A supercapacitor energy storage frequency modulation grid connection method and system

Abstract

The present invention discloses a method and system for frequency modulation and grid connection of supercapacitor energy storage, belonging to the technical field of wind speed prediction. The method includes: decomposing the original wind speed time series according to the decomposition level, adding the corresponding items of the similarity sequences of all subsequence pairs to obtain a linearity sequence, and adding the regular sequence values of all subsequences to obtain a total regularity sequence; calculating a fitness function by using the linearity sequence and the total regularity sequence; starting from the initial decomposition level, continuously increasing the value of the decomposition level to re-perform decomposition and calculate the fitness function until the decomposition level reaches the maximum, comparing the fitness functions to obtain an optimal subsequence set; inputting the optimal subsequence set into a prediction network for prediction to obtain the prediction results of each subsequence in the optimal subsequence set, and adding the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed for supercapacitor energy storage grid connection.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind speed prediction, and particularly relates to a supercapacitor energy storage frequency modulation grid connection method and system. Background Technique

[0002] Due to the randomness of wind speed changes, the grid connection of large-scale wind farms brings difficulties to power grid dispatching, thereby affecting the stability of the system. In order to improve the dispatchability of wind farms, a general method is to use the neural network method to predict the wind speed for the next day, calculate the power sent by the wind farm to the power grid through the predicted wind speed value, and submit the predicted value to the power grid dispatching agency. At the same time, a supercapacitor energy storage system is connected to the outlet of the wind farm to quickly respond to make up for the error between the actual power generated by the wind farm and the predicted power, thereby improving the credibility of the power grid for wind farm power generation dispatch based on wind power prediction and improving the coordinated operation ability between the wind farm and the power system.

[0003] However, traditional prediction methods generally use decomposition algorithms to decompose the original wind speed sequence, put the decomposed sequences into a neural network for prediction, and then superimpose the predicted results to obtain the final result. However, the existing decomposition methods currently have problems such as mode mixing and a large degree of linear correlation between each subsequence. These problems will lead to poor prediction accuracy, so there is an urgent need to propose a new method. Summary of the Invention

[0004] The purpose of the present invention is to overcome the problems of mode mixing and a large degree of linear correlation between each subsequence existing in the existing decomposition methods, and propose a supercapacitor energy storage frequency modulation grid connection method and system.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] In the first aspect, the present invention provides a supercapacitor energy storage frequency modulation grid connection method, including the following steps:

[0007] S1. Decompose the original wind speed time series according to the decomposition layer number to obtain a number of subsequences;

[0008] S2. Select any two subsequences from the number of subsequences as a subsequence pair, compare the similarity between any two subsequences to obtain the similarity sequence of all subsequence pairs, and add the corresponding items of the similarity sequence of all subsequence pairs to obtain the linearity sequence;

[0009] S3. Calculate the regular sequence of each subsequence through the regular sequence calculation method, and add the regular sequence values of all subsequences to obtain the total regularity sequence;

[0010] S4. Calculate the fitness function using the linearity sequence and the total regularity sequence;

[0011] S5. Starting from the initial decomposition level, continuously increase the value of the decomposition level to re - decompose and calculate the fitness function until the decomposition level reaches the maximum, and compare the fitness functions to obtain the optimal subsequence set;

[0012] S6. Input the optimal subsequence set into the prediction network to obtain the prediction results of each subsequence in the optimal subsequence set, and add up the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed for the super - capacity energy storage grid connection.

