Photovoltaic court output prediction method based on multiple environmental factors

Abstract

The invention discloses a photovoltaic transformer area output prediction method based on multiple environmental factors. The method comprises the following steps: S1, acquiring generated power data and environmental factor data of a photovoltaic transformer area; s2, performing normalization processing on the acquired power generation power data and the environmental factor data, performing arrangement on the basis of time, and performing processing to obtain a power generation data set; s3, presetting a time sequence model; and S4, inputting the power generation data set into a preset time sequence model, and performing output prediction of the photovoltaic transformer area. The method has the advantage of high prediction precision.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation station area output prediction, and in particular to a photovoltaic station area output prediction method based on multiple environmental factors. Background Art

[0002] As the global demand for renewable energy continues to grow, photovoltaic power generation, as a clean and sustainable form of energy, has received increasing attention.

[0003] However, photovoltaic power generation is affected by many factors such as meteorological conditions and geographical environment, and its output prediction faces many challenges. Traditional photovoltaic power generation output prediction methods mainly rely on a single time series model, and in terms of training data, only use the historical output data of photovoltaic power generation as a data set. Although it has a certain degree of accuracy in predicting photovoltaic output after a certain amount of training, when faced with more complex situations, the prediction results will deviate greatly from the actual results.

[0004] In addition, existing studies on photovoltaic power station output prediction methods generally lack comprehensive consideration of environmental factors, such as temperature, humidity, wind speed, etc. Some studies and methods consider a single environmental variable factor, and have achieved certain improvements in prediction accuracy, but if more environmental variables are integrated, the accuracy and stability of photovoltaic power station output prediction can be further improved.

[0005] Therefore, how to accommodate more environmental factor parameters in the time series prediction model and optimize the time series prediction model has become an important technical issue in the current photovoltaic power output prediction. In this regard, how to provide a method to solve the problem of inaccurate prediction and poor stability caused by the inability to consider environmental factors in multiple dimensions in the existing photovoltaic power output prediction method has become an urgent problem to be solved. Summary of the invention

[0006] The purpose of the present invention is to provide a photovoltaic area output prediction method based on multiple environmental factors, which has the advantage of high prediction accuracy.

[0007] To achieve the above-mentioned purpose, the present invention provides a photovoltaic area output prediction method based on multiple environmental factors, which comprises: step S1, obtaining the power generation data and environmental factor data of the photovoltaic area; step S2, normalizing the obtained power generation data and environmental factor data, and arranging them based on time to obtain a power generation data set; step S3, presetting a time series model; step S4, inputting the power generation data set into the preset time series model to predict the output of the photovoltaic area.

[0008] Preferably, the power generation data is the operating efficiency of the photovoltaic area; the environmental factor data includes irradiance data, temperature data, humidity data, wind speed data, and weather type data.

[0009] Preferably, the weather type data is quantized and normalized to the interval [0,1], and the expression of the weather type data is converted into an impact value on the photovoltaic area.

[0010] Preferably, a linear interpolation formula is used to align the power generation data and the environmental factor data in time:

[0011] y=y1+((y2-y1) / (x2-x1))*(x-x1);

[0012] Among them, (x, y) is the unknown data point that needs to be estimated, and (x1, y1) and (x2, y2) are two different known data points.

[0013] Preferably, the step S2 comprises:

[0014] Step S21, normalizing the acquired power generation data and environmental factor data one by one;

[0015]

[0016] Among them, X is the data that needs to be normalized, X normalize is the normalized value of X, min is the minimum value in this type of data, X max is the maximum value in this type of data;

[0017] Step S22, arranging the normalized data based on time to obtain a power generation data set.

[0018] Preferably, the step S22 includes:

[0019] Step S221, constructing six-dimensional data points at each time point,

[0020] P(time)=(W,I,T,H,V,WW);

[0021] Among them, P(time) is the six-dimensional data point at a certain time point, W is the power generation data collected by 96-point measurement at the corresponding time point, I is the irradiance data at the corresponding time point, T is the air temperature, H is the humidity data at the corresponding time point, V is the wind speed data at the corresponding time point, and WW is the weather type data at the corresponding time point;

[0022] Step S222, arranging the six-dimensional data points based on time to obtain a power generation data set.

[0023] Preferably, the step S3 includes: step S31, selecting an initial time series model; step S32, training the initial time series model.

[0024] Preferably, in step S31, the initial time series model includes: an ARIMA model, an LSTM model, a SARIMA model, an XGBoost model, a TCN model or a RNN model or a combination of one or more thereof.

