A type-2 fuzzy confidence self-evolution energy storage control method driven by operation deviation

Abstract

This application relates to a Type-2 fuzzy confidence self-evolving energy storage control method driven by operational deviation. This method constructs an uncertainty vector based on real-time and predicted data of the new energy storage system. Based on the feasible control domain and objective function of the uncertainty vector at any given scheduling time, it calculates the optimal control command within the corresponding feasible control domain. The optimal control command is executed, and the control error between the operating result and the corresponding operating objective is calculated. The operational deviation between the uncertainty vector and the corresponding nominal value is also calculated. The comprehensive confidence level is updated based on the control error. Based on the operational deviation correction weight set, the comprehensive confidence level and the corrected weight set are returned to the optimal control command calculation step to calculate the optimal control command for the next scheduling time. This energy storage control method achieves long-term adaptive evolution of the energy storage control strategy, significantly improving the system's economy, robustness, and safety under high uncertainty environments.

Description

Technical Field

[0001] This application relates to the field of energy storage control technology, and in particular to a Type-2 fuzzy confidence self-evolutionary energy storage control method driven by operational deviation. Background Technology

[0002] Existing energy storage control methods face the following technical challenges:

[0003] 1. Existing Type-2 fuzzy control generally adopts static membership functions, whose upper and lower membership intervals remain fixed during operation. Therefore, it is impossible to adjust the shape of the membership function online according to the operation deviation, which leads to the decay of the model's expressive ability when the uncertainty of new energy operation changes over time.

[0004] 2. Existing energy storage control strategies generally adopt offline parameters or empirical rules, which cannot automatically adjust control parameters according to operational deviations and cannot achieve continuous adaptive optimization, resulting in performance degradation under long-term operation.

[0005] 3. Existing technologies lack a self-correction mechanism for control strategies driven by operational feedback, which makes it impossible to dynamically suppress control errors and makes it difficult to ensure the long-term operational safety of energy storage systems.

[0006] Therefore, there is an urgent need for an energy storage control method that can model and evolve uncertainties with operational deviations and has an online self-correcting control strategy. Summary of the Invention

[0007] Based on this, it is necessary to provide a Type-2 fuzzy confidence self-evolutionary energy storage control method driven by operational deviation, including:

[0008] S1: Collect real-time and predicted data of the operation of new energy storage systems within a historical time period, and construct an uncertainty vector based on the error between real-time and predicted data at any scheduling time.

[0009] S2: Construct the corresponding initial fuzzy confidence set based on the uncertainty vector at any scheduling time and calculate the corresponding initial comprehensive confidence. Based on the initial fuzzy confidence set, real-time data, and initial comprehensive confidence, construct the feasible control domain and objective function for the corresponding scheduling time, and calculate the optimal control command of the objective function within the corresponding feasible control domain.

[0010] S3: Apply the optimal control command at any scheduling moment to the new energy storage system and record the corresponding operation results; calculate the control error between the operation results and the corresponding operation target, and calculate the operation deviation between the uncertainty vector and the corresponding nominal value;

[0011] S4: Update the confidence interval width of the initial fuzzy confidence set based on the running deviation, and update the initial comprehensive confidence at the corresponding scheduling time based on the updated confidence interval width of the fuzzy confidence set to obtain the comprehensive confidence.

[0012] S5: Construct a self-correcting cost function for the weights based on the control error and the weight set in the objective function. Calculate the gradient of the self-correcting cost function with respect to the weight set. Correct the weight set in the objective function based on the gradient. Return the combined confidence and corrected weight set to S2. Calculate the optimal control command for the next scheduling time.

[0013] Beneficial effects: This method constructs an uncertainty vector based on real-time and predicted data of the new energy storage system. Based on the feasible control domain and objective function of the uncertainty vector at any scheduling time, it calculates the optimal control command within the corresponding feasible control domain. The optimal control command is executed, and the control error between the operating result and the corresponding operating objective is calculated. The operating deviation between the uncertainty vector and the corresponding nominal value is also calculated. The comprehensive confidence level is updated based on the control error, and the weight set is corrected based on the operating deviation. The comprehensive confidence level and the corrected weight set are then returned to the optimal control command calculation step to calculate the optimal control command for the next scheduling time. This energy storage control method achieves long-term adaptive evolution of the energy storage control strategy, significantly improving the system's economy, robustness, and safety under high uncertainty environments. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the deviation-driven Type-2 fuzzy confidence self-evolutionary energy storage control method in the embodiments of this application. Detailed Implementation

[0016] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application are described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0017] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0018] like Figure 1 As shown, this embodiment provides a Type-2 fuzzy confidence self-evolving energy storage control method driven by operational deviation. This method can be applied to scenarios such as energy storage dispatch control, operational deviation correction, and uncertainty adaptive management under power system operation conditions involving new energy sources. The energy storage control method includes:

[0019] S1: Historical time period data collection The system collects real-time and predicted data of the internal new energy storage system and constructs an uncertainty vector based on the error between the real-time and predicted data at any scheduling time.

