A molten salt energy storage frequency modulation instruction prediction method and system

Abstract

The application discloses a noise-removed molten salt energy storage frequency modulation instruction prediction method, acquires an original frequency modulation instruction signal sequence, finds out a plurality of suspected noise sequence groups based on the original frequency modulation instruction signal sequence, acquires a final noise sequence group by using the suspected noise sequence groups, carries out neural network prediction by using the difference sequence of the original frequency modulation instruction signal sequence and the final noise sequence group, and applies the prediction result to power regulation of a molten salt energy storage system, so that the noise of the original frequency modulation sequence is found out, the noise is removed, and the prediction accuracy is improved.

Description

Technical Field

[0001] This disclosure relates to the field of power grid frequency regulation technology, and in particular to a method and system for predicting frequency regulation commands using molten salt energy storage with noise removal. Background Technology

[0002] Molten salt energy storage frequency regulation command prediction refers to establishing a predictive model by analyzing historical frequency regulation commands and related data to predict future frequency regulation command demands, thereby optimizing the control strategy of the hybrid energy storage system and improving its frequency regulation performance and efficiency. Its characteristics include the following:

[0003] I. Improve system performance

[0004] • Fast response: It can quickly adjust the output power in a short time, respond rapidly to changes in grid frequency, effectively balance the supply and demand differences in the power system, and maintain grid frequency stability.

[0005] • Improved regulation accuracy: It can track frequency modulation commands more accurately, reduce regulation errors, and improve the power supply quality and stability of the power grid.

[0006] • Enhance system stability: By rationally allocating the power and energy of different energy storage devices, the overall risk of the system can be reduced, the reliability and stability of the system can be improved, and the risk of system failure due to the failure or performance degradation of a single energy storage device can be reduced.

[0007] II. Extending Equipment Lifespan

[0008] • Optimize charging and discharging strategies: Based on the prediction results, reasonable charging and discharging strategies can be formulated to avoid overcharging and discharging or frequent charging and discharging of energy storage devices, thereby extending their service life.

[0009] • Balanced equipment usage: Rationally allocate the usage frequency and load of different energy storage devices to ensure that the aging of each device is relatively balanced, thereby reducing equipment maintenance costs and replacement frequency.

[0010] III. Cost Reduction

[0011] • Improved economic efficiency: By improving frequency regulation performance and efficiency, hybrid energy storage systems can obtain more frequency regulation benefits while reducing power outage losses and equipment damage costs caused by grid frequency instability.

[0012] • Optimize investment costs: Based on the forecast results, the capacity and quantity of different energy storage devices can be rationally allocated to avoid over-investment and waste of resources, thereby improving the return on investment.

[0013] Traditional artificial intelligence methods (such as those based on statistical models, shallow machine learning, or classical signal processing techniques) often fail to predict frequency-modulated (FM) sequences due to the presence of noise in the FM sequence itself. Noise in FM sequences is characterized by random fluctuations or interference unrelated to the true signal, which can be caused by various factors. Here's a detailed explanation: Noise is almost always present in actual measurement data (such as FM sequences). Even with high-precision sensors, environmental interference, instrument errors, or data transmission problems can introduce noise. Characteristics: Noise typically manifests as high-frequency random fluctuations superimposed on the true signal (such as the periodic variations or trends of the frequency modulation). Noise may originate from: Measurement equipment errors: Sensor electronic noise (such as thermal noise, quantization errors). Calibration deviations or insufficient sensitivity. Environmental interference: Electromagnetic interference (such as nearby power equipment). Temperature and humidity changes cause fluctuations in sensor performance. Noise disrupts traditional prediction models, such as linear superposition and periodic decomposability, while amplifying the system's sensitivity to initial conditions and dependence on external disturbances, leading to inaccurate prediction results. Summary of the Invention

[0014] To address the shortcomings of existing technologies, this invention proposes a noise-removing prediction method. This method identifies and removes noise from the original frequency modulation sequence, thereby improving prediction accuracy.

[0015] A noise-removing method for predicting frequency modulation commands in molten salt energy storage includes:

[0016] S101. Obtain the original frequency modulation command signal sequence;

[0017] S102. Based on the original frequency modulation command signal sequence, find several suspected noise sequence groups;

[0018] S103. Obtain the final noise sequence group using several suspected noise sequence groups;

[0019] S104. The difference between the original frequency modulation command signal sequence and the final noise sequence is used to perform neural network prediction, and the prediction results are applied to the power regulation of the molten salt energy storage system.

