A method for storing wind power generation data

Abstract

The present invention relates to the technical field of data storage, and in particular to a method for storing data of wind power generation. The method comprises: performing curve fitting on the actual power generation and predicted power generation at all acquisition moments in the current storage cycle to obtain the actual power generation curve and the predicted power generation curve; dividing the current storage cycle into multiple time periods based on the peaks and troughs of the actual power generation curve; obtaining the importance according to the actual power generation at each acquisition moment in each time period, the predicted power generation, the slope corresponding to each actual power generation, and the slope corresponding to the predicted power generation; obtaining the fitting difficulty according to the actual power generation at all acquisition moments in each time period and the predicted power generation at all acquisition moments, and then obtaining the coding priority index corresponding to each time period; encoding and storing the actual power generation of each time period based on the coding priority index. The present invention can ensure that important power generation data of a wind farm is not lost, and effectively improve the prediction accuracy of subsequent power generation.

Description

Technical Field

[0001] The present invention relates to the technical field of data storage, and in particular to a method for storing wind power generation data. Background Art

[0002] After wind power generation was incorporated into the national grid, the necessity of storing its data increased. In order to ensure the safe and stable operation of the system, wind farms need to report their power generation plans to the grid. If the prediction error exceeds a certain range, the wind farm must pay a fine to the grid operator. At the same time, in order to predict subsequent data, the data storage of wind farms is also very important. However, the data generated by wind farms is constantly updated. Whether it is prediction data or real-time measurement data, it is necessary to store data information for a long period of time in the system. Therefore, the amount of data that needs to be stored is extremely large. The conventional data storage method is to average the data within a fixed period of time and store it directly. In this way, the amount of data stored is still large, and some important data information is lost, resulting in a reduction in its reference value when used for prediction models. Summary of the invention

[0003] In order to solve the problem of some important information loss in the existing method when storing power generation data, the purpose of the present invention is to provide a method for storing wind power generation data, and the technical solution adopted is as follows:

[0004] The present invention provides a method for storing wind power generation data, the method comprising the following steps:

[0005] Obtain the actual power generation of the wind farm at each collection moment in the current storage cycle; obtain the predicted power generation at each collection moment in the current storage cycle based on the actual power generation at each collection moment in the historical storage cycle;

[0006] Performing curve fitting on the actual power generation at all acquisition moments and the predicted power generation at all acquisition moments in the current storage period respectively, to obtain the actual power generation curve and the predicted power generation curve corresponding to the current storage period; obtaining the peaks and troughs of the actual power generation curve, and dividing the current storage period into time periods based on the peaks and troughs;

[0007] According to the actual power generation at each collection moment in each time period, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment, the importance of the power generation data in each time period is obtained; according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments, the number of target collection moments in each time period is counted; according to the number of target collection moments, the fitting difficulty corresponding to the power generation data in each time period is obtained; the slope corresponding to the predicted power generation is obtained according to the predicted power generation curve, and the slope corresponding to the actual power generation is obtained according to the actual power generation curve;

[0008] Based on the importance and the fitting difficulty, a coding priority index corresponding to each time period is obtained; and based on the coding priority index, the actual power generation in each time period is encoded and stored.

[0009] Preferably, counting the number of target collection moments in each time period according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments includes:

[0010] The target acquisition time includes a first target acquisition time and a second target acquisition time;

[0011] For any time period within the current storage cycle:

[0012] The actual power generation at each collection time in the time period is fitted to obtain the function of the actual power generation with respect to the collection time, and the derivative is derived to obtain the derivative value corresponding to each collection time, and the collection time with a derivative value of 0 is recorded as the first target collection time, and the number of the first target collection time is counted;

[0013] The predicted power generation at each collection time within the time period is fitted to obtain the function of the predicted power generation with respect to the collection time, and the derivative is derived to obtain the derivative value corresponding to each collection time. The collection time with a derivative value of 0 is recorded as the second target collection time, and the number of the second target collection time is counted.

