A Method for Predicting Frequency Modulation Commands of a Sodium-Ion Supercapacitor Energy Storage System and Related Equipment

Abstract

The present invention relates to the technical field of supercapacitor energy storage, and specifically relates to a method for predicting frequency modulation commands of a sodium-ion supercapacitor energy storage system and related devices. First, obtain the non-linear sequence in the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system; respectively compensate the frequency modulation command data in the non-linear sequence through a compensation coefficient group to obtain a compensated sequence, and the compensated sequence is a linear sequence; according to the compensated sequence, use the constructed prediction model to predict the frequency modulation command to obtain a corresponding prediction result, obtain a reduction coefficient group based on the compensation coefficient group, and use the reduction coefficients in the reduction coefficient group to correspondingly restore the frequency modulation data in the prediction result, and finally obtain the predicted frequency modulation sequence of the sodium-ion supercapacitor energy storage system. This method can improve the prediction accuracy of frequency modulation commands.

Description

Technical Field

[0001] The present invention relates to the technical field of supercapacitor energy storage systems, and particularly to a method for predicting frequency modulation commands of a sodium-ion supercapacitor energy storage system and related equipment. Background Art

[0002] Frequency modulation of thermal power units is an important technical means to ensure the safe and stable operation of the power grid. For thermal power units, long-term frequency modulation will lead to increased coal consumption, reduced reliability, and shortened service life of the units. On the other hand, high-quality and efficient frequency modulation power sources are scarce. Currently, coal-fired thermal power units are still the main frequency modulation power sources. Coupled with the large-scale integration demand of new energy, environmental protection pressure restricts the regulation ability of the units, and problems such as "electricity determined by heat" for heating units further increase the demand for power frequency modulation. However, the performance of the current auxiliary frequency modulation energy storage equipment supporting thermal power plants cannot meet the efficiency and reliability requirements of the units themselves, affecting the acquisition of grid subsidy benefits and having low economic benefits.

[0003] The prediction of frequency modulation commands of the sodium-ion supercapacitor energy storage system in thermal power units is also an important link in maintaining the stable operation of the power system. The introduction of the sodium-ion supercapacitor energy storage system can significantly improve the frequency modulation response speed of the power system, enabling the system to achieve supply-demand balance faster. However, the frequency modulation commands of the sodium-ion supercapacitor energy storage system often have the characteristics of non-linearity and non-regularity, which cannot be ignored during the prediction process. Traditional prediction methods train all historical frequency modulation commands through a network model to predict the next frequency modulation command, or filter out the non-linear and non-regular data for prediction. The accuracy of the prediction results obtained by the existing methods is not high. Summary of the Invention

[0004] Aiming at the deficiencies in the existing technology, the present invention provides a method for predicting frequency modulation commands of a sodium-ion supercapacitor energy storage system and related equipment, which solves the problem of inaccurate prediction of frequency modulation commands of thermal power units by removing non-linear commands from the frequency modulation commands of thermal power units.

[0005] The object of the present invention is achieved by the following technical solutions:

[0006] A method for predicting frequency modulation commands of a sodium-ion supercapacitor energy storage system, comprising:

[0007] Obtaining non-linear sequences in the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system;

[0008] Compensating the frequency modulation command data in the non-linear sequences respectively through a compensation coefficient group to obtain a compensation sequence, and the compensation sequence is a linear sequence;

[0009] According to the compensation sequence, the frequency modulation command is predicted by using the constructed prediction model to obtain the corresponding prediction result. The prediction model is a trained deep learning network model, and the training data is historical frequency modulation sequence data;

[0010] Based on the compensation coefficient group, a reduction coefficient group is obtained, and the reduction coefficient in the reduction coefficient group is used to correspondingly reduce the frequency modulation data in the prediction result, and finally the predicted frequency modulation sequence of the sodium ion supercapacitor energy storage system is obtained.

[0011] Further, before obtaining the non-linear sequence in the frequency modulation sequence data of the sodium ion supercapacitor energy storage system, it includes:

[0012] The frequency modulation sequence data of the sodium ion supercapacitor energy storage system is obtained, and it is judged whether there is a non-linear sequence in the frequency modulation sequence. The judgment method is: judge whether there is a subsequence that satisfies the first condition in the frequency modulation sequence. When the subsequence satisfies any one of the situations in the first condition, the subsequence is a non-linear sequence, otherwise it is a linear sequence. The first condition is:

[0013]

[0014] Among them, the first situation is:

[0015]

[0016] The second situation is:

[0017]

[0018] In the formula, x i is an element in the frequency modulation sequence; i is the number of elements; [x i , x i+1 , x i+2 …, x i+Q , x i+Q+1 is the frequency modulation sequence and belongs to the non-linear sequence; Q is the length parameter of the subsequence in the frequency modulation sequence, and N is the length of the frequency modulation sequence of the sodium ion supercapacitor energy storage system.