[0013] Furthermore, the original wind speed time series is decomposed according to the decomposition level in the following way:

[0014]

[0015] where, is the i - th subsequence, i is an intermediate variable, is the K - th subsequence, K is the decomposition level, rand() is a random function, x(t)= is the original wind speed time series, t is time, is the historical wind speed data, N is the total number of data points in the sequence, and softplus(), sigmoid(), tanh() are all activation functions;

[0016] The softplus() activation function is:

[0017]

[0018] The sigmoid() activation function is:

[0019]

[0020] The tanh() activation function is:

[0021]

[0022] where, x is the input of the activation function, e is the natural constant, is the exponential function with base e and exponent x, ln(1 + ) is the natural logarithm of 1 + ; is the exponential function with base e and exponent - x;

[0023] The sigmoid() activation function maps any real number to the interval (0, 1), and the tanh() activation function maps any real number to the interval (-1, 1) and is symmetric about the origin.

[0024] Further, each element in the similarity sequence of the subsequence pairs is the ratio of the corresponding elements in the subsequence pairs;

[0025] The method for comparing the similarity between any two subsequences to obtain the similarity sequence of all subsequence pairs is:

[0026] P 1 =a 1 ×q 1 ,P 2 =a 2 ×q 2 ,P 3 =a 3 ×q 3 ,...,P N =a N ×q N

[0027] where S i-1 is the (i - 1)-th subsequence, q 1 ,q 2 ,q 3 ,..,q N are the elements in the (i - 1)-th subsequence, S i is the i-th subsequence, P 1 ,P 2 ,P 3 ,...,P N are the elements in the i-th subsequence, S i-1 and S i form a subsequence pair, a 1 ,a 2 ,a 3 ,...,a N is the similarity sequence between the (i - 1)-th subsequence and the i-th subsequence, and N is the total number of data points in the sequence.

[0028] Further, the calculation method of the regular sequence is:

[0029]

[0030] where rand() is a random function, N is the total number of data points in the sequence, K is the decomposition level, S i is the i-th subsequence, P 1 ,P 2 ,P 3 ,P N-1 ,...,P N are the elements in the i-th subsequence, and sigmoid() and tanh() are both activation functions;

[0031] The sigmoid() activation function is:

[0032]

[0033] The tanh() activation function is as follows:

[0034]

[0035] where x is the input of the activation function, and e is the natural constant. is an exponential function with base e and exponent -x. is an exponential function with base e and exponent x.

[0036] The sigmoid() activation function maps any real number to the interval (0, 1), and the tanh() activation function maps any real number to the interval (-1, 1) and is symmetric about the origin.

[0037] Furthermore, the calculation of the fitness function using the linearity sequence and the total regularity sequence is specifically as follows:

[0038]

[0039] where is the fitness function, β(x) is the linearity sequence, c 1 , c 2 , c 3 ,.., c N are the elements in the linearity sequence, α(x) is the total regularity sequence, d 1 , d 2 , d 3 ,.., d N are the elements in the total regularity sequence, and N is the total number of data points in the sequence.

[0040] Furthermore, the obtaining of the optimal subsequence set by comparing the fitness functions is specifically as follows:

[0041] In each iteration, compare the fitness functions at different decomposition levels;

[0042] The decomposition level value with the largest fitness function is the optimal decomposition level value, and the subsequence set of the optimal decomposition level value is the optimal subsequence set.

[0043] Furthermore, the prediction network adopts a GRU network.

[0044] In a second aspect, the present invention provides a supercapacitor energy storage frequency modulation grid-connected system, based on the above-mentioned supercapacitor energy storage frequency modulation grid-connected method, including:

[0045] A decomposition module, configured to decompose the original wind speed time series according to the decomposition level to obtain a plurality of subsequences;

[0046] A linearity sequence obtaining module, configured to select any two subsequences from several subsequences as a subsequence pair, compare the similarity between any two subsequences to obtain a similarity sequence of all subsequence pairs, and add the corresponding items of the similarity sequences of all subsequence pairs to obtain a linearity sequence;

[0047] A total regularity sequence obtaining module, configured to calculate the regularity sequence of each subsequence through a regularity sequence calculation method, and add the regularity sequence values of all subsequences to obtain a total regularity sequence;

[0048] A fitness function obtaining module, configured to calculate a fitness function by using the linearity sequence obtained by the linearity sequence obtaining module and the total regularity sequence obtained by the total regularity sequence obtaining module;

