Distributed photovoltaic cluster output prediction method based on data driving

Abstract

The invention discloses a distributed photovoltaic cluster output prediction method based on data driving, and relates to the technical field of photovoltaic output prediction. Comprising the following steps: defining a mean value of reference output values respectively corresponding to each node power station in each prediction cluster as calculated output values respectively corresponding to a first prediction cluster, a second prediction cluster and a third prediction cluster; and calculating to obtain a predicted output value of the distributed photovoltaic cluster, outputting the predicted output value, better reflecting the fluctuation condition of actual output by introducing a deviation value between the reference output value and the calibrated output value of the node power station, and comprehensively considering the spatial position and the output fluctuation. The problems of uneven spatial distribution and insufficient output fluctuation processing during output prediction of an existing distributed photovoltaic cluster are solved, and the prediction accuracy and reliability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic output prediction, and in particular is a distributed photovoltaic cluster output prediction method based on data-driven. Background Art

[0002] With the rapid development of renewable energy technology, distributed photovoltaic systems have been widely used around the world. Distributed photovoltaic systems have the advantages of flexible installation, local power supply, and reduced transmission losses, which are of great significance for promoting energy structure transformation and achieving sustainable development. However, the output of distributed photovoltaic systems is affected by many factors, such as solar radiation intensity, temperature, humidity, wind speed, power station equipment performance, operation and maintenance status, etc., resulting in volatility and intermittent output. This characteristic poses a great challenge to the dispatching and power quality management of the power grid. Distributed photovoltaic clusters are composed of multiple node power stations distributed in different locations, and their output is uncertain and volatile. Accurately predicting the output of distributed photovoltaic clusters is of great significance for the stable operation of the power grid, the reasonable dispatching of energy, and the optimal management of distributed photovoltaic systems.

[0003] However, in practical applications, the output characteristics of each node power station in a distributed photovoltaic cluster are different due to differences in geographical location, installation conditions, equipment performance and other factors. If this difference in spatial distribution is not taken into account during prediction, the prediction results will deviate from reality. Traditional prediction methods often regard all power stations as a whole for prediction, ignoring the differences between power stations and making it difficult to accurately reflect the output of each power station. Based on this, a data-driven distributed photovoltaic cluster output prediction method is proposed. Summary of the invention

[0004] The purpose of the present invention is to provide an information interaction method for power field operations, which solves the technical problem that if the differences in spatial distribution of each node power station in a distributed photovoltaic cluster are not taken into account during prediction, the prediction results will deviate from the actual ones.

[0005] The data-driven distributed photovoltaic cluster output prediction method includes the following steps: Step 1: The average of the daily output data of each node power station in the area where the distributed photovoltaic cluster is located within the preset number of days n is taken as the benchmark output value Bi corresponding to each node power station, where i refers to different node power stations, i=1, 2, ..., a, a refers to the number of node power stations in the distributed photovoltaic cluster, n and a are both positive integers, a≥1, n≥1; Step 2: Analyze the benchmark output values ​​corresponding to each node power station of the distributed photovoltaic cluster to obtain the calibrated output value corresponding to the distributed photovoltaic cluster; Step 3: According to the benchmark output values ​​corresponding to each node power station and the calibrated output values ​​corresponding to the distributed photovoltaic cluster, the deviation values ​​between the benchmark output values ​​of each node power station and the calibrated output values ​​are calculated; Step 4: Analyze the coordinates corresponding to the regional center of the area where the distributed photovoltaic cluster is located and each node power station in the area, and obtain the distance between each node power station and the regional center; Step 5: The sum of the distances between the coordinates of each node power station and the center of the region, the deviations between the benchmark output values ​​of each node power station and the calibrated output values, and the products of the preset fixed coefficients β1 and β2 are taken as the allocation values ​​corresponding to each node power station, and the preset fixed coefficients β1 and β2 satisfy 1=β1+β2 and β1>β2; Step 6: Divide each node power station into a first prediction cluster, a second prediction cluster, and a third prediction cluster according to the allocation values ​​corresponding to each node power station; Step 7: define the average of the benchmark output values ​​corresponding to each node power station in each prediction cluster as the calculated output values ​​H1, H2 and H3 corresponding to the first prediction cluster, the second prediction cluster and the third prediction cluster respectively; Step 8: According to the prediction clusters corresponding to the power stations with the expected output nodes on that day, the predicted output value of the distributed photovoltaic cluster is calculated and output.