[0019] Further, the step of respectively compensating the frequency modulation command data in the non-linear sequence through the compensation coefficient group to obtain a compensation sequence includes: when the non-linear sequence is the first situation, the first compensation coefficient in the first compensation coefficient group is used to correspondingly compensate the frequency modulation data in the non-linear sequence to obtain the first compensation sequence. The calculation expression of the first compensation coefficient is:

[0020]

[0021] In the formula, is the first compensation coefficient, is the first activation function, is the second activation function, a t is the frequency modulation sequence parameter, rand is a random number, is the frequency modulation instruction data related to the frequency modulation sequence time t; the frequency modulation sequence parameter is:

[0022]

[0023] In the formula, a1, a2,..., a i+Q are respectively the frequency modulation sequence parameters corresponding to the frequency modulation data in the non-linear sequence.

[0024] Furthermore, obtaining the reduction coefficient group based on the compensation coefficient group, and using the reduction coefficients in the reduction coefficient group to correspondingly restore the frequency modulation data in the prediction result includes:

[0025] Obtaining the first reduction coefficient group according to the first compensation coefficient group and the prediction result obtained based on the first compensation sequence, and respectively restoring the frequency modulation data in the prediction result according to the first reduction coefficients in the first reduction coefficient group. The specific restoration method is: subtracting the corresponding first reduction coefficients from each frequency modulation data in the prediction result to obtain the restored predicted frequency modulation data; the calculation expression of the first reduction coefficient is:

[0026]

[0027] In the formula, k t is the first reduction coefficient, is the first compensation coefficient, is the reduction parameter, N is the number of data in the frequency modulation sequence, and the reduction parameter is:

[0028]

[0029] In the formula, Swish() is the fourth activation function.

[0030] Furthermore,

[0031] Obtaining the reduction coefficient group based on the compensation coefficient group, and using the reduction coefficients in the reduction coefficient group to correspondingly restore the frequency modulation data in the prediction result also includes:

[0032] Obtaining the second reduction coefficient group according to the second compensation coefficient group and the prediction result obtained based on the second compensation sequence, and respectively restoring the frequency modulation data in the prediction result according to the second reduction coefficients in the second reduction coefficient group. The specific restoration method is: subtracting the corresponding second reduction coefficients from each frequency modulation data in the prediction result to obtain the restored predicted frequency modulation data; the calculation expression of the second reduction coefficient is:

[0033] ;

[0034] In the formula, is the second reduction coefficient, is the second compensation coefficient, is the reduction parameter, N is the number of data in the frequency modulation sequence, and the reduction parameter is:

[0035]

[0036] In the formula, Swish() is the fourth activation function.

[0037] Furthermore, compensating the frequency modulation command data in the non-linear sequence respectively through the compensation coefficient group to obtain a compensation sequence, including: when the non-linear sequence is the second case, then compensating the frequency modulation data in the non-linear sequence through the second compensation coefficient in the second compensation coefficient group to obtain a second compensation sequence, and the calculation expression of the second compensation coefficient is:

[0038]

[0039] In the formula, is the second compensation coefficient, tanh() is the third activation function, a t is the frequency modulation sequence parameter, is the frequency modulation command data related to time t; the frequency modulation sequence parameter is:

[0040]

[0041] In the formula, a1, a2,..., a i+Q are respectively the frequency modulation sequence parameters corresponding to the frequency modulation data in the non-linear sequence.

[0042] Furthermore, the basic model of the prediction model adopts any one of a long short-term memory neural network model, an RNN neural network model, and a bidirectional long short-term memory network model.

