A photovoltaic power generation load prediction method based on prediction accuracy

Abstract

The application provides a photovoltaic power generation load prediction method based on prediction accuracy, comprising the following steps: (1) calling load values of the previous 7 days of the photovoltaic power generation prediction load to form a sample set; (2) establishing a decision hyperplane; (3) calculating distances of sample points to the decision hyperplane; (4) finding an optimal hyperplane by calculating distances of each sample point to the decision hyperplane, so that the sample points farthest from the decision hyperplane on both sides of the hyperplane are closest to the hyperplane; (5) calculating the photovoltaic power generation prediction load; (6) calculating average prediction accuracy of the previous 7 days of the photovoltaic power generation prediction load; and (7) correcting the photovoltaic power generation prediction load according to the average prediction accuracy to obtain a final load prediction value. The application provides a photovoltaic power generation load prediction method based on prediction accuracy, which can more accurately predict photovoltaic power generation load.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of photovoltaic power generation load prediction, and particularly relates to a photovoltaic power generation load prediction method based on prediction accuracy. BACKGROUND

[0002] With the continuous development of science and technology, in order to protect traditional energy and ecological environment, people begin to widely use new energy such as solar power generation. However, due to the addition of new energy to the power grid, how to better allocate electric energy to ensure its economy and reliability needs to be properly distributed. In order to better allocate energy, if the solar power generation can be accurately predicted, the energy allocation work will be greatly facilitated and reasonable allocation will be ensured.

[0003] The application provides a photovoltaic power generation load prediction method based on prediction accuracy, a decision hyperplane is established according to historical data, an optimal hyperplane is found, the sample points farthest from the hyperplane on both sides of the decision hyperplane are farthest from the hyperplane, and then the load on the day is predicted according to the load before the photovoltaic power generation prediction load; the average prediction accuracy of the load 7 days before the photovoltaic power generation prediction load is calculated, the photovoltaic power generation prediction load is corrected according to the average prediction accuracy, and a more accurate prediction load is obtained. SUMMARY

[0004] The application provides a photovoltaic power generation load prediction method based on prediction accuracy, which can more accurately predict the photovoltaic power generation load.

[0005] The application specifically provides a photovoltaic power generation load prediction method based on prediction accuracy, and the photovoltaic power generation load prediction method comprises the following steps:

[0006] Step (1): the load values 7 days before the photovoltaic power generation prediction load are called to constitute a sample set;

[0007] Step (2): a decision hyperplane is established;

[0008] Step (3): the distance of the sample points from the decision hyperplane is calculated;

[0009] Step (4): an optimal hyperplane is found by calculating the distance of each sample point from the decision hyperplane, so that the sample points farthest from the hyperplane on both sides of the decision hyperplane are farthest from the hyperplane;

[0010] Step (5): the photovoltaic power generation prediction load is calculated out;

[0011] Step (6): the average prediction accuracy of the load 7 days before the photovoltaic power generation prediction load is calculated;

[0012] Step (7): correcting the photovoltaic power generation predicted load according to the average prediction accuracy to obtain a final load prediction value.

[0013] The method for establishing the decision hyperplane is f(x) = w T x + b, wherein w T is a weight, (x, y) is the sample set, and b is a displacement phase.

[0014] The algorithm for calculating the distance of a sample point from the decision hyperplane is y i - f(x) = y i - w T x - b, wherein i = 1, 2, …, 7.

[0015] The specific method for calculating the photovoltaic power generation predicted load comprises:

[0016] (1) calculating the photovoltaic power generation predicted load from the load value of the day before the photovoltaic power generation predicted load;

[0017] (2) calculating the photovoltaic power generation predicted load from the load value of the seventh day before the photovoltaic power generation predicted load;

[0018] (3) calculating the average value of the photovoltaic power generation predicted load from the load value of the day before the photovoltaic power generation predicted load and the photovoltaic power generation predicted load from the load value of the seventh day before the photovoltaic power generation predicted load as the photovoltaic power generation predicted load.

[0019] The algorithm for calculating the photovoltaic power generation predicted load is: The constraint condition is wherein C is a penalty factor, and ξ i is a relaxation factor.

[0020] The algorithm for calculating the average prediction accuracy of the load of the seventh day before the photovoltaic power generation predicted load is: wherein y i is the actual load value of the seventh day before the photovoltaic power generation predicted load.

[0021] The method for correcting the photovoltaic power generation predicted load according to the average prediction accuracy is:

[0022]

[0023] Compared with the existing technology, the beneficial effects are: the photovoltaic power generation load forecasting method establishes a decision hyperplane based on historical data, finds the optimal hyperplane, makes the sample points on both sides of the decision hyperplane closest to the hyperplane the farthest away, and then predicts the load on the current day based on the load 1 day and 7 days before the photovoltaic power generation forecast load; then calculates the average forecast accuracy of the load 7 days before the photovoltaic power generation forecast load, and corrects the photovoltaic power generation forecast load according to the average forecast accuracy to obtain a more accurate forecast load. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The present invention is a workflow diagram of a photovoltaic power generation load prediction method based on prediction accuracy. DETAILED DESCRIPTION

[0025] The following describes in detail a specific implementation of a photovoltaic power generation load prediction method based on prediction accuracy according to the present invention with reference to the accompanying drawings.