[0025] Preferably, the step S32 includes:

[0026] Step S321, training multiple different initial time series models simultaneously, so as to select the best result as the preset time series model;

[0027] Step S322, calculate the mean square error and the root mean square error:

[0028]

[0029] Where MSE stands for mean square error, RMSE stands for root mean square error, n is the total number of data points in the training data set, i = 1, 2...n, y i is the observed value of the i-th data point, y i' is the predicted value of the i-th data point;

[0030] Step S323, calculate the coefficient of determination:

[0031]

[0032] Among them, R represents the coefficient of determination, y i is the observed value of the i-th data point, y i' is the predicted value of the i-th data point, y average is the mean of the observed values;

[0033] Step S324, comprehensively evaluate various parameters and output a preset time series model.

[0034] Preferably, the preset time series model is a combination of a TCN model and a LSTM model.

[0035] In summary, compared with the prior art, the photovoltaic area output prediction method based on multiple environmental factors provided by the present invention has the following beneficial effects:

[0036] First, by combining historical photovoltaic area output data with temperature, humidity, wind speed, solar radiation, weather type and other data to form a training data set of multiple environmental factors, the photovoltaic area output is predicted using a time series prediction model, which is multi-faceted; improving the training model's ability to adapt to nonlinear and non-stationary data in photovoltaic power generation output prediction.

[0037] Second, the present invention can improve the prediction accuracy, stability and adaptability of photovoltaic power station output, which has significant advantages and contributes to the efficient utilization and stable operation of photovoltaic power generation.

[0038] Third, the present invention represents data such as weather types in a quantitative form, mines the potential information in the environmental data of the photovoltaic area, and improves the prediction model's ability to capture complex dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A flow chart of a photovoltaic area output prediction method based on multiple environmental factors.

[0040] Figure 2 This is a quantitative assignment table for the impact of weather type data on photovoltaic power generation.

[0041] Figure 3 This is a schematic diagram of the structure of the model that combines the TCN model and the LSTM model.

[0042] Figure 4 In step S3, when the time series model is preset, the prediction result evaluation index table is generated.

[0043] Figure 5 It is a comparison chart of the predicted results and the actual output under the embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the attached embodiment of the present invention Figure 1 ~Attached Figure 5 , the technical solutions, structural features, objectives achieved and effects in the embodiments of the present invention are described in detail.

[0045] It should be noted that the drawings are in a very simplified form and use non-precise proportions. They are only used to conveniently and clearly assist in explaining the embodiments of the present invention, and are not used to limit the conditions for the implementation of the present invention. Therefore, they have no substantive technical significance. Any structural modification, change in proportional relationship or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical content disclosed by the present invention.

[0046] It should be noted that, in the present invention, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only the elements explicitly listed, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0047] The present invention provides a photovoltaic area output prediction method based on multiple environmental factors, by collecting a variety of data, such as weather data, power generation data, etc., and then performing corresponding processing; the processed data is input into a preset time series model, so as to obtain the power data of the photovoltaic area at the corresponding time; the focus of the present invention is how to extract and process a variety of data, and what kind of time series model to choose.

[0048] like Figure 1 As shown, the method includes:

[0049] Step S1, obtaining power generation data and environmental factor data of the photovoltaic area;

[0050] Among them, the power generation data is the operating efficiency of the photovoltaic area, and the environmental factor data refers to the surrounding environmental factors around the photovoltaic area, including irradiance data, temperature data, humidity data, wind speed data, and weather type data.

[0051] For the problem that some of the aforementioned data are meaningless when expressed in an objective form, such as temperature data, humidity data, etc., the "influence value" method can be introduced to solve it; taking weather type data as an example, because weather type data tends to be an objective quantity; in order to solve this problem, the expression of weather type data is converted into an influence value on the photovoltaic area; specifically, the weather type data is quantified and normalized to the interval of [0,1]. At this time, the weather types are divided into: sunny, cloudy, overcast, rainy, snowy, foggy, and thunderstorms, and then the influence value on the photovoltaic area is obtained, generating the following Figure 2 Parameters shown.

[0052] Step S2, normalizing the acquired power generation data and environmental factor data, and arranging them based on time to obtain a power generation data set;

[0053] Step S3, presetting a time series model;

[0054] That is, the data model used to process the power generation data set.

[0055] Step S4, inputting the power generation data set into a preset time series model to perform output prediction of the photovoltaic area;

[0056] Among them, how to bring data into the model for calculation in step S4 is a prior art. The focus of this article is on the processing of the aforementioned data and the selection of the model, and step S4 will not be described in detail.