[0014] Preferably, obtaining the fitting difficulty corresponding to the power generation data in each time period according to the number of the target collection moments includes:

[0015] For any time period within the current storage cycle:

[0016] The ratio of the number of the first target collection moments to the number of all collection moments in the time period is calculated as the first target collection moment proportion; the ratio of the number of the second target collection moments to the number of all collection moments in the time period is calculated as the second target collection moment proportion, and the sum of the first target collection moment proportion and the second target collection moment proportion is taken as the fitting difficulty corresponding to the power generation data of the time period.

[0017] Preferably, obtaining the importance of power generation data in each time period according to the actual power generation at each collection moment in each time period, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment includes:

[0018] For any time period in the current storage cycle:

[0019] For any collection moment in the time period: calculate the absolute value of the difference between the actual power generation at the collection moment and the predicted power generation, record it as the first absolute value, perform normalization on the first absolute value, and use the normalization result as the power generation difference; calculate the absolute value of the difference between the slope corresponding to the actual power generation at the collection moment and the slope corresponding to the predicted power generation, as the slope difference, calculate the product of the amplitude adjustment coefficient and the absolute value of the slope corresponding to the actual power generation at the collection moment, record it as the first product, and use the value of the exponential function with the natural constant as the base and the negative of the first product as the exponent as the adjustment weight; calculate the product of the power generation difference, the slope difference and the adjustment weight as the importance of the power generation data at the collection moment;

[0020] The sum of the importance of the power generation data at all the collection moments in the time period is calculated as the importance of the power generation data in the time period.

[0021] Preferably, obtaining the coding priority index corresponding to each time period based on the importance and the fitting difficulty includes:

[0022] For any time period within the current storage cycle:

[0023] The importance of the power generation data for the time period and the fitting difficulty corresponding to the power generation data for the time period are normalized respectively, and the sum of the normalized value of the importance of the power generation data for the time period and the normalized value of the fitting difficulty corresponding to the power generation data for the time period is calculated as the coding priority indicator corresponding to the time period.

[0024] Preferably, the current storage cycle is divided based on the peaks and troughs to obtain each time period, including: the collection moment corresponding to each peak is a division point, the collection moment corresponding to each trough is a division point, and the current storage cycle is divided based on all division points to obtain each time period.

[0025] Preferably, obtaining the predicted power generation at each collection moment in the current storage period based on the actual power generation at each collection moment in the historical storage period includes:

[0026] Based on the actual power generation at each collection moment in the historical storage period, a power generation data sequence corresponding to the historical storage period is constructed; the power generation data sequence corresponding to the historical storage period is input into the prediction model to obtain the predicted power generation at each collection moment in the current storage period.

[0027] The encoding and storing of the actual power generation in each time period based on the encoding priority index includes: encoding and storing the actual power generation in each time period in the current storage cycle respectively in descending order of the encoding priority index.

[0028] The present invention has at least the following beneficial effects:

[0029] 1. The present invention takes into account that when storing power generation data, the existing data storage method is likely to cause the loss of some important power generation data. In order to ensure the safe and stable operation of the system, the wind farm needs to report its power generation plan to the power grid. If its prediction error exceeds a certain range, the wind farm needs to pay a fine to the power grid operator. Therefore, if some important power generation data is lost, the prediction accuracy of subsequent power generation data will be reduced. The present invention obtains the importance of power generation data in each time period according to the actual power generation at each collection moment in each time period in the current storage cycle, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment. The greater the importance, the more appropriate the power generation data in the corresponding time period is. Suitable for predicting subsequent power generation; the present invention also counts the number of target collection moments in each time period according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments, and then obtains the fitting difficulty corresponding to the power generation data of each time period. The greater the fitting difficulty, the greater the proportion of important data in the corresponding time period. Therefore, when encoding and storing, the power generation data in the corresponding time period should be encoded first; based on this, this embodiment obtains the coding priority index corresponding to each time period based on the importance and fitting difficulty, and encodes and stores the actual power generation of each time period in the current storage cycle based on the coding priority index, so that important power generation data of the wind farm is not lost, which can effectively improve the prediction accuracy of subsequent power generation and reduce the number of fines.