[0020] A noise-removing molten salt energy storage frequency modulation command prediction system, comprising:

[0021] The instruction acquisition module acquires the original frequency modulation instruction signal sequence;

[0022] The search module identifies several suspected noise sequence groups based on the original frequency modulation command signal sequence.

[0023] The module acquires the final noise sequence group by utilizing several suspected noise sequence groups.

[0024] The power regulation module uses a neural network to predict the difference between the original frequency modulation command signal sequence and the final noise sequence, and applies the prediction results to the power regulation of the molten salt energy storage system.

[0025] The beneficial effects of this invention are: it provides a method and system for predicting frequency regulation commands for molten salt energy storage by removing noise, which can respond promptly when the power plant grid frequency fluctuates. By identifying and removing the noise in the original frequency regulation sequence, the accuracy of prediction is improved, and the magnitude of the frequency regulation command can be predicted in advance for energy storage regulation, thereby improving response accuracy and frequency regulation benefits. Attached Figure Description

[0026] Figure 1 This is a diagram illustrating the method steps. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0028] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0029] It should be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0030] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0031] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0032] This disclosure provides a noise-removing method for predicting frequency modulation commands in molten salt energy storage, such as... Figure 1 As shown, it includes:

[0033] S101. Obtain the original frequency modulation command signal sequence;

[0034] S102. Based on the original frequency modulation command signal sequence, find several suspected noise sequence groups;

[0035] S103. Obtain the final noise sequence group using several suspected noise sequence groups;

[0036] S104. The difference between the original frequency modulation command signal sequence and the final noise sequence is used to perform neural network prediction, and the prediction results are applied to the power regulation of the molten salt energy storage system.

[0037] Step S101, obtaining the original frequency modulation command signal sequence, includes:

[0038] The frequency modulation command signal sequence is P t = [X1,X2,X3,…,X] i ,…,X N The original frequency modulation command signal sequence P t The value is a function of time t, the subscript indicates the number of samples, and the sampling interval is 1 second.

[0039] The above S102 involves identifying several suspected noise sequence groups based on the original frequency modulation command signal sequence, including:

[0040] Let the frequency modulation command be P t = [X1,X2,X3,…,X] i ,…,X N ], where N>1000.

[0041] Noise typically manifests as high-frequency random fluctuations, superimposed on the real FM signal. Noise is usually a high-frequency sequence.

[0042] Step 1: Let Z be a randomly generated set of suspected noise sequences. t1 =[Z 11 Z 12 Z 13 ,…,Z 1i ,…,Z 1N This sequence satisfies the following characteristics:

[0043] 1) That is, Z 1i The random range of values. 2)

[0045] 3)

[0047]

[0048] make

[0049] In the above example, median() represents the median, and tanh() represents the activation function.

[0050] Step 2: Generate 100 sets of suspected noise sequences Z based on the three characteristics mentioned above. t1 Z t2 Z t3 ,…,Z t100

[0051] S103. Obtain the final noise sequence group using several suspected noise sequence groups; including:

[0052] Step 1: Use the original frequency modulation command signal sequence P t Subtract Z respectively t1 Z t2 Z t3 ,…,Z t100 Get P t1 ,P t2 ,…,P t100 .

[0053] For example, P t2 =[X1-Z 21 X2-Z 22 X3-Z 23 ,…,X i -Z 2i ,…,X N -Z 2N ].

[0054] Step 2, P t1 ,P t2 ,…,P t100 The first 80% of each value is fed into a GRU to predict the last 20% of the values. The corresponding prediction errors (selected as MAPE (%)) are then obtained: W1, W2, ..., W 100

[0055] Step 3: Select W1, W2, ..., W 100 W with the largest mean error max The corresponding P tmax The corresponding Z in tmax It is the first suspected noise sequence.

[0056] Select W1, W2, ..., W 100 W, the second largest in terms of mean square error max1 The corresponding P tmax1 The corresponding Z in tmax1 It is the second suspected noise sequence.

