSE), and mean absolute percentage error (MAPE) were calculated to evaluate the accuracy of the prediction model. The formulas for calculating these errors are as follows:

[0036]

[0037] The present invention also provides a photovoltaic output prediction model construction system based on environmental variable correlation analysis, comprising the following:

[0038] Correlation coefficient calculation module: obtain various environmental data of the photovoltaic installation site and the maximum photovoltaic power output data P max , calculate the correlation coefficient r between environmental variables and photovoltaic power generation;

[0039] Model building module: select predictive model variables based on the correlation coefficient r and create a regression analysis model between different variables; if |r| is greater than the set value, the variable itself is used in the regression analysis model, otherwise the variable is multiplied by the POA value;

[0040] Verification module: Analyze the mean absolute error, root mean square error and mean absolute percentage error indicators to verify the effectiveness of the prediction model.

[0041] Furthermore, the specific process of the step correlation coefficient calculation module is as follows:

[0042] (11) Obtain the total horizontal radiation GHI and ambient temperature T at the photovoltaic installation site a , wind speed WS and relative humidity RH and other meteorological data, and measure the temperature T of the photovoltaic device m , array plane irradiance POA and maximum photovoltaic power output data P max ;

[0043] (12) Determine P max The correlation coefficient r between the two variables x i and i The degree of correlation between them is as follows:

[0044]

[0045] Among them, x i is the maximum photovoltaic power output data P max ,y i are other environmental variables, n is the number of samples; For variable x i The mean of is the variable y i The mean of

[0046] Thus we can deduce:

[0047]

[0048] in, and n are the mean and sample size respectively, {x i ,y i} is the i-th sample point.

[0049] Furthermore, the coefficient calculation process of the regression analysis model in the model building module is as follows: For the multiple regression analysis model, the dependent variable is described as:

[0050]

[0051] The superordinate and independent variables can be described by the following formula:

[0052] y i =β0+β1x 1i +β2x 2i +β3x 3i +…+e i (4)

[0053] Among them, y is the parent variable, that is, the maximum photovoltaic power output data Pmax, x is the independent variable, that is, other environmental variables, β is the coefficient, and e is the error term;

[0054] The error term has a mean of zero, a variance that follows a normal distribution, and a constant variance across all measurements; the coefficient β is estimated by least squares regression as follows:

[0055]

[0056] Y=(y1,y2,…,yn)T

[0057] Where X' is the transpose of X and Y' is the transpose of Y. In this case, the row equation is solved by setting the partial differentials of the squared error and L with respect to the coefficients to zero. This process yields an estimate of the regression coefficient β. The coefficient of determination is used for linear regression as follows:

[0058]

[0059] (23) For the univariate simple linear regression analysis model, the determination coefficient is defined as follows:

[0060] 1) The ratio of the variance of the error in the regression analysis model to the total variance of the dependent variable;

[0061] 2) The square of the correlation coefficient between the observed dependent variable and the regression analysis model;

[0062] The above determination coefficient definition can be expressed as:

[0063]

[0064] Where SSTO is the total sum of squares. The greater the change in , the greater the SSTO; SSE represents the error sum of squares, SSR represents the regression sum of squares, R 2 is the determination coefficient.

[0065] Furthermore, the specific verification process of the verified module is as follows:

[0066] The mean absolute error (AAD), root mean square error (RMSE), and mean absolute percentage error (MAPE) were calculated to evaluate the accuracy of the prediction model. The formulas for calculating these errors are as follows:

[0067]

[0068] The present invention also provides a method for predicting photovoltaic output using the above prediction model.

[0069] The present invention also provides a processing device, comprising at least one processor and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor can execute the above method by calling the program instructions.

[0070] The advantages of the present invention are:

[0071] In the context of large-scale access of distributed photovoltaics to the power system, the present invention provides a new photovoltaic output prediction model and establishes a correlation analysis method between photovoltaic output and environmental variables, which can more accurately capture the key factors affecting photovoltaic output, thereby improving the accuracy of the prediction model, which is important for optimizing the operation of the power system and improving the dispatching efficiency of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0073] Figure 2 The error comparison of the model under different POA irradiance ranges of the embodiment of the present invention;

[0074] Figure 3 The figure is a comparison of the errors of the models in different months according to the embodiment of the present invention. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] like Figure 1 As shown, a method for constructing a photovoltaic output prediction model based on environmental variable correlation analysis includes the following steps:

[0077] (1) Obtain various environmental data and maximum photovoltaic power output data at the photovoltaic installation site through monitoring devices, and conduct correlation analysis between environmental variables and photovoltaic power generation;

[0078] (2) Selecting prediction model variables based on the correlation coefficient between photovoltaic output and environmental variables to obtain a prediction model for photovoltaic power generation;

[0079] (3) Analyze indicators such as mean absolute error, root mean square error and mean absolute percentage error to verify the effectiveness of the model.

