A method and system for predicting frequency modulation commands for energy storage based on chaotic phase injection

Abstract

This invention relates to the field of power system energy storage equipment technology, specifically to a method and system for predicting energy storage frequency regulation commands based on chaotic phase injection. The method includes selecting candidate periods to segment the original frequency regulation command into a set of subsequences; calculating the dynamic time warping distance of all extracted sequence pairs and obtaining the similarity integral of the candidate periods; screening the true periods through standardized significance testing; selecting the optimal period with the largest significance value; calculating the consistency index based on the permutation entropy of the optimal period subsequence set; reconstructing the original frequency regulation command based on chaotic phase injection; and inputting the reconstructed sequence into a pre-trained GRU network to output predicted values. This invention solves the nonlinear time sequence alignment problem through dynamic time warping distance, utilizes PCI to control the chaotic intensity and periodic perturbation, achieves decoupling of noise-resistant period detection and chaotic features, and ultimately improves prediction accuracy and eliminates response lag.

Description

Technical Field

[0001] This invention relates to the field of energy storage equipment technology for power systems, and in particular to a method and system for predicting energy storage frequency regulation commands based on chaotic phase injection. Background Technology

[0002] The core purpose of predicting frequency regulation commands for energy storage is to solve the three major contradictions of regulation capacity mismatch, equipment life loss, and limited economic benefits under the traditional passive response mode. By predicting the dynamic characteristics of grid frequency regulation commands in advance, it can realize the pre-control of the SOC (State of Charge) of the energy storage system, the adaptive optimization of charging and discharging strategies, and the precise tracking of regulation accuracy, thereby improving the reliability and economy of energy storage participating in power system frequency regulation.

[0003] The existing AI-based energy storage frequency regulation command prediction has the following problems: (1) Timing alignment deviation causes strategy oscillation: There is usually a dynamic time delay between the frequency regulation command and the energy storage response (typical value 3-5 seconds). Existing time series prediction algorithms such as TCN (Temporal Convolutional Network) and Seq2Seq (Sequence to Sequence) adopt a fixed sliding window mechanism, which cannot achieve dynamic phase matching of subsequences. Furthermore, simulations and timing misalignments can lead to frequent switching of charging and discharging strategies; (2) Insufficient decoupling of chaotic features reduces prediction accuracy: Frequency regulation commands usually contain chaotic characteristics of multiple sources such as grid inertia and new energy penetration rate. The maximum Lyapunov exponent is >0.15, but traditional AI models only extract explicit correlations through attention mechanisms and fail to quantify the injected chaotic phase components; (3) Lack of multi-timescale collaboration: Existing methods usually independently construct minute-level and second-level prediction models. The lack of cross-scale feature interaction mechanisms increases the probability of conflict between SOC planning and real-time command response, which can easily trigger capacity over-limit alarms. Summary of the Invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide a method and system for predicting energy storage frequency regulation commands based on chaotic phase injection, so as to solve the problems of low prediction accuracy and response lag caused by time alignment deviation, insufficient decoupling of chaotic features and lack of multi-timescale coordination in the prior art.

[0005] This invention discloses a method for predicting frequency modulation commands for energy storage based on chaotic phase injection, comprising:

[0006] A set of candidate periods is selected, and the collected original frequency modulation commands are segmented using each candidate period to obtain a set of subsequences that correspond one-to-one with the candidate periods.

[0007] The subsequences in the subsequence set are combined pairwise to obtain sequence pairs. The dynamic time regularization distance of each sequence pair is calculated, and the similarity integral of the candidate period is calculated based on the dynamic time regularization distance of all sequence pairs.

[0008] The significance of the candidate period is obtained by calculating the similarity integral of the candidate period, and the true period is selected from all the candidate periods based on the significance of the candidate period. The true period with the maximum significance value is determined as the optimal period.

[0009] The permutation entropy of the subsequence set corresponding to the optimal period is calculated, and the consistency index of the optimal period is calculated by combining the significance of the optimal period and the permutation entropy.

[0010] Based on the consistency index of the optimal period, the original frequency modulation command is reconstructed using the chaotic phase injection method to obtain the reconstructed sequence.

[0011] The reconstructed sequence is input into a pre-trained GRU network, and the output is the predicted value of the original frequency modulation command at a preset prediction time.

