A new energy power forecast error distribution analysis method considering the differences in new energy consumption trends

Abstract

The present invention relates to a new energy power prediction error distribution analysis method that takes into account the differences in new energy consumption trends. The method includes a two-dimensional binning method for new energy power based on the new energy prediction value level and the new energy consumption trend level, and a mapping method for prediction error samples to prediction error probability distribution quantiles based on the Cornish-Fisher expansion technology. On the basis of taking into account the differences in new energy power prediction error distribution under different new energy prediction value levels, the present invention fully considers the impact of the differences in new energy consumption trends on the new energy power prediction error distribution, making the establishment of the new energy power prediction error distribution model more accurate, which is conducive to improving the accuracy of the scheduling model and further promoting the consumption of new energy.

Description

Technical Field

[0001] The present invention relates to the field of new energy prediction error modeling, and in particular to a new energy power prediction error distribution analysis method taking into account differences in new energy consumption situations. Background Art

[0002] Amidst an overall decline in global energy demand, the development and utilization of renewable energy has demonstrated greater resilience. It is projected that between 2020 and 2030, renewable energy demand will grow by two-thirds, accounting for approximately 80% of the increase in global electricity demand. Wind and solar power, as representatives of renewable energy, have gained widespread favor among energy industry practitioners. As wind and photovoltaic power generation technologies mature and development costs decline, their penetration in my country's power grid continues to rise, and their installed capacity is also increasing annually.

[0003] However, due to the randomness and volatility of renewable energy output, the large-scale integration of wind and photovoltaic power into the grid has also impacted its stable operation. When the grid lacks sufficient flexible and rapid regulation resources, the integration of renewable energy can lead to wind and solar curtailment and load shedding, placing an additional economic burden on grid operations. Therefore, fully utilizing renewable energy and increasing its absorption capacity has become a current research hotspot. Establishing a probability distribution of renewable energy power forecast errors, enabling dispatchers to formulate more appropriate thermal power output plans, is an effective technique for improving renewable energy absorption. Most current research clusters and bins historical error data based on the level of renewable energy forecast values, thereby constructing corresponding forecast error probability distribution models. While this binning approach accounts for the impact of different forecast values on forecast error, it fails to consider the coupled relationship between the overall renewable energy absorption trend and forecast error.

[0004] The present invention is based on the mining and analysis of historical renewable energy consumption data. On the basis of binning according to the new energy output forecast value, it fully considers the impact of the differences in new energy consumption trends on the prediction error, and performs two-dimensional binning according to the forecast value level and the consumption trend level, thereby establishing a more accurate new energy power prediction error distribution model. Summary of the Invention

[0005] The present invention provides a new energy power prediction error distribution analysis method that takes into account the differences in new energy consumption trends. Its purpose is to consider the new energy consumption trends in the process of establishing the new energy power prediction error distribution, so that the prediction error distribution model can reflect the impact of different new energy consumption trends, thereby further improving the new energy consumption space.

[0006] The present invention is achieved through the following technical solutions:

[0007] A new energy power forecast error distribution analysis method considering the differences in new energy consumption trends, the technical route includes:

[0008] Set multiple new energy forecast value levels according to the installed capacity of new energy power plants;

[0009] Set multiple new energy consumption levels based on the historical new energy consumption data of the region where the new energy power plant belongs;

[0010] Furthermore, by mining and analyzing the historical forecast values and forecast error data of new energy power plants and the historical new energy consumption data of the regions where the new energy power plants are located, a new energy power forecast error distribution analysis method that takes into account the differences in new energy consumption trends is established, including:

[0011] Step 1: Collect historical forecast values and forecast error data of new energy power plants, bin the historical forecast error data set according to the set new energy forecast value level, and obtain the historical forecast error sample set of new energy power plants under each forecast value level;

[0012] Step 2: Collect the historical new energy consumption in the region where the new energy power plant is located. According to the set new energy consumption situation level, further bin the historical error sample set of the new energy power plant at each prediction value level obtained in step 1 to obtain the historical prediction error sample set of the new energy power plant at each prediction value level and each new energy consumption situation level.

[0013] Step 3: Perform data normalization on the historical error sample set obtained through two-dimensional binning in step 2;

[0014] Step 4: For the historical forecast error sample set of new energy power plants normalized in step 3, establish the forecast error probability distribution corresponding to the historical forecast error sample set based on the Cornish-Fisher series expansion technique. The calculation is performed using the Cornish-Fisher expansion of the fifth-order cumulant, and the specific formula is as follows:

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] In formulas (1)-(7), e j is an element in the historical error sample set, N d is the number of historical forecast error samples in each bin after the binning process in steps 1-3; μ and σ represent the mean and standard deviation of the historical forecast error sample set, respectively; u3, u4, and u5 are the 3rd to 5th order origin moments of the forecast error samples after data standardization; ξ(q) is the quantile of the standard normal distribution corresponding to the probability value q; F -1 (q) is the quantile of the probability value q corresponding to the forecast error distribution (CDF) obtained by expanding the historical forecast error sample through the Cornish-Fisher series after denormalization.