[0080] The technical solution of the present invention is described in detail below:

[0081] Step 1: Conduct correlation analysis between environmental variables and photovoltaic power generation. The specific implementation steps are as follows:

[0082] (11) The total horizontal radiation (GHI) and ambient temperature (T a ), wind speed (WS) and relative humidity (RH) and other meteorological data, and measure the temperature of the photovoltaic equipment (T m ), array plane irradiance (POA) and maximum photovoltaic power output (P max ).

[0083] (12) Further analysis of relevant factors to determine P max The correlation coefficient r represents the relationship between two variables x i and i The degree of correlation between them is as follows:

[0084]

[0085] in, For variable x i The mean of is the mean of the variable xi,

[0086] Thus we can deduce:

[0087]

[0088] in, and n are the mean and sample size respectively, {x i ,y i} is the i-th sample point.

[0089] There are two different methods to estimate the correlation and correlation coefficient between two variables, namely the Pearson method and the Spearman method. The Pearson correlation analysis method evaluates the linear relationship between two variables. If the change in one variable is proportional to the change in the other variable, there is a linear relationship between the two variables. The Spearman correlation analysis method evaluates the simple (ordinal or rank) relationship between two variables. In this simple relationship, the two variables tend to change together, but not necessarily in a proportional manner. The Spearman correlation coefficient is not based on the raw data, but on the ranked values ​​of each variable. The Pearson correlation analysis method is used here to analyze the correlation between photovoltaic output and environmental variables.

[0090] Step 2: Prediction model of photovoltaic power generation. The specific implementation steps are as follows:

[0091] (21) Based on the correlation coefficient between PV output and environmental variables, the prediction model variables are selected and a regression model between different variables is created. If |r| is greater than 0.5, the variable itself is used in the regression model, otherwise the variable is multiplied by the POA value, because the POA value is closely related to P max has the highest correlation coefficient.

[0092] (22) Based on the correlation analysis, the factors affecting photovoltaic power generation are selected and the photovoltaic output model is obtained through regression analysis. Next, the coefficients of the regression analysis model are estimated. For the multivariate regression model, the dependent variable is described as:

[0093]

[0094] The superordinate and independent variables can be described by the following formula:

[0095] y i =β0+β1x 1i +β2x 2i +β3x 3i +…+e i (4)

[0096] Among them, y is the superordinate variable, x is the independent variable, β is the coefficient, and e is the error term.

[0097] The error term has a mean of zero and a variance that follows a normal distribution, with the variance between all measurements being constant. The coefficient β is estimated by least squares regression as follows:

[0098]

[0099] At this point, the row equation is solved by setting the partial differentials of the error squared sum L with respect to the coefficients to zero. This process yields an estimate of the regression coefficient β, the coefficient of determination commonly used in linear regression, as shown below:

[0100]

[0101] (23) For a univariate simple linear regression model, that is, with only one independent variable and one dependent variable, the coefficient of determination is the square of the Pearson correlation coefficient between these variables. However, for a univariate multiple linear regression model, the coefficient of determination is defined as follows:

[0102] 1) The ratio of the variance of the error in the regression model to the total variance of the dependent variable.

[0103] 2) The square of the correlation coefficient between the observed dependent variable and the regression model.

[0104] The above determination coefficient definition can be expressed as:

[0105]

[0106] Where SSTO is the total sum of squares. The greater the change in , the greater the SSTO. SSE stands for the sum of squared errors, SSR stands for the sum of squared regression, and R 2 is the determination coefficient.