[0043] A frequency modulation command prediction system for a sodium-ion supercapacitive energy storage system, including:

[0044] A data acquisition module, configured to obtain the non-linear sequence in the frequency modulation sequence data of the sodium-ion supercapacitive energy storage system;

[0045] A data compensation module, configured to respectively compensate the frequency modulation command data in the non-linear sequence through a compensation coefficient group to obtain a compensation sequence, and the compensation sequence is a linear sequence;

[0046] A frequency modulation command prediction module, configured to predict a frequency modulation command according to the compensation sequence by using a constructed prediction model, and obtain a corresponding prediction result. The prediction model is a trained deep learning network model, and the training data is historical frequency modulation sequence data;

[0047] A data restoration module, configured to obtain a restoration coefficient group based on the compensation coefficient group, and use the restoration coefficients in the restoration coefficient group to correspondingly restore the frequency modulation data in the prediction result, and finally obtain a predicted frequency modulation sequence of the sodium-ion supercapacitor energy storage system.

[0048] An electronic device includes a processor and a memory. The processor is configured to execute a computer program stored in the memory to implement the above-mentioned method for predicting the frequency modulation command of the sodium-ion supercapacitor energy storage system.

[0049] A computer-readable storage medium stores at least one instruction, and when the at least one instruction is executed by a processor, the above-mentioned method for predicting the frequency modulation command of the sodium-ion supercapacitor energy storage system is implemented.

[0050] The beneficial effects of the present invention are as follows: The method for predicting the frequency modulation command of the sodium-ion supercapacitor energy storage system provided by the present invention fully considers the non-linearity and non-regularity of some frequency modulation commands, linearly compensates the frequency modulation commands by using compensation coefficients, and uses the linearly compensated data for prediction by a prediction model. In the frequency modulation sequence, due to the complexity of the system itself and the interference of the external environment, there may be non-linear sequences. By identifying these non-linear sequences and applying the compensation coefficient group for compensation, they can be converted into linear sequences, thereby simplifying the subsequent processing process and improving the accuracy of data processing. The obtained prediction result is then restored by using the restoration coefficient, and the restoration coefficient is obtained according to the compensation coefficient. This method can not only reduce the difficulty of prediction by the prediction model, but also effectively improve the accuracy of the prediction result. After obtaining the prediction result, the frequency modulation data in the prediction result is restored by the restoration coefficient group, which can ensure that the prediction result can truly reflect the actual frequency modulation demand of the supercapacitor energy storage system. This process is an important bridge from the model output of the prediction result to the actual application, ensuring the effectiveness and practicality of the prediction result. Description of the Drawings

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings.

[0052] Figure 1Schematic flow chart of the method for predicting frequency modulation commands of a sodium-ion ultracapacitive energy storage system in an embodiment of the present invention;

[0053] Figure 2 Schematic structural diagram of a sodium-ion ultracapacitive energy storage system frequency modulation command prediction system in an embodiment of the present invention;

[0054] Figure 3 Schematic structural diagram of an electronic device in an embodiment of the present invention. Detailed implementation manners

[0055] In order to make the objectives and technical solutions of the present invention clearer and easier to understand, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] The concept of the present invention is to provide a method for predicting frequency modulation commands of a sodium-ion ultracapacitive energy storage system, including:

[0057] Obtaining a non-linear sequence in the frequency modulation sequence data of the sodium-ion ultracapacitive energy storage system;

[0058] Compensating the frequency modulation command data in the non-linear sequence respectively through a compensation coefficient group to obtain a compensated sequence, and the compensated sequence is a linear sequence;

[0059] According to the compensated sequence, using the constructed prediction model to predict the frequency modulation command to obtain a corresponding prediction result, the prediction model is a trained deep learning network model, and the training data is the frequency modulation sequence data;

[0060] Based on the compensation coefficient group, a reduction coefficient group is obtained, and the reduction coefficients in the reduction coefficient group are used to reduce the corresponding frequency modulation data in the prediction result, and finally the predicted frequency modulation sequence of the sodium-ion ultracapacitive energy storage system is obtained.

[0061] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. Among them, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments.

[0062] Embodiment 1

[0063] As Figure 1 shown, a method for predicting frequency modulation commands of a sodium-ion ultracapacitive energy storage system in this embodiment adopts the following specific implementation manners.

[0064] S1. Obtain the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system, and determine whether there is a non-linear sequence in the frequency modulation sequence. In this embodiment, the number of frequency modulation command data in the obtained frequency modulation sequence is N, that is, the length of the frequency modulation sequence of the sodium-ion supercapacitor energy storage system, where N≥300, then the frequency modulation sequence data is [x1, x2, x3,..., x N .

