Extreme weather new energy power grid active early warning and collaborative control method

Abstract

The application relates to the technical field of power system security defense and new energy power grid operation control, and discloses a new energy power grid active early warning and cooperative control method under extreme weather. An extreme value shape-preserving prediction method is used to model the tail of wind power prediction residual, to calculate the lower limit of the extreme value of wind power prediction error, to construct a wind power output adaptive safety interval, to quantitatively classify the imbalance risk in combination with the power balance relationship of the power grid section, to take the risk classification result as an active emergency control trigger condition, to coordinate multiple types of regulation resources such as energy storage, photovoltaic, conventional units, flexible load and demand response resources to implement hierarchical cooperative control, and to realize the rapid regulation and control of the imbalance risk of the power grid section under the impact of extreme weather. The application can improve the accuracy of wind power sudden change risk identification and the foresight of active emergency control triggering, and enhance the safe operation guarantee capability of the power grid section under extreme weather conditions.

Description

Technical Field

[0001] This invention relates to the field of power system security defense and new energy grid operation control technology, specifically to a method for proactive early warning and collaborative control of new energy grids under extreme weather conditions. Background Technology

[0002] With the continuous advancement of new power system construction, the penetration rate of new energy sources such as wind power and photovoltaics in the power grid is constantly increasing, and the impact of extreme weather on power source-load balance and cross-sectional safety is becoming increasingly prominent. Extreme weather events such as cold waves, typhoons, severe convection, rainstorms, blizzards, and high temperatures can simultaneously cause sudden drops in wind power output, short-term load fluctuations, attenuation of available photovoltaic output, and limitations in the regulation of conventional units, thereby inducing a rapid expansion of cross-sectional power deficit and creating a short-term, high-risk common-cause failure risk. Against this backdrop, research on proactive early warning and proactive emergency control of cross-sectional areas in response to extreme weather impacts has become an important technical requirement for improving the power system's ability to defend against extreme disasters.

[0003] In existing technologies, new energy power prediction methods mostly focus on the point prediction accuracy or general confidence interval construction under normal operating conditions, and do not adequately consider the heavy tail, skewness, and abrupt change characteristics of wind power prediction errors under extreme weather impacts. When the actual wind power output is significantly lower than the predicted value, the traditional mean prediction results and symmetrical interval boundaries are difficult to accurately reflect the extreme deviation of the lower tail, which can easily lead to a lag in the identification of cross-sectional risks, making it difficult to support active emergency control in short-term high-risk common-cause failure scenarios.

[0004] On the other hand, existing uncertainty quantification methods typically remain at the prediction level, lacking close integration with cross-sectional power balance relationships, risk classification rules, and multi-resource collaborative control. Especially under extreme weather conditions, using fixed thresholds or single-limit-over-limit triggering control not only fails to reflect the dynamics and foresight of risk evolution but also hinders the orderly deployment of collaborative response boundaries across multiple resources, including energy storage, photovoltaics, conventional units, flexible loads, and demand response. Therefore, a technical solution is urgently needed that can directly map the quantification results of wind power downstream risks to cross-sectional risk levels and further trigger proactive emergency control of multiple resources.

[0005] To address this, the present invention proposes a proactive early warning and collaborative control method for new energy power grids under extreme weather conditions, in order to improve the accuracy of cross-sectional operation risk identification and the timeliness of control triggering under extreme weather conditions. Summary of the Invention

[0006] The purpose of this invention is to address the problems of difficulty in effectively characterizing the tail risk of wind power prediction errors under extreme weather conditions, the lack of unified quantitative basis for grid section imbalance risk, and the unclear criteria for triggering active emergency control. It proposes a proactive early warning and collaborative control method for new energy power grids under extreme weather conditions. This method, targeting short-term, high-risk common-cause failure scenarios, organically combines tail extreme value modeling of wind power prediction errors, section imbalance risk classification, and multi-resource active emergency control triggering. This improves the accuracy of grid section operation risk identification and the timeliness of control response under extreme weather conditions.