[0030] 2. The present invention takes into account that the amount of power generation data in a cycle is large, and there are several large fluctuation points in the power generation data in the entire cycle, and the data similarity between every two adjacent fluctuation points is relatively high. Therefore, curve fitting is performed on the actual power generation at all acquisition moments in the current storage cycle, and the peaks and troughs of the actual power generation curve are obtained. The current storage cycle is divided into multiple time periods based on the peaks and troughs, and the actual power generation in each time period is relatively similar. Subsequently, the power generation data in each time period will be encoded and stored respectively, and then the power generation data in each time period will be analyzed to obtain the coding priority index corresponding to each time period. The acquisition result of the coding priority index is more accurate, and the risk of losing important power generation data is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 The present invention provides a flow chart of a method for storing wind power generation data. DETAILED DESCRIPTION

[0033] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, a wind power generation data storage method proposed in accordance with the present invention is described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0034] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0035] The following describes in detail a specific scheme of a wind power generation data storage method provided by the present invention in conjunction with the accompanying drawings.

[0036] A method for storing wind power generation data:

[0037] This embodiment proposes a method for storing wind power generation data, such as Figure 1 As shown, a wind power generation data storage method of this embodiment includes the following steps:

[0038] Step S1, obtaining the actual power generation of the wind farm at each collection moment in the current storage period; obtaining the predicted power generation at each collection moment in the current storage period based on the actual power generation at each collection moment in the historical storage period.

[0039] The specific scenario targeted by this embodiment is: when the power generation data of a wind farm is directly stored, the amount of data is large, and the reference value of some data for subsequent prediction models is also low. The data for storage is obtained from the wind farm monitoring system, and the fluctuation of the acquired data is analyzed, each storage cycle is divided into multiple time periods, and then the collected data and the predicted data are analyzed, the coding priority index corresponding to each time period is determined, and the actual power generation of each time period is encoded and stored based on the coding priority index.

[0040] In this embodiment, the actual power generation of the wind farm is collected once every 1 minute, and a storage period is set. In this embodiment, the time length of each storage period is one month. In the following, the current storage period is used as an example for analysis. The current period is the most recent month. The actual power generation at each collection time in the current storage period is obtained. In order to ensure the safe and stable operation of the system, the wind farm needs to report its power generation plan to the power grid. If its prediction error exceeds a certain range, the wind farm must pay a fine to the power grid operator. Therefore, it is necessary to construct a power generation data sequence corresponding to the historical storage period based on the actual power generation at each collection time in the historical storage period of the wind farm, input the power generation data sequence corresponding to the historical storage period into the prediction model, and predict the power generation at each collection time in the current storage period. When predicting the power generation at each collection time in the current storage period, this embodiment uses the prediction model to perform ultra-short-term prediction on the power generation at each collection time in the current storage period, and obtain the predicted power generation at each collection time in the current storage period. The prediction model is an existing model, and the ultra-short-term prediction is an existing technology. Therefore, the process of obtaining the predicted power generation is an existing technology, which will not be described in detail here.

[0041] Step S2, respectively performing curve fitting on the actual power generation at all acquisition moments and the predicted power generation at all acquisition moments in the current storage period to obtain the actual power generation curve and the predicted power generation curve corresponding to the current storage period; obtaining the peaks and troughs of the actual power generation curve, and dividing the current storage period into time periods based on the peaks and troughs.

