Joint regulation and control method for flexibility resources in Saggob area based on wavelet decomposition

Abstract

The invention relates to the technical field of power system regulation and control, and discloses a Sagomean area flexibility resource joint regulation and control method based on wavelet decomposition, which comprises the following steps: acquiring power system net load time sequence data, and performing discrete wavelet transform and decomposition on the power system net load time sequence data to obtain an approximation coefficient and a detail coefficient under each scale; reconstructing based on the approximation coefficient and the detail coefficient to obtain a high-frequency component, an intermediate-frequency component and a low-frequency component; constructing a power confidence interval by each frequency band component; constructing an energy storage system model, a demand response resource model and a thermal power generating unit model; the method comprises the following steps of: constructing a joint optimization scheduling model considering input uncertainty, performing deterministic conversion on chance constraints, and solving the joint optimization scheduling model by adopting a mixed integer linear programming solver, and has the advantages of being capable of effectively adapting to uncertainty of new energy output and load fluctuation, and being high in practicability. And the flexibility, the stability and the economical efficiency of a power system in the Sagomean region are improved.

Description

A Flexible Resource Joint Regulation Method Based on Wavelet Decomposition in the Gobi Desert Region Technical Field

[0001] This invention relates to the field of power system control technology, and in particular to a method for joint control of flexible resources in desert and Gobi areas based on wavelet decomposition. Background Technology

[0002] my country has planned and constructed large-scale wind and solar power bases in desert, Gobi, and arid regions, aiming to build a clean and low-carbon new energy system. However, the high proportion of renewable energy installed capacity in these regions, coupled with the strong intermittency, randomness, and anti-peak-shaving characteristics of its output, causes severe and complex multi-frequency fluctuations in system net load at minute, hourly, and even diurnal scales. This poses an unprecedented challenge to the real-time power balance and safe and stable operation of the power system. The traditional coal-fired power-based regulation mode, due to its slow response speed, limited ramping capacity, and poor economic efficiency of frequent regulation, is no longer sufficient to meet the dual demands of flexibility and economy under a high proportion of renewable energy grid connection. Therefore, how to efficiently and collaboratively utilize various types of flexible resources such as energy storage, demand response, and flexible generating units to achieve precise smoothing and economical dispatch of net load fluctuations at multiple time scales has become a core scientific problem and engineering challenge that urgently needs to be solved for the safe and efficient operation of new power systems in desert and Gobi regions.

[0003] To address the above challenges, the academic and industrial communities have conducted extensive research on the coordinated scheduling of flexible resources. Reference [1] systematically reviews the optimal scheduling model of high-proportion renewable energy power systems, emphasizing the importance of introducing energy storage and demand response. However, its model still focuses on the day-ahead and intraday hourly scales, and has limited ability to smooth fluctuations at shorter time scales. Reference [2] proposes a two-stage robust optimization scheduling method that considers source-load uncertainty, which improves the robustness of the scheduling scheme. However, it has high computational complexity and does not explicitly distinguish and differentiate the fluctuation characteristics at different time scales. Reference [3] explores in depth the application of electrochemical energy storage in smoothing photovoltaic power fluctuations and proposes a real-time control strategy based on model predictive control, which effectively improves the utilization efficiency of single energy storage. Patent No. 2024102900323 discloses a method for smoothing wind power output of energy storage systems based on fuzzy logic. However, such methods are usually aimed at a single resource or a single fluctuation frequency band, and lack an overall architecture design for the coordinated operation of multiple types and multiple time scale resources within the system. Wavelet transform has attracted attention in this field due to its excellent time-frequency localization analysis capability. Reference [4] proposes to use wavelet packet decomposition to separate the net load fluctuation of the microgrid and allocate it to the hybrid energy storage system composed of batteries and supercapacitors, achieving a good smoothing effect. Reference [5] further combines wavelet analysis with empirical mode decomposition to identify the adjustable potential of the load. Patent No. 2023106120720 discloses a distribution network active power coordination control method based on wavelet transform, which decomposes the fluctuation and allocates it to different controllable units. Reference [6] uses the HP filter decomposition method to decompose the system net load of the wind and solar real-time scenario in the real-time stage, decomposes it into high-frequency components and trend components, and uses energy storage real-time adjustment to smooth the high-frequency components of the net load. Reference [7] proposes a net load interval prediction method based on width learning for photovoltaic, energy storage and charging stations in the small data set net load uncertainty prediction problem.

[0004] Based on the above analysis, existing research, when addressing the multi-timescale and highly random net load fluctuations caused by the high proportion of renewable energy grid connection in desert and Gobi regions, mostly focuses on single timescales or single types of resources in its scheduling methods. It lacks a systematic decomposition and collaborative mitigation mechanism for fluctuations across the entire frequency band from minute to hourly levels. Although time-frequency analysis techniques such as wavelet decomposition have been introduced, they mostly remain at the level of describing fluctuation characteristics or serving as a preprocessing stage for prediction models. They do not further deeply couple the decomposed multi-frequency components with resources such as energy storage, demand response, and thermal power units, nor do they consider the uncertainties in renewable energy output and load in desert and Gobi regions.

[0005] References:

[0006] [1] Li Mingjie, Chen Guoping, Dong Cun, et al. Review of key technologies for optimized operation of high-proportion renewable energy power systems [J]. Proceedings of the CSEE, 2022, 42(5): 1845-1865.