[0007] To achieve the above objectives, this invention proposes a method for proactive early warning and coordinated control of new energy power grids under extreme weather conditions, comprising the following steps: (1) Obtain historical wind power operation data, wind power prediction data and prediction residual sequence of the study area under extreme weather conditions, construct wind power prediction error sample set, and extract tail samples that characterize the extreme under-biased characteristics of wind power output.

[0008] (2) Based on the tail sample, construct the extreme value conformal prediction boundary of wind power prediction error, form the safe range of wind power output, and construct the imbalance risk index and risk level in combination with the power balance relationship of the grid section, and realize the active early warning of grid section imbalance risk under extreme weather impact in combination with the risk level.

[0009] (3) Trigger multi-resource active emergency coordinated control based on the cross-sectional imbalance risk level, coordinate adjustment resources such as energy storage, photovoltaic, conventional units, flexible loads and demand response resources to correct the cross-sectional imbalance risk, and output active emergency control results.

[0010] Preferably, in step (1), the wind power time series is represented as a superposition of a conditional mean term and a random disturbance term, and its expression is: in, For a moment The actual power output of wind power For a moment The conditional mean term, For a moment Conditional fluctuation term, For a moment The standardized random perturbation term.

[0011] Preferably, in step (1), a standardized prediction residual sequence is constructed based on the wind power prediction results, and the conditional volatility is estimated using the exponentially weighted moving average method, the expression of which is: in, For a moment Standardized prediction residuals For a moment The predicted wind power output For a moment The conditional volatility estimate, For exponential weighting coefficients, For a moment The original prediction error.

[0012] Preferably, in step (1), to address the downside risk caused by the actual wind power output being lower than the predicted value, a mirror transformation is performed on the standardized prediction residual sequence, and a tail sample set is extracted from the mirror tail samples exceeding a high threshold; the high threshold is determined based on the high quantile of the empirical distribution of the mirror tail samples, and its expression is: Preferably, The value ranges from 0.90 to 0.95.

[0013] Preferably, in step (2), the tail sample is fitted using a generalized Pareto distribution, and the tail quantiles of the wind power prediction error at the confidence level are calculated, with the following expression: in, Confidence level The estimated tail fractions, For high threshold, is the scale parameter of the generalized Pareto distribution. The shape parameter of the generalized Pareto distribution. To calibrate the total number of samples, For exceeding the threshold The number of tail samples, This represents the failure probability.

[0014] Preferably, in step (2), the lower bound of wind power prediction and the adaptive safety interval of wind power output are constructed based on the tail fractional estimate, and their expressions are as follows: in, For a moment The lower bound of wind power prediction For a moment The adaptive safety range of wind power output For a moment The predicted wind power output For a moment The conditional volatility estimate.

[0015] Preferably, in step (2), the cross-sectional imbalance value and the risk value of positive cross-sectional imbalance are constructed by combining the cross-sectional power balance relationship, and their expressions are as follows: in, For a moment The cross-sectional power loss measurement For a moment The risk value of positive imbalance in the cross section, For a moment The planned power output of the cross section For a moment The photovoltaic power available for use, For a moment The conventional unit's adjustable output For a moment The load demand power.

[0016] Preferably, in step (2), the risk level criterion is constructed using the relative change in wind power, maximum gradeability, average output level, and cross-sectional positive imbalance risk value, and their expressions are as follows: Based on the relative change in wind power, maximum gradeability, average output level, and cross-sectional positive imbalance risk value, the cross-sectional imbalance risk is divided into Level I, Level II, Level III, and Level IV.

[0017] Preferably, in step (3), the active emergency control trigger level and the multi-resource collaborative control power requirement are determined according to the risk level, and their expressions are as follows: in, For a moment Active emergency control trigger level, For a moment The risk level of cross-sectional imbalance For a moment Multi-resource collaborative control of total power demand. For a moment The risk value of positive imbalance in the cross section.