[0042] In step S1, the present embodiment obtains the actual power generation and predicted power generation at each collection moment in the current storage period of the wind farm. When the predicted power generation and the actual power generation are the same or within the specified difference range, the actual power generation in the current storage period is of little significance for the subsequent prediction model; on the contrary, when there is a difference between the predicted power generation and the actual power generation, the value brought and the significance of preservation are greater, and it is very necessary to pay the fine corresponding to the prediction error or to adjust the subsequent prediction model.

[0043] For the power generation data within a storage period, the whole is composed of multiple wavelets, some of which have small amplitudes and are relatively smooth overall, while some have large amplitudes and are uneven. Therefore, when the power generation data within a storage period is uneven, if the power generation data within the storage period is stored, the amount of stored data is large. When fitting the power generation data of all acquisition moments within the storage period, the fitting difficulty of different time periods is different. The more chaotic the power generation data of a certain time period within the storage period, the higher the fitting difficulty of the corresponding curve. Therefore, for the segments with higher fitting difficulty, the overall priority in the Huffman coding sequence is higher to maximize the guarantee that the overall coding sequence is shorter.

[0044] Next, this embodiment analyzes the actual power generation at all acquisition moments in the current storage cycle, divides the current storage cycle into multiple time periods, and then analyzes the power generation data of each time period respectively.

[0045] Specifically, the least squares method is used to perform curve fitting on the time series of the actual power generation at all acquisition moments in the current storage period to obtain a curve, which is recorded as the actual power generation curve corresponding to the current storage period; at the same time, the least squares method is used to perform curve fitting on the time series of the predicted power generation at all acquisition moments in the current storage period to obtain a curve, which is recorded as the predicted power generation curve corresponding to the current storage period; the least squares method curve fitting is a well-known technology and will not be described in detail here. For the actual power generation curve corresponding to the current storage period: obtain all the peaks and all the troughs of the actual power generation curve corresponding to the current storage period. Since the curve only passes through some data points when the least squares method is used to fit the curve, some small peak points and trough points cannot be reflected in the actual power generation curve, that is, the information of some peak points and trough points with smaller fluctuation amplitudes in the current storage period is hidden. At this time, the peaks and troughs obtained based on the actual power generation curve are some peak points and trough points with larger fluctuation amplitudes; divide the current storage period based on all the peaks and all the troughs, the collection time corresponding to each peak is a division point, and the collection time corresponding to each trough is a division point. Based on all the division points, the current storage period is divided into multiple time periods, that is, ,in, is the first acquisition moment in the current storage cycle. is the first division point in the current storage cycle. It is the second division point in the current storage cycle. is the m-1th partition point in the current storage cycle, It is the mth acquisition moment in the current storage cycle (the last acquisition moment in the current storage cycle).

[0046] At this point, the current storage cycle is divided into m time periods. Each collection moment in each time period has corresponding actual power generation and predicted power generation. The power generation data of each time period will be analyzed separately later.

[0047] Step S3, obtaining the importance of the power generation data of each time period according to the actual power generation at each collection moment in each time period, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment; counting the number of target collection moments in each time period according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments; obtaining the fitting difficulty corresponding to the power generation data of each time period according to the number of target collection moments; the slope corresponding to the predicted power generation is obtained according to the predicted power generation curve, and the slope corresponding to the actual power generation is obtained according to the actual power generation curve.

[0048] In step S2, the present embodiment obtains the actual power generation curve and the predicted power generation curve corresponding to the current storage period, and calculates the slope corresponding to the actual power generation at each collection moment based on the actual power generation curve corresponding to the current storage period. At the same time, based on the predicted power generation curve corresponding to the current storage period, calculates the slope corresponding to the predicted power generation at each collection moment. The method for obtaining the slope of the corresponding position on the curve is a prior art and will not be repeated here.