[0012] Optionally, the step of segmenting the collected original frequency modulation commands for each candidate period to obtain a set of subsequences corresponding one-to-one with the candidate periods includes:

[0013] The number of segments of the original frequency modulation command is determined based on the total number of samples of the original frequency modulation command and the candidate period. The functional expression for the number of segments of the original frequency modulation command is as follows:

[0014]

[0015] In the formula, K represents the number of subsequences after segmentation, N represents the total number of samples of the original frequency modulation command, and T represents the candidate period. Indicates rounding down;

[0016] Based on the determined number of segments, the original frequency modulation command is divided into a corresponding number of subsequences using the candidate period, and the multiple subsequences after segmentation are arranged in the segmentation order to form the subsequence set.

[0017] Optionally, the step of calculating the dynamic time-warped distance of each sequence pair and calculating the similarity integral of the candidate period based on the dynamic time-warped distances of all sequence pairs includes:

[0018] For each sequence, the dynamic time-warped distance is obtained by minimizing the cumulative path distance through dynamic programming;

[0019] The average similarity is calculated based on the dynamic time-normalized distance of all the sequence pairs, and the calculated average similarity is normalized to obtain the similarity integral of the candidate period within a preset range. The functional expression of the similarity integral is:

[0020]

[0021] In the formula, The similarity integral represents the candidate period. This represents the number of subsequences in the set of subsequences. Representing a subsequence The average of the sum of all values ​​in the range. Representing a subsequence The average of the sum of all values ​​in the range. Indicates the dynamic time-warped distance. represents the maximum value; L represents the normalization factor.

[0022] Optionally, the step of calculating the saliency of the candidate period based on the similarity integral of the candidate period includes:

[0023] The mean and standard deviation of the similarity integrals of all candidate periods are calculated, and the significance of the candidate periods is obtained by standardization based on the similarity integrals of the candidate periods. The functional expression of the significance of the candidate periods is as follows:

[0024]

[0025] In the formula, The significance value of the candidate period is represented. The similarity integral represents the candidate period. This represents the mean of the periodic similarity integrals for all candidate cycles. This represents the standard deviation of the integral of similarity across all candidate cycles. This represents a local minimum constant.

[0026] Optionally, the step of selecting the true period from all the candidate periods based on the significance of the candidate periods, and determining the true period with the maximum significance value as the optimal period, includes:

[0027] If the significance value of the candidate period is greater than 0, then the candidate period is determined to be the true period;

[0028] If the significance value of the candidate period is close to 0, then the candidate period is determined to be a non-true period.

[0029] From all the candidate periods that satisfy the significance value greater than 0, the actual period with the largest significance value is selected as the optimal period.

[0030] Optionally, the energy storage frequency regulation command prediction method further includes methods for calculating the consistency index of the optimal period and verifying the optimal period, including:

[0031] Combining the significance of the optimal period and the consistency index of the permutation entropy with respect to the optimal period, the functional expression of the optimal period consistency index is as follows:

[0032]

[0033] In the formula, The consistency index represents the optimal period. This represents the significance value of the optimal period. This represents the permutation entropy of the set of subsequences corresponding to the optimal period. Indicates the optimal period;

[0034] If the consistency index of the optimal period is close to 1, then the optimal period is determined to have a significant and simple pattern.

[0035] If the consistency index of the optimal period is close to 0, then the optimal period is determined to be insignificant or the pattern is too random.

[0036] Optionally, the reconstructing of the original frequency modulation command using a chaotic phase injection method based on the consistency index of the optimal period to obtain the reconstructed sequence includes:

[0037] Based on the consistency index of the optimal period, the original frequency modulation command is reconstructed using a chaotic phase injection method to obtain a reconstructed sequence. The functional expression for reconstructing the original frequency modulation command is as follows:

[0038]

[0039] In the formula, This represents the i-th element in the reconstructed sequence. This represents the i-th element in the original frequency modulation command. Represents the chaos intensity factor. Indicates the optimal period. This represents the consistency index of the optimal period. Represents the sine function. Represents pi (π). This represents the periodic disturbance coefficient. This represents standard Gaussian noise.

[0040] This invention also discloses a prediction system that employs the aforementioned energy storage frequency modulation command prediction method based on chaotic phase injection. The prediction system includes:

[0041] The frequency modulation command segmentation module is used to select a set of candidate periods, and segment the collected original frequency modulation commands using each candidate period to obtain a set of subsequences that correspond one-to-one with the candidate periods.

[0042] The sequence alignment module is used to combine the subsequences in the subsequence set into pairs to obtain sequence pairs, calculate the dynamic time regularization distance of each sequence pair, and calculate the similarity integral of the candidate period based on the dynamic time regularization distance of all the sequence pairs.