[0006] As a further solution of the present invention: obtaining the calibrated output value corresponding to the distributed photovoltaic cluster, the specific method is: The discrete value U of the benchmark output value corresponding to each node power station of the distributed photovoltaic cluster is obtained. When the discrete value U is less than the preset value Y1, the mean value Bp of Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster. When U is greater than or equal to the preset value Y1, the values ​​corresponding to Bi with larger deviations from the mean value Bp are deleted in order from small to large, and the number b of deleted values ​​is recorded. At the same time, the discrete degree value is recalculated each time the discrete value is deleted until the discrete value satisfies the preset value Y1, and the number b of deleted values ​​is analyzed to obtain the calibrated output value C corresponding to the distributed photovoltaic cluster.

[0007] As a further solution of the present invention: the specific method of analyzing the number of deleted values ​​b to obtain the calibrated output value corresponding to the distributed photovoltaic cluster is: When the number of deletions b is greater than the preset value Y2, the average of the maximum and minimum values ​​in Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster. When the number of deletions b is less than or equal to the preset value Y2, the average value Bp of Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster.

[0008] As a further solution of the present invention: the specific method of obtaining the deviation value between the benchmark output value of each node power station and the calibrated output value is: The absolute value of the difference between the benchmark output value and the calibrated output value corresponding to each node power station is obtained, and the product of the absolute value of the difference and the preset parameter θ1 is defined as the deviation value Gi between the benchmark output value and the calibrated output value of each node power station.

[0009] As a further solution of the present invention: the specific method of obtaining the distance between each node power station and the regional center is: The regional center coordinates of the distributed photovoltaic cluster area are marked as Q (Qx, Qy), and the corresponding coordinates of each node power station in the distributed photovoltaic cluster area are Di (Dix, Diy) respectively, through the distance calculation formula, , calculate and obtain the distance Li between each node power station coordinate Di (Dix, Diy) and the regional center Q (Qx, Qy).

[0010] As a further solution of the present invention: the specific method of obtaining the distance between each node power station and the regional center is: The sum of the products of the distance Li between the coordinates of each node power station and the center of the region and the deviation value Gi between the benchmark output value and the calibrated output value of each node power station and the preset fixed coefficients β1 and β2 is taken as the allocation value Pi corresponding to each node power station. The preset fixed coefficients β1 and β2 satisfy 1=β1+β2 and β1>β2.

[0011] As a further solution of the present invention, the specific manner of dividing each node power station into the first prediction cluster, the second prediction cluster and the third prediction cluster is: Set preset thresholds Z1 and Z2, Z1 is greater than Z2, and the node power station with an allocation value greater than the preset threshold Z1 is defined as the first prediction cluster, the node power station with an allocation value greater than the preset threshold Z2 and less than or equal to the preset threshold Z1 is defined as the second prediction cluster, and the node power station with an allocation value Pi less than or equal to the preset threshold Z2 is defined as the third prediction cluster.

[0012] As a further solution of the present invention: the specific method of obtaining the predicted output value of the distributed photovoltaic cluster is: According to the prediction set corresponding to the node power station with expected output on that day, the number of node power stations with expected output G1, G2 and G3 corresponding to the first prediction cluster, the second prediction cluster and the third prediction cluster are obtained respectively. By the calculation formula, M=H1×G1+H2×G2+H3×G3, the predicted output value M of the distributed photovoltaic cluster is calculated and output.

[0013] Compared with the prior art, the present invention has the following beneficial effects: The present invention, by considering the relative position of the node power station and the regional center, effectively makes up for the shortcomings of the existing methods in dealing with the impact of uneven spatial distribution, which helps to more accurately predict the output of node power stations in different geographical locations. By introducing the deviation value between the benchmark output value and the calibrated output value of the node power station, the fluctuation of the actual output is better reflected, thereby improving the accuracy and reliability of the prediction. By comprehensively considering the spatial position and output fluctuation, the problems of uneven spatial distribution and insufficient output fluctuation processing in the output prediction of the existing distributed photovoltaic clusters are solved, and the accuracy and reliability of the output prediction of the distributed photovoltaic clusters are effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the system flow of the present invention; Figure 2 A schematic diagram of a process of determining the first prediction cluster, the second prediction cluster and the third prediction cluster according to the present invention; Figure 3 It is a schematic diagram of a method for obtaining a predicted output value according to the present invention. DETAILED DESCRIPTION