[0065] The specific method for determining whether the frequency modulation sequence is non-linear sequence data is to determine whether the frequency modulation sequence satisfies the first condition, and determine whether there is a subsequence in the frequency modulation sequence that satisfies the first condition. When the subsequence satisfies any one of the situations in the first condition, the subsequence is a non-linear sequence, otherwise it is a linear sequence. And the first condition is:

[0066]

[0067] In the formula, x i is an element in the frequency modulation sequence; i is the number of elements; Q is the length parameter of the subsequence in the frequency modulation sequence; [x i , x i+1 , x i+2 …, x i+Q , x i+Q+1 is the frequency modulation sequence and belongs to the non-linear sequence. When a partial sequence in the frequency modulation sequence satisfies the above first situation or the second situation, then the subsequence is a non-linear sequence.

[0068] S2. If there is a non-linear sequence, compensate the frequency modulation command data in the non-linear sequence respectively through the compensation coefficient group to obtain a compensation sequence. Wherein the compensation sequence is a linear sequence.

[0069] Specifically, when a partial sequence satisfies the first situation, the frequency modulation data in the non-linear sequence is compensated accordingly through the first compensation coefficient in the first compensation coefficient group to obtain a first compensation sequence. And the first situation of the non-linear sequence is:

[0070]

[0071] Where the first compensation coefficient is defined as:

[0072]

[0073] In the formula, is the first compensation coefficient, is the first activation function, is the second activation function, a t is the frequency modulation sequence parameter, rand is the random number calculation, is the frequency modulation command data related to the time t of the frequency modulation sequence;.

[0074] Calculate respectively :

[0075]

[0076] Form a first compensation coefficient group through the first compensation coefficients corresponding to the frequency modulation data in the non - linear sequence, that is . By adding the first compensation coefficient to the corresponding frequency modulation data, the first compensation sequence can be obtained, that is .

[0077] Among them, the frequency modulation sequence parameters are defined as:

[0078]

[0079] In the formula, a1, a2,..., a i+Q are respectively the frequency modulation sequence parameters corresponding to the frequency modulation data in the non - linear sequence.

[0080] Similarly, when a partial sequence satisfies the second case, the frequency modulation data in the non - linear sequence is compensated accordingly through the second compensation coefficients in the second compensation coefficient group to obtain a second compensation sequence.

[0081] Among them, the second case in the non - linear sequence is:

[0082]

[0083] Then, the corresponding second compensation coefficient is defined as:

[0084]

[0085] In the formula, is the second compensation coefficient, tanh() is the third activation function, a t is the frequency modulation sequence parameter, and this frequency modulation sequence parameter has the same definition as the above.

[0086] Calculate respectively the second compensation coefficients corresponding to the frequency modulation data in the non - linear sequence, and form a second compensation coefficient group through the second compensation coefficients corresponding to the frequency modulation data in the non - linear sequence, that is . By adding the second compensation coefficient to the corresponding frequency modulation data, the second compensation sequence can be obtained, that is .

[0087] S3. According to the compensation sequence, use the constructed prediction model to predict the frequency modulation instruction to obtain the corresponding prediction result, where the prediction model is a trained deep - learning network model, and the training data is historical frequency modulation sequence data;

[0088] Specifically, the base model of the prediction model adopts any one of a long short-term memory neural network model, an RNN neural network model, and a bidirectional long short-term memory network model. The prediction model in this embodiment adopts a long short-term memory neural network model. The non-linear subsequences in the historical frequency modulation sequence data are de-linearized by the above method to obtain a linear sequence, and the linear sequence is input into the long short-term memory neural network model for training to obtain a trained prediction model.

[0089] The prediction model in this embodiment is a model existing in the prior art, and its information transmission process exists in the prior art and will not be described in detail here.

[0090] S4. Based on the compensation coefficient group, a reduction coefficient group is obtained, and the reduction coefficients in the reduction coefficient group are used to correspondingly reduce the frequency modulation data in the prediction result, and finally the predicted frequency modulation sequence of the sodium-ion over-capacity energy storage system is obtained.

[0091] Specifically, based on the non-linear sequence in the first case, a first reduction coefficient group is correspondingly obtained according to the first compensation coefficient group and the prediction result based on the first compensation sequence. The frequency modulation data in the prediction result are respectively reduced according to the first reduction coefficients in the first reduction coefficient group. The specific reduction method is to subtract the corresponding first reduction coefficients from the respective frequency modulation data in the prediction result to obtain the reduced predicted frequency modulation data.