[0057] In a specific embodiment, in step S1, the power generation data of the photovoltaic area is collected in the form of 96 points (24 hours a day, collected every 15 minutes, 96 points for the whole day). The corresponding environmental factor data of the photovoltaic area is generally collected in the form of equal hourly intervals. Therefore, this leads to the two types of data not being aligned in time. In order to solve this problem, it is proposed to align the data using a linear interpolation formula, because each type of data is arranged in the form of (x, y), that is, (x represents time, y represents value). The goal of linear interpolation is to estimate the y value corresponding to any x value between two different points of the same data by constructing a straight line.

[0058] The linear interpolation formula can be expressed by a straight line equation in the form of slope-intercept: y=y1+((y2-y1) / (x2-x1))*(x-x1) where (x, y) is the unknown data point to be estimated, and (x1, y1) and (x2, y2) are two different known data points.

[0059] For example, there are temperature data (15:00, b1) and (16:00, b2) and power generation data (15:00, a1kw), (15:30, a2kw) and (16:00, a3kw). If the temperature data is to be aligned with the power generation data, the temperature data at 15:30 can be obtained through the aforementioned linear interpolation formula to achieve alignment.

[0060] In a specific embodiment, the step S2 includes:

[0061] Step S21, normalizing the acquired power generation data and environmental factor data one by one;

[0062] Normalization formula:

[0063]

[0064] Among them, X is the data that needs to be normalized, X normalize is the normalized value of X, min is the minimum value in this type of data, X maxis the maximum value of this type of data; a detailed explanation is given here, because it has been mentioned above that there are 6 types of data, namely power generation data, irradiance data, temperature data, humidity data, wind speed data, and weather type data; so the X in the above min Then X max Refers to the maximum and minimum values ​​of the same type of data. By normalizing all data in the six types, the corresponding normalized values ​​are obtained.

[0065] Regarding the question of whether the data will be confused during the normalization process, as mentioned above, there is an x ​​in the data, namely "time", so the time of how individual data of the same type changes is unique. This is also the basis for the subsequent combination of different types of data at a single time point to form a six-dimensional data point at a single time point.

[0066] Step S22, arranging the normalized data based on time to obtain a power generation data set;

[0067] Step S221, constructing six-dimensional data points at each time point,

[0068] P(time)=(W,I,T,H,V,WW)

[0069] Wherein, P(time) is the six-dimensional data point at a certain time point (where time represents the time parameter), W is the power generation data collected by 96-point measurement at the corresponding time point, I is the irradiance data at the corresponding time point, T is the air temperature, H is the humidity data at the corresponding time point, V is the wind speed data at the corresponding time point, and WW is the weather type data at the corresponding time point. These six parameters constitute the six-dimensional data point.

[0070] Step S222, arranging the six-dimensional data points based on time to obtain a power generation data set.

[0071] In a specific embodiment, the step S3 includes:

[0072] Step S31, selecting an initial time series model;

[0073] The initial time series model can be selected from ARIMA model (Auto regressive Integrated Moving Average Model); LSTM model (Long Short Term Memory Network); TCN (Temporal Convolutional Network); SARIMA model (Seasonal Auto regressive Integrated Moving Average Model); XGBoost model (Gradient Boosting Tree); RNN model (Recurrent Neural Network, Chinese called recurrent neural network) and the like, or a combination thereof; Figure 3 As shown, the present invention adopts a combination of TCN model and LSTM model.

[0074] Step S32, training an initial time series model;

[0075] For example, the input is used for training data sets (all known data), where 80% of the data in the training data sets are randomly assigned to the training set and used to substitute into the initial time series model for computational training; 20% of the data are randomly assigned to the validation set to check the training quality and adjust the initial time series model for computation and optimization until the output data of the validation set tends to be stable.

[0076] In a preferred embodiment, multiple different initial time series models can be trained simultaneously, so as to select the best result as the preset time series model; specifically, the prediction result evaluation index is used to judge the advantages and disadvantages of different initial time series models;

[0077] Specifically, the prediction results of the time series model are evaluated using three indicators: mean square error (MSE), root mean square error (RMSE), and coefficient of determination (R-Squared);

[0078] The formulas for MSE and RMSE are:

[0079]

[0080] Where n is the number of samples in the training data set, that is, the total number of data points in the training data set, i = 1, 2...n, y i is the observed value of the i-th data point, that is, the actual true value, y i' is the predicted value of the i-th data point, that is, the value predicted by the time series model.

[0081] R-Square formula:

[0082]

[0083] In the formula, y i is the observed value of the i-th data point, y i' is the predicted value of the i-th data point, y average is the mean of the observations, representing the average level of all observations.