[0049] An optimal subsequence set obtaining module, configured to start from the initial decomposition level, continuously increase the value of the decomposition level to re-perform decomposition and calculate the fitness function until the decomposition level reaches the maximum, and compare the fitness functions to obtain an optimal subsequence set;

[0050] A predicted wind speed obtaining module, configured to input the optimal subsequence set into a prediction network for prediction to obtain the prediction results of each subsequence in the optimal subsequence set, and add the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed for the supercapacitor energy storage grid connection.

[0051] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for supercapacitor energy storage frequency modulation grid connection as described above is implemented.

[0052] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the method for supercapacitor energy storage frequency modulation grid connection as described above is implemented.

[0053] Compared with the prior art, the present invention has the following beneficial technical effects:

[0054] A supercapacitor energy storage frequency modulation grid connection method proposed by the present invention decomposes the original wind speed sequence into multiple subsequences and further optimizes these subsequences to capture the change characteristics of different frequencies and scales in the wind speed data. This multi-scale analysis method helps to more accurately simulate the complex dynamics of the wind speed, effectively avoid modal aliasing, and the optimal decomposition can reduce the linear correlation between subsequences and improve the prediction accuracy, jointly improving the accuracy and reliability of wind speed prediction and providing more accurate wind speed prediction data for supercapacitor energy storage grid connection. Through the decomposition and reconstruction process, the influence of adverse factors such as noise and non-stationarity on the prediction results can be reduced, making the model more robust. Even in the case of missing or outlier wind speed data, it can maintain good prediction performance. The present invention can not only improve the accuracy and reliability of wind speed prediction, but also optimize the operation strategy of the supercapacitor energy storage system, reduce the operation cost, and promote the utilization of renewable energy and the construction of smart grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The drawings described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure of the present invention in any way. Additionally, the shapes and proportional dimensions of the components in the drawings are only schematic and are used to assist in understanding the present invention, rather than specifically defining the shapes and proportional dimensions of the components of the present invention. In the drawings:

[0056] Figure 1 is a flowchart of a supercapacitor energy storage frequency modulation grid connection method of the present invention.

[0057] Figure 2 is a structural diagram of a supercapacitor energy storage frequency modulation grid connection system of the present invention.

[0058] Figure 3 is a diagram of an electronic device for a supercapacitor energy storage frequency modulation grid connection method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] In order to enable those skilled in the art of the present technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a 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 those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0060] Embodiment 1

[0061] See Figure 1 , a supercapacitor energy storage frequency modulation grid connection method, includes the following steps:

[0062] S1. Decompose the original wind speed time series according to the decomposition level to obtain a number of subsequences;

[0063] The decomposition of the original wind speed time series according to the number of decomposition layers in S1 is carried out in the following manner:

[0064]

[0065] Among them, is the i-th subsequence, where i is an intermediate variable, is the K-th subsequence, where K is the number of decomposition layers, rand() is a random function, and x(t) = is the original wind speed time series, where t is time, is the historical wind speed data, N is the total number of data points in the sequence, and softplus(), sigmoid(), and tanh() are all activation functions;

[0066] The softplus() activation function is:

[0067]

[0068] The sigmoid() activation function is:

[0069]

[0070] The tanh() activation function is:

[0071]

[0072] Among them, x is the input of the activation function, e is the natural constant, is the exponential function with e as the base and x as the exponent, ln(1 + ) is the natural logarithm of 1 + , is the exponential function with e as the base and -x as the exponent;

[0073] The sigmoid() activation function maps any real number to the interval (0, 1), and the tanh() activation function maps any real number to the interval (-1, 1) and is symmetric about the origin.