[0043] The optimal period determination module is used to calculate the significance of the candidate periods based on the similarity integral of the candidate periods, select the true period from all the candidate periods based on the significance of the candidate periods, and determine the true period with the maximum significance value as the optimal period.

[0044] The consistency check module is used to calculate the permutation entropy of the subsequence set corresponding to the optimal period, and to calculate the consistency index of the optimal period by combining the significance of the optimal period and the permutation entropy.

[0045] The sequence reconstruction module is used to reconstruct the original frequency modulation command based on the consistency index of the optimal period using a chaotic phase injection method to obtain a reconstructed sequence.

[0046] The frequency modulation command prediction module is used to input the reconstructed sequence into a pre-trained GRU network and output the predicted value of the original frequency modulation command at a preset prediction time.

[0047] The present invention also discloses a computer-readable storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the above-described method for predicting energy storage frequency modulation commands based on chaotic phase injection.

[0048] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the above-mentioned energy storage frequency modulation command prediction method based on chaotic phase injection.

[0049] Compared with the prior art, the beneficial effects of the energy storage frequency modulation command prediction method and system based on chaotic phase injection provided in this embodiment of the invention are as follows:

[0050] By accurately quantifying the morphological similarity of subsequences through dynamic time warping distance, and obtaining the optimal period consistency index based on the saliency screening of candidate periods and permutation entropy calculation, the chaotic phase injection method is driven to reconstruct the original frequency modulation command and generate a low-complexity feature sequence. Combined with the adaptive learning capability of the GRU network, the timing alignment deviation is effectively eliminated, chaotic features are decoupled, and multi-scale characteristics are integrated, significantly improving the prediction accuracy of frequency modulation commands, solving the problems of strategy oscillation and capacity over-limit alarm caused by traditional methods, and enhancing the regulation reliability and economy of energy storage systems. Attached Figure Description

[0051] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0052] Figure 1 A schematic diagram illustrating the steps of the energy storage frequency regulation command prediction method provided in an embodiment of the present invention. Detailed Implementation

[0053] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0054] This invention discloses a method for predicting frequency modulation commands for energy storage based on chaotic phase injection, such as... Figure 1 As shown, it includes:

[0055] S1. Select a set of candidate periods, and use each candidate period to segment the collected original frequency modulation commands to obtain a set of subsequences that correspond one-to-one with the candidate periods.

[0056] S2. Combine the subsequences in the subsequence set into pairs to obtain sequence pairs, calculate the dynamic time regularization distance of each sequence pair, and calculate the similarity integral of the candidate period based on the dynamic time regularization distance of all sequence pairs.

[0057] S3. Calculate the significance of the candidate periods based on the similarity integral of the candidate periods, select the true periods from all candidate periods based on the significance of the candidate periods, and determine the true period with the maximum significance value as the optimal period.

[0058] S4. Calculate the permutation entropy of the subsequence set corresponding to the optimal period, and combine the significance of the optimal period and the permutation entropy to calculate the consistency index of the optimal period.

[0059] S5. Based on the consistency index of the optimal period, the chaotic phase injection method is used to reconstruct the original frequency modulation command to obtain the reconstructed sequence.

[0060] S6. Input the reconstructed sequence into the pre-trained GRU network and output the predicted value of the original frequency modulation command at the preset prediction time.

[0061] Through the implementation of the above-described energy storage frequency regulation command prediction method, the nonlinear morphological similarity of all sequence pairs in the subsequence set formed by candidate period segmentation is accurately quantified using Dynamic Time Warping Distance (DTW). This overcomes the insensitivity of traditional period detection to time axis scaling / compression, significantly improving the period capture capability of non-stationary frequency regulation commands. The statistical significance of candidate periods is calculated based on similarity integrals. Standardization eliminates the risk of misjudgment caused by noise interference, selecting the true period and locking in the optimal period with the highest significance value. The period consistency index is calculated by combining the permutation entropy of the subsequence set corresponding to the optimal period. An entropy weighting mechanism ensures that the period pattern simultaneously satisfies statistical significance and low complexity, completely eliminating the interference of complex random patterns on period selection. This paper employs a chaotic phase injection method to adaptively reconstruct the original frequency regulation command, generating a reconstructed sequence that combines principal periodic characteristics with chaotic fluctuations. Finally, a pre-trained GRU (Gated Recurrent Unit) network is used to achieve multi-timescale coordinated prediction, effectively addressing the problems of strategy oscillation caused by time-series alignment deviations, prediction distortion of abrupt transitions due to insufficient decoupling of chaotic features, and capacity over-limit alarms caused by the lack of multi-timescale coordination in traditional methods. Based on the reconstructed sequence, the nonlinear complexity of the original frequency regulation command can be significantly reduced, enabling the GRU network to accurately decouple the chaotic characteristics of grid inertia and renewable energy penetration rate. This simultaneously improves the coordination between second-level real-time response and minute-level SOC planning (State of Charge estimation), eliminates the dynamic time lag between frequency regulation commands and energy storage responses, ensures that the energy storage system accurately tracks grid frequency regulation needs in high-proportion renewable energy scenarios, avoids battery life loss due to frequent switching of charging and discharging strategies, and improves the accuracy of regulation capacity matching and response qualification rate. Therefore, while maintaining grid frequency stability, this extends the lifespan of energy storage devices and increases frequency regulation compensation benefits.