[0007] [2] Wei Zhinong, Zhang Side, Sun Guoqiang, et al. Two-stage partial blue bar optimization scheduling of power system considering source-load uncertainty [J]. Automation of Electric Power Systems, 2021, 45(10): 21-30.

[0008] [3] Jia Hongjie, Mu Yunfei, Wang Mingshen, et al. Optimal control strategy for electrochemical energy storage in photovoltaic power fluctuation smoothing[J]. Journal of Electrical Engineering, 2023, 38(2): 535-546.

[0009] [4] Li Peng, Wang Zixuan, Dou Xiaobo, et al. Coordinated control of hybrid energy storage in islanded microgrids based on wavelet packet decomposition [J]. Automation of Electric Power Systems, 2020, 44(3): 142-150.

[0010] [5] Wang Yi, Yang Zhifang, Wang Jianxue, et al. Identification of Adjustable Potential of Residential Load Based on VMD-WPT and Deep Learning [J]. Automation of Electric Power Systems, 2021, 45(23): 13-22.

[0011] [6] Zhang Haibo, Duan Rui, Shen Jie, et al. Two-stage stochastic optimization scheduling of thermal power units with real-time energy storage regulation [J]. Acta Energiae Solaris Sinica, 2025, 46(11):119-128.

[0012] [7] Cui Wei, Wang Junlong, Wang Yifeng, et al. Prediction method for net load interval of low-voltage distribution network with photovoltaic, energy storage and charging [J]. Computing Technology and Automation, 2025, 44(01):118-124. DOI:10.16339 / j.cnki.jsjsyzdh.202501021. Summary of the Invention

[0013] The purpose of this invention is to overcome the shortcomings of the prior art and provide a flexible resource joint regulation method for desert areas based on wavelet decomposition.

[0014] The objective of this invention is achieved through the following technical solution: a flexible resource joint regulation method for desert and Gobi areas based on wavelet decomposition, the method comprising,

[0015] S1: Obtain the time series data of the net load of the power system and perform discrete wavelet transform on it to decompose and obtain the approximation coefficients and detail coefficients at each scale;

[0016] S2: High-frequency components, mid-frequency components, and low-frequency components are reconstructed based on approximation coefficients and detail coefficients;

[0017] S3: Calculate the instantaneous error between the wavelet decomposition and reconstruction synthetic signal obtained by adding the high-frequency, medium-frequency and low-frequency components and the original net load time series data, and perform rolling statistics. Combine the error statistical characteristics with the standard deviation of the new energy prediction error to construct the power confidence interval for each frequency band component. Use the frequency band component with the confidence interval as the input of the subsequent scheduling model.

[0018] S4: Construct an energy storage system model based on high-frequency components with confidence intervals, construct a demand response resource model based on medium-frequency components with confidence intervals, and construct a thermal power unit model based on low-frequency components with confidence intervals.

[0019] S5: With the goal of minimizing the total operating cost of the system, a joint optimization scheduling model considering input uncertainties is constructed. The constraints of the joint optimization scheduling model include system power balance constraints and opportunity constraints.

[0020] S6: Perform a deterministic transformation on the chance constraints, and use a mixed-integer linear programming solver to solve the joint optimization scheduling model to obtain the optimal robust output plan for various flexible resources within the scheduling period.

[0021] Specifically, the steps of S1 are as follows:

[0022] For the net load discrete signal, its j-th level approximation coefficients and detail coefficients are calculated as follows:

[0023] ;

[0024] ;

[0025] ;

[0026] In the formula, is the approximation coefficient of the j-th layer; m is the convolution summation index; These are the approximation coefficients for the (j-1)th layer; Let J be the detail coefficients for the j-th layer; These are the coefficients of the low-pass filter; Here are the high-pass filter coefficients; k is the translation parameter. The net load discrete signal; represents the initial approximation coefficients, indicating the original signal before decomposition.

[0027] Specifically, the steps of S2 are as follows:

[0028] High-frequency components: obtained through reconstruction of detail coefficients from layer 1 to J1:

[0029] ;

[0030] In the formula, These are high-frequency components; To reconstruct the high-pass filter coefficients;

[0031] Intermediate frequency component: via the first Layer to the first The layer detail coefficients were reconstructed to obtain:

[0032] ;

[0033] In the formula, This is the mid-frequency component;

[0034] Low-frequency components: via the first Layer to the first The reconstructed layer approximation coefficients yielded:

[0035] ;

[0036] In the formula, It is a low-frequency component.

[0037] Specifically, the steps of S3 are as follows:

[0038] Calculate the instantaneous error between the reconstructed synthetic signal and the original net load signal:

[0039] ;

[0040] In the formula, Let be the wavelet decomposition and reconstruction error power at time t; The measured net load power of the system at time t; For the high-frequency components at time t; Let be the mid-frequency component at time t; The low-frequency component at time t;

[0041] Statistical characteristics of calculation error:

[0042] ;

[0043] ;

[0044] In the formula, This represents the mean of the recent error sequence; It is an instantaneous error sequence; is the standard deviation of the recent error sequence; N is the length of the rolling time window;

[0045] By combining the standard deviation of the recent error sequence with the standard deviation of the new energy prediction error, power confidence intervals are constructed for each frequency band component:

[0046] ;

[0047] ;

[0048] In the formula, Let be the standard deviation of the overall uncertainty of the i-th frequency band component at time t, where i is the index of high, medium, and low frequencies; Let be the standard deviation of the ultra-short-term prediction error of new energy at time t; , The error allocation coefficients assigned to the i-th frequency band satisfy the following conditions: ; Let be the lower and upper limits of the power confidence interval for the i-th frequency band component at time t; This represents the nominal power value of the i-th frequency band component; is the confidence level coefficient.