[0015] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0016] Embodiment 1 See also Figure 1-Figure 3 , the present application provides a distributed photovoltaic cluster output prediction method based on data-driven, comprising the following steps: Step 1: Obtain the output data of each node power station in the area where the distributed photovoltaic cluster is located for each day within a preset number of days n; Step 2: Calculate the average output data of each node power station within the preset number of days n, and obtain the average output data corresponding to each node power station within the preset number of days n, which is defined as the benchmark output value Bi corresponding to each node power station in the distributed photovoltaic cluster, where i refers to different node power stations, i=1, 2, ..., a, a refers to the number of node power stations in the distributed photovoltaic cluster, n and a are both positive integers, a≥1, n≥1; Step 3: Analyze the benchmark output values ​​corresponding to each node power station of the distributed photovoltaic cluster, and obtain the calibrated output value corresponding to the distributed photovoltaic cluster according to the analysis results. The specific method is as follows: By formula , where Bp is the mean value of Bi, Be is any value in Bi, and satisfies a≥e≥1. Obtain the discrete value U of the benchmark output value Bi corresponding to each node power station of the distributed photovoltaic cluster, and analyze the discrete value U. When the discrete value U is less than the preset value Y1, the mean value Bp of Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster; When U is greater than or equal to the preset value Y1, the values ​​with larger deviations from the mean value Bp are deleted in ascending order according to the values ​​corresponding to Bi, and the number of deleted values ​​b is recorded. At the same time, the discrete degree value is recalculated each time the value is deleted until the discrete value satisfies the preset value Y1. When the number of deletions b is greater than the preset value Y2, the average of the maximum and minimum values ​​in Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster, that is, C=(Bmax+Bmin) / 2 is used to calculate the calibrated output value C corresponding to the distributed photovoltaic cluster, where Bmax is the maximum value in Bi and Bmin is the minimum value in Bi. When the number of deletions b is less than or equal to the preset value Y2, the mean value Bp of Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster, where the specific values ​​of Y1 and Y2 are set by relevant personnel according to specific application scenarios and requirements; Step 4: According to the benchmark output values ​​corresponding to each node power station and the calibrated output values ​​corresponding to the distributed photovoltaic cluster, calculate the deviation values ​​between the benchmark output values ​​of each node power station and the calibrated output values. The specific method is as follows: Obtain the absolute value of the difference between the benchmark output value and the calibrated output value corresponding to each node power station, and define the product of the absolute value of the difference and the preset parameter θ1 as the deviation value Gi between the benchmark output value and the calibrated output value of each node power station, that is, through Gi=|Bi-C|×θ1, calculate the deviation value Gi between the benchmark output value and the calibrated output value of each node power station, where the specific value of the preset parameter θ1 is formulated by relevant personnel according to actual needs; Step 5: Obtain the coordinates of the regional center of the area where the distributed photovoltaic cluster is located and the corresponding coordinates of each node power station in the area, and analyze them. According to the analysis results, obtain the distance between each node power station and the regional center. The specific method is as follows: The coordinates of the regional center of the area where the distributed photovoltaic cluster is located are marked as Q (Qx, Qy), and the coordinates corresponding to each node power station in the area where the distributed photovoltaic cluster is located are marked as Di (Dix, Diy); According to the distance calculation formula, , calculate the distance Li between each node power station coordinate Di (Dix, Diy) and the regional center Q (Qx, Qy): Step 6: Comprehensively analyze the deviation between the benchmark output value of each node power station and the calibrated output value, as well as the distance between the coordinates Di (Dix, Diy) of each node power station and the regional center Q (Qx, Qy). According to the analysis results, the corresponding allocation values ​​of each node power station are obtained. The specific method is as follows: The sum of the products of the distance Li between the coordinates of each node power station and the regional center and the deviation value Gi between the benchmark output value and the calibrated output value of each node power station and the preset fixed coefficients β1 and β2 is taken as the distribution value Pi corresponding to each node power station, that is, through the formula, Pi=Li×β1+Gi×β2, where the specific values ​​of the preset fixed coefficients β1 and β2 are formulated by relevant personnel according to actual needs, satisfying 1=β1+β2 and β1>β2; Step 7: Divide each node power station into multiple different prediction clusters according to the corresponding allocation values. The specific method is as follows: Set preset thresholds Z1 and Z2, Z1 is greater than Z2, define the node power station whose allocation value Pi is greater than the preset threshold Z1 as the first prediction cluster, define the node power station whose allocation value Pi is greater than the preset threshold Z2 and less than or equal to the preset threshold Z1 as the second prediction cluster, and define the node power station whose allocation value Pi is less than or equal to the preset threshold Z2 as the third prediction cluster; The benchmark output value is obtained by calculating the mean value of the output data of each node power station within the preset number of days, which provides basic data for subsequent analysis, making the prediction more stable and reliable. The benchmark output value is analyzed for discrete values. When the discrete value is small, the mean value is used as the calibration output value to ensure accuracy when the data is relatively concentrated; when the discrete value is large, the value with large deviation is deleted and the degree of dispersion is recalculated until the conditions are met, or when the number of deletions exceeds the preset value, the mean of the maximum and minimum values ​​is used as the calibration output value, which effectively eliminates the interference of abnormal data and improves the accuracy of the prediction; A comprehensive analysis is performed based on the deviation between the benchmark output value and the calibrated output value of each node power station and the distance from the regional center to obtain the allocation value, and then divide the prediction cluster to make the prediction more precise and better adapt to the characteristics of different node power stations.