[0049] This embodiment divides the current storage cycle, that is, the power generation data within the current storage cycle is divided. In the same time period, the data collected at some collection moments are not only more suitable for the adjustment and optimization of the prediction model, but also related to whether the power generation connected to the power grid is the same as the predicted power generation, thereby determining whether to impose a penalty. Most of a segment of data is composed of multiple wavelets, some of which have larger and denser amplitudes and are unevenly distributed in the entire segment of data. For this type of data, more parameters such as peak points, trough points, and corresponding amplitudes need to be obtained, and the subsequent re-fitting of the curve is more difficult. Therefore, for this part of the data, if it is located at a later position in the Huffman coding sequence, it will result in a longer coding sequence, and the probability of losing data during transmission or storage is greater. Therefore, this part of the data should be shorter in the Huffman coding sequence.

[0050] For any time period in the current storage cycle:

[0051] For the rth collection moment within the time period, if the difference between the actual power generation at the collection moment and the predicted power generation is greater, then when encoding the power generation data for the time period, more attention should be paid to the actual power generation at the collection moment, and the greater the importance of the actual power generation at the collection moment; the greater the difference between the slope corresponding to the actual power generation at the collection moment and the slope corresponding to the predicted power generation at the collection moment, when encoding the power generation data for the time period, more attention should be paid to the actual power generation at the collection moment; the greater the absolute value of the slope corresponding to the actual power generation at the collection moment, the more likely it is that an emergency has occurred in the power generation at the collection moment, and the lower the reference value of the actual power generation data at the collection moment for the prediction model. Based on this, this embodiment calculates the absolute value of the difference between the actual power generation and the predicted power generation at the r-th collection moment, recorded as the first absolute value, normalizes the first absolute value, and uses the normalized result as the power generation difference. The absolute value of the difference between the slope corresponding to the actual power generation at the r-th collection moment and the slope corresponding to the predicted power generation at the r-th collection moment is calculated as the slope difference. The product of the amplitude adjustment coefficient and the absolute value of the slope corresponding to the actual power generation at the r-th collection moment is calculated, recorded as the first product. The value of the exponential function with the natural constant as the base and the negative first product as the exponent is used as the adjustment weight. The product of the power generation difference, the slope difference and the adjustment weight is calculated as the importance of the power generation data at the r-th collection moment in the time period. The specific calculation formula for the importance of the power generation data at the r-th collection moment in the time period is:

[0052]

[0053] in, is the importance of the power generation data at the rth collection moment in this time period, is the actual power generation at the rth collection moment in this time period, is the predicted power generation at the rth collection moment in this time period, is the slope corresponding to the actual power generation at the rth acquisition moment in this time period, is the slope corresponding to the predicted power generation at the rth collection moment in this time period, To take the absolute value, is a natural constant, The amplitude adjustment coefficient is set to 0.5 in this embodiment. In specific applications, the implementer may adjust it according to specific circumstances.

[0054] Characterizes the difference between the actual power generation and the predicted power generation at the rth collection moment, that is, the first absolute value; The difference between the actual power generation and the predicted power generation at the rth collection moment is normalized. The normalization is based on the consideration that the same difference has different meanings on a windy day and a weak wind day. Therefore, in this embodiment, the actual power generation at the corresponding collection moment is used as the amplitude to normalize the difference. It represents the difference between the slope corresponding to the actual power generation at the rth acquisition moment and the slope corresponding to the predicted power generation, that is, the slope difference; To adjust the weight, when the absolute value of the slope corresponding to the actual power generation at the r-th collection moment approaches infinity, it means that the wind force change at the r-th collection moment is more likely to be an unexpected situation, that is, an instantaneous change, and some difficult-to-predict situations have occurred. Therefore, the reference value of the actual power generation data at the r-th collection moment for the prediction model is reduced; the power generation difference and the slope difference are both positively correlated with the importance, and the slope corresponding to the actual power generation at the r-th collection moment is negatively correlated with the importance; when the difference between the actual power generation at the r-th collection moment and the predicted power generation is greater, the difference between the slope corresponding to the actual power generation at the r-th collection moment and the slope corresponding to the predicted power generation is greater, and the absolute value of the slope corresponding to the actual power generation at the r-th collection moment is smaller, it means that the power generation data at the r-th collection moment is more important, that is, the importance of the power generation data at the r-th collection moment is greater.

