A method and related equipment for predicting frequency regulation commands for energy storage based on time-mode entropy merging.

Abstract

This invention relates to the field of frequency regulation command prediction technology, and discloses a method and related equipment for predicting energy storage frequency regulation commands based on time-mode entropy merging. The method includes: calculating the local dynamic change rate of the original frequency regulation sequence to obtain the local fluctuation intensity of each sampling point; generating a mode entropy fingerprint based on the local fluctuation intensity of each sampling point, and generating a mode entropy fingerprint sequence based on the mode entropy fingerprint of all sampling points in the original frequency regulation sequence; calculating a fingerprint similarity measure for each sampling point within a set radius; determining an adaptive merging threshold based on the variance of the mode entropy fingerprint sequence; comparing the fingerprint similarity measure with the adaptive merging threshold, and merging similar fingerprint groups to obtain a calculated weighted merged value when the fingerprint similarity measure is greater than the adaptive merging threshold; obtaining the merged sequence based on the weighted merged value, and inputting the merged sequence into a GRU network for prediction. This invention solves the technical problem of inaccurate prediction results for energy storage frequency regulation commands.

Description

Technical Field

[0001] This invention relates to the field of frequency regulation command prediction technology, specifically to a method and related equipment for predicting frequency regulation commands for energy storage based on time pattern entropy merging. Background Technology

[0002] Power system frequency stability is a core indicator for ensuring the safe operation of the power grid and the quality of electricity for residential use. When the power grid encounters disturbances such as sudden load changes, fluctuations in new energy generation, or generator failures, the system frequency will deviate from the safe range (e.g., the Chinese power grid standard 50Hz ± 0.2Hz) at millisecond-level speeds. If the frequency deviation continues to exceed the safe threshold, it may trigger cascading equipment failures or even large-scale power outages, causing significant impacts on the national economy and social order.

[0003] As a "flexible regulation resource" for modern power grids, power plant energy storage systems must respond to dispatch commands within 100 milliseconds after a disturbance occurs, regulating active power through rapid charging and discharging to suppress frequency fluctuations and restore system balance. However, traditional energy storage regulation strategies based on real-time command responses have significant limitations: constrained by physical limitations such as battery charging and discharging rates and state of charge (SOC) management boundaries, relying solely on real-time commands can easily lead to regulation lag or overcompensation, making it difficult to meet the requirements for high-precision and high-time-efficiency frequency stability control.

[0004] To address the aforementioned issues, existing industry technologies propose achieving proactive regulation of energy storage systems by predicting the frequency modulation command curve 10-30 seconds in the future. Mainstream solutions employ neural network models such as Long Short-Term Memory (LSTM) and Transformer to directly predict the original frequency modulation command sequence. However, these methods suffer from fundamental technical flaws, specifically as follows:

[0005] Key signal overload: The original frequency modulation command sequence contains a large number of highly repetitive numerical segments (such as ±0.05Hz micro-oscillations lasting for tens of minutes), accounting for more than 70% of the total data. This redundant information dilutes the feature weights of key events such as steep rises and falls, and step jumps, making it difficult for the model to capture sudden disturbance features;

[0006] Model cognitive bias: The neural network overlearns redundant patterns, and the prediction error for sudden disturbance events is as high as 15%-25%, which seriously restricts the dynamic response accuracy of the energy storage system.

[0007] Waste of computing resources: The daily frequency modulation sequence data volume can reach tens of thousands of points. The inference latency of edge computing devices exceeds the 100ms threshold due to model complexity, which cannot meet the real-time requirements of frequency modulation commands.

[0008] In summary, existing technologies fail to effectively resolve the conflict between data noise and model efficiency, making it difficult to support the large-scale application of energy storage systems in high-reliability power grids. Therefore, there is an urgent need for a frequency regulation command prediction method that combines noise resistance and lightweight characteristics to improve the dynamic regulation performance of energy storage systems in response to grid frequency fluctuations. Summary of the Invention

[0009] In order to overcome the defects of the prior art, the purpose of this invention is to provide a method and related equipment for predicting energy storage frequency regulation commands based on time pattern entropy merging, so as to solve the technical problem of inaccurate prediction results of energy storage frequency regulation commands in the prior art.