[0018] Preferably, in step (3), energy storage, photovoltaic, conventional units, flexible loads, and demand response resources are subjected to hierarchical coordinated control based on the total demand power of multi-resource coordinated control. The control allocation expression is as follows: Through the above-described scheme, this invention addresses the problems of difficulty in effectively characterizing the tail risk of wind power prediction errors under extreme weather conditions, lack of unified quantitative basis for grid cross-sectional imbalance risk, and unclear criteria for triggering active emergency control. It establishes a complete technical chain from tail-end extreme value modeling of wind power prediction errors to cross-sectional risk quantification, and then to multi-resource active emergency control triggering and coordinated allocation. First, by introducing an extreme value conformal prediction boundary oriented towards tail risk, the lower-tail risk of wind power prediction errors under extreme weather conditions is characterized, improving the reliability of high-risk condition identification. Second, by combining cross-sectional power balance relationships and wind power fluctuation indicators to construct risk level criteria, a forward-looking quantitative assessment of cross-sectional imbalance risk under short-term high-risk common-cause failure scenarios is achieved. Finally, by directly mapping risk levels to active emergency control trigger signals and multi-resource coordinated control requirements, the linkage response capability and multi-resource coordinated control effect of energy storage, conventional units, photovoltaics, flexible loads, and demand response resources are enhanced. Detailed Implementation

[0019] The following provides a detailed description of specific embodiments of the present invention. It should be understood that the specific embodiments described herein are for illustrative and explanatory purposes only and are not intended to limit the scope of the invention.

[0020] Example The present invention provides a method for proactive early warning and coordinated control of new energy power grids under extreme weather conditions, comprising: (1) Obtain historical actual wind power sequence, wind power prediction sequence, planned power output sequence, load power sequence, and operational boundary data of energy storage, conventional units, photovoltaic, flexible loads and demand response resources under extreme weather conditions at the research section; establish an extreme value conformal prediction model based on the tail sample of wind power prediction error to obtain the lower bound of wind power prediction and the adaptive safe range of wind power output. (2) Based on the wind power prediction lower bound, the planned power output sequence of the section, the load power sequence and other adjustable power output, construct the section power imbalance value and the section positive imbalance risk value, and classify the section imbalance risk by combining the short-term fluctuation amplitude of wind power, the maximum ramp rate, the average output level and the section positive imbalance risk value, and realize the proactive early warning of grid section imbalance risk under extreme weather impact by combining the risk level; (3) Based on the cross-sectional imbalance risk level and the cross-sectional positive imbalance risk value, determine the active emergency coordination trigger level, implement hierarchical coordination control according to the response boundaries of energy storage, conventional units, photovoltaic, flexible loads and demand response resources, and output the cross-sectional power deficit correction result.

[0021] In step (1), a standardized prediction residual sequence is constructed based on the actual wind power and the predicted wind power, and its expression is: in, For a moment Standardized prediction residuals For a moment The predicted wind power output. To obtain the time-varying volatility adapted to extreme weather shocks, this embodiment uses the exponentially weighted moving average method to estimate the conditional volatility, the expression of which is: in, For a moment The conditional volatility estimate, The exponential weighting coefficient is preferably 0.94. For a moment The original prediction error satisfies .

[0022] Since this invention focuses on the down-skew risk caused by the actual wind power output being lower than the predicted value under extreme weather conditions, preferably, a mirror transformation is performed on the lower tail portion of the standardized prediction residuals to obtain the tail analysis variables: in, For a moment Mirror residual variables used for tail extreme value fitting. When The larger the value, the more significant the downward deviation of the actual wind power output from the predicted value.