[0092] The first reduction coefficient therein is:

[0093]

[0094] In the formula, k t is the first reduction coefficient, is the first compensation coefficient, is the reduction parameter, N is the number of data in the frequency modulation sequence, and both Q and N are positive integers. The reduction parameter is:

[0095]

[0096] In the formula, Swish() is the fourth activation function, .

[0097] When the prediction result is [x N+1 , x N+2 , x N+3 ,..., x N+0.05N , and the first reduction coefficient group is , then the reduced predicted frequency modulation data is:

[0098] [x N+1 -k1, x N+2 -k2, x N+3 -k3,..., xN+0.05N -k 0.05N-1 。

[0099] Based on the non-linear sequence in the second case, the second reduction coefficient group is obtained corresponding to the second compensation coefficient group and the prediction result obtained based on the second compensation sequence. The frequency modulation data in the prediction result is restored respectively according to the second reduction coefficients in the second reduction coefficient group. The specific restoration method is to subtract the corresponding second reduction coefficients from each frequency modulation data in the prediction result to obtain the restored predicted frequency modulation data.

[0100] Among them, the second reduction coefficient is:

[0101] 。

[0102] In the formula, is the second reduction coefficient, is the second compensation coefficient, is the reduction parameter, N is the number of data in the frequency modulation sequence, and both Q and N are positive integers. The reduction parameter is:

[0103]

[0104] In the formula, Swish() is the fourth activation function.

[0105] When the prediction result is [x N+1 , x N+2 , x N+3 ,..., x N+0.05N , and the first reduction coefficient group is , then the restored predicted frequency modulation data is:

[0106] [x N+1 - k`1, x N+2 - k`2, x N+3 - k`3,..., x N+0.05N - k` 0.05N-1 .

[0107] For the frequency modulation command prediction method of the sodium-ion supercapacitor energy storage system provided in this embodiment, implementation verification was carried out as follows:

[0108] The data set uses the historical frequency modulation data of a certain thermal power plant from January to June 2022; the historical frequency modulation data is divided and preprocessed; then the non-linear sequence in the frequency modulation data is obtained; then the calculation of the compensation and reduction parameters is carried out; according to the compensation sequence, the trained prediction model is used to predict the frequency modulation command; finally, the prediction result is restored in combination with the compensation coefficient to obtain the predicted frequency modulation sequence of the sodium-ion supercapacitor energy storage system. The verification data is shown in Table 1.

[0109] Table 1 is a comparison table of the method of the present invention and traditional prediction methods

[0110]

[0111] As can be seen from Table 1, the method of the present invention reduces the mean square error by 38.2% and the mean absolute error by 34.2% through nonlinear compensation; the linearity of the compensated sequence is increased to 92.3%; the prediction response time is shortened to 120 ms.

[0112] In summary, this embodiment fully considers the non-linearity and non-regularity of some frequency modulation commands, as well as the time series nature of the frequency modulation commands. The obtained frequency modulation sequence is a time series. The non-linear sequence in the frequency modulation sequence is linearly compensated using the compensation coefficient to reduce the difficulty of predicting non-linear data. The data after linear compensation is used for prediction using a prediction model, and the obtained prediction result is then restored using the restoration coefficient. The restoration coefficient is obtained based on the compensation coefficient. This method can not only reduce the difficulty of prediction by the prediction model, but also effectively improve the accuracy of the prediction result, and is more suitable for the actual situation of the sodium-ion over-capacity energy storage system.

[0113] Embodiment 2

[0114] As Figure 2 shown, a frequency modulation command prediction system for a sodium-ion over-capacity energy storage system in this embodiment includes:

[0115] A data acquisition module for obtaining the frequency modulation sequence data of the sodium-ion over-capacity energy storage system. In this embodiment, the number of frequency modulation command data in the obtained frequency modulation sequence is N, where N≥300, and the frequency modulation sequence data is [x1, x2, x3,..., x N .

[0116] A data judgment module for judging whether there is a non-linear sequence in the frequency modulation sequence based on the obtained frequency modulation sequence data of the sodium-ion over-capacity energy storage system.