[0026] like Figure 1 As shown, the photovoltaic power generation load prediction method of the present invention includes the following steps:

[0027] Step (1): retrieve the load values ​​of the photovoltaic power generation forecast load for the previous 7 days to form a sample set;

[0028] Step (2): Establish the decision hyperplane f(x) = w T x+b, where w T is the weight, (x, y) is the sample set, and b is the displacement phase;

[0029] Step (3): Calculate the distance y between the sample point and the decision hyperplane i -f(x)=y i -w T xb, where i = 1, 2, ..., 7;

[0030] Step (4): finding an optimal hyperplane by calculating the distance between each sample point and the decision hyperplane, so that the sample points closest to the hyperplane on both sides of the decision hyperplane are the farthest away;

[0031] Step (5): Calculating the photovoltaic power generation forecast load;

[0032] Step (6): Calculate the average load prediction accuracy of the photovoltaic power generation forecast load for the previous 7 days;

[0033] Step (7): Correct the photovoltaic power generation predicted load according to the average prediction accuracy to obtain a final load prediction value.

[0034] The specific method for calculating the photovoltaic power generation predicted load includes:

[0035] (1) According to The constraint condition is Wherein C is a penalty factor, ξ i is a relaxation factor, the photovoltaic power generation predicted load y pre-1 is calculated according to the load value of the day before the photovoltaic power generation predicted load.

[0036] (2) The photovoltaic power generation predicted load y pre-7 is calculated according to the load value of the 7 days before the photovoltaic power generation predicted load.

[0037] (3) The photovoltaic power generation predicted load is calculated as the average value of the photovoltaic power generation predicted load calculated according to the load value of the day before the photovoltaic power generation predicted load and the photovoltaic power generation predicted load calculated according to the load value of the 7 days before the photovoltaic power generation predicted load.

[0038] The algorithm for calculating the average prediction accuracy of the 7 days before the photovoltaic power generation predicted load is: Wherein y i is the actual load value of the 7 days before the photovoltaic power generation predicted load.

[0039] The method for correcting the photovoltaic power generation predicted load according to the average prediction accuracy is:

[0040]

[0041] Finally, it should be noted that the above embodiments only illustrate the technical solutions of the present application and do not limit the same. It should be understood by those skilled in the art that the specific embodiments of the present application can be modified or replaced by equivalents, and these modifications or changes are within the scope of protection of the claims.

Claims

1.A photovoltaic power generation load prediction method based on prediction accuracy, characterized by, The photovoltaic power generation load prediction method comprises the following steps: Step (1): retrieve the load values of the previous 7 days of the photovoltaic power generation prediction load to constitute a sample set; Step (2): establish a decision hyperplane; Step (3): calculate the distance of the sample points from the decision hyperplane; Step (4): find an optimal hyperplane by calculating the distance of each sample point from the decision hyperplane, so that the sample points farthest from the decision hyperplane on both sides of the hyperplane are closest to the hyperplane; Step (5): calculate the photovoltaic power generation prediction load, which comprises the following steps: (1) calculate the photovoltaic power generation prediction load obtained from the load value of the previous day of the photovoltaic power generation prediction load; (2) calculate the photovoltaic power generation prediction load obtained from the load value of the previous 7 days of the photovoltaic power generation prediction load; (3) calculate the average value of the photovoltaic power generation prediction load obtained from the load value of the previous day of the photovoltaic power generation prediction load and the photovoltaic power generation prediction load obtained from the load value of the previous 7 days of the photovoltaic power generation prediction load, as the photovoltaic power generation prediction load; The algorithm for calculating the photovoltaic power generation predicted load is: The constraint condition is Wherein C is a penalty factor, ξ i is a relaxation factor; Step (6): calculate the average prediction accuracy of the load of the previous 7 days of the photovoltaic power generation prediction load, and the specific algorithm is: where y i is the actual load value of the previous 7 days of the photovoltaic power generation load prediction; Step (7): correcting the photovoltaic power generation predicted load according to the average prediction accuracy to obtain a final load prediction value 2.The photovoltaic power generation load prediction method based on prediction accuracy according to claim 1, wherein, The method of establishing the decision hyperplane is: f(x) = w T x + b, where w T is a weight, (x, y) is the sample set, and b is a displacement phase. 3.The photovoltaic power generation load prediction method based on prediction accuracy according to claim 2, characterized in that, The algorithm to calculate the distance of the sample point from the decision hyperplane distance is: y i -f(x) = y i -w T x - b, where i = 1, 2, ···, 7.

Images