[0049] Specifically, the energy storage system model is as follows:

[0050] ;

[0051] ;

[0052] ;

[0053] ;

[0054] In the formula, Let s be the lifetime loss cost of the s-th energy storage unit at time t; Unit cycle cost; This represents the total number of cycles in the energy storage system. Let be the charging and discharging power of the s-th energy storage unit at time t. For discharge, To charge; This refers to the rated capacity of the energy storage. Let be the state of charge of the s-th energy storage unit at time t; , These are the lower and upper limits of the permissible state of charge, respectively; These are the charging and discharging efficiencies, respectively. , These are the maximum charging and discharging power of the energy storage system, respectively.

[0055] Specifically, the demand response resource model is as follows:

[0056] ;

[0057] ;

[0058] ;

[0059] ;

[0060] In the formula, Let be the actual adjustment power of the i-th type of demand response resource at time t; , These are the lower and upper limits of the adjustable power for this type of resource, respectively. It is a 0 / 1 variable, indicating whether the i-th type of resource is in a invoked state at time t; , These represent the maximum power change rate of the resource in the start-up and shutdown states, respectively; The minimum continuous runtime that a resource must maintain after it has been invoked; Let be the comprehensive adjustment cost of the i-th type of resource at time t; The penalty factor for deviation from baseline power; This is the penalty coefficient for the power fluctuation amplitude; This represents the baseline power of this type of load at time t.

[0061] Specifically, the thermal power unit model is as follows:

[0062] ;

[0063] ;

[0064] ;

[0065] ;

[0066] ;

[0067] In the formula, Let g be the total operating cost of the g-th generating unit at time t; , , These are the coefficients of the unit's fuel cost curve; Price per unit of carbon emissions; The carbon emission intensity of the unit; Let g be the output of the g-th thermal power unit at time t; The variable is 0 / 1, representing the time point at which the g-th thermal power unit is located. Start-up and shutdown status; The variable is 0 or 1, representing the start-up / shutdown status of the unit at time t; , The lower and upper limits of the technology required for generating unit output; , The maximum ramp rate for the generator unit; This represents the minimum continuous operating time of the g-th generating unit. This represents the minimum continuous downtime of the g-th unit.

[0068] Specifically, the objective function of the joint optimization scheduling model is:

[0069] ;

[0070] In the formula, T is the total number of time periods in the scheduling cycle; , , These represent the total number of thermal power units, demand response resource aggregators, and energy storage systems, respectively. Let g be the total operating cost of the g-th generating unit at time t; Let be the comprehensive adjustment cost of the i-th type of resource at time t; This represents the lifetime loss cost of the energy storage at time t.

[0071] Specifically, the constraints of the joint optimization scheduling model include:

[0072] System power balance constraints:

[0073] ;

[0074] In the formula, Let t be the net system load at time t;

[0075] Opportunity Constraints:

[0076] ;

[0077] ;

[0078] ;

[0079] In the formula, The probability of the event within the parentheses occurring; , , These are the allowable tracking error tolerances for high, medium, and low frequency components, respectively. , The preset confidence level represents the minimum probability that resource output is required to track the corresponding component within its respective tolerance.

[0080] Specifically, the deterministic transformation of the opportunity constraint includes:

[0081] The opportunity constraint for energy storage systems to track high-frequency components is transformed into:

[0082] ;

[0083] ;

[0084] ;

[0085] ;

[0086] In the formula, It is a positive number; For high-frequency component binary auxiliary variables, when The time indicates that the high-frequency tracking constraint at time t is strictly satisfied; This is the floor function operator; This represents the total number of time periods within the scheduling cycle.

[0087] The opportunity constraint for the mid-frequency component of demand response resource tracking is transformed into:

[0088] ;

[0089] ;

[0090] ;

[0091] ;

[0092] In the formula, For the intermediate frequency component binary auxiliary variable, when The time indicates that the intermediate frequency tracking constraint at time t is strictly satisfied;

[0093] For thermal power units tracking low-frequency components, the conversion is as follows:

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] In the formula, For low-frequency component binary auxiliary variables, when The time indicates that the low-frequency tracking constraint at time t is strictly satisfied.

[0099] The present invention has the following advantages:

[0100] This invention establishes refined models for energy storage systems, demand response resources, and thermal power units for load characteristics in each frequency band. It also introduces frequency band matching guidance constraints for high-frequency-energy storage, medium-frequency-demand response, and low-frequency-thermal power in the joint dispatch model. By introducing frequency band-resource matching constraints and tracking error opportunity constraints, and considering the prediction error and decomposition uncertainty of new energy sources, it achieves adaptive and accurate tracking of resource output and load fluctuations. This minimizes the overall operating cost and improves the overall regulation economy while meeting system power balance and various operating constraints.