[0017] Embodiment 2 As the second embodiment of the present invention, when the present application is implemented, compared with the first embodiment, the technical solution of the second embodiment is different from the first embodiment only in that the second embodiment further includes step eight and step nine; Step 8: According to the benchmark output values ​​corresponding to the node power stations in each prediction cluster, the average of the benchmark output values ​​corresponding to the node power stations in each prediction cluster is defined as the calculated output values ​​corresponding to each prediction cluster, thereby obtaining the calculated output values ​​H1, H2 and H3 corresponding to the first prediction cluster, the second prediction cluster and the third prediction cluster respectively; Step 9: According to the predicted clusters corresponding to the power stations at each expected output node on that day, the predicted output value of the distributed photovoltaic cluster is calculated. The specific method is as follows: According to the prediction set corresponding to the node power station with the expected output on that day, the number of node power stations with the expected output corresponding to the first prediction cluster, the second prediction cluster and the third prediction cluster are obtained respectively. The predicted output value M of the distributed photovoltaic cluster is calculated and output by the calculation formula M=H1×G1+H2×G2+H3×G3; By considering the relative position of the node power station and the regional center, the shortcomings of the existing methods in dealing with the impact of uneven spatial distribution are effectively compensated, which helps to more accurately predict the output of node power stations in different geographical locations. By introducing the deviation value between the benchmark output value and the calibrated output value of the node power station, the fluctuation of the actual output is better reflected, thereby improving the accuracy and reliability of the prediction. By comprehensively considering the spatial position and output fluctuation, the problems of uneven spatial distribution and insufficient output fluctuation processing in the output prediction of the existing distributed photovoltaic cluster are solved, the accuracy and reliability of the prediction are improved, and the accuracy and reliability of the output prediction of the distributed photovoltaic cluster based on data-driven is effectively improved.

[0018] Embodiment 3 As the third embodiment of the present invention, when the present application is specifically implemented, compared with the first and second embodiments, the technical solution of this embodiment is to combine the solutions of the first and second embodiments mentioned above for implementation.

[0019] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0020] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A data-driven distributed photovoltaic cluster output prediction method, characterized in that: The following steps are involved: Step 1: The average of the daily output data of each node power station in the area where the distributed photovoltaic cluster is located within the preset number of days n is taken as the benchmark output value Bi corresponding to each node power station, where i refers to different node power stations, i=1, 2, ..., a, a refers to the number of node power stations in the distributed photovoltaic cluster, n and a are both positive integers, a≥1, n≥1; Step 2: Analyze the benchmark output values ​​corresponding to each node power station of the distributed photovoltaic cluster to obtain the calibrated output value corresponding to the distributed photovoltaic cluster; Step 3: According to the benchmark output values ​​corresponding to each node power station and the calibrated output values ​​corresponding to the distributed photovoltaic cluster, the deviation values ​​between the benchmark output values ​​of each node power station and the calibrated output values ​​are calculated; Step 4: Analyze the coordinates corresponding to the regional center of the area where the distributed photovoltaic cluster is located and each node power station in the area, and obtain the distance between each node power station and the regional center; Step 5: The sum of the distances between the coordinates of each node power station and the center of the region, the deviations between the benchmark output values ​​of each node power station and the calibrated output values, and the products of the preset fixed coefficients β1 and β2 are taken as the allocation values ​​corresponding to each node power station, and the preset fixed coefficients β1 and β2 satisfy 1=β1+β2 and β1>β2; Step 6: Divide each node power station into a first prediction cluster, a second prediction cluster, and a third prediction cluster according to the allocation values ​​corresponding to each node power station; Step 7: define the average of the benchmark output values ​​corresponding to each node power station in each prediction cluster as the calculated output values ​​H1, H2 and H3 corresponding to the first prediction cluster, the second prediction cluster and the third prediction cluster respectively; Step 8: According to the prediction clusters corresponding to the power stations with the expected output nodes on that day, the predicted output value of the distributed photovoltaic cluster is calculated and output.