[0020] Specifically, the real-time data includes actual renewable energy output, actual load power, real-time electricity price, and current energy storage status of charge.

[0021] The forecast data includes renewable energy forecast output, load forecast power, and electricity price forecast.

[0022] For any given scheduling time, calculate the error between the actual output of renewable energy and the predicted output of renewable energy, the error between the actual load power and the predicted load power, and the error between the real-time electricity price and the predicted electricity price. Use each type of error as a different uncertainty vector for the corresponding scheduling time.

[0023] Furthermore, before constructing the uncertainty vector, the collected real-time data is preprocessed to reduce the impact of measurement noise, communication errors, and abnormal sampling on the accuracy of energy storage control; the preprocessing includes at least one or more of the following: noise reduction, outlier removal, and time alignment.

[0024] In this embodiment, the preprocessing includes:

[0025] Noise reduction processing: The actual output of renewable energy, actual load power, and real-time electricity price are smoothed by using moving average filtering, exponential weighted filtering, or Kalman filtering methods.

[0026] Outlier handling: Based on threshold discrimination or statistical distribution characteristics, real-time data that deviates significantly from the normal operating range is removed or corrected;

[0027] Time alignment processing: Time synchronization and interpolation compensation are performed for data with different sampling periods.

[0028] After completing the above preprocessing, an uncertainty vector is constructed based on the error between the predicted data and the preprocessed real-time data.

[0029] S2: Construct the corresponding initial fuzzy confidence set based on the uncertainty vector at any scheduling time and calculate the corresponding initial comprehensive confidence. Based on the initial fuzzy confidence set, real-time data, and initial comprehensive confidence, construct the feasible control domain and objective function at the corresponding scheduling time, and calculate the optimal control instruction of the objective function within the corresponding feasible control domain.

[0030] Specifically, the steps include:

[0031] S2.1: Calculate the mean and standard deviation of the uncertainty vector at any scheduling time. Based on the mean, standard deviation, and preset initial confidence contraction coefficient, calculate the upper membership function and lower membership function of the corresponding uncertainty vector. Based on the upper membership function and lower membership function, calculate the confidence index of the corresponding uncertainty vector. Based on the confidence index, calculate the initial comprehensive confidence of the corresponding scheduling time.

[0032] Furthermore, the formulas for calculating the mean and standard deviation are:

[0033] ;

[0034] ;

[0035] in, Indicates the first The scheduling time of the first... The mean of a vector of uncertainties. Indicates the first The scheduling time of the first... The standard deviation of an uncertainty vector Indicates a historical period. Indicates the first The scheduling time of the first... An uncertain vector.

[0036] Furthermore, the initial composite confidence level is calculated as follows:

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] in, Indicates the first The initial overall confidence level at each scheduling time. Indicates the first The number of uncertainty vectors at each scheduling moment Indicates the first Preset weights for an uncertain vector, Indicates the first The scheduling time of the first... A confidence index for an uncertainty vector. Represents the normalization constant. Indicates the first The scheduling time of the first... Membership function of an uncertain vector. Indicates the first The scheduling time of the first... The membership function of an uncertain vector Indicates the first An uncertain vector, express The range of values, Indicates the first The scheduling time of the first... The mean of a vector of uncertainties. Indicates the first The scheduling time of the first... The standard deviation of an uncertainty vector Indicates the first The scheduling time of the first... The initial confidence contraction coefficient of an uncertainty vector. This is used to indicate the degree of contraction of the uncertainty range at a higher confidence level.

[0042] S2.2: Define the corresponding energy storage control strategy, wherein the energy storage control strategy is the charging and discharging power when the real-time data and weight set at any scheduling time conform to a Poisson distribution. The energy storage control strategy is expressed as: ;in, Indicates the first The charging and discharging power at each scheduling moment. Represents the Poisson distribution. Indicates the first Real-time data at each scheduling moment. Indicates the first The set of weights for each scheduling moment.

[0043] S2.3: Combine the mean, standard deviation and initial confidence shrinkage coefficient of the uncertainty vector at any scheduling time to construct the corresponding initial fuzzy confidence set.