[0101] The control method of this invention is not only applicable to high-proportion renewable energy power systems in desert and Gobi areas, but can also be extended to other regional power grids with high renewable energy penetration and complex load fluctuations. The wavelet basis function, decomposition level, and frequency band division threshold used can be adjusted according to the actual system characteristics. Resource types such as energy storage, demand response, and thermal power can also be expanded to other flexible resources (such as gas turbine units, pumped storage, etc.), demonstrating good technical adaptability and scalability. Attached Figure Description

[0102] Figure 1 is a schematic diagram of the combined regulation method of the present invention;

[0103] Figure 2 is a schematic diagram of wavelet decomposition;

[0104] Figure 3 is a schematic diagram of the time-frequency operation of continuous wavelet transform;

[0105] Figure 4 is a schematic diagram of flexible resource component tracking. Detailed Implementation

[0106] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described embodiments are merely some embodiments of the invention, and not all embodiments. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0107] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0108] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0109] The present invention will be further described below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following description.

[0110] As shown in Figures 1 to 4, a flexible resource joint regulation method based on wavelet decomposition is proposed for desert areas. This method includes...

[0111] S1: Acquire time-series data of the net load of the power system, with a sampling interval of 5-15 minutes. Select the Daubechies 4th order wavelet (db4) as the basis function, and use the Mallat algorithm to perform discrete wavelet transform on it to decompose and obtain the approximation coefficients and detail coefficients at each scale; process the discrete signal of the net load. The approximation coefficients and detail coefficients of its j-th layer are calculated as follows:

[0112] ;

[0113] ;

[0114] ;

[0115] In the formula, is the approximation coefficient of the j-th layer, corresponding to the low-frequency components of the signal; m is the convolution summation index; These are the approximation coefficients for the (j-1)th layer; Let be the detail coefficients of the j-th layer, corresponding to the high-frequency components of the signal; These are the coefficients of the low-pass filter, which are related to the scaling function; represents the high-pass filter coefficients, which are related to the wavelet function; k is the translation parameter, representing the index of the coefficient at scale j; The net load discrete signal; represents the initial approximation coefficients, indicating the original signal before decomposition.

[0116] This invention employs the Daubechies wavelet series for decomposition, whose tight support and smoothness facilitate effective characterization of the load signal in both the time and frequency domains. The fourth-order Daubechies wavelet (db4) is chosen as the decomposition basis function because its moderate support length achieves a good balance between frequency resolution and temporal locality. Furthermore, its sufficiently high-order vanishing moments enable more accurate extraction of the smoothing trend components in the net load, while ensuring the orthogonality of the decomposition and avoiding component information redundancy. This lays a reliable foundation for subsequent hierarchical and precise matching and collaborative scheduling of multiple resource types.

[0117] Based on the main timescale and sampling frequency of load fluctuations, the number of decomposition layers is determined, typically covering typical fluctuation frequency bands from minutes to several hours. The number of decomposition layers is calculated as follows:

[0118] ;

[0119] In the formula, The signal sampling frequency; The minimum fluctuation frequency; This is for rounding down; in practical applications, it is usually rounded down to the nearest integer. To cover typical load fluctuations ranging from minutes to hours.

[0120] S2: High-frequency components, mid-frequency components, and low-frequency components are reconstructed based on approximation coefficients and detail coefficients;

[0121] To verify the correctness of the decomposition process and support subsequent independent analysis of fluctuation components at different time scales, the original signal can be completely reconstructed using the approximation coefficients and detail coefficients obtained from the decomposition through inverse discrete wavelet transform (IDWT):

[0122] ;

[0123] In the formula, and These are the coefficients for reconstructing the low-pass and high-pass filters, respectively. For reconstructing the signal; k is the translation parameter, representing the index of the coefficient at scale j;

[0124] This invention performs partial reconstruction of specific frequency bands to extract load fluctuation components suitable for tracking different types of resources, specifically including:

[0125] High-frequency component: Obtained through the detailed coefficients from layer 1 to J1, corresponding to rapid random fluctuations on the minute level. This high-frequency component fluctuates frequently and has a relatively small amplitude, which matches the rapid power regulation characteristics of the energy storage system and is suitable for tracking and smoothing by energy storage resources.

[0126] ;

[0127] In the formula, These are high-frequency components; To reconstruct the high-pass filter coefficients;

[0128] Intermediate frequency component: via the first Layer to the first The reconstructed layer detail coefficients correspond to periodic fluctuations ranging from 30 minutes to 2 hours. The rate of change of this mid-frequency component is moderate, matching the adjustment capability and response delay characteristics of demand response resources, making it suitable for tracking and adjustment by demand response resources.

[0129] ;

[0130] In the formula, This is the mid-frequency component;

[0131] Low-frequency components: via the first Layer to the first The reconstructed layer approximation coefficients correspond to long-term trends and slow changes over a period of more than 2 hours. The low-frequency component changes slowly with a large amplitude, which matches the large capacity and slow regulation characteristics of thermal power units, making it suitable for tracking the basic output plan of thermal power units.

[0132] ;

[0133] In the formula, It is a low-frequency component.

[0134] S3: Calculate the instantaneous error between the wavelet decomposition and reconstruction synthetic signal obtained by adding the high-frequency, medium-frequency and low-frequency components and the original net load time series data, and perform rolling statistics. Combine the error statistical characteristics with the standard deviation of the new energy prediction error to construct the power confidence interval for each frequency band component. Use the frequency band component with the confidence interval as the input of the subsequent scheduling model.