[0023] In step (1), tail samples exceeding a high threshold are extracted from the mirror residual variable and fitted using a generalized Pareto distribution. To ensure the feasibility of threshold selection, the high threshold is determined based on the high quantile of the empirical distribution of the mirror tail samples, and its expression is: in, For high threshold, For the empirical distribution of the mirror tail samples quantiles, The optimal value is 0.90. When the extreme weather impact is stronger and the calibration sample is more abundant, it can also be... We set it to 0.95 to further highlight the extreme nature of the tail samples.

[0024] In step (2), based on the generalized Pareto distribution fitting result of the mirrored tail samples, the estimated tail quantile value at a given confidence level is obtained, and its expression is: in, Confidence level The estimated tail fractions, For scale parameters, For shape parameters, To calibrate the total number of samples, For exceeding the threshold The number of tail samples, This represents the failure probability. This step is used to describe the thick-tailed characteristics of wind power forecasting errors under extreme weather conditions, thereby improving the ability to characterize the down-biased risk boundary.

[0025] In step (2), the lower bound of wind power prediction and the adaptive safety interval of wind power output are constructed based on the tail fractional estimate, and their expressions are as follows: in, For a moment The lower bound of wind power prediction For a moment The adaptive safety range for wind power output. This safety range is used to characterize the range of potential downward deviation in wind power output under a given risk tolerance level.

[0026] In step (2), the lower bound of wind power prediction obtained in step (1) is substituted into the cross-sectional power balance relationship to construct the cross-sectional power imbalance value and the cross-sectional positive imbalance risk value, the expression of which is: in, For a moment The cross-sectional power loss measurement For a moment The risk value of positive imbalance in the cross section, For a moment The planned power output of the cross section For a moment The photovoltaic power available for use, For a moment The conventional unit's adjustable output For a moment The equivalent load demand power. This indicator is used to quantify the actual power gap that a section needs to compensate for under the influence of wind power downside risk.

[0027] In step (2), to characterize the impact of wind power fluctuations on the operational safety of the section under extreme weather conditions, the short-term fluctuation amplitude, maximum gradeability, and average output relative level of wind power are further calculated, and their expressions are as follows: in, For a moment The corresponding short-term fluctuation range of wind power within a 1-hour sliding window. For a moment The maximum ramp rate within a 1-hour sliding window. For a moment The average output level within the corresponding 1-hour sliding window is relatively horizontal. and These represent the maximum and minimum wind power output within the sliding window, respectively. The average wind power output within the sliding window. This refers to the installed capacity of wind power.

[0028] In step (2), the cross-sectional imbalance risk is classified based on the comparison results of the cross-sectional positive imbalance risk value, the short-term fluctuation amplitude of wind power, the maximum ramp rate, and the relative level of average output with the preset threshold. Preferably, the risk level criterion is as follows: Among them, when A level I rating indicates that the cross-section is in the highest risk state, requiring the triggering of enhanced active emergency control; when... A level II rating indicates that the cross-section is in a high-risk state, requiring the triggering of rapid coordinated control; when A level III rating indicates that the cross-section is in a state of early warning risk, requiring the timely activation of fast response resources; when A value equal to Level IV indicates that the cross-section is in a monitorable state.

[0029] In step (3), the active emergency control trigger level is determined based on the cross-sectional imbalance risk level, and the total power demand for multi-resource coordinated control is determined based on the cross-sectional positive imbalance risk value, the expression of which is: in, For a moment Active emergency control trigger level, For a moment The risk level of cross-sectional imbalance For a moment Multi-resource collaborative control of total power demand. When When it is Level IV, only risk monitoring status is maintained; when When it reaches Level III, activate the early warning level control; when When it is Level II, initiate rapid coordinated control; when When the emergency level is Level I, enhanced emergency control measures are activated.