[0010] This invention is achieved through the following technical solution:

[0011] In a first aspect, the present invention provides a method for predicting energy storage frequency regulation commands based on time-mode entropy merging, comprising:

[0012] The local fluctuation intensity at each sampling point is obtained by calculating the local dynamic rate of change of the original frequency modulation sequence;

[0013] A mode entropy fingerprint is generated based on the local fluctuation intensity of each sampling point, and a mode entropy fingerprint sequence is generated based on the mode entropy fingerprint of all sampling points in the original frequency modulation sequence.

[0014] For each sampling point, calculate the fingerprint similarity measure within the defined radius.

[0015] The adaptive merging threshold is determined based on the variance of the pattern entropy fingerprint sequence;

[0016] The fingerprint similarity measure is compared with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain a weighted merged value.

[0017] The merged sequence is obtained based on the weighted merging value, and then the merged sequence is input into the GRU network for prediction, outputting the final frequency modulation command prediction result.

[0018] Preferably, the formula for calculating the local fluctuation intensity at each sampling point is as follows:

[0019] Where i = 2, 3, ..., N-1

[0020] in, These are sampling points; when the boundary point i=0 or i=N, .

[0021] Preferably, the formula for generating the pattern entropy fingerprint sequence is as follows:

[0022]

[0023] in, To prevent division by zero; This represents the local fluctuation intensity at each sampling point.

[0024] The preferred expression for fingerprint similarity measurement is as follows:

[0025]

[0026] Where σ is the similarity bandwidth parameter σ=0.1·Std(Pt), Std(Pt) is the standard deviation of the original sequence Pt; j is the value of the sampling point within the set radius; The pattern entropy fingerprint sequence; It is a sequence of values ​​of the sampling points within the set radius.

[0027] The preferred expression for the adaptive merging threshold is as follows:

[0028] θ = α·Var(F);

[0029]

[0030] in, The pattern entropy fingerprint sequence; α is the mean of the pattern entropy fingerprint sequence; α = 0.5 is the scaling factor.

[0031] The preferred formula for calculating the weighted combined value is as follows:

[0032]

[0033] in, For similar fingerprint groups; k is the original value of the similar fingerprint group; Entropy weights; for The original values ​​of all similar points within the range.

[0034] Preferably, the expression for the entropy weight is as follows:

[0035]

[0036] in, Let k be the local dynamic rate of change at point k. is the mean of the pattern entropy fingerprint sequence.

[0037] Secondly, the present invention also provides a storage frequency regulation command prediction system based on time pattern entropy merging, comprising:

[0038] The local fluctuation intensity calculation module is used to calculate the local dynamic change rate of the original frequency modulation sequence to obtain the local fluctuation intensity of each sampling point.

[0039] The pattern entropy fingerprint sequence generation module is used to generate pattern entropy fingerprints based on the local fluctuation intensity of each sampling point, and to generate pattern entropy fingerprint sequences based on the pattern entropy fingerprints of all sampling points in the original frequency modulation sequence.

[0040] The fingerprint similarity measurement calculation module is used to calculate the fingerprint similarity measurement for each sampling point within a set radius.

[0041] An adaptive merging threshold determination module is used to determine the adaptive merging threshold based on the variance of the pattern entropy fingerprint sequence.

[0042] The weighted merge value calculation module is used to compare the fingerprint similarity measure with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain the calculated weighted merge value.

[0043] The frequency modulation command prediction module is used to obtain the merged sequence based on the weighted merging value, input the merged sequence into the GRU network for prediction, and output the final frequency modulation command prediction result.

[0044] Thirdly, the present invention also provides a mobile terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the energy storage frequency modulation command prediction method based on time pattern entropy merging as described above.

[0045] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the energy storage frequency modulation command prediction method based on time pattern entropy merging as described above.