[0135] First, online extraction and statistical characterization of the decomposition error are performed: at each scheduling time t, the high-frequency components obtained through partial reconstruction in step S2 are... Intermediate frequency components and low frequency components Summation to calculate the measured net system load at the same time. Instantaneous error between high, medium, and low frequency summation The error sequence was then subjected to rolling statistical analysis. The formula for calculating the instantaneous reconstruction error is:

[0136] ;

[0137] In the formula, Let be the wavelet decomposition and reconstruction error power at time t; Let t be the net system load power measured at time t.

[0138] The system continuously stores the error sequences of the most recent N scheduling moments. And calculate its statistical characteristics in real time:

[0139]

[0140]

[0141] In the formula, The mean of the recent error sequence reflects the systematic bias in wavelet decomposition. It is an instantaneous error sequence; is the standard deviation of the recent error sequence, reflecting the fluctuation amplitude and uncertainty of the decomposed error; N is the length of the rolling time window, which is set according to the system sampling frequency and usually covers several hours to capture the time-varying characteristics of the error.

[0142] Then, the standard deviation of the obtained recent error series is... The standard deviation of the new energy prediction error is integrated with the standard deviation of the new energy prediction error. Treating both as independent random variables, a weighted synthesis is used to construct a confidence interval for the power value of each frequency band component, replacing a single deterministic nominal value. The formula for calculating the comprehensive uncertainty of each frequency band is as follows:

[0143]

[0144] In the formula, Let be the standard deviation of the overall uncertainty of the i-th frequency band component at time t, where i is the index of high, medium, and low frequencies; Let t be the standard deviation of the ultra-short-term prediction error of new energy sources (wind power, photovoltaics); , The error allocation coefficients assigned to the i-th frequency band satisfy the following conditions: and Its value can be determined based on the energy proportion of each frequency band.

[0145] The formula for constructing the confidence interval of a frequency band component is:

[0146] ;

[0147] In the formula, Let be the power confidence interval of the i-th frequency band component at time t, indicating that the true value of the component falls within this range with a high probability; Let be the lower and upper limits of the power confidence interval for the i-th frequency band component at time t; This represents the nominal power value of the i-th frequency band component; is the confidence level coefficient.

[0148] When using wavelet transform for multi-scale decomposition, inherent boundary effects and mode aliasing are amplified, resulting in non-negligible decomposition errors. Directly assigning these erroneous decomposition components as deterministic commands to flexible resources leads to a mismatch between resource actions and actual demand, increasing regulation costs and potentially causing power imbalances that threaten system safety. Therefore, this invention combines decomposition errors from signal processing with uncertain power system scheduling, forming a closed-loop control architecture with online identification and adaptive compensation capabilities, thus improving the robustness of regulation under severe fluctuations during desertification.

[0149] S4: Based on high-frequency components with confidence intervals Construct an energy storage system model based on the mid-frequency component with confidence intervals. Construct a demand response resource model based on low-frequency components with confidence intervals. A thermal power unit model is constructed. To achieve accurate and economical tracking and mitigation of load fluctuations in each frequency band after decomposition, mathematical modeling of the physical characteristics, operational constraints, and economics of the main flexibility resources in the system is required. This invention matches various resources with load components of different frequency bands based on their response speed, regulation capacity, and duration. The energy storage system is suitable for tracking high-frequency components. Demand response resources are suitable for tracking intermediate frequency components. Thermal power units are suitable for tracking low-frequency components. .

[0150] The energy storage system model is constructed as follows:

[0151] Lifetime loss cost model

[0152] ;

[0153] Power and capacity constraints: At any time t, the charging and discharging power of the energy storage system must satisfy the following:

[0154] ;

[0155] In the formula, , These are the maximum charging and discharging power of the energy storage system, respectively.

[0156] Energy storage system capacity status Must meet:

[0157] ;

[0158] The formula for dynamic capacity update is:

[0159] ;

[0160] In the formula, Let s be the lifetime loss cost of the s-th energy storage unit at time t; Unit cycle cost; This represents the total number of cycles in the energy storage system. Let be the charging and discharging power of the s-th energy storage unit at time t. For discharge, To charge; This refers to the rated capacity of the energy storage. Let be the state of charge of the s-th energy storage unit at time t; , These are the lower and upper limits of the permissible state of charge, respectively; These are the charging and discharging efficiencies, respectively. .

[0161] The demand response resource model is as follows:

[0162] Introducing a penalty function to quantify the impact of moderating behavior on user comfort and incentive costs:

[0163] ;

[0164] Assume the adjustable power of the i-th type of flexible load at time t is Its adjustment range is limited:

[0165] ;

[0166] The power change rate of demand response resources is limited by ramping capability and must meet minimum sustained settling time requirements:

[0167] ;

[0168] ;

[0169] In the formula, Let be the actual adjustment power of the i-th type of demand response resource at time t; , These are the lower and upper limits of the adjustable power for this type of resource, respectively. The variable is 0 or 1, indicating whether the i-th type of resource is in a call state at time t (1 for yes, 0 for no). , These represent the maximum power change rate of the resource in the start-up and shutdown states, respectively; The minimum continuous runtime that a resource must maintain after it has been invoked; Let be the comprehensive adjustment cost of the i-th type of resource at time t; The penalty factor for deviation from baseline power; This is the penalty coefficient for the power fluctuation amplitude; This represents the baseline power of this type of load at time t.