[0062] Furthermore, the collected original frequency modulation commands are segmented for each candidate period to obtain a set of subsequences corresponding one-to-one with the candidate periods, including:

[0063] The number of segments in the original frequency modulation command is determined based on the total number of samples and the candidate period. The functional expression for the number of segments in the original frequency modulation command is as follows:

[0064]

[0065] In the formula, K represents the number of subsequences after segmentation, N represents the total number of samples of the original frequency modulation command, and T represents the candidate period. Indicates rounding down;

[0066] Based on the determined number of segments, the original frequency modulation command is divided into a corresponding number of subsequences using candidate periods, and the multiple subsequences after segmentation are arranged in the order of segmentation to form a subsequence set.

[0067] Through the implementation of the above-described energy storage frequency regulation command prediction method embodiment, the number of subsequences is accurately determined dynamically based on relevant functions during the segmentation stage. The floor operation completely avoids the problem of incomplete tail subsequences caused by non-integer periods, ensuring that the subsequence set generated for each candidate period T has a complete structure. This quantization segmentation mechanism ensures that each subsequence strictly contains T consecutive sampling points (except for the last segment, which automatically adapts to a length of N for subsequences shorter than T), eliminating the fixed-length deviation of traditional sliding windows from the source. The set of subsequences arranged in the segmentation order completely preserves the phase continuity of the frequency modulation command in the time dimension, providing an alignment benchmark for subsequent dynamic time warping distance calculation. Compared to the edge point truncation distortion caused by traditional equal-length segmentation, the method in this embodiment ensures that each subsequence independently carries a local timing pattern through dynamic segmentation and down-rounding control of the candidate period T. This enables subsequent period detection to identify the short-period frequency modulation characteristics caused by inertia differences in the high-penetration renewable energy grid, ultimately significantly reducing the action delay and strategy switching frequency of energy storage response.

[0068] For example, given the original frequency modulation command P=[X1,X2,X3……X N ], N is the number of sampling times; select a set of candidate period numbers The number of subsequences is between 7 and 15. For each candidate period T, the original frequency modulation command is divided into K subsequences, resulting in a set of subsequences. .

[0069] Furthermore, the dynamic time-warped distance of each sequence pair is calculated, and the similarity integral of the candidate period is calculated based on the dynamic time-warped distance of all sequence pairs, including:

[0070] For each sequence, the dynamic time-warped distance is obtained by minimizing the cumulative path distance through dynamic programming.

[0071] The average similarity is calculated based on the dynamic time-normalized distance of all sequence pairs, and the calculated average similarity is normalized to the similarity integral of the candidate periods within a preset range. The functional expression of the similarity integral is as follows:

[0072]

[0073] In the formula, The similarity integral represents the candidate period. This represents the number of subsequences in the set of subsequences. Representing a subsequence The average of the sum of all values ​​in the range. Representing a subsequence The average of the sum of all values ​​in the range. Indicates the dynamic time-warped distance. represents the maximum value; L represents the normalization factor.

[0074] By implementing the above-described energy storage frequency regulation command prediction method, the Dynamic Time Warped Distance (DTW) is used to replace traditional linear correlation calculations, solving the problem of period detection under nonlinear time deformation. Specifically, it involves dividing each candidate period T into subsequence sets containing sequence pairs. The DTW value is obtained by strictly minimizing the cumulative path distance according to the dynamic programming principle, effectively overcoming the incompatibility of fixed phase matching with time axis scaling / compression. Based on this, a similarity integral function expression is established, and the dynamic time warping distance of 2 / K(K-1) non-repeating sequence pairs is normalized through a double summation mechanism. The dynamic time warping distance at the numerator of the core operator accurately quantifies the sequence morphological similarity, while the denominator uses the average of the normalization factor L and the sum of all values ​​of the subsequences (i.e.,...). , The product of these factors forms an adaptive scaling benchmark, which eliminates interference from amplitude differences and enhances compatibility with subsequences of unequal length through a maximum value function. This enables the computational process to simultaneously perform multi-scale periodic analysis. For example, the above operations can be performed in parallel for 7–15 candidate periods T, with each T independently generating K subsequences. By traversing the similarity integral C(T) under different T values, the system covers potential periodic patterns from the second to the minute level. Thus, a dual synergy is formed through dynamic time warping distance calculation and multi-candidate period traversal. The former ensures the accuracy of morphological similarity measurement at each scale, while the latter avoids feature omissions caused by a single period preset. Together, they make the similarity integral C(T) of the candidate periods a robust indicator of periodic significance, thereby solving the problem of false periodic detection in nonlinear time deformation scenarios using traditional methods.