[0074] In this embodiment, the original wind speed sequence is obtained. The wind speed data is obtained from weather stations, wind power plants, or public datasets, and the wind speed data is cleaned, such as checking and processing missing values, outliers, etc. Decompose the wind speed sequence into equal subsequences. The purpose of decomposition is to better understand the composition of the wind speed and possibly perform different processing or modeling on different parts. Optimal decomposition further analyzes or processes the decomposed subsequences to select or generate a set of subsequences that are most useful for prediction. It includes feature selection techniques such as screening based on statistical characteristics and principal component analysis. The optimization depends on the specific requirements of the prediction task and the performance evaluation criteria.

[0075] S2. Take any two subsequences among a number of subsequences as a subsequence pair, compare the similarity between any two subsequences to obtain the similarity sequence of the subsequence pair, and add the corresponding items of the similarity sequences of all subsequence pairs to obtain the linearity sequence;

[0076] Each element in the similarity sequence of the subsequence pair described in S2 is the ratio of the corresponding elements in the subsequence pair;

[0077] The comparison method for comparing the similarity between any two subsequences to obtain the similarity sequence of all subsequence pairs described in S2 is:

[0078] P 1 =a 1 ×q 1 ,P 2 =a 2 ×q 2 ,P 3 =a 3 ×q 3 ,...,P N =a N ×q N

[0079] Among them, S i-1 is the (i - 1)-th subsequence, q 1 ,q 2 ,q 3 ,..,q N are the elements in the (i - 1)-th subsequence, S i is the i-th subsequence, P 1 ,P 2 ,P 3 ,...,P N are the elements in the i-th subsequence, S i-1 and S i form a subsequence pair, a 1 ,a 2 ,a 3 ,...,a Nis the similarity sequence of the (i - 1)-th subsequence and the i-th subsequence, and N is the total number of data points in the sequence.

[0080] S3. Calculate the regular sequence of each subsequence through the regular sequence calculation method, and add up the regular sequence values of all subsequences to obtain the total regularity sequence;

[0081] The regular sequence calculation method described in S3 is as follows:

[0082]

[0083] where rand() is a random function, N is the total number of data points in the sequence, K is the decomposition level, S i is the i-th subsequence, and P 1 , P 2 , P 3 , P N-1 ,..., P N are the elements in the i-th subsequence, and sigmoid() and tanh() are both activation functions;

[0084] The sigmoid() activation function is:

[0085]

[0086] The tanh() activation function is:

[0087]

[0088] where x is the input of the activation function, e is the natural constant, is the exponential function with e as the base and -x as the exponent, is the exponential function with e as the base and x as the exponent;

[0089] The sigmoid() activation function maps any real number to the interval (0, 1), and the tanh() activation function maps any real number to the interval (-1, 1) and is symmetric about the origin.

[0090] S4. Calculate the fitness function using the linearity sequence obtained in S2 and the total regularity sequence obtained in S3;

[0091] The specific method for calculating the fitness function using the linearity sequence and the total regularity sequence described in S4 is:

[0092]

[0093] where, is the fitness function, β(x) is the linearity sequence, and c 1 , c 2 , c 3,..,c N are elements in the linearity sequence, α(x) is the total regularity sequence, d 1 ,d 2 ,d 3 ,..,d N are elements in the total regularity sequence, and N is the total number of data points in the sequence.

[0094] S5. Starting from the initial decomposition level, continuously increase the value of the decomposition level to re - decompose and calculate the fitness function until the decomposition level reaches the maximum, and compare the fitness functions to obtain the optimal subsequence set;

[0095] In S5, comparing the fitness functions to obtain the optimal subsequence set is specifically as follows:

[0096] In each iteration, compare the fitness functions under different decomposition levels;

[0097] The decomposition level value with the maximum fitness function is the optimal decomposition level value, and the subsequence set corresponding to the optimal decomposition level value is the optimal subsequence set.

[0098] In this embodiment, the calculation process of the fitness function helps to evaluate the advantages and disadvantages of the subsequence sets under different decomposition levels. By continuously optimizing the decomposition level, the system can find the most suitable decomposition method for the current wind speed data, thereby improving the adaptability and flexibility of the system.