[0057] Step 4

[0058] Let the first suspected noise sequence Z tmax =[Z max1 Z max2 Z max3 ,.Zmaxi ..,Z maxN ]

[0059] Let the second suspected noise sequence Z tmax1 =[Z max1 1 Z max1 2 Z max1 3 ,.Z max1 i ..,Z max1 N ]

[0060] The final noise sequence is

[0061] Z final =Z tmax

[0062]

[0063] This represents the superposition of tensor products, and γ represents the entropy at the iteration position.

[0064]

[0065] sigmoid represents the activation function.

[0066] The above-mentioned S104 involves performing neural network prediction on the difference between the original frequency modulation command signal sequence and the final noise sequence, and applying the prediction result to the power regulation of the molten salt energy storage system; including:

[0067] The original frequency modulation sequence P t Subtract Z final The difference sequence [X1-Z] was then obtained. final1 X2-Z final2 X3-Z final3 ,.X i -Z finali ..,X N -Z finalN The difference sequence is fed into the GRU network for prediction.

[0068] Power regulation in molten salt energy storage systems is achieved using forecast results. Real-time monitoring of grid frequency deviations allows for dynamic allocation of power commands, rapidly smoothing out frequency fluctuations.

[0069] With the above Figure 1 Correspondingly, embodiments of the present invention also provide a noise-removing molten salt energy storage frequency modulation command prediction system, comprising:

[0070] The instruction acquisition module acquires the original frequency modulation instruction signal sequence;

[0071] The search module identifies several suspected noise sequence groups based on the original frequency modulation command signal sequence.

[0072] The module acquires the final noise sequence group by utilizing several suspected noise sequence groups.

[0073] The power regulation module uses a neural network to predict the difference between the original frequency modulation command signal sequence and the final noise sequence, and applies the prediction results to the power regulation of the molten salt energy storage system.

[0074] A power regulation module is used to regulate the power of different energy storage units in a hybrid energy storage system based on a predicted frequency modulation command sequence. Traditional prediction methods often fail to produce accurate results when directly fed into neural networks due to the high-frequency noise in the original frequency modulation command signal sequence. To address this drawback, this invention proposes a novel method. First, the noise in the original frequency modulation sequence is identified and removed. Then, a replacement sequence is used to perform the prediction, yielding the final prediction result. This proposed method significantly reduces noise and further improves prediction accuracy.

[0075] To further verify the advantages of the present invention, the present invention uses the method of the present invention and the GRU prediction method to predict the frequency modulation sequence, and the results are shown in Table 1 and Table 2 below.

[0076] The frequency modulation commands for the experiment were acquired from a power plant in Inner Mongolia. Frequency modulation sequence 1 was collected from 4:00 AM to 6:00 PM on a certain day in November 2024, with one data point collected every second. Frequency modulation sequence 2 was collected from 12:00 AM to 10:00 PM on a certain day in February 2025, with one data point collected every second.

[0077] Table 1:

[0078]

[0079]

[0080] Table 2: Four Evaluation Indicators

[0081]

[0082] N represents the sample size, y n and These represent the actual value and the predicted value at time n, respectively.

[0083] The experimental results show that all four evaluation indicators have decreased, indicating that the proposed model can significantly improve prediction accuracy.

[0084] This application provides a computer device, including a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps provided in any embodiment of this application.

[0085] The computer device provided in this application includes a processor, a memory, and a bus. The memory, also known as internal memory, stores execution instructions and includes main memory and external memory. The main memory temporarily stores data processed by the processor, as well as data exchanged with external storage devices such as hard disks. The processor exchanges data with external storage devices through main memory. When the electronic device is running, the processor and memory communicate via the bus, enabling the processor to execute the following instructions:

[0086] Obtain the original frequency modulation command signal sequence;

[0087] Several suspected noise sequence groups were identified based on the original frequency modulation command signal sequence;

[0088] The final noise sequence group is obtained by using several suspected noise sequence groups;

[0089] The difference between the original frequency modulation command signal sequence and the final noise sequence is used to make neural network predictions, and the prediction results are applied to the power regulation of the molten salt energy storage system.

[0090] This application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps provided in any embodiment of this application. The storage medium can be either volatile or non-volatile computer-readable storage.

[0091] Through the above description of the embodiments, those skilled in the art can clearly understand that the embodiments of this disclosure can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a read-only optical disc, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this disclosure.

[0092] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the drawings are not necessarily essential for implementing this disclosure.