[0170] The thermal power unit model is constructed as follows:

[0171] Fuel cost and carbon emission model:

[0172] ;

[0173] The output range and climbing ability constraints of thermal power units are as follows:

[0174] ;

[0175] ;

[0176] To ensure stable unit operation, minimum start-up and shutdown time constraints must be met:

[0177] ;

[0178] ;

[0179] In the formula, Let g be the total operating cost of the g-th generating unit at time t; , , These are the coefficients of the unit's fuel cost curve; Price per unit of carbon emissions; The carbon emission intensity of the unit; Let g be the output of the g-th thermal power unit at time t; The variable is 0 / 1, representing the time point at which the g-th thermal power unit is located. Start-up and shutdown status; The variable is 0 or 1, representing the start-up and shutdown status of the unit at time t (1 for running, 0 for shutting down). , The lower and upper limits of the technology required for generating unit output; , The maximum ramp rate for the generator unit; This represents the minimum continuous operating time of the g-th generating unit. This represents the minimum continuous downtime of the g-th unit.

[0180] S5: With minimizing the total system operating cost as the core objective, construct a joint optimization scheduling model that considers input uncertainty. This model transforms the uncertainty quantified in step S3 into a quantitative requirement for the robustness of the scheduling scheme by introducing chance constraints;

[0181] The objective function of the joint optimization scheduling model considering input uncertainty is:

[0182] ;

[0183] In the formula, T is the total number of time periods in the scheduling cycle; , , These represent the total number of thermal power units, demand response resource aggregators, and energy storage systems, respectively. Let g be the total operating cost of the g-th generating unit at time t; Let be the comprehensive adjustment cost of the i-th type of resource at time t; This represents the lifetime loss cost of the energy storage at time t.

[0184] The constraints of the joint optimization scheduling model include:

[0185] System power balance constraint: at any given time, the sum of the regulating power of all flexible resources must be balanced with the net load.

[0186] ;

[0187] In the formula, Let be the net system load at time t, and be the difference between the original load and the output of renewable energy.

[0188] To construct a robust control model that adapts to uncertainty, based on opportunity-constrained frequency band matching robustness constraints, this invention introduces the following opportunity constraints to replace deterministic tracking requirements, ensuring the reliability of the scheduling scheme under various potential scenarios:

[0189] ;

[0190] ;

[0191] ;

[0192] In the formula, The probability of the event within the parentheses occurring; , , These are the allowable tracking error tolerances for high, medium, and low frequency components, respectively. , The preset confidence level represents the minimum probability that resource output is required to track the corresponding component within its respective tolerance.

[0193] S6: Perform a deterministic transformation on the chance constraints, and use the mixed integer linear programming (Gurobi) solver to solve the joint optimization scheduling model considering uncertainty, so as to obtain the optimal robust output plan of various flexible resources within the scheduling cycle.

[0194] To solve the above probabilistic model, the chance constraints need to be transformed into a deterministic mathematical programming form.

[0195] Deterministic transformation of high-frequency component tracking constraints:

[0196] For the opportunity constraint of energy storage systems tracking high-frequency components, its confidence interval is used. Perform deterministic equivalent transformations and introduce auxiliary binary variables. :

[0197] ;

[0198] ;

[0199] ;

[0200] ;

[0201] In the formula, It is a sufficiently large positive number; For high-frequency component binary auxiliary variables, when When, it means that the high-frequency tracking constraint at time t is strictly satisfied; This is the floor function operator; This represents the total number of time periods within the scheduling cycle.

[0202] Deterministic transformation of mid-frequency component tracking constraints:

[0203] For the opportunity constraint of the mid-frequency component in demand response resource tracking, its confidence interval is used. Perform deterministic equivalent transformations and introduce auxiliary binary variables. :

[0204] ;

[0205] ;

[0206] ;

[0207] ;

[0208] In the formula, For the intermediate frequency component binary auxiliary variable, when The time interval t indicates that the intermediate frequency tracking constraint is strictly satisfied.

[0209] Deterministic transformation of low-frequency component tracking constraints:

[0210] For thermal power units tracking low-frequency components, their confidence intervals are used. Perform deterministic equivalent transformations and introduce auxiliary binary variables. :

[0211] ;

[0212] ;

[0213] ;

[0214] ;

[0215] In the formula, For low-frequency component binary auxiliary variables, when The time indicates that the low-frequency tracking constraint at time t is strictly satisfied.

[0216] Taking actual net load data from a region in Northwest China as the research object, the number of layers was calculated to be 6. The decomposition layer was performed using the Daubechies 4th order wavelet (db4) for 6 layers. The data time span was 10 days, the sampling interval was 5 minutes, and a total of 2880 data points were obtained. Through multi-layer decomposition, detail components D1~D6 covering the period range of 5 minutes to 5.3 hours and approximate component A6 reflecting the long-period trend were obtained. The complete wavelet decomposition results are shown in Figure 2. As can be seen from Figure 2, each component exhibits obvious differences in frequency characteristics: detail components D1 (5-10 minutes) and D2 (10-20 minutes) correspond to ultra-high frequency and high frequency fluctuations, with dense waveforms and small amplitudes, mainly caused by random load disturbances and short-term equipment switching. The detailed components D3 (20-40 minutes) and D4 (40-80 minutes) correspond to mid-frequency variations, reflecting the periodic adjustment characteristics of electricity consumption behavior. The detailed components D5 (1.3-2.7 hours), D6 (2.7-5.3 hours), and the approximate component A6 (>5.3 hours) represent low-frequency trends and base load patterns, reflecting the overall daily load variation. Energy distribution analysis shows that low-frequency components contribute the majority of the total load energy, while high-frequency components, although relatively small in amplitude, exhibit drastic changes, placing higher demands on the system's rapid adjustment capabilities. The continuous wavelet transform time-frequency diagram is shown in Figure 3.