[0046] Compared with the prior art, the present invention has the following beneficial technical effects:

[0047] This invention provides a method for predicting energy storage frequency regulation commands based on time-mode entropy merging. It calculates the local dynamic change rate of the original frequency regulation sequence to obtain the local fluctuation intensity of each sampling point, enabling detailed analysis of the dynamic changes near each sampling point in the frequency regulation sequence and more accurately capturing local fluctuation characteristics. Mode entropy fingerprints, as a unique feature representation, can comprehensively reflect the dynamic characteristics of the sampling point and its surrounding area. This method transforms complex frequency regulation sequences into more representative and discriminative fingerprint sequences, facilitating clearer identification of different patterns and features within the frequency regulation sequence and providing a robust data structure for subsequent similarity measurement and merging operations. By comparing the similarity of fingerprints within a neighborhood, fingerprint groups with similar features can be accurately identified, avoiding local similarities that might be overlooked due to global comparisons, thus contributing to more accurate identification of repetitive or similar patterns in the frequency regulation sequence. An adaptive merging threshold is determined based on the variance of the mode entropy fingerprint sequence, allowing the merging threshold to automatically adjust according to the characteristics of the input data. By using a weighted merging method, the contributions of each fingerprint within a similar fingerprint group are comprehensively considered, making the merged value more accurately reflect the common features of the fingerprint group. Compared to the original frequency-modulated sequence, the merged sequence significantly reduces the amount of data while retaining key feature information. This greatly reduces the computational burden on the GRU network when processing data, enabling faster training and prediction, and improving prediction efficiency. Attached Figure Description

[0048] Figure 1 This is a flowchart of the energy storage frequency regulation command prediction method in this invention;

[0049] Figure 2 This is a schematic diagram of the energy storage frequency regulation command prediction system in this invention;

[0050] In the diagram: 1. Local fluctuation intensity calculation module; 2. Pattern entropy fingerprint sequence generation module; 3. Fingerprint similarity measurement calculation module; 4. Adaptive merging threshold determination module; 5. Weighted merging value calculation module; 6. Frequency modulation command prediction module. Detailed Implementation

[0051] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0052] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0053] The purpose of this invention is to provide a method and related equipment for predicting energy storage frequency regulation commands based on time pattern entropy merging, so as to solve the technical problem of inaccurate prediction results of energy storage frequency regulation commands in the prior art.

[0054] The present invention will now be described in further detail with reference to the accompanying drawings:

[0055] Example 1

[0056] See Figure 1 In one embodiment of the present invention, a method for predicting energy storage frequency regulation commands based on time-mode entropy merging is provided. This method aims to handle nonlinear and irregular frequency regulation sequences Pt=[X1,X2,X3,...X... i ..,X N This method captures "regular similarity" by calculating the local pattern entropy features of each point, and then merges values ​​based on feature similarity. The core idea is to generate pattern fingerprints using the local dynamic characteristics of the sequence (such as rate of change and information entropy), and then merge the values ​​corresponding to similar fingerprints using an adaptive threshold. The specific process includes:

[0057] Step 1: Calculate the local dynamic change rate of the original frequency modulation sequence to obtain the local fluctuation intensity of each sampling point;

[0058] Specifically, let the original frequency modulation sequence be Pt=[X1,X2,X3,.X... i ..,X N (The subscript indicates the number of samples, which is taken once per second).

[0059] The formula for calculating the local fluctuation intensity at each sampling point is as follows:

[0060] Where i = 2, 3, ..., N-1

[0061] in, These are sampling points; when the boundary point i=0 or i=N, .

[0062] In this embodiment, point X is calculated. i The average of the absolute changes at the previous and subsequent points is used as the local fluctuation intensity.

[0063] Step 2: Generate mode entropy fingerprints based on the local fluctuation intensity of each sampling point, and generate mode entropy fingerprint sequences based on the mode entropy fingerprints of all sampling points in the original frequency modulation sequence.

[0064] Specifically, the formula for generating the pattern entropy fingerprint sequence is as follows:

[0065]

[0066] in, To prevent division by zero; The local fluctuation intensity at each sampling point;

[0067] In this embodiment, the rate of change Convert to information entropy form, A high value indicates that the point is in a high-variable region, and a low value indicates a stable region. The pattern of the nonlinear sequence is encoded as a scalar fingerprint.