[0107] Step 3: Verify the effectiveness of the model. The specific implementation steps are as follows:

[0108] (31) The mean absolute error (AAD), root mean square error (RMSE), and mean absolute percentage error (MAPE) were calculated to evaluate the accuracy of the prediction model. The formulas for calculating these errors are as follows:

[0109]

[0110] (32) AAD uses the same scale as the data being evaluated and is often used for error prediction in time series analysis. RMSE analysis is proportional to the square of the error and is therefore sensitive to outliers. For MAPE, accuracy is usually expressed as a percentage. The disadvantage of this form is that if the prediction is too high, there is no upper limit to the error rate. In order to compensate for the shortcomings of each error analysis method, the prediction accuracy of the models estimated by the three methods is jointly evaluated. Specifically, the coefficient of determination, coefficient β, and coefficient β of the regression model are calculated and compared. i and errors.

[0111] For the photovoltaic output prediction model based on environmental variable correlation analysis proposed above, the applicability of this method is evaluated through specific case analysis.

[0112] The method was verified using a real photovoltaic power station, and six regression prediction models were constructed. Model P1 used POA and T m , Model P2 uses POA and RH, Model P3 uses POA, T mand RH, Model P4 uses POA, WS, and T a , Model P5 uses POA, T a , WS and RH, Model P6 uses POA, T m , WS and RH, as follows:

[0113]

[0114]

[0115] The regression coefficient values ​​for each prediction model are as follows:

[0116]

[0117] The coefficient of determination and error of each prediction model are shown below:

[0118]

[0119] It can be seen that the model with the smallest AAD, RMSE, and MAPE during the entire monitoring period is P5, so the prediction value of model P5 is the most accurate in terms of the overall determination coefficient and error.

[0120] Depend on Figure 2 From the error comparison diagram of the models under different POA irradiance ranges, it can be seen that within the entire irradiance range, the error of model P5 is smaller than that of model P4. It can be seen that model P5 uses the best combination of environmental variables of POA, Ta, WS and RH, and can obtain the predicted value with the smallest error. Figure 3 The error comparison of the model in different months of the embodiment of the present invention shows that the error rate is highest from November to February. It can be seen that high humidity, low radiation and low temperature in winter will increase the error rate. It can also be obtained from the error analysis that when POA and T m When it is higher, the model error is lower, and T a or WS and P max The correlation is low and has little impact on the results.

[0121] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A photovoltaic output prediction model construction method based on environmental variable correlation analysis is characterized in that: The method comprises the following steps: (1) Obtain various environmental data of the photovoltaic installation site and the maximum photovoltaic power output data P max , calculate the correlation coefficient r between environmental variables and photovoltaic power generation; (2) Select the prediction model variables based on the correlation coefficient r and create a regression analysis model between different variables; if |r is greater than the set value, use the variable itself in the regression analysis model, otherwise multiply the variable by the POA value; (3) Analyze the mean absolute error, root mean square error, and mean absolute percentage error indicators to verify the effectiveness of the prediction model.

2. The method for constructing a photovoltaic output prediction model based on environmental variable correlation analysis according to claim 1, characterized in that: Step (1) is specifically as follows: (11) Obtain the total horizontal radiation GHI and ambient temperature T at the photovoltaic installation site a , wind speed WS and relative humidity RH and other meteorological data, and measure the temperature T of the photovoltaic device m , array plane irradiance POA and maximum photovoltaic power output data P max ; (12) Determine P max The correlation coefficient r between the two variables x i and i The degree of correlation between them is as follows: Among them, x i is the maximum photovoltaic power output data P max ,y i are other environmental variables, n is the number of samples; For variable x i The mean of is the variable y i The mean of Thus we can deduce: in, and n are the mean and sample size respectively, {x i ,y i } is the i-th sample point.

3. The method for constructing a photovoltaic output prediction model based on environmental variable correlation analysis according to claim 2, characterized in that: Step (2) The coefficient calculation process of the regression analysis model is as follows: For the multiple regression analysis model, the dependent variable is described as: The superordinate and independent variables can be described by the following formula: y i =β0+β1x 1i +β2x 2i +β3x 3i +…+e i (4) Among them, y is the parent variable, that is, the maximum photovoltaic power output data Pmax, x is the independent variable, that is, other environmental variables, β is the coefficient, and e is the error term; The error term has a mean of zero, a variance that follows a normal distribution, and a constant variance across all measurements; the coefficient β is estimated by least squares regression as follows: Y=(y1,y2,…,yn)T Where X' is the transpose of X and Y' is the transpose of Y. In this case, the row equation is solved by setting the partial differentials of the squared error and L with respect to the coefficients to zero. This process yields an estimate of the regression coefficient β. The coefficient of determination is used for linear regression as follows: (23) For the univariate simple linear regression analysis model, the determination coefficient is defined as follows: 1) The ratio of the variance of the error in the regression analysis model to the total variance of the dependent variable; 2) The square of the correlation coefficient between the observed dependent variable and the regression analysis model; The above determination coefficient definition can be expressed as: Where SSTO is the total sum of squares. The greater the change in , the greater the SSTO; SSE represents the error sum of squares, SSR represents the regression sum of squares, R 2 is the determination coefficient.