[0075] Furthermore, the saliency of candidate periods is obtained by calculating the similarity integral of the candidate periods, including:

[0076] Calculate the mean and standard deviation of the similarity integrals of all candidate periods, and standardize them by combining the similarity integrals of the candidate periods to obtain the significance of the candidate periods. The functional expression for the significance of the candidate periods is:

[0077]

[0078] In the formula, The significance value of the candidate period is represented. The similarity integral represents the candidate period. This represents the mean of the periodic similarity integrals for all candidate cycles. This represents the standard deviation of the integral of similarity across all candidate cycles. This represents a local minimum constant.

[0079] Furthermore, the true period is selected from all candidate periods based on their significance, and the true period with the highest significance value is determined as the optimal period, including:

[0080] If the significance value of the candidate period is greater than 0, then the candidate period is judged to be the true period;

[0081] If the significance value of the candidate period is close to 0, then the candidate period is judged to be a non-true period.

[0082] From all candidate periods that satisfy the significance value greater than 0, select the actual period with the largest significance value as the optimal period.

[0083] Through the implementation of the above-described energy storage frequency regulation command prediction method embodiment, the mean of the similarity integral C(T) of all candidate cycles is used. The standard deviation is the baseline level for random noise. The overall volatility intensity is quantified by converting the single-period C(T) into a significance value PSS(T) using a standardization operation. This overcomes the limitations of traditional absolute thresholds, i.e., the denominator... + This forms an adaptive noise filtering layer, enabling molecules to... - Only periodic signals that significantly exceed the population mean are retained. Among these, the numerator negative value (C(T) < Automatically identified as invalid noise, eliminating reliance on manually set fixed thresholds. A statistical significance criterion is established based on the screening condition PSS(T) > 0: only when PSS(T) is significantly greater than 0 is the similarity of the period significantly higher than random fluctuations, potentially indicating a true period; when PSS(T) is close to 0, the periodicity is considered comparable to noise levels, not an approximate period of the sequence. This excludes values ​​below [a certain threshold]. This method mitigates random fluctuations in frequency modulation commands and locks in the optimal period based on the maximum value of PSS(T), forming a dual statistical verification closed loop. This significantly improves robustness to short-term fluctuation noise in frequency modulation commands: for example, when a candidate period experiences an abnormally high similarity integral due to a sudden load change, Synchronous expansion keeps PSS(T) stable; conversely, in low-noise scenarios... Contraction enhances the period resolution sensitivity to avoid the dual defects of the fixed threshold scheme: missing effective periods in high-noise environments and misrepresenting pseudo periods in low-noise environments.

[0084] For example, for each candidate period The corresponding similarity integral is calculated. Calculate all candidate cycles The mean and standard deviation are calculated, and the results are standardized for each candidate period to obtain the significance for each candidate period. The optimal period is the true period with the largest significance value. The corresponding candidate period is the optimal period.

[0085] Furthermore, the energy storage frequency regulation command prediction method also includes methods for calculating the consistency index of the optimal period and verifying the optimal period, including:

[0086] Combining the significance of the optimal period and the consistency index of the permutation entropy with respect to the optimal period, the functional expression of the optimal period consistency index is as follows:

[0087]

[0088] In the formula, This represents the consistency index of the optimal period. This represents the significance value of the optimal period. This represents the permutation entropy of the set of subsequences corresponding to the optimal period. Indicates the optimal period;

[0089] If the consistency index of the optimal period is close to 1, then the optimal period is determined to be significant and the pattern is simple.

[0090] If the consistency index of the optimal period is close to 0, then the optimal period is determined to be insignificant or the pattern is too random.