[0044] During the initialization phase of the method, the weight parameters are set to initial values. Through the above process, the initial fuzzy confidence set, the initial confidence index, and the initial comprehensive confidence are obtained. These results will serve as the basis for subsequent control command generation and self-evolution and self-correction.

[0045] S2.4: For any scheduling time, construct the feasible control domain for the corresponding scheduling time based on the initial fuzzy confidence set or the updated fuzzy confidence set; construct the objective function for the corresponding scheduling time based on real-time data, energy storage control strategy, initial comprehensive confidence or updated comprehensive confidence.

[0046] Furthermore, the expression for the feasible control domain is:

[0047] ;

[0048] in, Indicates the first The feasible control domain at each scheduling moment This represents the constraint function, which includes the energy storage state of charge constraint ( , This represents the minimum state of charge of the energy storage. This indicates the maximum state of charge of the energy storage. Indicates the first Energy storage state of charge at each scheduling moment), charging and discharging power constraints ( , Indicates the maximum charging power. Indicates the maximum discharge power. Indicates the first The charging and discharging power at each scheduling moment), and safety constraints (including voltage safety constraints, current safety constraints, and line power safety constraints). Indicates the first Real-time data at each scheduling moment. Indicates the first The initial or updated fuzzy confidence set at each scheduling time. This represents the control decision variable (control input). This represents the perturbation vector (or uncertainty), whose value range comes from the fuzzy confidence set. .

[0049] The expression for the objective function is:

[0050] ;

[0051] in, Indicates the first The objective function at each scheduling time, Indicates the first Real-time electricity price at each scheduling moment. Indicates the first Energy storage control strategy at each scheduling moment Indicates the first weight. Indicates the expected risk cost. This represents the risk cost function (used to describe risks such as voltage over-limit and over-discharge of energy storage state of charge). Indicates the second weight. Indicates the first The initial overall confidence level at each scheduling moment.

[0052] In this embodiment, the risk costs include voltage exceeding limits, SOC exceeding limits, and temperature exceeding limits. The expression for the risk cost function is:

[0053] ;

[0054] in, Represents the risk cost function. Indicates the first The system state vector at each scheduling moment Indicates the first Energy storage control strategy at each scheduling moment This represents the perturbation vector. Indicates disturbance Next Bus voltage at each scheduling moment Indicates the voltage reference value. Indicates the allowable voltage deviation threshold. Indicates disturbance Next The state of charge of energy storage at each scheduling moment This represents the minimum state of charge of the energy storage. This indicates the maximum state of charge of the energy storage. Indicates disturbance Next Battery temperature at each scheduling moment, Indicates the upper limit of safe temperature. , , These represent the penalty weighting coefficients for voltage over-limit risk, SOC over-limit risk, and temperature over-limit risk, respectively.

[0055] S2.5: Calculate the optimal control command for the objective function within the corresponding feasible control domain. The calculation formula is:

[0056] ;

[0057] in, Indicates the first The optimal control command at each scheduling time. Indicates the first The feasible control domain at each scheduling moment Indicates the first Each scheduling time is based on control strategy parameters. The constructed objective function.

[0058] S3: Apply the optimal control command at any scheduling time to the new energy storage system and record the corresponding operation results; calculate the control error between the operation results and the corresponding operation target, and calculate the operation deviation between the uncertainty vector and the corresponding nominal value.

[0059] Specifically, the steps include:

[0060] Apply the optimal control command at any given scheduling time to the new energy storage system. If the value of the optimal control command is greater than 0, it is considered that the energy storage is discharging, and the discharge power is set to the value of the optimal control command, while the charging power is set to 0. If the value of the optimal control command is less than 0, it is considered that the energy storage is charging, and the charging power is set to the negative value of the optimal control command, while the discharge power is set to 0. If the value of the optimal control command is 0, it is considered that the energy storage is in standby mode, and both the charging power and the discharge power are set to 0, while maintaining the energy storage working mode at the current scheduling time.

[0061] Energy storage control is performed based on the set charging and discharging power, and the corresponding operating results are recorded. The operating results include the energy storage state of charge, bus voltage, grid power, and economic benefits at the next scheduling time.

[0062] The first difference between the operating results and the corresponding operating objectives (including the target energy storage state of charge, voltage reference value, grid power reference value, and expected economic benefits) is calculated and used as the control error.

[0063] Calculate the second difference between the uncertainty vector and the corresponding nominal value, and use the absolute value of the second difference as the operating deviation.