[0117] The specific method for judging whether the frequency modulation sequence is non-linear sequence data is to judge whether the frequency modulation sequence satisfies the first condition, and to judge whether there is a subsequence in the frequency modulation sequence that satisfies the first condition. When the subsequence satisfies any one of the situations in the first condition, the subsequence is a non-linear sequence, otherwise it is a linear sequence. And the first condition is:

[0118]

[0119] In the formula, [x i , x i+1 , x i+2 …, x i+Q , x i+Q+1is a frequency modulation sequence, belonging to a non-linear sequence. When a partial sequence in the frequency modulation sequence satisfies the above first case or the second case, then this sub-sequence is a non-linear sequence.

[0120] A data compensation module, used to respectively compensate the frequency modulation command data in the non-linear sequence through a compensation coefficient group to obtain a compensation sequence, where the compensation sequence is a linear sequence.

[0121] Specifically, when the partial sequence satisfies the first case, the frequency modulation data in this non-linear sequence is correspondingly compensated through the first compensation coefficient in the first compensation coefficient group to obtain a first compensation sequence. And the first case of the non-linear sequence is:

[0122]

[0123] Where the first compensation coefficient is defined as:

[0124]

[0125] In the formula, is the first compensation coefficient, is the first activation function, is the second activation function, a t is the frequency modulation sequence parameter, and rand is random number calculation.

[0126] And the first compensation coefficient group is correspondingly formed through the first compensation coefficient, that is . By adding the first compensation coefficient to the corresponding frequency modulation data, the first compensation sequence can be obtained, that is .

[0127] Where the frequency modulation sequence parameter is defined as:

[0128]

[0129] In the formula, a1, a2,..., a i+Q are respectively the frequency modulation sequence parameters corresponding to the frequency modulation data in the non-linear sequence.

[0130] Similarly, when the partial sequence satisfies the second case, the frequency modulation data in this non-linear sequence is correspondingly compensated through the second compensation coefficient in the second compensation coefficient group to obtain a second compensation sequence.

[0131] Among them, the second case of the non-linear sequence is:

[0132]

[0133] Then, the corresponding second compensation coefficient is defined as:

[0134]

[0135] In the formula, is the second compensation coefficient, tanh() is the third activation function, and a t is the frequency modulation sequence parameter, and this frequency modulation sequence parameter is the same as the above definition.

[0136] The frequency modulation command prediction module is used to predict the frequency modulation command according to the compensation sequence by using the constructed prediction model to obtain the corresponding prediction result. The prediction model is a trained deep learning network model, and the training data is historical frequency modulation sequence data; specifically, the basic model of the prediction model adopts any one of the long short-term memory neural network model, the RNN neural network model, and the bidirectional long short-term memory network model. In this embodiment, the long short-term memory neural network model is adopted; among them, the RNN neural network model, that is, the Recurrent Neural Network, is a neural network used to process sequence data.

[0137] The data restoration module is used to obtain a restoration coefficient group based on the compensation coefficient group, and use the restoration coefficients in the restoration coefficient group to restore the corresponding frequency modulation data in the prediction result, and finally obtain the predicted frequency modulation sequence of the sodium ion ultra-capacitive energy storage system.

[0138] Based on the non-linear sequence of the first case, the first restoration coefficient group is obtained corresponding to the first compensation coefficient group and the prediction result obtained based on the first compensation sequence. The corresponding frequency modulation data in the prediction result is restored respectively according to the first restoration coefficients in the first restoration coefficient group. The specific restoration method is to subtract the corresponding first restoration coefficient from each frequency modulation data in the prediction result to obtain the restored predicted frequency modulation data.

[0139] The first restoration coefficient among them is:

[0140]

[0141] In the formula, k t is the first restoration coefficient, is the first compensation coefficient, is the restoration parameter, and N is the number of data in the frequency modulation sequence. The restoration parameter is:

[0142]

[0143] In the formula, Swish() is the fourth activation function, .

[0144] When the prediction result is [x N+1 , x N+2 , x N+3 ,..., x N+0.05N , the first restoration coefficient group is , then the restored predicted frequency modulation data is:

[0145] [x N+1 -k1,x N+2 -k2,x N+3 -k3,...,x N+0.05N -k 0.05N-1 .

[0146] Based on the non-linear sequence of the second case, the second restoration coefficient group is obtained corresponding to the prediction result based on the second compensation coefficient group and the second compensation sequence. The frequency modulation data in the prediction result is restored respectively according to the second restoration coefficients in the second restoration coefficient group. The specific restoration method is to subtract the corresponding second restoration coefficient from each frequency modulation data in the prediction result to obtain the restored predicted frequency modulation data.