[0091] Through the implementation of the above-described energy storage frequency regulation command prediction method embodiment, the significance value of the optimal period is achieved. Using the base value, through the reciprocal operator 1 / (1+ ( The permutation entropy of the set of subsequences corresponding to the optimal period. ( This is transformed into an inverse weighting coefficient, forming a dual-constraint fusion logic: when the permutation entropy approaches 0 (the subsequence pattern is highly ordered), the weighting coefficient approaches 1, making PCI ≈ To preserve the original periodic significance, the weight coefficients geometrically decay when the permutation entropy increases significantly (subsequence complexity increases), forcing the PCI to converge to 0. This method ensures that the periodic consistency index simultaneously satisfies... The statistical significance of the representation and ( The low complexity of the metric allows for precise implementation of the defined decision criterion: if PCI→1, then the following conditions must be met simultaneously. Approaching the theoretical peak and ( ) ≈0 (the optimal cycle exists significantly and the pattern is simple); if PCI→0, it reflects The periodicity is suppressed by high entropy values ​​(the optimal period is not significant or the pattern is too random). Therefore, based on the entropy weighting mechanism, the periodic pattern is ensured to simultaneously satisfy significance and low complexity, so as to quantify the overall periodicity.

[0092] Furthermore, based on the consistency index of the optimal period, a chaotic phase injection method is used to reconstruct the original frequency modulation command to obtain a reconstructed sequence, including:

[0093] Based on the consistency index of the optimal period, a chaotic phase injection method is used to reconstruct the original frequency modulation command to obtain the reconstructed sequence. The functional expression for reconstructing the original frequency modulation command is as follows:

[0094]

[0095] In the formula, This represents the i-th element in the reconstructed sequence. This represents the i-th element in the original frequency modulation command. Represents the chaos intensity factor. Indicates the optimal period. This represents the consistency index of the optimal period. Represents the sine function. Represents pi (π). This represents the periodic disturbance coefficient. This represents standard Gaussian noise.

[0096] Through the implementation of the above-described energy storage frequency regulation command prediction method, the sine function sin() is strictly based on the optimal period. A periodic modulation signal is generated for the reference frequency. The mathematical constant pi (π) is coupled with the index i to form a phase transition operator, accurately preserving the main periodic characteristics of the power grid frequency regulation command. This is based on three collaborative mechanisms: the optimal periodic consistency index PCI is directly coupled with the chaos intensity factor. A product constraint is constructed, causing the chaos intensity to dynamically decay with the PCI value. Specifically, when PCI→1 (strong significant periodicity), chaotic interference is suppressed, and when PCI→0 (weak periodicity), phase perturbation is enhanced, forming an adaptive anti-overfitting barrier; standard Gaussian noise. Through the periodic perturbation coefficient Linear coupling to the denominator term generates dynamic jitter of ±30% period length to overcome the sensitivity of a fixed-period preset to grid inertia fluctuations; product term A nonlinear superposition relationship is established between the original frequency modulation command and the chaotic phase, maintaining the amplitude distribution characteristics while injecting spectral perturbations. Under this reconstruction method, the basic sequence... It retains static characteristics, decouples the dynamic chaotic properties of sinusoidal modulation signals, and includes standard Gaussian noise terms. The three methods of generating frequency-domain anti-aliasing noise work together to generate a low-complexity reconstructed sequence. This significantly improves the predictive adaptability of GRU networks to non-steady-state frequency modulation commands, and completely eliminates the capacity over-limit alarms caused by policy oscillations and lack of multi-scale coordination due to timing alignment deviations in traditional methods.

[0097] For example, the original frequency modulation command P=[X1,X2,X3……X N Reconstructed as =[X1 、 X2 、 X3 、 ...X N 、 ],Bundle =[X1 、 X2 、 X3 、 ...X N 、 The input is fed into a pre-trained GRU network for prediction, resulting in the predicted value [X]. N+1 ,X N +2,X N+3 ...X N+S This predicted value is the predicted value of the original frequency modulation command.

[0098] To further verify the advantages of this invention, the method of this embodiment and the direct prediction method using a GRU network were respectively used to predict frequency modulation sequences. Four evaluation metrics—MAE (Mean Absolute Error), SSE (Sum of Squares Error), RMSE (Root Mean Square Error), and MAPE (Mean Absolute Percentage Error)—were used to evaluate the prediction results. The selection of these evaluation metrics can be found in the evaluation metric table.