[0064] Furthermore, the state of charge (SBC) of the energy storage at the next scheduling moment is obtained by updating the SBC of the current scheduling moment using the set charging power, discharging power, and energy storage efficiency of the new energy storage system. The update formula is as follows:

[0065] ;

[0066] in, Indicates the first The state of charge of energy storage at each scheduling moment Indicates the first The state of charge of energy storage at each scheduling moment Indicates charging efficiency. Indicates the number after setting The charging power at each scheduling moment. Indicates the time step. Indicates discharge efficiency. Indicates the number after setting The discharge power at each scheduling moment.

[0067] S4: Update the confidence interval width of the initial fuzzy confidence set based on the running deviation, and update the initial comprehensive confidence level at the corresponding scheduling time based on the updated confidence interval width of the fuzzy confidence set to obtain the comprehensive confidence level.

[0068] Specifically, the steps include:

[0069] The confidence interval width of the initial fuzzy confidence set is updated based on the operational deviation, and the update formula is as follows:

[0070] ;

[0071] in, Indicates the first The scheduling time of the first... Each uncertainty vector corresponds to the confidence interval width of the updated fuzzy confidence set. Indicates the first The scheduling time of the first... Each uncertainty vector corresponds to the confidence interval width of the updated fuzzy confidence set. Indicates the first The learning rate corresponding to each uncertainty vector , Indicates the first The scheduling time of the first... The running deviation corresponding to each uncertainty vector Indicates the first The deviation threshold corresponding to each uncertainty vector; when Greater than If the confidence interval is positive, it indicates that the current characterization of uncertainty is too optimistic and the confidence interval needs to be appropriately widened; otherwise, it indicates that the current confidence interval is too conservative and can be appropriately narrowed to improve the model's identification accuracy.

[0072] In this embodiment, to avoid unbounded expansion or excessive contraction, the implementation can be modified accordingly. Set upper and lower limit constraints , Indicates the first The lower bound of the width of an uncertain vector. Indicates the first An upper limit on the width of an uncertain vector is set to ensure the stability of the self-evolutionary process.

[0073] Based on the confidence index of the uncertainty vector and the corresponding expected confidence level, the initial confidence contraction coefficient is updated using the following formula:

[0074] ;

[0075] in, Indicates the first The scheduling time of the first... The confidence contraction coefficient of an uncertainty vector. Indicates the first The scheduling time of the first... The initial confidence contraction coefficient of an uncertainty vector. Indicates the first The confidence contraction adjustment step size corresponding to each uncertainty vector. Indicates the first The expected confidence level of an uncertainty vector. Indicates the first The scheduling time of the first... Confidence index for an uncertainty vector;

[0076] When the confidence index is lower than the expected confidence level The size of the membership function will increase, thereby imposing a stronger contraction constraint on its shape at the next scheduling time, improving the model's sensitivity to extreme deviations. Conversely, the contraction granularity will be appropriately reduced to avoid the model becoming overly conservative.

[0077] Based on the updated fuzzy confidence set, the confidence interval width and confidence contraction coefficient are used to update the initial comprehensive confidence at the corresponding scheduling time, thus obtaining the comprehensive confidence.

[0078] Furthermore, based on the confidence interval width and confidence contraction coefficient of the updated fuzzy confidence set, the upper and lower membership functions of Type-2 for the next scheduling time are constructed. In this embodiment, a Gaussian Type-2 fuzzy set is used as an example, and its upper and lower membership functions are expressed as follows:

[0079] ;

[0080] ;

[0081] in, Indicates the first The scheduling time of the first... Membership function of an uncertain vector. Indicates the first The scheduling time of the first... The membership function of an uncertain vector Indicates the first An uncertain vector, Indicates the first The scheduling time of the first... The mean of a vector of uncertainties. Indicates the first The scheduling time of the first... The standard deviation of an uncertainty vector Indicates the first The scheduling time of the first... The confidence contraction coefficient of an uncertainty vector.

[0082] The width and shape of the uncertainty boundary band formed between the two membership function curves are directly determined by the operating deviation, the confidence interval width, and the confidence contraction coefficient, thus realizing the characteristic of uncertainty itself dynamically evolving with operation.

[0083] Furthermore, the overall confidence level is calculated based on the reconstructed upper and lower membership functions. The formula for calculating the overall confidence level is as follows:

[0084] ;

[0085] ;

[0086] in, Indicates the first The overall confidence level at each scheduling moment. Indicates the first The number of uncertainty vectors at each scheduling moment Indicates the first The scheduling time of the first... The preset weights of the uncertainty vectors can be adjusted online according to the importance or sensitivity of the variables. Indicates the first The scheduling time of the first... A confidence index for an uncertainty vector. Represents the normalization constant. Indicates the first The scheduling time of the first... Membership function of an uncertain vector. Indicates the first The scheduling time of the first... The membership function of an uncertain vector Indicates the first An uncertain vector, express The range of values, Indicates the first The scheduling time of the first... The mean of a vector of uncertainties. Indicates the first The scheduling time of the first... The standard deviation of an uncertainty vector.