[0099] S6. Input the optimal subsequence set into the prediction network for prediction to obtain the prediction results of each subsequence in the optimal subsequence set, and add up the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed for the ultra - capacity energy storage grid connection.

[0100] The prediction network described in S6 uses a GRU network.

[0101] In this embodiment, the GRU (Gated Recurrent Unit) network is a variant of the recurrent neural network, suitable for processing sequence data. Take the optimal subsequence set as the input feature and the corresponding part of the original wind speed sequence as the target output to train the GRU network. Adjust hyperparameters such as the network structure (such as the number of layers, the number of hidden units), learning rate, and optimizer through methods such as cross - validation to improve the model performance.

[0102] Use the trained GRU model to predict new wind speed data to obtain the predicted wind speed sequence. According to the predicted wind speed sequence, the impact on the power grid can be further analyzed, especially the synergy with the ultra - capacity energy storage system. The predicted wind speed can help determine the charging and discharging strategies of the energy storage system to optimize energy utilization and power grid stability. Using the GRU network for prediction can make full use of the long - term dependencies in the time - series data and further improve the prediction accuracy.

[0103] The high-precision predicted wind speed obtained by the method of this embodiment can provide strong support for the charge and discharge strategies, power distribution, and scheduling of the supercapacitor energy storage system, thereby optimizing the operating efficiency and economic benefits of the system. The real-time and accuracy of the predicted wind speed can help the supercapacitor energy storage system better respond to grid demands and improve the stability and reliability of the grid. Accurate wind speed prediction can reduce the standby capacity requirements of the supercapacitor energy storage system and lower the operating costs of the system. By optimizing the charge and discharge strategies of the energy storage system, the energy utilization efficiency can be further improved and energy waste can be reduced.

[0104] Embodiment 2

[0105] See Figure 2 , a supercapacitor energy storage frequency modulation grid-connected system, using a supercapacitor energy storage frequency modulation grid-connected method described in Embodiment 1, including:

[0106] A decomposition module for decomposing the original wind speed time series according to the decomposition level to obtain a number of subsequences;

[0107] A linearity sequence obtaining module for selecting any two subsequences from the number of subsequences as a subsequence pair, comparing the similarity between any two subsequences to obtain the similarity sequence of all subsequence pairs, and adding the corresponding terms of the similarity sequence of all subsequence pairs to obtain a linearity sequence;

[0108] A total regularity sequence obtaining module for calculating the regularity sequence of each subsequence through a regularity sequence calculation method, and adding the regularity sequence values of all subsequences to obtain a total regularity sequence;

[0109] A fitness function obtaining module for calculating a fitness function by using the linearity sequence obtained by the linearity sequence obtaining module and the total regularity sequence obtained by the total regularity sequence obtaining module;

[0110] An optimal subsequence set obtaining module for starting from the initial decomposition level, continuously increasing the value of the decomposition level to re-decompose and calculate the fitness function until the decomposition level reaches the maximum, and comparing the fitness function to obtain an optimal subsequence set;

[0111] A predicted wind speed obtaining module for inputting the optimal subsequence set into a prediction network for prediction to obtain the prediction results of each subsequence in the optimal subsequence set, and adding the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed for supercapacitor grid connection.

[0112] In this embodiment, the supercapacitor energy storage grid-connected system decomposes the original wind speed time series into multiple subsequences and considers the similarity and regularity between these subsequences. The system can capture the characteristics of wind speed changes more meticulously, allowing the system to analyze wind speed data from multiple perspectives and levels, thereby improving the accuracy and reliability of wind speed prediction, enhancing the precision of wind speed prediction, strengthening the adaptability of the system, and helping the supercapacitor energy storage system to better cooperate with the power grid by providing accurate wind speed prediction results, thus improving the stability and efficiency of the entire power system. By predicting wind speed changes in advance, the system can more effectively schedule wind power generation equipment, reduce energy waste, and improve resource utilization efficiency. It can reduce power grid failures and downtime caused by wind speed fluctuations, lower operation and maintenance costs, and also improve the overall reliability and security of the power grid.