[0068] Step 3: Calculate the fingerprint similarity metric for each sampling point within the defined radius; Pt=[X1,X2,X3,.X i ..,X N All values ​​in [F1, F2, F3, ..., F] are generated based on the pattern entropy fingerprint to produce the corresponding values. i ..,F N ].

[0069] Specifically, the expression for fingerprint similarity measurement is as follows:

[0070]

[0071] Where σ is the similarity bandwidth parameter σ=0.1·Std(Pt), Std(Pt) is the standard deviation of the original sequence Pt; j is the value of the sampling point within the set radius; The pattern entropy fingerprint sequence; It is a sequence of values ​​of the sampling points within the set radius.

[0072] In this embodiment, the fingerprint similarity metric is calculated for all values ​​within the radius of Xi. For example, when the radius is 5, X... i =X7, from X2 to X 12 Both need to calculate the fingerprint similarity measure with X7.

[0073] Step 4: Determine the adaptive merging threshold based on the variance of the pattern entropy fingerprint sequence;

[0074] Specifically, the expression for the adaptive merging threshold is as follows:

[0075] θ = α·Var(F);

[0076]

[0077] in, The pattern entropy fingerprint sequence; α is the mean of the pattern entropy fingerprint sequence; α = 0.5 is the scaling factor.

[0078] In this embodiment, the threshold θ adapts to the sequence complexity: when the variance is high (irregular), θ is larger, reducing merging; when the variance is low, θ is smaller, resulting in less merging.

[0079] Step 5: Compare the fingerprint similarity measure with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, merge similar fingerprint groups to obtain the calculated weighted merge value.

[0080] Specifically, the formula for calculating the weighted combined value is as follows:

[0081]

[0082] in, For similar fingerprint groups; k is the original value of the similar fingerprint group; Entropy weights are used to give higher weights to low-entropy points (stable modes); for The original values ​​of all similar points within the range.

[0083] The expression for the entropy weight is as follows:

[0084]

[0085] in, Let k be the local dynamic rate of change at point k. is the mean of the pattern entropy fingerprint sequence.

[0086] Step 6: Obtain the merged sequence based on the weighted merging value, input the merged sequence into the GRU network for prediction, and output the final frequency modulation command prediction result.

[0087] To further verify the advantages of the present invention, the present invention uses both the method of the present invention and the GRU direct prediction method to predict frequency modulation sequences. The prediction results are shown in Table 1.

[0088] Frequency modulation sequence 1 is derived from frequency modulation data from a power plant in Gansu Province from 3 PM to midnight on March 6, 2025.

[0089] Frequency modulation sequence 2 is from frequency modulation data of a power plant in Gansu Province from 3 PM to midnight on March 7, 2025.

[0090] Table 1 shows the prediction results for frequency modulation sequences 1 and 2.

[0091]

[0092] Table 2 Four evaluation indicators

[0093]

[0094] Where N represents the sample size, and These represent the actual value and the predicted value at time n, respectively.

[0095] As shown in Table 2, the experimental results indicate that all four evaluation indicators decreased, demonstrating that the decomposition method proposed in this invention can significantly reduce the nonlinearity of the original sequence compared to the traditional GRU prediction method, thereby further improving prediction accuracy. This helps power plants improve their frequency regulation response capabilities and ultimately increase their profitability.

