Wind power generation power anomaly prediction and early warning method and system under extreme weather

Abstract

The application discloses a wind power generation power abnormality prediction and early warning method and system under extreme weather, and belongs to the technical field of renewable energy power generation. By collecting meteorological condition variables and operation parameters of a wind power generation device, an initial operation data set is constructed; dynamic correlation between wind speed change and power output is extracted, response delay time of wind speed change on power fluctuation is determined, wind speed change trend and mechanical inertia response indexes are obtained; power is abnormally predicted, when a prediction result exceeds a first threshold value, a corresponding time period is marked as a potential power abnormality event; an interaction influence degree between the mechanical inertia response and the meteorological condition variables is quantified, and a corresponding power abnormality event strength evaluation index is generated; in combination with a preset classification rule, an early warning level is determined, an early warning signal is triggered and output; by quantifying dynamic interaction between wind speed fluctuation and turbine mechanical inertia, the application improves the accuracy of wind power generation power abnormality prediction under extreme weather.

Description

Technical Field

[0001] This invention relates to the field of renewable energy power generation technology, specifically a method for predicting and warning of abnormal wind power generation under extreme weather conditions. Background Technology

[0002] As a crucial pillar of clean energy, the efficient utilization of wind power is vital for achieving energy transition and addressing climate change. With the rapid growth of installed wind power capacity, the stability of wind power output directly impacts the safe operation of the power grid and the efficient dispatch of energy. However, wind power output often fluctuates significantly due to weather conditions and the inherent characteristics of wind turbines, especially during wind power ramp-up events under extreme weather conditions—phenomena characterized by rapid power changes within a short period. Failure to provide timely warnings for such events can lead to grid imbalances and even blackouts. Therefore, researching how to accurately predict and provide early warnings for wind power ramp-up events has become a critical issue that urgently needs to be addressed in the wind power sector.

[0003] Existing research on wind power ramp-up events often relies too heavily on single meteorological data or simple power fluctuation statistics, neglecting the complex interaction between atmospheric motion and the physical characteristics of wind turbines. This makes it difficult for prediction models to capture the dynamic correlation between wind speed changes and power fluctuations under complex weather conditions. For example, in severe convective weather, rapid changes in wind speed may cause turbine response lag, but existing methods struggle to accurately quantify the specific impact of this lag on power ramp-up. Furthermore, existing methods often suffer from low prediction efficiency when processing multidimensional feature data due to feature redundancy or interference from irrelevant variables, making it difficult to meet the needs of real-time early warning. The core technical challenge lies in accurately characterizing the dynamic response relationship between wind speed changes and the mechanical inertia of wind turbines. Rapid fluctuations in wind speed directly affect turbine speed and power output, but the turbine blade mass and rotational inertia cause a time delay in its response to wind speed changes. This delay makes it difficult to accurately determine the magnitude and duration of power ramp-up. For example, in a sudden strong wind event, wind speed may jump from a low value to a high value within minutes, but the turbine speed cannot keep up immediately due to inertia, resulting in deviations in the prediction of the start time and magnitude of power ramp-up. Therefore, how to accurately quantify the dynamic interaction between wind speed fluctuations and turbine mechanical inertia, and based on this, divide the ramp-up event characteristics of different power fluctuation ranges, has become a key issue in wind power ramp-up event early warning research. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for predicting and warning of abnormal wind power generation under extreme weather conditions, so as to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A method for predicting and warning of abnormal wind power generation under extreme weather conditions, comprising the following steps:

[0007] Real-time acquisition and preprocessing of meteorological condition variables and operating parameters of wind power generation devices under extreme weather conditions to construct an initial operating dataset containing time series data;

[0008] Based on the initial operating dataset, the dynamic correlation between wind speed change and power output is extracted, and the response delay time of wind speed change to power fluctuation is determined to obtain the wind speed change trend and mechanical inertial response index.

[0009] Based on the response delay time and operating parameters, power anomaly prediction is performed. When the prediction result exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0010] For the set of event candidates, the degree of interaction between the mechanical inertial response and meteorological condition variables is quantified to generate corresponding power anomaly event intensity assessment indicators;

[0011] Based on the intensity assessment indicators and the preset classification rules, the warning level is determined;

[0012] Based on the aforementioned warning level, a warning signal is triggered and output.