[0084] like Figure 4 As shown, it is a table of prediction result evaluation indicators of several time series models. It is mentioned above that the present invention adopts a combination of TCN model and LSTM model. The analysis results show that the model adopted by the present invention has a significant improvement in prediction accuracy. This method can maintain high stability and prediction accuracy, and can maintain excellent prediction effects under various complex environmental influences.

[0085] like Figure 5 As shown, for the output results of the present invention; it can be seen that the present invention fully considers various complex environmental factors of photovoltaic power generation in order to give a more accurate prediction of the output of the future station. Because this method takes into account various complex environmental factors affecting photovoltaic power generation and the power generation characteristics of the photovoltaic power generation equipment itself, it can maintain a high level of accuracy and precision in the prediction results, and ultimately enhances the ability of photovoltaic stations to process predictions, making important contributions to the operation and management of photovoltaic power generation systems and renewable energy.

[0086] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A photovoltaic area output prediction method based on multiple environmental factors, characterized in that: The method comprises: Step S1, obtaining power generation data and environmental factor data of the photovoltaic area; Step S2, normalizing the acquired power generation data and environmental factor data, and arranging them based on time to obtain a power generation data set; Step S3, presetting a time series model; Step S4, inputting the power generation data set into a preset time series model to perform output prediction of the photovoltaic area.

2. A photovoltaic area output prediction method based on multiple environmental factors according to claim 1, characterized in that: The power generation data is the operating efficiency of the photovoltaic area; The environmental factor data includes irradiance data, temperature data, humidity data, wind speed data, and weather type data.

3. The photovoltaic area output prediction method based on multiple environmental factors according to claim 2 is characterized in that: The weather type data is quantified and normalized to the interval of [0,1], and the expression of weather type data is converted into the impact value on the photovoltaic area.

4. The photovoltaic area output prediction method based on multiple environmental factors according to claim 3 is characterized in that: The linear interpolation formula is used to align the power generation data and environmental factor data in time: y=y1+((y2-y1) / (x2-x1))*(x-x1) Among them, (x, y) is the unknown data point that needs to be estimated, and (x1, y1) and (x2, y2) are two different known data points.

5. The photovoltaic area output prediction method based on multiple environmental factors according to claim 4 is characterized in that: The step S2 comprises: Step S21, normalizing the acquired power generation data and environmental factor data one by one; Among them, X is the data that needs to be normalized, X normalize is the normalized value of X, min is the minimum value in this type of data, X max is the maximum value in this type of data; Step S22, arranging the normalized data based on time to obtain a power generation data set.

6. A photovoltaic area output prediction method based on multiple environmental factors according to claim 5, characterized in that: The step S22 comprises: Step S221, constructing six-dimensional data points at each time point, P(time)=(W,I,T,H,V,WW) Among them, P(time) is the six-dimensional data point at a certain time point, W is the power generation data collected by 96-point measurement at the corresponding time point, I is the irradiance data at the corresponding time point, T is the air temperature, H is the humidity data at the corresponding time point, V is the wind speed data at the corresponding time point, and WW is the weather type data at the corresponding time point; Step S222, arranging the six-dimensional data points based on time to obtain a power generation data set.

7. A photovoltaic area output prediction method based on multiple environmental factors according to claim 6, characterized in that: The step S3 comprises: Step S31, selecting an initial time series model; Step S32, training the initial time series model.

8. The photovoltaic area output prediction method based on multiple environmental factors according to claim 7 is characterized in that: In step S31, the initial time series model includes: an ARIMA model, an LSTM model, a SARIMA model, an XGBoost model, a TCN model or a RNN model or a combination of one or more.

9. A photovoltaic area output prediction method based on multiple environmental factors according to claim 8, characterized in that: The step S32 comprises: Step S321, training multiple different initial time series models simultaneously, so as to select the best result as the preset time series model; Step S322, calculate the mean square error and the root mean square error: Where MSE stands for mean square error, RMSE stands for root mean square error, n is the total number of data points in the training data set, i = 1, 2...n, y i is the observed value of the ith data point, y i' is the predicted value of the i-th data point; Step S323, calculate the coefficient of determination: Among them, R represents the coefficient of determination, y i is the observed value of the i-th data point, y i' is the predicted value of the i-th data point, y average is the mean of the observed values; Step S324, comprehensively evaluate various parameters and output a preset time series model.

10. A photovoltaic area output prediction method based on multiple environmental factors according to claim 9, characterized in that: The preset time series model is a combination of a TCN model and a LSTM model.

Images