[0096] In summary, this invention provides a method for predicting energy storage frequency regulation commands based on time-mode entropy merging. It calculates the local dynamic change rate of the original frequency regulation sequence to obtain the local fluctuation intensity of each sampling point. This allows for detailed analysis of the dynamic changes near each sampling point in the frequency regulation sequence, while more accurately capturing local fluctuation characteristics. Mode entropy fingerprints, as a unique feature representation, can comprehensively reflect the dynamic characteristics of the sampling point and its surrounding area. This method transforms complex frequency regulation sequences into more representative and discriminative fingerprint sequences, facilitating clearer identification of different patterns and features within the frequency regulation sequence and providing a robust data structure for subsequent similarity measurement and merging operations. By comparing the similarity of fingerprints within a neighborhood, fingerprint groups with similar features can be accurately identified, avoiding local similarities that might be overlooked due to global comparisons, thus contributing to more accurate identification of repetitive or similar patterns in the frequency regulation sequence. An adaptive merging threshold is determined based on the variance of the mode entropy fingerprint sequence, allowing the merging threshold to automatically adjust according to the characteristics of the input data. By using a weighted merging method, the contributions of each fingerprint within a similar fingerprint group are comprehensively considered, making the merged value more accurately reflect the common features of the fingerprint group. Compared to the original frequency-modulated sequence, the merged sequence significantly reduces the amount of data while retaining key feature information. This greatly reduces the computational burden on the GRU network when processing data, enabling faster training and prediction, and improving prediction efficiency.

[0097] This invention pioneers a "point-driven neighborhood" mechanism: by dynamically capturing local fluctuation characteristics solely through the instantaneous change relationship between each sampling point and its immediate neighbors, it completely eliminates window dependency. This design preserves the microscopic dynamics of the sequence while avoiding the subjectivity of pre-defined segmentation. It also integrates information entropy with signal dynamics, pioneering the concept of "pattern entropy fingerprint": using the local rate of change as the entropy source to quantify the disorder intensity of each point; high-entropy fingerprints identify regions of violent fluctuation (such as abrupt peaks), while low-entropy fingerprints identify stable regions (such as stable baselines), transforming invisible temporal patterns into computable entropy scalars.

[0098] This invention introduces a "time-skip similarity" mechanism: directly comparing the pattern entropy fingerprints of any two points (even if time is discontinuous); similar points are automatically clustered into the same group, achieving "global pattern capture" (e.g., similar oscillation peaks scattered in power plant frequency regulation can be merged into a representative waveform). Simultaneously, this invention proposes a "variance-driven threshold": automatically adjusting the merging sensitivity based on the disorder (variance) of the entire fingerprint; high-disorder sequences (e.g., strong noise) raise the merging threshold to preserve details; low-disorder sequences (e.g., stable segments) lower the threshold to improve compression ratio. This invention designs an "entropy credibility weighting" strategy: assigning high weights to low-entropy points (stable patterns) and low weights to high-entropy points (chaotic patterns); weighted output representative values ​​suppress noise interference and strengthen the expression of regularity.

[0099] Example 2

[0100] according to Figure 2 As shown, the present invention also provides a storage frequency regulation command prediction system based on time pattern entropy merging, comprising:

[0101] Local fluctuation intensity calculation module 1 is used to calculate the local dynamic change rate of the original frequency modulation sequence to obtain the local fluctuation intensity of each sampling point;

[0102] The pattern entropy fingerprint sequence generation module 2 is used to generate pattern entropy fingerprints based on the local fluctuation intensity of each sampling point, and to generate a pattern entropy fingerprint sequence based on the pattern entropy fingerprints of all sampling points in the original frequency modulation sequence.

[0103] The fingerprint similarity measurement calculation module 3 is used to calculate the fingerprint similarity measurement for each sampling point within a set radius.

[0104] The adaptive merging threshold determination module 4 is used to determine the adaptive merging threshold based on the variance of the pattern entropy fingerprint sequence.

[0105] The weighted merging value calculation module 5 is used to compare the fingerprint similarity measure with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain the calculated weighted merging value.

[0106] The frequency modulation command prediction module 6 is used to obtain the merged sequence based on the weighted merging value, input the merged sequence into the GRU network for prediction, and output the final frequency modulation command prediction result.

[0107] Example 3

[0108] The present invention also provides a mobile terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor, such as a storage frequency modulation command prediction program based on time pattern entropy merging.

[0109] When the processor executes the computer program, it implements the steps of the above-described energy storage frequency modulation command prediction method based on time pattern entropy merging, for example:

[0110] The local fluctuation intensity at each sampling point is obtained by calculating the local dynamic rate of change of the original frequency modulation sequence;

[0111] A mode entropy fingerprint is generated based on the local fluctuation intensity of each sampling point, and a mode entropy fingerprint sequence is generated based on the mode entropy fingerprint of all sampling points in the original frequency modulation sequence.