4. The method for constructing a photovoltaic output prediction model based on environmental variable correlation analysis according to claim 3, characterized in that: Step (3) is specifically as follows: The mean absolute error (AAD), root mean square error (RMSE), and mean absolute percentage error (MAPE) were calculated to evaluate the accuracy of the prediction model. The formulas for calculating these errors are as follows:

5. A photovoltaic output prediction model construction system based on environmental variable correlation analysis, characterized in that: Includes the following: Correlation coefficient calculation module: obtain various environmental data of the photovoltaic installation site and the maximum photovoltaic power output data P max , calculate the correlation coefficient r between environmental variables and photovoltaic power generation; Model building module: select predictive model variables based on the correlation coefficient r and create a regression analysis model between different variables; if |r| is greater than the set value, the variable itself is used in the regression analysis model, otherwise the variable is multiplied by the POA value; Verification module: Analyze the mean absolute error, root mean square error and mean absolute percentage error indicators to verify the effectiveness of the prediction model.

6. The photovoltaic output prediction model construction system based on environmental variable correlation analysis according to claim 5, characterized in that: The specific process of the step correlation coefficient calculation module is as follows: (11) Obtain the total horizontal radiation GHI and ambient temperature T at the photovoltaic installation site a , wind speed WS and relative humidity RH and other meteorological data, and measure the temperature T of the photovoltaic device m , array plane irradiance POA and maximum photovoltaic power output data P max ; (12) Determine P max The correlation coefficient r between the two variables x i and i The degree of correlation between them is as follows: Among them, x i is the maximum photovoltaic power output data P max ,y i are other environmental variables, n is the number of samples; For variable x i The mean of is the variable y i The mean of Thus we can deduce: in, and n are the mean and sample size respectively, {x i ,y i } is the i-th sample point.

7. The photovoltaic output prediction model construction system based on environmental variable correlation analysis according to claim 6, characterized in that: The coefficient calculation process of the regression analysis model in the model building module is as follows: For the multiple regression analysis model, the dependent variable is described as: The superordinate and independent variables can be described by the following formula: y i =β0+β1x 1i +β2x 2i +β3x 3i +…+e i (4) Among them, y is the parent variable, that is, the maximum photovoltaic power output data Pmax, x is the independent variable, that is, other environmental variables, β is the coefficient, and e is the error term; The error term has a mean of zero, a variance that follows a normal distribution, and a constant variance across all measurements; the coefficient β is estimated by least squares regression as follows: Y=(y1,y2,…,yn)T Where X' is the transpose of X and Y' is the transpose of Y. In this case, the row equation is solved by setting the partial differentials of the squared error and L with respect to the coefficients to zero. This process yields an estimate of the regression coefficient β. The coefficient of determination is used for linear regression as follows: (23) For the univariate simple linear regression analysis model, the determination coefficient is defined as follows: 1) The ratio of the variance of the error in the regression analysis model to the total variance of the dependent variable; 2) The square of the correlation coefficient between the observed dependent variable and the regression analysis model; The above determination coefficient definition can be expressed as: Where SSTO is the total sum of squares. The greater the change in , the greater the SSTO; SSE represents the error sum of squares, SSR represents the regression sum of squares, R 2 is the determination coefficient.

8. The method for constructing a photovoltaic output prediction model based on environmental variable correlation analysis according to claim 7, characterized in that: The specific verification process of the verified module is as follows: The mean absolute error (AAD), root mean square error (RMSE), and mean absolute percentage error (MAPE) were calculated to evaluate the accuracy of the prediction model. The formulas for calculating these errors are as follows:

9. Use the prediction model described in any one of claims 1 to 4 to predict photovoltaic output.

10. A processing device, characterized in that: It comprises at least one processor and at least one memory in communication with the processor, wherein the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute any method according to claims 1 to 4.

Images