[0055] By adopting the above method, the importance of the power generation data at each collection moment in the time period can be obtained, and the sum of the importance of the power generation data at all collection moments in the time period can be calculated as the importance of the power generation data in the time period.

[0056] Next, this embodiment will analyze the actual power generation at each collection moment in the time period and the predicted power generation at each collection moment in the time period. Considering that when fitting a curve, most of the key points are peak points or trough points. Once a peak point or trough point is lost, it will have a great impact on the trend of a section of data after it is refitted into a curve. Therefore, for data with more peak points or trough points, not only is it difficult to fit, but the degree to which it needs to be retained is also greater. Based on this, this embodiment will analyze the peaks and troughs of the actual power generation in the time period and the peaks and troughs of the predicted power generation. Fit the actual power generation at each collection moment in the time period to obtain a function of the actual power generation with respect to the collection moment, and derive the derivative value corresponding to each collection moment. The collection moment with a derivative value of 0 is the peak or trough of the actual power generation in the time period. The collection moment with a derivative value of 0 is recorded as the first target collection moment, and the number of the first target collection moments is counted. At the same time, the predicted power generation at each collection time in the time period is fitted to obtain the function of the predicted power generation with respect to the collection time, and the derivative value corresponding to each collection time is obtained by derivation. The collection time with a derivative value of 0 is the peak or trough of the predicted power generation in the time period. The collection time with a derivative value of 0 is recorded as the second target collection time, and the number of the second target collection time is counted. It should be noted that all the peaks and troughs obtained at this time are all the real peaks and troughs, that is, the total number of peaks and troughs of actual power generation in this time period and the total number of peaks and troughs of predicted power generation are obtained, and the number of the first target acquisition time is calculated. The ratio of the number of all collection moments in this time period is used as the first target collection moment ratio. At the same time, the number of second target collection moments is calculated. The ratio of the number of all collection moments in the time period is taken as the second target collection moment ratio, and the sum of the first target collection moment ratio and the second target collection moment ratio is taken as the fitting difficulty corresponding to the power generation data of the time period; the specific calculation formula for the fitting difficulty corresponding to the power generation data of the time period is:

[0057]

[0058] in, is the fitting difficulty corresponding to the power generation data in this time period, is the number of first target acquisition moments in this time period, is the number of second target acquisition moments in this time period, is the number of all collection moments in this time period.

[0059] Characterizes the proportion of the first target collection time, that is, the proportion of the peak and trough of the actual power generation in this time period, Characterizes the proportion of the second target collection time, that is, the proportion of the peak and trough of the predicted power generation in this time period; when the proportion of the first target collection time and the proportion of the second target collection time in this time period are both larger, it means that the number of key points of the power generation data in this time period is more, that is, the corresponding fitting difficulty of the power generation data in this time period is greater; when the proportion of the first target collection time and the proportion of the second target collection time in this time period are both smaller, it means that the number of key points of the power generation data in this time period is fewer, that is, the corresponding fitting difficulty of the power generation data in this time period is smaller.

[0060] By adopting the above method, the importance of the power generation data in each time period and the corresponding fitting difficulty can be obtained.

[0061] Step S4, based on the importance and the fitting difficulty, obtaining a coding priority index corresponding to each time period; encoding and storing the actual power generation in each time period based on the coding priority index.

[0062] In step S3, this embodiment obtains the importance of the power generation data of each time period in the current storage cycle and the corresponding fitting difficulty, and normalizes the importance of the power generation data of each time period and the corresponding fitting difficulty respectively. For any time period in the current storage cycle: the sum of the normalized value of the importance of the power generation data of the time period and the normalized value of the fitting difficulty corresponding to the power generation data of the time period is calculated as the coding priority index corresponding to the time period, and the value range of the coding priority index is between 0 and 2. By adopting the above method, the coding priority index corresponding to each time period in the current storage cycle can be obtained.