[0087] Through the above process, the shape and confidence interval width of the Type-2 fuzzy confidence set are automatically adjusted based on operational deviations within each scheduling cycle, thereby enabling the uncertainty model to possess adaptive learning and self-evolution capabilities. This operational deviation-driven Type-2 fuzzy confidence self-evolution mechanism, which differs from traditional static fuzzy modeling and fixed Type-2 range settings, is the key innovation of this invention in uncertainty representation, laying a dynamic and reliable foundation for subsequent self-correction of control strategies.

[0088] S5: Construct a self-correcting cost function for the weights based on the control error and the weight set in the objective function. Calculate the gradient of the self-correcting cost function with respect to the weight set. Correct the weight set in the objective function based on the gradient. Return the combined confidence and corrected weight set to S2. Calculate the optimal control command for the next scheduling time.

[0089] Specifically, the steps include:

[0090] A self-calibrating cost function is constructed based on the control error and the weight set in the objective function. The expression for the self-calibrating cost function is as follows:

[0091] + ;

[0092] in, Indicates the first The self-correcting cost function of the weight set at each scheduling time. Indicates the first Control error at each scheduling moment Describing the L2 norm, Indicates the first The overall confidence level at each scheduling moment. Indicates the first The weight set for each scheduling time, including the real-time electricity price, the first weight, and the second weight for the corresponding scheduling time. express The corresponding reference value, This indicates the weight of the control error term. This indicates the weight of the confidence penalty term. This indicates the weight of the parameter deviating from the penalty term. Indicates the weight of the regularization term; This represents a regularization term used to suppress excessive fluctuations in the weight set during online self-calibration, thereby enhancing the stability and convergence of the parameter update process.

[0093] Calculate the gradient of the self-correcting cost function with respect to the weight set, and correct the weight set in the objective function based on the gradient. The correction formula is as follows:

[0094] ;

[0095] ;

[0096] in, Indicates the first The set of weights for each scheduling moment. Indicates the learning rate parameter. , This represents the gradient of the self-correcting cost function with respect to the weight set. This represents the partial derivative; in practical implementation, the chain rule can be used to reduce the sensitivity of the control error to the weight set. The indirect impact of confidence on the weight set is incorporated into the gradient estimation and calculated using numerical approximation or online identification methods.

[0097] In this embodiment, it is also possible to The algorithm is decomposed into economic weight sub-vectors, risk weight sub-vectors, and fuzzy rule parameter sub-vectors, and updated using different learning rates and constraints, making the trade-off between economic and safety objectives more flexible and controllable.

[0098] The weight set of the comprehensive confidence and correction is returned to S2 to calculate the optimal control command at the next scheduling time, so as to realize the integrated closed loop of "fuzzy confidence evolution - strategy parameter adjustment - control command regeneration".

[0099] Based on the "fuzzy confidence self-evolution," a control strategy self-correction mechanism driven jointly by operating error and comprehensive confidence is introduced to achieve a closed-loop evolution of "strategy—execution—deviation—confidence—re-strategy." This mechanism enables the energy storage control strategy parameters to be adjusted in real time according to the system operating state and uncertainty evolution, significantly enhancing the adaptability and robustness of the control strategy.

[0100] The method of this invention is applied to a microgrid energy storage system in an industrial park. The system includes a photovoltaic power generation unit, a load unit, and a centralized energy storage unit. The photovoltaic installed capacity is 5MW, the rated capacity of the energy storage system is 2MWh, the maximum charging and discharging power is 1MW, and the dispatch time scale is 15 minutes.

[0101] The system collects real-time data on photovoltaic output, load power, electricity price, and energy storage state of charge. Before constructing the uncertainty vector, it performs noise reduction and outlier processing on the real-time data. Based on the proposed operational deviation-driven Type-2 fuzzy confidence self-evolutionary method, it generates energy storage charging and discharging control commands and verifies them over continuous operating cycles.

[0102] The advantages of this invention in reducing SOC overruns and voltage overruns, as well as overall operational economy, are compared with the following: the method of this invention, traditional Type-2 fuzzy control (static membership function), and strategies without self-correction. Data source: 72 hours of continuous operation of the microgrid (288 scheduling moments). The SOC overrun frequency and voltage overrun duration of each control method are shown in Table 1 below, and the overall operational economy of each control method is shown in Table 2 below.