[0013] As a preferred embodiment of the wind power anomaly prediction and early warning method under extreme weather conditions in this invention, the specific implementation process of extracting the dynamic correlation between wind speed changes and power output based on the initial operating dataset, determining the response delay time of wind speed changes to power fluctuations, and obtaining wind speed change trends and mechanical inertial response indicators includes:

[0014] Based on meteorological condition variables, the short-term wind speed change rate is extracted to obtain the wind speed change trend;

[0015] Based on the operating parameters of the wind power generation device, obtain the power with timestamps;

[0016] Based on the time series data in the preliminary operational dataset, extract the time series features of wind speed changes and power output;

[0017] Based on the time series characteristics of the wind speed change and the time series characteristics of the power output, the correlation coefficient between wind speed change and power output under different lag times is calculated.

[0018] Based on the correlation coefficient between wind speed change and power output at different lag times, the lag time with the largest correlation coefficient is selected as the response delay time of wind speed change to power fluctuation, thus obtaining the mechanical inertial response index.

[0019] As a preferred embodiment of the wind power anomaly prediction and early warning method under extreme weather conditions in this invention, the specific implementation process of predicting power anomalies based on the response delay time and operating parameters, and marking the corresponding time period as a potential power anomaly event when the prediction result exceeds a first threshold to form an event candidate set includes:

[0020] Based on the response delay time and operating parameters, the abnormal predicted value of power is calculated using the following formula:

[0021] ;

[0022] in, This represents the predicted abnormal power value at time t, where T represents the length of the time series. This represents the weight coefficient at time i. Indicates delay time The wind speed value after that, This represents the reference wind speed, and β represents the attenuation coefficient. This indicates the response time of wind speed changes to power fluctuations;

[0023] When the predicted abnormal value exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0024] As a preferred embodiment of the wind power anomaly prediction and early warning method under extreme weather conditions in this invention, the specific implementation process of quantifying the degree of interaction between the mechanical inertial response and meteorological condition variables for the event candidate set and generating corresponding power anomaly event intensity assessment indicators includes:

[0025] For each potential power anomaly event in the event candidate set, multi-dimensional data of each potential power anomaly event is extracted, including time-series data of mechanical inertial response index, time-series data of meteorological condition variables, and time-series data of operating parameters;

[0026] Based on the aforementioned multi-dimensional data, the degree of interaction between mechanical inertial response and various meteorological condition variables is quantified, specifically including:

[0027] Key statistical features were extracted from the time series data of mechanical inertial response index and meteorological condition variables, respectively, to obtain the mechanical inertial response characteristics and meteorological condition characteristics.

[0028] Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of each meteorological condition, and obtain the interaction influence quantity y, which characterizes the degree of interaction between the mechanical inertial response and the meteorological condition variables, by weighted averaging the correlation coefficients.

[0029] By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value p of the historical normal pattern, the final power anomaly event intensity assessment index I is generated, and the generation formula is as follows:

[0030] ;

[0031] Where s represents the duration of the potential power anomaly event. , , This represents the weighting coefficient.

[0032] As a preferred embodiment of the wind power anomaly prediction and early warning method under extreme weather conditions in this invention, the preprocessing includes: time alignment of the collected data, outlier removal and correction, missing value imputation, noise reduction and standardization; the initial running dataset is a multi-dimensional time series matrix containing meteorological condition variables, power output and operating parameters.

[0033] As a preferred embodiment of the wind power anomaly prediction and early warning method under extreme weather conditions in this invention, the preset classification rule is to set multiple threshold intervals based on the power anomaly event intensity assessment index and power ramp rate, and each threshold interval corresponds to an early warning level; the early warning level includes at least attention level, early warning level, alarm level and emergency level.

[0034] A wind power generation anomaly prediction and early warning system under extreme weather conditions, the system comprising a data acquisition module, a response module, an anomaly prediction module, an evaluation module, and an early warning module;

[0035] The acquisition module is used to collect and preprocess meteorological condition variables and operating parameters of wind power generation devices under extreme weather conditions in real time, and construct an initial operating dataset containing time series data. The preprocessing includes: time alignment of the collected data, outlier removal and correction, missing value imputation, noise reduction, and standardization. The initial operating dataset is a multi-dimensional time series matrix containing meteorological condition variables, power output, and operating parameters.

[0036] The response module is used to extract the dynamic correlation between wind speed change and power output based on the initial running dataset, determine the response delay time of wind speed change to power fluctuation, and obtain the wind speed change trend and mechanical inertial response index.