[0063] Next, this embodiment will use Huffman coding to encode the actual power generation of each time period in the current storage cycle based on the coding priority index corresponding to each time period in the current storage cycle, and encode the actual power generation of each time period in the order of coding priority index from large to small. For the first time period in the current storage cycle: the first actual power generation data in the time period is used as the data at the top of the Huffman tree, and the second and third actual power generation data in the time period are taken out in order as the data at the top of the Huffman tree, and at the same time, the left side of each binary tree is designated as 0 and the right side is designated as 1, and the code corresponding to its path is obtained along the top of the binary tree to each character; repeat the above steps until the actual power generation of all time periods in the current storage cycle is filled in, that is, the encoding of the actual power generation data in the current storage cycle is completed, and the encoded power generation data is stored to realize the storage of wind power generation data. Huffman coding is a prior art and will not be described in detail here.

[0064] This embodiment takes into account that when storing power generation data, the existing data storage method is likely to cause the loss of some important power generation data. In order to ensure the safe and stable operation of the system, the wind farm needs to report its power generation plan to the power grid. If its prediction error exceeds a certain range, the wind farm needs to pay a fine to the power grid operator. Therefore, if some important power generation data is lost, the prediction accuracy of subsequent power generation data will be reduced. This embodiment obtains the importance of power generation data in each time period based on the actual power generation at each collection moment in each time period in the current storage cycle, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment. The greater the importance, the more appropriate the power generation data in the corresponding time period is. Suitable for predicting subsequent power generation; this embodiment also counts the number of target collection moments in each time period according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments, and then obtains the fitting difficulty corresponding to the power generation data of each time period. The greater the fitting difficulty, the greater the proportion of important data in the corresponding time period. Therefore, when encoding and storing, the power generation data in the corresponding time period should be encoded first; based on this, this embodiment obtains the coding priority index corresponding to each time period based on the importance and fitting difficulty, and encodes and stores the actual power generation of each time period in the current storage cycle based on the coding priority index, so that important power generation data of the wind farm is not lost, which can effectively improve the prediction accuracy of subsequent power generation and reduce the number of fines. This embodiment takes into account that the amount of power generation data in one cycle is large, and there are several large fluctuation points in the power generation data in the entire cycle, and the data similarity between every two adjacent fluctuation points is relatively high. Therefore, curve fitting is performed on the actual power generation at all acquisition moments in the current storage cycle, and the peaks and troughs of the actual power generation curve are obtained. The current storage cycle is divided into multiple time periods based on the peaks and troughs, and the actual power generation in each time period is relatively similar. Subsequently, the power generation data in each time period will be encoded and stored respectively, and then the power generation data in each time period will be analyzed to obtain the coding priority index corresponding to each time period. The acquisition result of the coding priority index is more accurate, and the risk of losing important power generation data is reduced.