[0030] In step (3), the target regulation power of each type of resource is allocated based on the response speed and operating boundary of energy storage, photovoltaic, conventional units, flexible loads, and demand response resources. The expression for this allocation is as follows: in, , , , and These are the target adjustment powers for energy storage, photovoltaics, conventional generating units, flexible loads, and demand response resources, respectively. , , , and These represent the maximum adjustable capacity of the corresponding resources. , , and These represent the remaining demand power after energy storage, photovoltaic, conventional units, and flexible loads have successively undertaken the regulation tasks. This step directly maps the quantification results of cross-sectional risks into control commands for multiple types of resources, thereby achieving a closed-loop connection from risk identification to proactive emergency control.

[0031] In step (3), after the multi-resource collaborative control is completed, the following power balance constraint is satisfied: When the above inequality holds true, it indicates that the current cross-sectional power deficit has been effectively corrected; when the above inequality does not hold true, the current control trigger state is maintained, and the wind power prediction boundary, cross-sectional imbalance risk value, and multi-resource control command are updated again in the next scheduling cycle.

[0032] The method described in this embodiment achieves a closed-loop analysis from the input of uncertain power generation from renewable energy sources under extreme weather conditions to the identification of grid cross-section risks and the output of control commands through a sequential processing flow of "extreme value conformal prediction boundary construction - cross-sectional imbalance risk quantification and proactive early warning - multi-resource proactive emergency coordinated regulation". Specifically, the extreme value conformal prediction boundary construction stage improves the characterization capability of tail-end risks under wind power conditions; the risk quantification stage achieves a comprehensive assessment of the degree of cross-sectional power deficit and fluctuation characteristics; and the proactive emergency coordinated control stage enables the hierarchical response and coordinated deployment of multiple resources, including energy storage, conventional units, photovoltaics, flexible loads, and demand response resources. This method can provide support for proactive safety defense of grid cross-sections, proactive emergency control triggering judgment, and multi-resource coordinated regulation under extreme weather conditions.

[0033] The above are merely preferred embodiments of the present invention; however, the present invention is not limited thereto. Within the scope of the technical concept of the present invention, various simple modifications can be made to the technical solution of the present invention, all of which should fall within the protection scope of the present invention.

Claims

1. A method for proactive early warning and coordinated control of new energy power grids under extreme weather conditions, characterized in that, The method includes: (1) Obtain historical wind power operation data, wind power prediction data and prediction residual sequence of the study area under extreme weather conditions, construct a wind power prediction error sample set under extreme disturbance conditions, and extract tail risk information reflecting the characteristics of sudden drop in wind power output. (2) The extreme value conformal prediction method is used to model the tail of the wind power prediction residual, calculate the extreme lower bound of the wind power prediction error, construct the wind power output adaptive safety range, and combine the power balance relationship of the grid section to quantify and classify the imbalance risk, and combine the risk level to realize the active early warning of the grid section imbalance risk under extreme weather impact. (3) The risk classification results are used as the triggering conditions for active emergency control. The energy storage, photovoltaic, conventional units, flexible loads and demand response resources are coordinated to implement hierarchical collaborative control in order to achieve rapid regulation of grid section imbalance risk under extreme weather impact.

2. The method according to claim 1, characterized in that, In step (1), the wind power data is represented as a superposition of the conditional mean term and the random disturbance term, and its expression is: in, For a moment The actual power output of wind power For a moment The conditional mean term is used to characterize the deterministic trend of wind power variation. For a moment The conditional fluctuation term is used to characterize the intensity of time-varying fluctuations in wind power. For a moment The standardized random perturbation term.

3. The method according to claim 2, characterized in that, In step (1), a standardized prediction residual sequence is constructed based on the wind power point prediction results, and the conditional volatility is estimated using the exponentially weighted moving average method, the expression of which is: in, For a moment Standardized prediction residuals For a moment The actual power output of wind power For a moment The predicted wind power output For a moment The conditional volatility estimate, This is the exponential weighting coefficient, preferably ranging from 0.90 to 0.