[0103] Table 1. Comparison of SOC overrun frequency and voltage overrun duration for each control method

[0104] Control methods SOC outage frequency (times / 72h) Voltage over-limit duration (min / 72h) Traditional Type-2 fuzzy control 8 35 No self-calibration strategy 12 48 Method of the present invention 2 10

[0105] Table 2. Comprehensive Economic Comparison of Various Control Methods

[0106] Control methods Revenue from electricity sales (RMB) Charge / discharge loss (kWh) Overall net income (RMB) Traditional Type-2 fuzzy control 12800 620 11950 No self-calibration strategy 11500 780 10420 Method of the present invention 13500 450 12820

[0107] As shown in Tables 1 and 2, under the circumstances of fluctuations in new energy output and prediction errors, the method of the present invention can effectively reduce the risks of energy storage state of charge exceeding limits and voltage exceeding limits, while improving the overall economic efficiency of system operation, thus verifying the feasibility and effectiveness of the method in actual engineering scenarios.

[0108] The Type-2 fuzzy confidence self-evolving energy storage control method driven by operational deviation provided in this embodiment has the following beneficial effects:

[0109] 1. A proposed Type-2 fuzzy confidence modeling mechanism that can dynamically evolve with operation: Fuzzy control often uses fixed membership functions, and its fuzzy interval does not dynamically adjust with actual operating deviations, making it difficult to accurately express the "uncertainty of uncertainty itself" in new energy scenarios. This invention is the first to construct a Type-2 fuzzy confidence modeling mechanism driven by operating deviations. It jointly characterizes the uncertainty range and confidence level through upper and lower membership functions, and automatically shrinks or widens the fuzzy interval according to the actual deviation, enabling the fuzzy model to have adaptive evolution capabilities, thereby breaking through the inherent limitations of traditional fuzzy control "static modeling".

[0110] 2. Constructed Confidence-Driven Dynamic Feasible Domain and Objective Function for Energy Storage Strategy: In traditional energy storage scheduling, confidence is only used for risk assessment and rarely directly participates in the construction of control strategies. This invention introduces "confidence" as a dynamic control variable into the feasible domain constraint and strategy objective function of energy storage control for the first time. This allows the feasible domain to automatically scale with the confidence level, thereby realizing a dynamic trade-off mechanism of "prioritizing returns when confidence is high and prioritizing safety when confidence is low," thus achieving real-time coupled optimization of risk and return.

[0111] 3. The proposed closed-loop self-correction mechanism of strategy-execution-deviation-confidence-re-strategy: Most existing energy storage control systems are one-time strategies or offline optimizations, lacking the ability to continuously learn and self-correct during operation. This invention constructs a self-correction cost function driven by operating deviation and confidence index, enabling the energy storage control strategy to automatically adjust parameters according to the deviation after execution and regenerate control commands, realizing a complete closed-loop iterative strategy generation → execution → error feedback → confidence self-evolution → strategy self-correction → new strategy generation.

[0112] In summary, this method introduces operational deviation as a driving force for uncertainty evolution, constructs a Type-2 fuzzy confidence model that can be adjusted in real time according to actual operational deviation, and further combines it with a closed-loop self-correction mechanism for energy storage strategy based on confidence feedback to achieve continuous adaptive evolution of energy storage control strategy under high uncertainty operating environment.