[0099] Table 1 Evaluation Indicators

[0100]

[0101] As shown in Table 1, among the four evaluation metrics, N represents the sample size. and These represent the actual and predicted values ​​at time n, respectively. The results of evaluating the above predictions using four evaluation metrics are shown in the frequency modulation sequence prediction table:

[0102] Table 2 Frequency Modulation Sequence Prediction Table

[0103]

[0104] As shown in Table 2, frequency modulation sequence 1 comes from frequency modulation data of a power plant in Hulunbuir from 0:00 to 20:00 on December 1, 2024; frequency modulation sequence 2 comes from frequency modulation data of a power plant in Hulunbuir from 2:00 to 22:00 on December 2, 2024. The evaluation results in Table 2 show that all four evaluation indicators have decreased, indicating that the decomposition method proposed in this invention can reduce the nonlinearity of the original sequence compared to the traditional GRU prediction method, further improving the prediction accuracy and effectively helping power plants improve their frequency modulation response capability, thereby further increasing the power plant's revenue.

[0105] This invention also discloses a prediction system that employs the aforementioned energy storage frequency modulation command prediction method based on chaotic phase injection. The prediction system includes:

[0106] The frequency modulation command segmentation module is used to select a set of candidate periods, and to segment the collected original frequency modulation commands using each candidate period to obtain a set of subsequences that correspond one-to-one with the candidate periods.

[0107] The sequence alignment module is used to combine subsequences in the subsequence set into pairs to obtain sequence pairs, calculate the dynamic time regularization distance of each sequence pair, and calculate the similarity integral of the candidate period based on the dynamic time regularization distance of all sequence pairs.

[0108] The optimal period determination module is used to calculate the significance of candidate periods based on the similarity integral of candidate periods, select the true period from all candidate periods based on the significance of candidate periods, and determine the true period with the maximum significance value as the optimal period.

[0109] The consistency test module is used to calculate the permutation entropy of the set of subsequences corresponding to the optimal period, and to calculate the consistency index of the optimal period by combining the significance of the optimal period and the permutation entropy.

[0110] The sequence reconstruction module is used to reconstruct the original frequency modulation command based on the consistency index of the optimal period and the chaotic phase injection method to obtain the reconstructed sequence.

[0111] The frequency modulation command prediction module is used to input the reconstructed sequence into a pre-trained GRU network and output the predicted value of the original frequency modulation command at a preset prediction time.

[0112] The present invention also discloses a computer-readable storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the above-described method for predicting energy storage frequency modulation commands based on chaotic phase injection.

[0113] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the above-mentioned energy storage frequency modulation command prediction method based on chaotic phase injection.

[0114] This invention is described based on flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to specific embodiments. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the flowcharts and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0115] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0116] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] It should be understood that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Those skilled in the art can modify the technical solutions described in the above embodiments, or make equivalent substitutions for some of the technical features; and all such modifications and substitutions should fall within the protection scope of the present invention.

Claims

1. A chaos phase injection-based energy storage frequency modulation instruction prediction method, characterized in that, The energy storage frequency modulation instruction prediction method comprises: selecting a group of candidate periods, and using each candidate period to segment the collected original frequency modulation instruction to obtain a subsequence set corresponding to the candidate period one by one; combining the subsequences in the subsequence set two by two to obtain sequence pairs, calculating the dynamic time warping distance of each sequence pair, and calculating the similarity integral of the candidate period according to the dynamic time warping distance of all sequence pairs; obtaining the significance of the candidate period according to the similarity integral of the candidate period, screening the real period from all the candidate periods according to the significance of the candidate period, and determining the real period with the maximum significance value as the optimal period; calculating the permutation entropy of the subsequence set corresponding to the optimal period, and calculating the consistency index of the optimal period in combination with the significance and permutation entropy of the optimal period; based on the consistency index of the optimal period, using the chaotic phase injection method to reconstruct the original frequency modulation instruction to obtain a reconstructed sequence; inputting the reconstructed sequence into a pre-trained GRU network to output the predicted value of the original frequency modulation instruction in the preset prediction time length.

2. The chaos phase injection based energy storage frequency modulation instruction prediction method according to claim 1, characterized in that, The method comprises the following steps: determining the segmentation number of the original frequency modulation instruction according to the total sampling number of the original frequency modulation instruction and the candidate period, and the function expression of the original frequency modulation instruction segmentation number is: In the formula, K represents the number of sub-sequences after division, N represents the total sampling number of original frequency modulation instructions, T represents the candidate period, represents the floor function. segmenting the original frequency modulation instruction into a corresponding number of subsequences using the candidate period according to the determined segmentation number, and arranging the segmented subsequences in sequence to form the subsequence set.