[0147] Among them, the second restoration coefficient is:

[0148] .

[0149] In the formula, is the second restoration coefficient, is the second compensation coefficient, is the restoration parameter, and N is the number of data in the frequency modulation sequence. The restoration parameter is:

[0150]

[0151] In the formula, Swish() is the fourth activation function.

[0152] Example 3

[0153] Such as Figure 3The present embodiment shown provides an electronic device for implementing the frequency modulation command prediction method of the sodium-ion supercapacitor energy storage system in Embodiment 1 above. The electronic device 100 includes at least one processor 102, a memory 101, a computer program 103 stored in the memory 101 and executable on at least one processor 102, and at least one communication bus 104. The memory 101 can be used to store the computer program 103. The processor 102 realizes the steps of the frequency modulation command prediction method of the sodium-ion supercapacitor energy storage system in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 may mainly include a program storage area and a data storage area. Among them, the program storage area may store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area may store data created according to the use of the electronic device 100 (such as audio data, etc.). In addition, the memory 101 may include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a SmartMedia Card (SMC), a Secure Digital (SD) card, a Flash Card, at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices.

[0154] At least one processor 102 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or the processor 102 may also be any conventional processor, etc. The processor 102 is the control center of the electronic device 100 and connects various parts of the entire electronic device 100 through various interfaces and lines.

[0155] The memory 101 in the electronic device 100 stores multiple instructions to implement a frequency modulation command prediction method for a sodium-ion supercapacitor energy storage system. The processor 102 can execute multiple instructions to implement:

[0156] Obtain the non-linear sequence in the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system;

[0157] Compensate the FM command data in the non-linear sequence respectively according to the compensation coefficient group to obtain a compensated sequence, and the compensated sequence is a linear sequence;

[0158] According to the compensated sequence, use the constructed prediction model to predict the FM command to obtain the corresponding prediction result. The prediction model is a trained deep learning network model, and the training data is historical FM sequence data;

[0159] Based on the compensation coefficient group, obtain a reduction coefficient group, and use the reduction coefficients in the reduction coefficient group to correspondingly restore the FM data in the prediction result, and finally obtain the predicted FM sequence of the sodium-ion supercapacitor energy storage system.

[0160] Embodiment 4

[0161] If the modules / units integrated in the electronic device 100 are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods of the present invention can also be completed by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory and read-only memory (ROM, Read-Only Memory).