[0113] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0114] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A Type-2 fuzzy confidence self-evolutionary energy storage control method driven by operational deviation, characterized in that, include: S1: Collect real-time and predicted data of the operation of new energy storage systems within a historical time period, and construct an uncertainty vector based on the error between real-time and predicted data at any scheduling time. S2: Construct the corresponding initial fuzzy confidence set based on the uncertainty vector at any scheduling time and calculate the corresponding initial comprehensive confidence. Based on the initial fuzzy confidence set, real-time data, and initial comprehensive confidence, construct the feasible control domain and objective function for the corresponding scheduling time, and calculate the optimal control command of the objective function within the corresponding feasible control domain. S3: Apply the optimal control command at any scheduling moment to the new energy storage system and record the corresponding operation results; calculate the control error between the operation results and the corresponding operation target, and calculate the operation deviation between the uncertainty vector and the corresponding nominal value; S4: Update the confidence interval width of the initial fuzzy confidence set based on the running deviation, and update the initial comprehensive confidence level at the corresponding scheduling time based on the updated confidence interval width of the fuzzy confidence set to obtain the comprehensive confidence level, including: Based on the updated fuzzy confidence set's confidence interval width and confidence contraction coefficient, the Type-2 upper and lower membership functions for the next scheduling time are constructed, respectively expressed as: ； ； in, Indicates the first The scheduling time of the first... Membership function of an uncertain vector. Indicates the first The scheduling time of the first... The membership function of an uncertain vector Indicates the first An uncertain vector, Indicates the first The scheduling time of the first... The mean of a vector of uncertainties. Indicates the first The scheduling time of the first... The standard deviation of an uncertainty vector Indicates the first The scheduling time of the first... The confidence contraction coefficient of an uncertainty vector. Indicates the first The scheduling time of the first... Each uncertainty vector corresponds to a confidence interval width of the updated fuzzy confidence set; The overall confidence level is calculated based on the reconstructed upper and lower membership functions. The formula for calculating the overall confidence level is as follows: ； ； in, Indicates the first The overall confidence level at each scheduling moment. Indicates the first The number of uncertainty vectors at each scheduling moment Indicates the first The scheduling time of the first... The preset weights of the uncertainty vectors can be adjusted online according to the importance or sensitivity of the variables. Indicates the first The scheduling time of the first... A confidence index for an uncertainty vector. Represents the normalization constant. express The range of values ​​for ; S5: Construct a self-correcting cost function for the weights based on the control error and the weight set in the objective function. Calculate the gradient of the self-correcting cost function with respect to the weight set. Correct the weight set in the objective function based on the gradient. Return the combined confidence and corrected weight set to S2. Calculate the optimal control command for the next scheduling time.

2. The method according to claim 1, characterized in that, In S1, The real-time data includes actual renewable energy output, actual load power, real-time electricity price, and current energy storage status of charge. The forecast data includes renewable energy forecast output, load forecast power, and electricity price forecast. For any given scheduling time, calculate the error between the actual output of renewable energy and the predicted output of renewable energy, the error between the actual load power and the predicted load power, and the error between the real-time electricity price and the predicted electricity price. Use each type of error as a different uncertainty vector for the corresponding scheduling time.

3. The method according to claim 2, characterized in that, S2 include: S2.1: Calculate the mean and standard deviation of the uncertainty vector at any scheduling time. Based on the mean, standard deviation and preset initial confidence contraction coefficient, calculate the upper membership function and lower membership function of the corresponding uncertainty vector. Based on the upper membership function and lower membership function, calculate the confidence index of the corresponding uncertainty vector. Based on the confidence index, calculate the initial comprehensive confidence of the corresponding scheduling time. S2.2: Define the corresponding energy storage control strategy, wherein the energy storage control strategy is the charging and discharging power when the real-time data and weight set at any scheduling time conform to the Poisson distribution; S2.3: Combine the mean, standard deviation and initial confidence shrinkage coefficient of the uncertainty vector at any scheduling time to construct the corresponding initial fuzzy confidence set; S2.4: For any scheduling time, construct the feasible control domain for the corresponding scheduling time based on the initial fuzzy confidence set or the updated fuzzy confidence set; construct the objective function for the corresponding scheduling time based on real-time data, energy storage control strategy, initial comprehensive confidence or updated comprehensive confidence. S2.5: Calculate the optimal control command for the objective function within the corresponding feasible control domain. The calculation formula is: ； in, Indicates the first The optimal control command at each scheduling time. Indicates the first The feasible control domain at each scheduling moment Indicates the first Each scheduling time is based on control strategy parameters. The constructed objective function.

4. The method according to claim 3, characterized in that, The formula for calculating the initial overall confidence level is: ； ； ； ； in, Indicates the first The initial overall confidence level at each scheduling time. Indicates the first The number of uncertainty vectors at each scheduling moment Indicates the first Preset weights for an uncertain vector, Indicates the first The scheduling time of the first... A confidence index for an uncertainty vector. Represents the normalization constant. Indicates the first The scheduling time of the first... Membership function of an uncertain vector. Indicates the first The scheduling time of the first... The membership function of an uncertain vector Indicates the first An uncertain vector, express The range of values, Indicates the first The scheduling time of the first... The mean of a vector of uncertainties. Indicates the first The scheduling time of the first... The standard deviation of an uncertainty vector Indicates the first The scheduling time of the first... The initial confidence contraction coefficient of an uncertainty vector.