[0112] For each sampling point, calculate the fingerprint similarity measure within the defined radius.

[0113] The adaptive merging threshold is determined based on the variance of the pattern entropy fingerprint sequence;

[0114] The fingerprint similarity measure is compared with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain a weighted merged value.

[0115] The merged sequence is obtained based on the weighted merging value, and then the merged sequence is input into the GRU network for prediction, outputting the final frequency modulation command prediction result.

[0116] Alternatively, when the processor executes the computer program, it implements the functions of each module in the above system, for example:

[0117] Local fluctuation intensity calculation module 1 is used to calculate the local dynamic change rate of the original frequency modulation sequence to obtain the local fluctuation intensity of each sampling point;

[0118] The pattern entropy fingerprint sequence generation module 2 is used to generate pattern entropy fingerprints based on the local fluctuation intensity of each sampling point, and to generate a pattern entropy fingerprint sequence based on the pattern entropy fingerprints of all sampling points in the original frequency modulation sequence.

[0119] The fingerprint similarity measurement calculation module 3 is used to calculate the fingerprint similarity measurement for each sampling point within a set radius.

[0120] The adaptive merging threshold determination module 4 is used to determine the adaptive merging threshold based on the variance of the pattern entropy fingerprint sequence.

[0121] The weighted merging value calculation module 5 is used to compare the fingerprint similarity measure with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain the calculated weighted merging value.

[0122] The frequency modulation command prediction module 6 is used to obtain the merged sequence based on the weighted merging value, input the merged sequence into the GRU network for prediction, and output the final frequency modulation command prediction result.

[0123] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the mobile terminal.

[0124] For example, the computer program can be divided into a local fluctuation intensity calculation module 1, a pattern entropy fingerprint sequence generation module 2, a fingerprint similarity measurement calculation module 3, an adaptive merging threshold determination module 4, a weighted merging value calculation module 5, and a frequency modulation command prediction module 6.

[0125] The specific functions of each module are as follows:

[0126] Local fluctuation intensity calculation module 1 is used to calculate the local dynamic change rate of the original frequency modulation sequence to obtain the local fluctuation intensity of each sampling point;

[0127] The pattern entropy fingerprint sequence generation module 2 is used to generate pattern entropy fingerprints based on the local fluctuation intensity of each sampling point, and to generate a pattern entropy fingerprint sequence based on the pattern entropy fingerprints of all sampling points in the original frequency modulation sequence.

[0128] The fingerprint similarity measurement calculation module 3 is used to calculate the fingerprint similarity measurement for each sampling point within a set radius.

[0129] The adaptive merging threshold determination module 4 is used to determine the adaptive merging threshold based on the variance of the pattern entropy fingerprint sequence.

[0130] The weighted merging value calculation module 5 is used to compare the fingerprint similarity measure with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain the calculated weighted merging value.

[0131] The frequency modulation command prediction module 6 is used to obtain the merged sequence based on the weighted merging value, input the merged sequence into the GRU network for prediction, and output the final frequency modulation command prediction result.

[0132] The mobile terminal can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The mobile terminal may include, but is not limited to, a processor and memory.

[0133] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the mobile terminal, connecting various parts of the mobile terminal via various interfaces and lines.

[0134] The memory can be used to store the computer program and / or module. The processor implements various functions of the mobile terminal by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0135] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function (such as sound playback, image playback, etc.); the data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). Furthermore, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, SmartMediaCards (SMC), Secure Digital (SD) cards, FlashCards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0136] Example 4

[0137] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the energy storage frequency modulation command prediction method based on time pattern entropy merging.

[0138] If the modules / units integrated in the mobile terminal are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0139] Based on this understanding, all or part of the processes in the above method can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of the above-described aggregated reinforcement learning resource scheduling method. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate form.