[0113] Embodiment Three

[0114] See Figure 3 , an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the supercapacitor energy storage frequency modulation grid-connected method described in Embodiment One.

[0115] Embodiment Four

[0116] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, it implements the supercapacitor energy storage frequency modulation grid-connected method described in Embodiment One.

[0117] Embodiment Five

[0118] In this embodiment, taking the decomposition layer number K = 5 as an example, a supercapacitor energy storage frequency modulation grid-connected method is implemented, and the specific steps are as follows:

[0119] Step 1: First, decompose the original wind speed time series x(t) to obtain subsequences S 1 , S 2 ,.., S K .

[0120] Let the original wind speed time series x(t) = [x 1 , x 2 , x 3 ,..., x N . The decomposition layer number is K, and the specific decomposition method is:

[0121]

[0122] Among them, is the i-th subsequence, and i is an intermediate variable. is the K-th subsequence, where K is the decomposition level, rand() is a random function, and x(t) = is the original wind speed time series, t is time, is the historical wind speed data, N is the total number of data points in the sequence, and softplus(), sigmoid(), and tanh() are all activation functions;

[0123] The softplus() activation function is:

[0124]

[0125] The sigmoid() activation function is:

[0126]

[0127] The tanh() activation function is:

[0128]

[0129] where x is the input of the activation function, e is the natural constant, is the exponential function with base e and exponent x, ln(1 + ) is the natural logarithm of 1 + , is the exponential function with base e and exponent -x;

[0130] The sigmoid() activation function maps any real number to the interval (0, 1), and the tanh() activation function maps any real number to the interval (-1, 1) and is symmetric about the origin.

[0131] Step 2: Generate the linearity sequence β(x) based on the subsequence Let the (i - 1)-th subsequence S

[0132] i-1 =[q 1 , q 2 , q 3 ,.., q N and the i-th subsequence S i 1 =[P, P 2 , P 3 ,..., P N 1 ;

[0133] If the sequence [a 1 , a 2 , a 3 ,..., a N satisfies the relation

[0134] P1 =a 1 ×q 1 ,P 2 =a 2 ×q 2 ,P 3 =a 3 ×q 3 ,...,P N =a N ×q N ,

[0135] Then the sequence [a 1 ,a 2 ,a 3 ,...,a N is the similarity sequence between the (i - 1)-th subsequence S i-1 and the i-th subsequence S i .

[0136] There are K(K - 1) / 2 similarity sequences between every two of the subsequences S 1 ,S 2 ,..,S K , where K is the decomposition level. Adding up each item of these similarity sequences gives the linearity sequence β(x).

[0137] Step 3. Calculate the total regularity sequence α(x) based on the subsequences S 1 ,S 2 ,..,S K

[0138] Let the i-th subsequence S i =[P 1 ,P 2 ,P 3 ,...,P N . The regularity sequence of this sequence is:

[0139]

[0140] where rand() is a random function, N is the total number of data points in the sequence, K is the decomposition level, S i is the i-th subsequence, and P 1 ,P 2 ,P 3 ,P N-1 ,...,P N are the elements in the i-th subsequence, and sigmoid() and tanh() are both activation functions;

[0141] Calculate the regularity sequence of each subsequence and then sum them up to obtain the total regularity sequence α(x).

[0142] ​Step 4: Obtain the fitness function F(x).

[0143] Let the linearity sequence β(x) = [c 1 , c 2 , c 3 ,.., c N , and the overall regularity sequence α(x) = [d 1 , d 2 , d 3 ,.., d N . Then the fitness function is:

[0144]

[0145] where is the fitness function, β(x) is the linearity sequence, c 1 , c 2 , c 3 ,.., c N are the elements in the linearity sequence, α(x) is the overall regularity sequence, d 1 , d 2 , d 3 ,.., d N are the elements in the overall regularity sequence, and N is the total number of data points in the sequence.