[0217] Based on the aforementioned wavelet decomposition results, a multi-resource collaborative scheduling optimization model for energy storage systems, demand response, and thermal power units was constructed, and the simulation results are shown in Figure 4.

[0218] As shown in Figure 4, the three types of flexible resources achieved accurate tracking and effective regulation of load components in different frequency bands. Regarding energy storage tracking of high-frequency components, the energy storage system, with its rapid response capability, effectively suppressed short-term random fluctuations, and the residual curve remained within a relatively small range, indicating that energy storage has a good smoothing effect on high-frequency disturbances. Regarding demand response tracking of mid-frequency components, the demand response resources achieved good tracking of load fluctuations with a period of 20-80 minutes. Although there were some response delays and tracking errors, the overall regulation effect was significant, and the residual amplitude was significantly smaller than that of the original mid-frequency component. Regarding thermal power tracking of low-frequency components, the thermal power units fully utilized their advantages of large capacity and sustainable regulation, effectively tracking low-frequency trend changes of more than 1.3 hours. The residual curve was smooth and the amplitude was small, verifying the applicability of thermal power units in undertaking base load and slow trend regulation.

[0219] The coordinated operation of the three types of resources effectively decomposes and absorbs the complex multi-frequency fluctuations of the original load. Each resource plays a regulatory role on its preferred time scale, avoiding the efficiency loss and cost increase caused by frequent cross-time scale adjustments by a single resource. Residual analysis shows that after multi-resource joint scheduling, the variance of the system's net load fluctuation is significantly reduced compared to the original load, verifying the effectiveness and superiority of the frequency-layered scheduling strategy based on wavelet decomposition.

[0220] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any person skilled in the art can make many possible variations and modifications to the technical solution of the present invention, or modify it into equivalent embodiments, without departing from the scope of the present invention. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technology of the present invention without departing from the scope of the present invention are within the protection scope of the present invention.

Claims

1. A flexible resource joint regulation method based on wavelet decomposition in desert and Gobi areas, characterized in that: The method includes, S1: acquiring the net load time series data of the power system and performing discrete wavelet transform on it to decompose and obtain the approximation coefficients and detail coefficients at each scale; S2: High-frequency components, mid-frequency components, and low-frequency components are reconstructed based on approximation coefficients and detail coefficients; S3: Calculate the instantaneous error between the wavelet decomposition reconstructed synthetic signal obtained by adding the high-frequency, mid-frequency, and low-frequency components and the original net load time-series data, and perform rolling statistics. Combine the error statistical characteristics with the standard deviation of the new energy prediction error to construct power confidence intervals for each frequency band component; S4: Construct an energy storage system model based on the high-frequency components with confidence intervals, a demand response resource model based on the mid-frequency components with confidence intervals, and a thermal power unit model based on the low-frequency components with confidence intervals; S5: Construct a joint optimization scheduling model considering input uncertainty with the goal of minimizing the total system operating cost. The constraints of the joint optimization scheduling model include system power balance constraints and opportunity constraints; S6: Perform a deterministic transformation on the opportunity constraints, and use a mixed-integer linear programming solver to solve the joint optimization scheduling model to obtain the optimal robust output plan for various flexible resources within the scheduling cycle.

2. The method for flexible resource joint regulation in desert and Gobi areas based on wavelet decomposition according to claim 1, characterized in that: The specific steps of S1 are as follows: For the net load discrete signal, the approximation coefficients and detail coefficients of the j-th layer are calculated as follows: ； ； In the formula, is the approximation coefficient of the j-th layer; m is the convolution summation index; These are the approximation coefficients for the (j-1)th layer; Let J be the detail coefficients for the j-th layer; These are the coefficients of the low-pass filter; Here are the high-pass filter coefficients; k is the translation parameter. The net load discrete signal; represents the initial approximation coefficients, indicating the original signal before decomposition.

3. The method for flexible resource joint regulation in desert and Gobi areas based on wavelet decomposition according to claim 2, characterized in that: The specific steps of S2 are as follows: High-frequency components: obtained by reconstructing detail coefficients from layer 1 to layer J1: In the formula, These are high-frequency components; To reconstruct the high-pass filter coefficients; Intermediate frequency component: via the first Layer to the first The layer detail coefficients were reconstructed to obtain: In the formula, This is the mid-frequency component; Low-frequency components: via the first Layer to the first The reconstructed layer approximation coefficients yielded: In the formula, It is a low-frequency component.