3. The chaos phase injection based energy storage frequency modulation instruction prediction method according to claim 1, characterized in that, The method comprises the following steps: minimizing the cumulative path distance of each sequence through dynamic programming to obtain the dynamic time warping distance; calculating the average similarity according to the dynamic time warping distance of all sequence pairs, and normalizing the calculated average similarity to obtain the similarity integral of the candidate period within a preset range, and the function expression of the similarity integral is: wherein denotes the similarity integral of a candidate period, denotes the number of subsequences in the set of subsequences, denotes the subsequence the average of the sum of all values in the subsequence denotes the subsequence the average of the sum of all values in the subsequence denotes the dynamic time warping distance, denotes the maximum value; L denotes the normalization factor.

4. The chaos phase injection based energy storage frequency modulation instruction prediction method according to claim 1, characterized in that, The method comprises the following steps: calculating the mean and standard deviation of the similarity integral of all candidate periods, and normalizing the similarity integral of the candidate period to obtain the significance of the candidate period, and the function expression of the significance of the candidate period is: wherein represents a significance value of a candidate period, represents a similarity integral of a candidate period, represents a mean value of all candidate period similarity integrals, represents a standard deviation of all candidate period similarity integrals, represents a minimum value constant.

5. The chaos phase injection based energy storage frequency modulation instruction prediction method according to claim 4, characterized in that, The method comprises the following steps: if the significance value of the candidate period is greater than 0, the candidate period is determined to be a real period; if the significance value of the candidate period is close to 0, the candidate period is determined to be a non-real period; selecting the real period with the maximum significance value from all the candidate periods with a significance value greater than 0 as the optimal period.

6. The chaos phase injection based energy storage frequency modulation command prediction method of claim 1, wherein, The energy storage frequency modulation instruction prediction method further comprises a method for calculating a consistency index of the optimal period and verifying the optimal period, comprising: The consistency index of the optimal period is combined with the significance and permutation entropy of the optimal period, and a function expression of the optimal period consistency index is: In the formula, represents the consistency index of the optimal period, represents the significance value of the optimal period, represents the permutation entropy of the optimal period corresponding subsequence set, represents the optimal period; If the consistency index of the optimal period is close to 1, it is determined that the optimal period is significant and the mode is simple; If the consistency index of the optimal period is close to 0, it is determined that the optimal period is not significant or the mode is too random.

7. The chaos phase injection based energy storage frequency modulation instruction prediction method according to claim 1, characterized in that, Based on the consistency index of the optimal period, a reconstructed sequence is obtained by reconstructing the original frequency modulation instruction using a chaotic phase injection method, comprising: Based on the consistency index of the optimal period, a reconstructed sequence is obtained by reconstructing the original frequency modulation instruction using a chaotic phase injection method, and a function expression for reconstructing the original frequency modulation instruction is: wherein, denotes the i-th element in the reconstructed sequence, denotes the i-th element in the original frequency modulation command, denotes the chaos intensity factor, denotes the optimal period, denotes the consistency index of the optimal period, denotes the sine function, denotes the pi, denotes the periodic perturbation coefficient, denotes the standard Gaussian noise.

8. A prediction system employing the chaos phase injection based energy storage frequency modulation instruction prediction method of any one of claims 1-7, characterized in that, The prediction system comprises: A frequency modulation instruction segmentation module is configured to select a group of candidate periods, segment the collected original frequency modulation instruction using each candidate period to obtain a subsequence set corresponding to each candidate period; A sequence alignment module is configured to combine the subsequences in the subsequence set to obtain sequence pairs, calculate the dynamic time warping distance of each sequence pair, and calculate the similarity integral of the candidate periods according to the dynamic time warping distance of all sequence pairs; An optimal period determination module is configured to calculate the significance of the candidate periods according to the similarity integral of the candidate periods, filter out the real periods from all candidate periods according to the significance of the candidate periods, and determine the real period with the maximum significance value as the optimal period; A consistency verification module is configured to calculate the permutation entropy of the subsequence set corresponding to the optimal period, and calculate the consistency index of the optimal period in combination with the significance and permutation entropy of the optimal period; A sequence reconstruction module is configured to reconstruct the original frequency modulation instruction using a chaotic phase injection method based on the consistency index of the optimal period to obtain a reconstructed sequence; A frequency modulation instruction prediction module is configured to input the reconstructed sequence into a pre-trained GRU network to output the predicted value of the original frequency modulation instruction in a preset prediction time.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the energy storage frequency modulation instruction prediction method based on chaotic phase injection according to any one of claims 1-7.

10. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the energy storage frequency modulation instruction prediction method based on chaotic phase injection according to any one of claims 1-7.

Images