[0146] Step 5: Continuously iterate and update the decomposition, and output the subsequence S 1 , S 2 ,.., S K .

[0147] The initial decomposition level K = 5. Then generate a subsequence to obtain a fitness function F(x). Next, the decomposition level K = 6, and generate its corresponding fitness function F(x), until the maximum decomposition level K = 30 is reached. Output the subsequence corresponding to the maximum fitness function among all fitness functions, which is the optimal subsequence used for predicting the wind speed.

[0148] Step 6: Put the optimal subsequence obtained in Step 5 into the GRU network for prediction, and finally sum up the prediction results to obtain the predicted wind speed for the ultra-capacitor energy storage grid connection.

[0149] To further verify the advantages of this embodiment, the method of this embodiment and a general prediction method are respectively used to predict the wind speed of the ultra-capacitor energy storage grid connection using the test data set, and the MAPE (Mean Absolute Percentage Error) is compared. The results are shown in Table 1 below:

[0150] Table 1 Comparison of MAPE of the method of this embodiment and a general prediction method for predicting wind speed

[0151]

[0152] It can be seen that the MAPE value generated by the optimal wind speed sequence for the grid connection of over-capacity energy storage obtained by the wind speed prediction method provided in this embodiment is significantly lower than that of the traditional wind speed prediction method. A lower MAPE value means that the wind speed prediction result is more accurate, thus being able to better guide the charge and discharge strategies of the over-capacity energy storage grid connection system.

[0153] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0154] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0155] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0156] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1Steps of the functions specified in one or more boxes. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: it is still possible to modify the specific implementation manners of the present invention or make equivalent replacements, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the present invention.

Claims

1. A supercapacitor energy storage frequency modulation grid-connected method, characterized in that: The following steps are involved: S1, decompose the original wind speed time series according to the number of decomposition layers to obtain several subsequences; S2. Select any two subsequences from a number of subsequences as a subsequence pair, compare the similarity between any two subsequences to obtain the similarity sequence of all subsequence pairs, add the corresponding items of the similarity sequence of all subsequence pairs to obtain a linear sequence; S3, calculating the regular sequence of each subsequence by a regular sequence calculation method, adding the regular sequence values ​​of all subsequences to obtain a total regular sequence; The regular sequence calculation method is: Among them, rand() is a random function, N is the total number of data points in the sequence, K is the number of decomposition levels, S i is the i-th subsequence, P1, P2, P3, P N-1 ,...,P N is the element in the i-th subsequence, sigmoid() and tanh() are both activation functions; The sigmoid() activation function is: The tanh() activation function is: Among them, x is the input of the activation function, e is a natural constant, is an exponential function with e as base and x as exponent. It is an exponential function with e as base and x as exponent; The sigmoid() activation function maps any real number to the interval (0,1), and the tanh activation function maps any real number to the interval (-1,1), and is symmetric about the origin; S4, using the linearity sequence and the total regularity sequence to calculate the fitness function; The fitness function calculated by using the linearity sequence and the total regularity sequence is specifically: in, is the fitness function, β(x) is the linear sequence, c1,c2,c3,..,c N is an element in the linear sequence, α(x) is the total regularity sequence, d1,d2,d3,..,d N is the element in the total regularity sequence, and N is the total number of data points in the sequence; S5, starting from the initial decomposition level, continuously increase the decomposition level to re-decompose and calculate the fitness function until the decomposition level reaches the maximum, and compare the fitness function to obtain the optimal subsequence set; S6. Input the optimal subsequence set into the prediction network for prediction to obtain the prediction result of each subsequence in the optimal subsequence set, and add the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed of the super-capacity energy storage grid-connected.