Claims

1. A method for storing wind power generation data, characterized in that: The method comprises the following steps: Obtain the actual power generation of the wind farm at each collection moment in the current storage cycle; obtain the predicted power generation at each collection moment in the current storage cycle based on the actual power generation at each collection moment in the historical storage cycle; Performing curve fitting on the actual power generation at all acquisition moments and the predicted power generation at all acquisition moments in the current storage period respectively, to obtain the actual power generation curve and the predicted power generation curve corresponding to the current storage period; obtaining the peaks and troughs of the actual power generation curve, and dividing the current storage period into time periods based on the peaks and troughs; According to the actual power generation at each collection moment in each time period, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment, the importance of the power generation data in each time period is obtained; according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments, the number of target collection moments in each time period is counted; according to the number of target collection moments, the fitting difficulty corresponding to the power generation data in each time period is obtained; the slope corresponding to the predicted power generation is obtained according to the predicted power generation curve, and the slope corresponding to the actual power generation is obtained according to the actual power generation curve; Based on the importance and the fitting difficulty, a coding priority index corresponding to each time period is obtained; based on the coding priority index, an actual power generation in each time period is encoded and stored; The importance of power generation data in each time period is obtained according to the actual power generation at each collection moment in each time period, the predicted power generation at each collection moment, the slope corresponding to the actual power generation at each collection moment, and the slope corresponding to the predicted power generation at each collection moment, including: For any time period in the current storage cycle: For any collection moment in the time period: calculate the absolute value of the difference between the actual power generation at the collection moment and the predicted power generation, record it as the first absolute value, perform normalization on the first absolute value, and use the normalization result as the power generation difference; calculate the absolute value of the difference between the slope corresponding to the actual power generation at the collection moment and the slope corresponding to the predicted power generation, as the slope difference, calculate the product of the amplitude adjustment coefficient and the absolute value of the slope corresponding to the actual power generation at the collection moment, record it as the first product, and use the value of the exponential function with the natural constant as the base and the negative of the first product as the exponent as the adjustment weight; calculate the product of the power generation difference, the slope difference and the adjustment weight as the importance of the power generation data at the collection moment; Calculate the sum of the importance of the power generation data at all collection moments in the time period as the importance of the power generation data in the time period; The method of counting the number of target collection moments in each time period according to the actual power generation at all collection moments in each time period and the predicted power generation at all collection moments includes: The target acquisition time includes a first target acquisition time and a second target acquisition time; For any time period within the current storage cycle: The actual power generation at each collection time in the time period is fitted to obtain the function of the actual power generation with respect to the collection time, and the derivative is derived to obtain the derivative value corresponding to each collection time, and the collection time with a derivative value of 0 is recorded as the first target collection time, and the number of the first target collection time is counted; The predicted power generation at each collection time in the time period is fitted to obtain the function of the predicted power generation with respect to the collection time, and the derivative is derived to obtain the derivative value corresponding to each collection time, and the collection time with a derivative value of 0 is recorded as the second target collection time, and the number of the second target collection time is counted; According to the number of target collection moments, the fitting difficulty corresponding to the power generation data in each time period is obtained, including: For any time period within the current storage cycle: The ratio of the number of the first target collection time to the number of all collection times in the time period is calculated as the first target collection time proportion; the ratio of the number of the second target collection time to the number of all collection times in the time period is calculated as the second target collection time proportion, and the sum of the first target collection time proportion and the second target collection time proportion is used as the fitting difficulty corresponding to the power generation data of the time period; The obtaining of the coding priority index corresponding to each time period based on the importance and the fitting difficulty includes: For any time period within the current storage cycle: The importance of the power generation data for the time period and the fitting difficulty corresponding to the power generation data for the time period are normalized respectively, and the sum of the normalized value of the importance of the power generation data for the time period and the normalized value of the fitting difficulty corresponding to the power generation data for the time period is calculated as the coding priority indicator corresponding to the time period.

2. A method for storing wind power generation data according to claim 1, characterized in that: The current storage cycle is divided based on the peaks and troughs to obtain various time periods, including: the collection moment corresponding to each peak is a division point, the collection moment corresponding to each trough is a division point, and the current storage cycle is divided based on all division points to obtain various time periods.

3. A method for storing wind power generation data according to claim 1, characterized in that: The method of obtaining the predicted power generation at each collection moment in the current storage period based on the actual power generation at each collection moment in the historical storage period includes: Based on the actual power generation at each collection moment in the historical storage period, a power generation data sequence corresponding to the historical storage period is constructed; the power generation data sequence corresponding to the historical storage period is input into the prediction model to obtain the predicted power generation at each collection moment in the current storage period.

4. A method for storing wind power generation data according to claim 1, characterized in that: The encoding and storing of the actual power generation in each time period based on the encoding priority index includes: encoding and storing the actual power generation in each time period in the current storage cycle respectively in descending order of the encoding priority index.

Images