[0023] Step 5: During day-ahead scheduling, collect the next day's renewable energy power forecast and renewable energy consumption status values, find their corresponding renewable energy power forecast level and renewable energy consumption status level, and obtain the corresponding renewable energy power forecast error probability distribution. This forecast error probability distribution is then superimposed on the renewable energy power forecast to obtain the renewable energy power probability distribution. Based on the renewable energy power probability distribution, unit combinations are performed to develop the optimal thermal power output plan.

[0024] Therefore, the present invention has the following advantages: on the basis of taking into account the differences in the distribution of new energy power prediction errors under different new energy prediction value levels, the present invention fully considers the impact of the differences in new energy consumption trends on the distribution of new energy power prediction errors, making the establishment of the new energy power prediction error distribution model more accurate, which is conducive to improving the accuracy of the scheduling model and promoting the consumption of new energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Attachment Figure 1 It is a schematic flow chart of the method of the present invention.

[0026] Attachment Figure 2 3. It is a schematic diagram comparing the prediction error distribution model results with and without considering the differences in new energy consumption trends in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The technical solution of the present invention will be further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0028] Example:

[0029] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings and examples. It should be noted that the examples are exemplary and are used to explain the present invention, but should not be interpreted as limiting the present invention.

[0030] The process of establishing a new energy power forecast error distribution analysis method considering the differences in new energy consumption trends is as follows: Figure 1 shown.

[0031] Taking the renewable energy consumption data from June to December 2017 in a certain region of China as an example, a wind farm in this region was selected to conduct wind power forecast error modeling considering the differences in renewable energy consumption trends. The basic parameters are shown in Table 1:

[0032] Table 1 Basic parameters

[0033]

[0034] Based on the parameters in Table 1, the wind power forecast value level and new energy consumption level are formulated according to Table 2:

[0035] Table 2 Classification

[0036]

[0037]

[0038] The predicted value and prediction error samples of the wind farm are counted. All the prediction error samples are divided into 25 boxes according to Table 2. For each box, the corresponding prediction error probability distribution is established using the Cornish-Fisher expansion technique described by equations (1)-(7). The following uses the results of wind power prediction value level 1 as an example to illustrate the necessity of considering the new energy consumption situation and dividing the boxes according to the new energy consumption level:

[0039] Depend on Figure 2 The bold black line represents the result of binning by wind power forecast level, ignoring the impact of renewable energy consumption trends. The remaining curves represent the results of binning by renewable energy consumption level, based on the wind power forecast level. As can be seen from the figure, due to differences in renewable energy consumption trends, the distribution of wind power forecast errors at different consumption levels also exhibits significant differences. This also illustrates the necessity of considering the impact of different renewable energy consumption trends when establishing renewable energy forecast error distribution models.

[0040] After obtaining the wind power forecast error distribution, refer to step 5 to collect relevant data on the day before and obtain the corresponding wind power probability distribution for day-ahead scheduling calculation.

[0041] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Persons skilled in the art may make various modifications, additions, or substitutions to the described specific embodiments without departing from the spirit of the present invention or exceeding the scope of the appended claims.

Claims

1. A new energy power forecast error distribution analysis method considering the differences in new energy consumption trends, characterized by: include Step 1: Collect historical forecast values and forecast error data of new energy power plants, bin the historical forecast error data set according to the set new energy forecast value level, and obtain the historical forecast error sample set of new energy power plants under each forecast value level; Step 2: Collect the historical new energy consumption in the region where the new energy power plant is located. According to the set new energy consumption situation level, further bin the historical error sample set of the new energy power plant at each prediction value level obtained in step 1 to obtain the historical prediction error sample set of the new energy power plant at each prediction value level and each new energy consumption situation level. Step 3: Perform data normalization on the historical error sample set obtained through two-dimensional binning in step 2; Step 4: For the historical forecast error sample set of new energy power plants normalized in step 3, establish the forecast error probability distribution corresponding to the historical forecast error sample set based on the Cornish-Fisher series expansion technology; use the Cornish-Fisher expansion of the fifth-order cumulant for calculation, and the specific formula is as follows: In formulas (1)-(7), e j is an element in the historical error sample set, N d is the number of historical forecast error samples in each bin after the binning process in steps 1-3; μ and σ represent the mean and standard deviation of the historical forecast error sample set, respectively; u3, u4, and u5 are the 3rd to 5th order origin moments of the forecast error samples after data standardization; ξ(q) is the quantile of the standard normal distribution corresponding to the probability value q; F -1 (q) is the quantile of the probability value q corresponding to the forecast error distribution (CDF) obtained by expanding the historical forecast error sample through the Cornish-Fisher series after denormalization; Step 5: In the day-ahead scheduling, collect the new energy power forecast value and new energy consumption status value for the next day, find the new energy power forecast level and new energy consumption status level to which it belongs, obtain the corresponding new energy power forecast error probability distribution, and superimpose the forecast error probability distribution on the new energy power forecast value to obtain the new energy power probability distribution; based on the probability distribution of new energy power, combine the units to formulate the optimal thermal power output plan.

Images