2. A supercapacitor energy storage frequency modulation grid-connected method according to claim 1, characterized in that: The original wind speed time series is decomposed according to the number of decomposition layers in the following manner: in, is the i-th subsequence, i is the intermediate variable, is the K-th subsequence, K is the number of decomposition layers, rand() is a random function, x(t)= is the original wind speed time series, t is the time, is the historical wind speed data, N is the total number of data points in the sequence, softplus(), sigmoid(), and tanh() are all activation functions; The softplus() activation function is: The sigmoid() activation function is: The tanh() activation function is: Among them, x is the input of the activation function, e is a natural constant, It is an exponential function with e as base and x as exponent, ln(1+ ) is 1+ The natural logarithm of It is an exponential function with e as base and x as exponent; The sigmoid() activation function maps any real number to the interval (0,1), and the tanh() activation function maps any real number to the interval (-1,1) and is symmetric about the origin.

3. A supercapacitor energy storage frequency modulation grid-connected method according to claim 1, characterized in that: Each element in the similarity sequence of the subsequence pair is the ratio of the corresponding elements in the subsequence pair; The comparison method for comparing the similarity between any two subsequences to obtain the similarity sequence of all subsequence pairs is: P1=a1×q1,P2=a2×q2,P3=a3×q3,...,P N =a N ×q N Among them, S i-1 is the i-1th subsequence, q1,q2,q3,..,q N is the element in the i-1th subsequence, S i is the i-th subsequence, P1, P2, P3, ..., P N is the element in the i-th subsequence, S i-1 and S i is a subsequence pair, a1,a2,a3,...,a N is the similarity sequence between the i-1th subsequence and the i-th subsequence, and N is the total number of data points in the sequence.

4. A supercapacitor energy storage frequency modulation grid-connected method according to claim 1, characterized in that: The comparison fitness function obtains the optimal subsequence set, specifically: In each iteration, the fitness functions at different decomposition levels are compared; The decomposition layer value with the largest fitness function is the optimal decomposition layer value, and the subsequence set of the optimal decomposition layer value is the optimal subsequence set.

5. A supercapacitor energy storage frequency modulation grid-connected method according to claim 1, characterized in that: The prediction network adopts the GRU network.

6. A supercapacitor energy storage frequency modulation grid-connected system, based on a supercapacitor energy storage frequency modulation grid-connected method as described in any one of claims 1 to 5, characterized in that: include: The decomposition module is used to decompose the original wind speed time series according to the number of decomposition layers to obtain several subsequences; A linearity sequence obtaining module is used to select any two subsequences from a number of subsequences as a subsequence pair, compare the similarity between any two subsequences to obtain similarity sequences of all subsequence pairs, and add corresponding items of the similarity sequences of all subsequence pairs to obtain a linearity sequence; A total regularity sequence obtaining module is used to calculate the regularity sequence of each subsequence by a regularity sequence calculation method, and add the regularity sequence values ​​of all subsequences to obtain the total regularity sequence; A fitness function obtaining module is used to calculate the fitness function using the linearity sequence obtained in the linearity sequence obtaining module and the total regularity sequence obtained in the total regularity sequence obtaining module; The module for obtaining the optimal subsequence set is used to start from the initial decomposition layer number, continuously increase the value of the decomposition layer number, re-decompose and calculate the fitness function until the decomposition layer number reaches the maximum, and compare the fitness function to obtain the optimal subsequence set; The predicted wind speed obtaining module is used to input the optimal subsequence set into the prediction network for prediction to obtain the prediction result of each subsequence in the optimal subsequence set, and add the prediction results of all subsequences in the optimal subsequence set to obtain the predicted wind speed of super-capacity energy storage grid connection.

7. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a supercapacitor energy storage frequency modulation and grid-connected method as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, a supercapacitor energy storage frequency modulation and grid-connected method as described in any one of claims 1 to 5 is implemented.

Images