2. The data-driven distributed photovoltaic cluster output prediction method according to claim 1 is characterized in that: The calibrated output value corresponding to the distributed photovoltaic cluster is obtained in the following way: The discrete value U of the benchmark output value corresponding to each node power station of the distributed photovoltaic cluster is obtained. When the discrete value U is less than the preset value Y1, the mean value Bp of Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster. When U is greater than or equal to the preset value Y1, the values ​​corresponding to Bi with larger deviations from the mean value Bp are deleted in order from small to large, and the number b of deleted values ​​is recorded. At the same time, the discrete degree value is recalculated each time the discrete value is deleted until the discrete value satisfies the preset value Y1, and the number b of deleted values ​​is analyzed to obtain the calibrated output value C corresponding to the distributed photovoltaic cluster.

3. The data-driven distributed photovoltaic cluster output prediction method according to claim 2 is characterized in that: The specific method of analyzing the number of deleted values ​​b to obtain the calibrated output value corresponding to the distributed photovoltaic cluster is: When the number of deletions b is greater than the preset value Y2, the average of the maximum and minimum values ​​in Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster. When the number of deletions b is less than or equal to the preset value Y2, the average value Bp of Bi is used as the calibrated output value C corresponding to the distributed photovoltaic cluster.

4. The data-driven distributed photovoltaic cluster output prediction method according to claim 3 is characterized in that: The specific method of obtaining the deviation value between the benchmark output value of each node power station and the calibrated output value is as follows: The absolute value of the difference between the benchmark output value and the calibrated output value corresponding to each node power station is obtained, and the product of the absolute value of the difference and the preset parameter θ1 is defined as the deviation value Gi between the benchmark output value and the calibrated output value of each node power station.

5. The method for predicting output of distributed photovoltaic clusters based on data drive according to claim 4 is characterized in that: The specific method of obtaining the distance between each node power station and the regional center is: The coordinates of the regional center of the area where the distributed photovoltaic cluster is located are marked as Q (Qx, Qy), and the coordinates corresponding to each node power station in the area where the distributed photovoltaic cluster is located are marked as Di (Dix, Diy); According to the distance calculation formula, , calculate and obtain the distance Li between each node power station coordinate Di (Dix, Diy) and the regional center Q (Qx, Qy).

6. The method for predicting output of distributed photovoltaic clusters based on data drive according to claim 5 is characterized in that: The specific method of obtaining the distance between each node power station and the regional center is: The sum of the products of the distance Li between the coordinates of each node power station and the center of the region and the deviation value Gi between the benchmark output value and the calibrated output value of each node power station and the preset fixed coefficients β1 and β2 is taken as the allocation value Pi corresponding to each node power station. The preset fixed coefficients β1 and β2 satisfy 1=β1+β2 and β1>β2.

7. The method for predicting output of distributed photovoltaic clusters based on data drive according to claim 6 is characterized in that: The specific method of dividing each node power station into the first prediction cluster, the second prediction cluster and the third prediction cluster is: Set preset thresholds Z1 and Z2, Z1 is greater than Z2, and the node power station with an allocation value greater than the preset threshold Z1 is defined as the first prediction cluster, the node power station with an allocation value greater than the preset threshold Z2 and less than or equal to the preset threshold Z1 is defined as the second prediction cluster, and the node power station with an allocation value Pi less than or equal to the preset threshold Z2 is defined as the third prediction cluster.

8. The data-driven distributed photovoltaic cluster output prediction method according to claim 7 is characterized in that: The specific method for obtaining the predicted output value of the distributed photovoltaic cluster is: According to the prediction set corresponding to the node power station with expected output on that day, the number of node power stations with expected output G1, G2 and G3 corresponding to the first prediction cluster, the second prediction cluster and the third prediction cluster are obtained respectively. By the calculation formula, M=H1×G1+H2×G2+H3×G3, the predicted output value M of the distributed photovoltaic cluster is calculated and output.

Images