[0162] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate a device for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0163] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one or more of the processes and / or blocks Figure 1 in one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0164] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the processes and / or blocks Figure 1 in one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0165] 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 them. 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: modifications or equivalent replacements can still be made to the specific embodiments of the present invention. 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 method for predicting frequency modulation commands of a sodium-ion supercapacitor energy storage system, characterized in that Including: Obtaining the non-linear sequence in the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system; Compensating the frequency modulation command data in the non-linear sequence respectively through a compensation coefficient group to obtain a compensation sequence, and the compensation sequence is a linear sequence; According to the compensation sequence, using the constructed prediction model to predict the frequency modulation command to obtain the corresponding prediction result, and the prediction model is a trained deep learning network model, and the training data is historical frequency modulation sequence data; Obtaining a reduction coefficient group based on the compensation coefficient group, and using the reduction coefficients in the reduction coefficient group to correspondingly reduce the frequency modulation data in the prediction result, and finally obtaining the predicted frequency modulation sequence of the sodium-ion supercapacitor energy storage system; Before obtaining the non-linear sequence in the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system, it includes: Obtaining the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system, and judging whether there is a non-linear sequence in the frequency modulation sequence. The judgment method is: judging whether there is a subsequence that satisfies the first condition in the frequency modulation sequence. When the subsequence satisfies any one of the situations in the first condition, the subsequence is a non-linear sequence, otherwise it is a linear sequence. The first condition is: Among them, the first situation is: The second situation is: Where x i is an element in the frequency modulation sequence; i is the number of elements; [x i , x i+1 , x i+2 …, x i+Q , x i+Q+1 is the frequency modulation sequence, which belongs to the non-linear sequence; Q is the length parameter of the subsequence in the frequency modulation sequence, and N is the length of the frequency modulation sequence of the sodium-ion supercapacitor energy storage system; When the non-linear sequence is the first situation, the frequency modulation data in the non-linear sequence is correspondingly compensated through the first compensation coefficients in the first compensation coefficient group to obtain a first compensation sequence. The calculation expression of the first compensation coefficient is: Δ t = rand * -0.2x t * leakyRelu(a t ), -0.1x t * Relu(a t ), t = 1, 2, 3, …, i + 1, i + 2, …, i + Q Where, Δ t is the first compensation coefficient, leakyRelu() is the first activation function, Relu is the second activation function, a t is the frequency modulation sequence parameter, rand is a random number, x t is the frequency modulation command data related to the frequency modulation sequence time t; Obtaining a first reduction coefficient group corresponding to the first compensation coefficient group and the prediction result based on the first compensation sequence, and respectively reducing the frequency modulation data in the prediction result according to the first reduction coefficients in the first reduction coefficient group. The specific reduction method is: subtracting the corresponding first reduction coefficients from each frequency modulation data in the prediction result to obtain the reduced predicted frequency modulation data; the calculation expression of the first reduction coefficient is: where k t is the first reduction coefficient, Δ t is the first compensation coefficient, β t is the reduction parameter, N is the number of data in the frequency modulation sequence, and the reduction parameter is: In the formula, Swish() is the fourth activation function; When the non-linear sequence is the second situation, the frequency modulation data in the non-linear sequence is correspondingly compensated through the second compensation coefficients in the second compensation coefficient group to obtain a second compensation sequence. The calculation expression of the second compensation coefficient is: Δ′ t = rand(+0.1x t *tanh(a t ), +0.2x t *Relu(a t ), t = 1, 2, 3,..., i + 1, i + 2,..., i + Q where Δ t ′ is the second compensation coefficient, tanh() is the third activation function, a t is the frequency modulation sequence parameter, and x t is the frequency modulation command data related to time t; Obtaining a second reduction coefficient group corresponding to the second compensation coefficient group and the prediction result based on the second compensation sequence, and respectively reducing the frequency modulation data in the prediction result according to the second reduction coefficients in the second reduction coefficient group. The specific reduction method is: subtracting the corresponding second reduction coefficients from each frequency modulation data in the prediction result to obtain the reduced predicted frequency modulation data; the calculation expression of the second reduction coefficient is: where k t ′ is the second reduction coefficient, Δ t ′ is the second compensation coefficient, β t ′ is the reduction parameter, N is the number of data in the frequency modulation sequence, and the reduction parameter is: In the formula, Swish() is the fourth activation function.

2. The frequency modulation command prediction method for the sodium-ion supercapacitor energy storage system according to claim 1, wherein The frequency modulation sequence parameters are: Where a1, a2, …, a i+Q are respectively the frequency modulation sequence parameters corresponding to the frequency modulation data in the non-linear sequence.

3. The frequency modulation command prediction method for the sodium-ion supercapacitor energy storage system according to claim 1, wherein The basic model of the prediction model adopts any one of a long short-term memory neural network model, an RNN neural network model, and a bidirectional long short-term memory network model.

4. A frequency modulation command prediction system for a sodium-ion supercapacitor energy storage system, characterized in that, The steps for implementing the sodium-ion supercapacitor energy storage system frequency modulation command prediction method according to any one of claims 1-3 include: A data acquisition module for obtaining the non-linear sequence in the frequency modulation sequence data of the sodium-ion supercapacitor energy storage system; A data compensation module, which is used to compensate the frequency modulation command data in the non-linear sequence respectively through a compensation coefficient group to obtain a compensation sequence, and the compensation sequence is a linear sequence; A frequency modulation command prediction module, which is used to predict the frequency modulation command according to the compensation sequence by using the constructed prediction model to obtain the corresponding prediction result, the prediction model is a trained deep learning network model, and the training data is historical frequency modulation sequence data; A data restoration module, which is used to obtain a restoration coefficient group based on the compensation coefficient group, and use the restoration coefficients in the restoration coefficient group to restore the corresponding frequency modulation data in the prediction result, and finally obtain the predicted frequency modulation sequence of the sodium-ion supercapacitor energy storage system.

5. An electronic device, characterized in that, It includes a processor and a memory. The processor is used to execute the computer program stored in the memory to implement the frequency modulation command prediction method of the sodium-ion supercapacitor energy storage system as described in any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, and when the at least one instruction is executed by the processor, it implements the frequency modulation command prediction method of the sodium-ion supercapacitor energy storage system as described in any one of claims 1-3.

Images