99. For a moment The original prediction error.

4. The method according to claim 3, characterized in that, In step (1), to address the down-bias risk caused by the actual wind power output being lower than the predicted value, a mirror transformation is performed on the standardized prediction residual sequence to obtain mirror tail variables. A tail sample set is then extracted from the mirror tail samples exceeding a preset high threshold, and a generalized Pareto distribution is used for tail fitting to characterize the down-tail risk information of wind power prediction errors under extreme weather conditions. The expressions for the mirror transformation and high threshold selection are: in, For a moment Mirror residual variables used for tail extreme value fitting For high threshold, For the empirical distribution of the mirror tail samples quantiles, The value range is from 0.90 to 0.

95.

5. The method according to claim 4, characterized in that, In step (2), based on the tail sample fitting results, the extreme quantiles of the wind power prediction residuals at the confidence level are calculated, and their expression is: in, Confidence level The estimated tail fractions, To preset a high threshold, is the scale parameter of the generalized Pareto distribution. The shape parameter of the generalized Pareto distribution. To calibrate the total number of samples, For exceeding the threshold The number of tail samples, This represents the probability of failure. At the confidence level, and and satisfy .

6. The method according to claim 5, characterized in that, In step (2), the lower limit of the extreme value of the wind power prediction error and the adaptive safety interval of wind power output are constructed based on the extreme value quantiles. The expression is as follows: in, For a moment The lower bound of wind power prediction For a moment The adaptive safety range of wind power output For a moment The predicted wind power output For a moment The conditional volatility estimate, Confidence level The estimated value of the tail fractions.

7. The method according to claim 6, characterized in that, In step (2), the cross-sectional imbalance value and the risk value of positive imbalance are constructed based on the power balance relationship of the power grid section. The expression is as follows: in, For a moment The cross-sectional power loss measurement For a moment The risk value of positive imbalance in the cross section, For a moment The planned power output of the cross section For a moment The lower bound of wind power prediction For a moment The photovoltaic power available for use, For a moment The conventional unit's adjustable output For a moment The equivalent load demand power.

8. The method according to claim 7, characterized in that, In step (2), the cross-sectional imbalance risk is classified according to the relative change in wind power, the maximum ramp rate within the window, the average output level, and the positive imbalance risk value of the cross-section. The indicators and risk level criterion expressions are as follows: in, This represents the relative change in wind power within a 1-hour sliding window. and These represent the maximum and minimum wind power outputs within the sliding window, respectively. For wind power installed capacity, The maximum ramp rate within the sliding window. This represents the actual wind power output at the previous moment. The average output force within the sliding window is relatively horizontal. The average wind power within the sliding window, , , , , , , , , and , , These are the classification thresholds for the risk value of positive imbalance at the cross section, the relative change in wind power, the maximum gradeability, and the relative level of average output, respectively, with Level I being the highest risk level and Level IV being the lowest risk level.

9. The method according to claim 8, characterized in that, In step (3), the risk level is used as the trigger condition for active emergency control, and the multi-resource collaborative control requirements are determined based on the cross-sectional power loss measurement, the expression of which is: in, For a moment Active emergency control trigger level, For a moment The risk level, For a moment Multi-resource collaborative control of total power demand. For a moment The risk value of positive imbalance in the cross section; when Active emergency control is not triggered when it is Level IV. When it reaches Level III, the early warning level control is activated. When it reaches Level II, rapid cooperative control is initiated. Enhanced emergency control will be activated when the level is I.

10. The method according to claim 9, characterized in that, In step (3), based on the total demand power of the multi-resource collaborative control, hierarchical collaborative control is implemented according to the response speed and adjustment boundary of energy storage, photovoltaic, conventional units, flexible loads, and demand response resources. The control allocation expression is as follows: in, For a moment Energy storage target adjustment power, For a moment The target adjustment power of photovoltaic power, For a moment The target regulating power of conventional units, For a moment Flexible load target adjustment power, For a moment Demand response target adjustment power, , , , and These represent the maximum adjustable capacity of the corresponding resources. , , and These are the remaining control power requirements after being deducted level by level according to priority.

Images