[0140] The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0141] It should be noted that the content contained in the computer-readable medium may be appropriately added to or subtracted from the content as required by the legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium may not include electrical carrier signals and telecommunication signals.

[0142] 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 skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting frequency regulation commands for energy storage based on time-mode entropy merging, characterized in that, include: The local fluctuation intensity at each sampling point is obtained by calculating the local dynamic rate of change of the original frequency modulation sequence; A mode entropy fingerprint is generated based on the local fluctuation intensity of each sampling point, and a mode entropy fingerprint sequence is generated based on the mode entropy fingerprint of all sampling points in the original frequency modulation sequence. The formula for generating the pattern entropy fingerprint sequence is as follows: in, To prevent division by zero; The local fluctuation intensity at each sampling point; For each sampling point, calculate the fingerprint similarity measure within the defined radius. The adaptive merging threshold is determined based on the variance of the pattern entropy fingerprint sequence; The fingerprint similarity measure is compared with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain a weighted merged value. The formula for calculating the weighted combined value is as follows: in, For similar fingerprint groups; k is the original value of the similar fingerprint group; Entropy weights; for The original values ​​of all similar points within the range; The merged sequence is obtained based on the weighted merging value, and then the merged sequence is input into the GRU network for prediction, outputting the final frequency modulation command prediction result.

2. The method for predicting energy storage frequency regulation commands based on time pattern entropy merging according to claim 1, characterized in that, The formula for calculating the local fluctuation intensity at each sampling point is as follows: Where i = 2, 3, ..., N-1 in, These are sampling points; when the boundary point i=0 or i=N, .

3. The method for predicting energy storage frequency regulation commands based on time pattern entropy merging according to claim 1, characterized in that, The expression for fingerprint similarity measurement is as follows: Where σ is the similarity bandwidth parameter σ=0.1·Std(Pt), Std(Pt) is the standard deviation of the original sequence Pt; j is the value of the sampling point within the set radius; The pattern entropy fingerprint sequence; It is a sequence of values ​​of the sampling points within the set radius.

4. The method for predicting energy storage frequency regulation commands based on time pattern entropy merging according to claim 1, characterized in that, The expression for the adaptive merging threshold is as follows: θ = α·Var(F); in, The pattern entropy fingerprint sequence; α is the mean of the pattern entropy fingerprint sequence; α = 0.5 is the scaling factor.

5. The method for predicting energy storage frequency regulation commands based on time pattern entropy merging according to claim 1, characterized in that, The expression for entropy weight is as follows: in, Let k be the local dynamic rate of change at point k. This is the mean of the pattern entropy fingerprint sequence.

6. A frequency regulation command prediction system for energy storage based on time-mode entropy merging, characterized in that, A method for predicting energy storage frequency regulation commands based on time-mode entropy merging as described in any one of claims 1-5 includes: The local fluctuation intensity calculation module is used to calculate the local dynamic change rate of the original frequency modulation sequence to obtain the local fluctuation intensity of each sampling point. The pattern entropy fingerprint sequence generation module is used to generate pattern entropy fingerprints based on the local fluctuation intensity of each sampling point, and to generate pattern entropy fingerprint sequences based on the pattern entropy fingerprints of all sampling points in the original frequency modulation sequence. The fingerprint similarity measurement calculation module is used to calculate the fingerprint similarity measurement for each sampling point within a set radius. An adaptive merging threshold determination module is used to determine the adaptive merging threshold based on the variance of the pattern entropy fingerprint sequence. The weighted merge value calculation module is used to compare the fingerprint similarity measure with the adaptive merging threshold. When the fingerprint similarity measure is greater than the adaptive merging threshold, similar fingerprint groups are merged to obtain the calculated weighted merge value. The frequency modulation command prediction module is used to obtain the merged sequence based on the weighted merging value, input the merged sequence into the GRU network for prediction, and output the final frequency modulation command prediction result.

7. A mobile terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the energy storage frequency regulation command prediction method based on time pattern entropy merging as described in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the energy storage frequency regulation command prediction method based on time pattern entropy merging as described in any one of claims 1-5.

Images