4. The method for flexible resource joint regulation in desert and Gobi areas based on wavelet decomposition according to claim 1, characterized in that: The specific steps of S3 are as follows: Calculate the instantaneous error between the reconstructed synthetic signal and the original net load signal: In the formula, Let be the wavelet decomposition and reconstruction error power at time t; The measured net system load power at time t; For the high-frequency components at time t; Let be the mid-frequency component at time t; The low-frequency component at time t; Statistical characteristics of calculation error: ； In the formula, This represents the mean of the recent error sequence; It is an instantaneous error sequence; The standard deviation of the recent error sequence is given; N is the length of the rolling time window; the standard deviation of the recent error sequence is combined with the standard deviation of the new energy prediction error to construct the power confidence interval for each frequency band component: ； In the formula, Let be the standard deviation of the overall uncertainty of the i-th frequency band component at time t, where i is the index of high, medium, and low frequencies; Let be the standard deviation of the ultra-short-term prediction error of new energy at time t; 、 The error allocation coefficients assigned to the i-th frequency band satisfy the following conditions: ； Let be the lower and upper limits of the power confidence interval for the i-th frequency band component at time t; This represents the nominal power value of the i-th frequency band component; is the confidence level coefficient.

5. A flexible resource joint regulation method based on wavelet decomposition in desert and Gobi areas according to claim 1, characterized in that: The energy storage system model is as follows: ； ； ； In the formula, Let s be the lifetime loss cost of the s-th energy storage unit at time t; Unit cycle cost; This represents the total number of cycles in the energy storage system. Let be the charging and discharging power of the s-th energy storage unit at time t. For discharge, To charge; This refers to the rated capacity of the energy storage. Let be the state of charge of the s-th energy storage unit at time t; 、 These are the lower and upper limits of the permissible state of charge, respectively; These are the charging and discharging efficiencies, respectively. 、 These are the maximum charging and discharging power of the energy storage system, respectively.

6. The method for flexible resource joint regulation in desert and Gobi areas based on wavelet decomposition according to claim 1, characterized in that: The demand response resource model is as follows: ； ； ； In the formula, Let be the actual adjustment power of the i-th type of demand response resource at time t; 、 These are the lower and upper limits of the adjustable power for this type of resource, respectively. It is a 0 / 1 variable, indicating whether the i-th type of resource is in a invoked state at time t; 、 These represent the maximum power change rate of the resource in the start-up and shutdown states, respectively; The minimum continuous runtime that a resource must maintain after it has been invoked; Let be the comprehensive adjustment cost of the i-th type of resource at time t; The penalty factor for deviation from baseline power; This is the penalty coefficient for the power fluctuation amplitude; This represents the baseline power of this type of load at time t.

7. A flexible resource joint regulation method based on wavelet decomposition in desert and Gobi areas according to claim 1, characterized in that: The thermal power unit model is as follows: ； ； ； ； In the formula, Let g be the total operating cost of the g-th generating unit at time t; 、 、 These are the coefficients of the unit's fuel cost curve; Price per unit of carbon emissions; The carbon emission intensity of the unit; Let g be the output of the g-th thermal power unit at time t; The variable is 0 / 1, representing the time point at which the g-th thermal power unit is located. Start-up and shutdown status; The variable is 0 or 1, representing the start-up and shutdown status of the unit at time t; 、 The lower and upper limits of the technology required for generating unit output; 、 The maximum ramp rate for the generator unit; This represents the minimum continuous operating time of the g-th generating unit. This represents the minimum continuous downtime of the g-th unit.

8. A method for joint regulation of flexible resources in desert and Gobi areas based on wavelet decomposition as described in claim 1, characterized in that: The objective function of the joint optimization scheduling model is: In the formula, T represents the total number of time periods in the scheduling cycle; 、 、 These represent the total number of thermal power units, demand response resource aggregators, and energy storage systems, respectively. Let g be the total operating cost of the g-th generating unit at time t; Let be the comprehensive adjustment cost of the i-th type of resource at time t; This represents the lifetime loss cost of the energy storage at time t.

9. A method for flexible resource joint regulation in desert and Gobi areas based on wavelet decomposition as described in claim 8, characterized in that: The constraints of the joint optimization scheduling model include: system power balance constraints: In the formula, Let be the net system load at time t; chance constraints: ； ； In the formula, The probability of the event within the parentheses occurring; 、 、 These are the allowable tracking error tolerances for high, medium, and low frequency components, respectively. 、 The preset confidence level represents the minimum probability that resource output is required to track the corresponding component within its respective tolerance.

10. A method for joint regulation of flexible resources in desert and Gobi areas based on wavelet decomposition as described in claim 8, characterized in that: The deterministic transformation of the opportunity constraint includes: for the opportunity constraint of the energy storage system tracking high-frequency components, the transformation is as follows: ； ； ； In the formula, It is a positive number; For high-frequency component binary auxiliary variables, when The time indicates that the high-frequency tracking constraint at time t is strictly satisfied; This is the floor function operator; The total number of time periods within the scheduling cycle; the opportunity constraint for the mid-frequency component of demand response resource tracking is converted to: ； ； ； In the formula, For the intermediate frequency component binary auxiliary variable, when The time interval t indicates that the intermediate frequency tracking constraint is strictly satisfied; for thermal power units tracking the low-frequency component, it is converted to: ； ； ； In the formula, For low-frequency component binary auxiliary variables, when The time indicates that the low-frequency tracking constraint at time t is strictly satisfied.

Images