5. The method according to claim 3, characterized in that, In S2.4, the expression for the feasible control domain is: ； in, Indicates the first The feasible control domain at each scheduling moment This represents the constraint functions, which include energy storage state of charge constraints, charge / discharge power constraints, and safety constraints. Indicates the first Real-time data at each scheduling moment. Indicates the first The initial or updated fuzzy confidence set at each scheduling time. Indicates the control decision variable. Represents the perturbation vector; The expression for the objective function is: ； in, Indicates the first The objective function at each scheduling time, Indicates the first Real-time electricity price at each scheduling moment. Indicates the first Energy storage control strategy at each scheduling moment Indicates the first weight. Indicates the expected risk cost. Represents the risk cost function. Indicates the second weight. Indicates the first The initial overall confidence level at each scheduling moment.

6. The method according to claim 3, characterized in that, S3 include: Apply the optimal control command at any given scheduling time to the new energy storage system. If the value of the optimal control command is greater than 0, it is considered that the energy storage is discharging, and the discharge power is set to the value of the optimal control command, while the charging power is set to 0. If the value of the optimal control command is less than 0, it is considered that the energy storage is charging, and the charging power is set to the negative value of the optimal control command, while the discharge power is set to 0. If the value of the optimal control command is 0, it is considered that the energy storage is in standby mode, and both the charging power and the discharge power are set to 0, while maintaining the energy storage working mode at the current scheduling time. Energy storage control is performed based on the set charging and discharging power, and the corresponding operating results are recorded. The operating results include the energy storage state of charge, bus voltage, grid power, and economic benefits at the next scheduling time. Calculate the first difference between the running result and the corresponding running target, and use the first difference as the control error; Calculate the second difference between the uncertainty vector and the corresponding nominal value, and use the absolute value of the second difference as the operating deviation.

7. The method according to claim 6, characterized in that, The state of charge (SBC) of the energy storage system at the next scheduling moment is obtained by updating the SBC of the current scheduling moment using the set charging power, discharging power, and energy storage efficiency of the new energy storage system. The update formula is as follows: ； in, Indicates the first The state of charge of energy storage at each scheduling moment Indicates the first The state of charge of energy storage at each scheduling moment Indicates charging efficiency. Indicates the number after setting The charging power at each scheduling moment, Indicates the time step. Indicates discharge efficiency. Indicates the number after setting The discharge power at each scheduling moment.

8. The method according to claim 4, characterized in that, S4 include: The confidence interval width of the initial fuzzy confidence set is updated based on the operational deviation, and the update formula is as follows: ； in, Indicates the first The scheduling time of the first... Each uncertainty vector corresponds to the confidence interval width of the updated fuzzy confidence set. Indicates the first The scheduling time of the first... Each uncertainty vector corresponds to the confidence interval width of the updated fuzzy confidence set. Indicates the first The learning rate corresponding to each uncertainty vector Indicates the first The scheduling time of the first... The running deviation corresponding to each uncertainty vector Indicates the first The deviation threshold corresponding to each uncertainty vector; Based on the confidence index of the uncertainty vector and the corresponding expected confidence level, the initial confidence contraction coefficient is updated using the following formula: ； in, Indicates the first The scheduling time of the first... The confidence contraction coefficient of an uncertainty vector. Indicates the first The scheduling time of the first... The initial confidence contraction coefficient of an uncertainty vector. Indicates the first The confidence contraction adjustment step size corresponding to each uncertainty vector. Indicates the first The expected confidence level of an uncertainty vector. Indicates the first The scheduling time of the first... Confidence index for an uncertainty vector; Based on the updated fuzzy confidence set, the confidence interval width and confidence contraction coefficient are used to update the initial comprehensive confidence at the corresponding scheduling time, thus obtaining the comprehensive confidence.

9. The method according to claim 5, characterized in that, S5 include: A self-calibrating cost function is constructed based on the control error and the weight set in the objective function. The expression for the self-calibrating cost function is as follows: ＋ ； in, Indicates the first The self-correcting cost function of the weight set at each scheduling time. Indicates the first Control error at each scheduling moment Represents the L2 norm. Indicates the first The overall confidence level at each scheduling moment. Indicates the first The weight set for each scheduling time, including the real-time electricity price, the first weight, and the second weight for the corresponding scheduling time. express The corresponding reference value, This indicates the weight of the control error term. This indicates the weight of the confidence penalty term. This indicates the weight of the parameter deviating from the penalty term. Indicates the weight of the regularization term; Calculate the gradient of the self-correcting cost function with respect to the weight set, and correct the weight set in the objective function based on the gradient. The correction formula is as follows: ； ； in, Indicates the first The set of weights for each scheduling moment. Indicates the learning rate parameter. This represents the gradient of the self-correcting cost function with respect to the weight set. Indicates partial derivative; Return the combined confidence level and the corrected weight set to S2 to calculate the optimal control command for the next scheduling time.

Images