[0093] Those skilled in the art will understand that the modules in the apparatus of the embodiments can be distributed in the apparatus of the embodiments as described in the embodiments, or they can be located in one or more devices different from this embodiment with corresponding changes. The modules of the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0094] The sequence numbers of the embodiments disclosed above are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0095] Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the claims of this disclosure and their equivalents, this disclosure is also intended to include such modifications and variations.

[0096] Finally, it should be noted that the above description is merely an explanation of the present invention and is not intended to limit the invention. Although the present invention has been described in detail, those skilled in the art can still modify the technical solutions described above or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting frequency modulation commands in molten salt energy storage with noise removal, characterized in that, include: S101. Obtain the original frequency modulation command signal sequence; S102. Based on the original frequency modulation command signal sequence, find several suspected noise sequence groups; S103. Obtain the final noise sequence group using several suspected noise sequence groups; S104. The difference between the original frequency modulation command signal sequence and the final noise sequence is used to make neural network predictions, and the prediction results are applied to the power regulation of the molten salt energy storage system. Step S102 includes: Original frequency modulation command signal sequence P t =[X1,X2,X3,…,X i ,…,X N ], where N > 1000; Step 1: Randomly generate a suspected noise sequence Z that meets certain conditions based on the original frequency modulation command signal sequence. t1 =[Z 11 Z 12 Z 13 ,…,Z 1i ,…,Z 1N ]; The conditions that a suspected noisy sequence must meet are: 1) Z 1i A random range of values; 2） ； 3） ； Step 2: Continue generating several suspected noise sequences in the same manner as in Step 1; Generate 100 sets of suspected noise sequences Z according to the conditions in step one. t1 Z t2 Z t3 ,…,Z t100 ; S103. Obtain the final noise sequence group using several suspected noise sequence groups; including: Step 1: Use the original frequency modulation command signal sequence P t Subtract Z respectively t1 Z t2 Z t3 ,…,Z t100 Get P t1 ,P t2 ,…,P t100 ; Step 2, P t1 ,P t2 ,…,P t100 The first 80% of their lengths are fed into a GRU to predict the last 20% of the values, yielding the corresponding prediction errors W1, W2, ..., W. 100 ; Step 3: Select W1, W2, ..., W 100 W with the largest mean error max The corresponding P tmax The corresponding Z in tmax This is the first suspected noise sequence; Select W1, W2, ..., W 100 W, the second largest in terms of mean square error max1 The corresponding P tmax1 The corresponding Z in tmax1 This is the second suspected noise sequence; Step 4 Let the first suspected noise sequence Z tmax =[Z max1 Z max2 Z max3 ,.Z maxi ..,Z maxN ] , Let the second suspected noise sequence Z tmax1 =[Z max1 1 Z max1 2 Z max1 3 ,.Z max1 i ..,Z max1 N ] , The final noise sequence is WITH final =Z tmax , ， This represents the superposition of tensor products. Represents the iterative position entropy. 。 2. The noise-removing molten salt energy storage frequency modulation command prediction method according to claim 1, characterized in that, In step S101, The frequency modulation command signal sequence is P t =[X1,X2,X3,…,X i ,…,X N The original frequency modulation command signal sequence P t The value is a function of time t, the subscript indicates the number of samples, and the sampling interval is 1 second.

3. The noise-removing molten salt energy storage frequency modulation command prediction method according to claim 1, characterized in that, The above-mentioned S104 uses a neural network to predict the difference between the original frequency modulation command signal sequence and the final noise sequence, and applies the prediction result to the power regulation of the molten salt energy storage system. include: The original frequency modulation sequence P t Subtract Z final The difference sequence [X1-Z] was then obtained. final1 X2-Z final2 X3-Z final3 ,.X i -Z finali ..,X N -Z finalN The difference sequence is fed into the GRU network for prediction.

4. A noise-removing molten salt energy storage frequency modulation command prediction system employing the prediction method of claim 1, characterized in that, include: The instruction acquisition module acquires the original frequency modulation instruction signal sequence; The search module identifies several suspected noise sequence groups based on the original frequency modulation command signal sequence. The module acquires the final noise sequence group by utilizing several suspected noise sequence groups. The power regulation module uses a neural network to predict the difference between the original frequency modulation command signal sequence and the final noise sequence, and applies the prediction results to the power regulation of the molten salt energy storage system.

Images