[0037] The anomaly prediction module is used to predict power anomalies based on the response delay time and operating parameters. When the prediction result exceeds a first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0038] The evaluation module is used to quantify the degree of interaction between the mechanical inertial response and meteorological condition variables for the event candidate set, and generate corresponding power anomaly event intensity evaluation indicators.

[0039] The early warning module is used to determine the early warning level based on the intensity assessment index and a preset classification rule; and to trigger and output an early warning signal based on the early warning level.

[0040] The preset classification rule is to set multiple threshold intervals based on the power anomaly event intensity assessment index and power ramp rate, with each threshold interval corresponding to a warning level; the warning levels include at least attention level, warning level, alarm level and emergency level.

[0041] As a preferred embodiment of the wind power generation anomaly prediction and early warning system under extreme weather conditions in this invention, the response module includes: extracting the short-term wind speed change rate based on meteorological condition variables to obtain the wind speed change trend;

[0042] Based on the operating parameters of the wind power generation device, obtain the power with timestamps;

[0043] Based on the time series data in the preliminary operational dataset, extract the time series features of wind speed changes and power output;

[0044] Based on the time series characteristics of the wind speed change and the power output, the correlation coefficients between wind speed change and power output under different lag times are calculated.

[0045] Based on the correlation coefficient between wind speed change and power output at different lag times, the lag time with the largest correlation coefficient is selected as the response delay time of wind speed change to power fluctuation, thus obtaining the mechanical inertial response index.

[0046] As a preferred embodiment of the wind power generation anomaly prediction and early warning system under extreme weather conditions in this invention, the anomaly prediction module includes:

[0047] Based on the response delay time and operating parameters, the abnormal predicted value of power is calculated using the following formula:

[0048] ;

[0049] in, This represents the predicted abnormal power value at time t, where T represents the length of the time series. This represents the weight coefficient at time i. Indicates delay time The wind speed value after that, This represents the reference wind speed, and β represents the attenuation coefficient. This indicates the response time of wind speed changes to power fluctuations;

[0050] When the predicted abnormal value exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0051] As a preferred embodiment of the wind power generation anomaly prediction and early warning system under extreme weather conditions in this invention, the evaluation module includes:

[0052] For each potential power anomaly event in the event candidate set, multi-dimensional data of each potential power anomaly event is extracted, including time-series data of mechanical inertial response index, time-series data of meteorological condition variables, and time-series data of operating parameters;

[0053] Based on the aforementioned multi-dimensional data, the degree of interaction between mechanical inertial response and various meteorological condition variables is quantified, specifically including:

[0054] Key statistical features were extracted from the time series data of mechanical inertial response index and meteorological condition variables, respectively, to obtain the mechanical inertial response characteristics and meteorological condition characteristics.

[0055] Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of each meteorological condition, and obtain the interaction influence quantity y, which characterizes the degree of interaction between the mechanical inertial response and the meteorological condition variables, by weighted averaging the correlation coefficients.

[0056] By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value p of the historical normal pattern, the final power anomaly event intensity assessment index I is generated, and the generation formula is as follows:

[0057] ;

[0058] Where s represents the duration of the potential power anomaly event. , , This represents the weighting coefficient.

[0059] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: The method and system for predicting and warning of wind power anomalies under extreme weather conditions provided by this invention construct an initial operating dataset by collecting meteorological condition variables and operating parameters of the wind power generation device; extracting the dynamic correlation between wind speed changes and power output, determining the response delay time of wind speed changes to power fluctuations, and obtaining wind speed change trends and mechanical inertial response indicators; predicting power anomalies, and marking the corresponding time period as a potential power anomaly event when the prediction result exceeds a first threshold; quantifying the degree of interaction between the mechanical inertial response and meteorological condition variables, generating corresponding power anomaly event intensity assessment indicators; and determining the warning level by combining preset classification rules, triggering and outputting warning signals. This invention improves the accuracy of predicting wind power anomalies under extreme weather conditions by quantifying the dynamic interaction between wind speed fluctuations and turbine mechanical inertia. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0061] Figure 1 This is a schematic diagram of the system structure in an embodiment of the present invention; Figure 2 This is a schematic diagram of the method steps in an embodiment of the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Please see Figures 1-2 In this first embodiment: a method for predicting and warning of abnormal wind power generation under extreme weather conditions is provided, the method comprising:

[0064] Step 1: Real-time acquisition and preprocessing of meteorological condition variables and wind power generation unit operating parameters under extreme weather conditions to construct an initial operating dataset containing time series data. Time series data is acquired or received in real-time through meteorological stations and wind measurement towers deployed within the wind farm, or by accessing regional weather forecast services. Meteorological condition variables include at least: wind speed (m / s), wind direction (°), ambient temperature (°C), atmospheric pressure (hPa), and relative humidity (%). The data sampling frequency is typically once every 1-10 minutes. The operating status time series data of each wind turbine is acquired in real-time through the wind turbine monitoring and data acquisition system. Operating parameters include at least: generator active power (kW), rotor speed (rpm), generator torque (kN·m), blade pitch angle (°), and nacelle vibration acceleration (m / s²). The data sampling frequency is typically on the order of seconds to minutes. The above data is uniformly transmitted to the central data processing server through a dedicated industrial communication network (such as OPC UA, Modbus TCP) or a secure API interface. The server runs a data receiving service, parsing and temporarily storing the raw data stream according to the preset data format and protocol. The processed data is arranged in chronological order to form a structured multidimensional time series matrix, which is the initial running dataset;

[0065] Step 2: Based on the initial running dataset, extract the dynamic correlation between wind speed changes and power output, determine the response delay time of wind speed changes to power fluctuations, and obtain the wind speed change trend and mechanical inertial response index.

[0066] Step 3: Based on the response delay time and operating parameters, perform power anomaly prediction. When the prediction result exceeds the first threshold, mark the corresponding time period as a potential power anomaly event and form an event candidate set.

[0067] Step 4: For the event candidate set, quantify the degree of interaction between the mechanical inertial response and meteorological condition variables, and generate corresponding power anomaly event intensity assessment indicators;

[0068] Step 5: Determine the warning level based on the intensity assessment index and the preset classification rules; trigger and output a warning signal based on the warning level.

[0069] Specifically, the process of extracting the dynamic correlation between wind speed changes and power output based on the initial operating dataset, determining the response delay time of wind speed changes to power fluctuations, and obtaining the wind speed change trend and mechanical inertial response index includes:

[0070] Based on meteorological condition variables, the short-term wind speed change rate is extracted as a quantitative indicator of the wind speed change trend.

[0071] Based on the operating parameters of the wind power generation device, obtain the power with timestamps;

[0072] Based on the time series data in the preliminary operational dataset, extract the time series features of wind speed changes and power output;

[0073] Based on the time series characteristics of the wind speed change and the time series characteristics of the power output, the correlation coefficient between wind speed change and power output under different lag times (such as lag of 0 seconds, 1 second, 2 seconds, etc.) is calculated.

[0074] Based on the correlation coefficient between wind speed changes and power output at different lag times, the lag time with the largest correlation coefficient is selected as the response delay time of wind speed changes to power fluctuations. This time reflects the mechanical inertial response characteristics of the wind turbine from sensing wind speed changes to adjusting power output, and is used as a mechanical inertial response index. Wind power generation exhibits an inertial delay in response to wind speed changes, rather than an instantaneous response. This step aims to quantitatively characterize this key physical characteristic; under extreme weather conditions, excessively long delays or abnormal response patterns (such as a sudden drop in correlation coefficient) are often precursors to failures or performance degradation. This index provides a dynamic benchmark for anomaly detection.

[0075] Traditional correlation analysis (such as the Pearson correlation coefficient) measures the linear relationship between two variables at the same moment. However, due to the mechanical inertia of wind turbines, changes in wind speed take a delay (ranging from a few seconds to tens of seconds) before they are fully reflected in power output.

[0076] Therefore, we need to analyze the correlation between the wind speed sequence and the power sequence at different time offsets (lags). This analytical method is called cross-correlation analysis or lag correlation analysis. Its goal is to find the time offset that makes the correlation between the two the strongest, which is the response delay time.

[0077] Specifically, the process of predicting power anomalies based on the response delay time and operating parameters, and marking the corresponding time period as a potential power anomaly event when the prediction result exceeds a first threshold, to form an event candidate set, includes:

[0078] Based on the response delay time and operating parameters, the abnormal predicted value of power is calculated using the following formula:

[0079]

[0080] in, This represents the predicted abnormal power value at time t, where T represents the length of the time series. This represents the weight coefficient at time i. Indicates delay time The wind speed value after that, This represents the reference wind speed, and β represents the attenuation coefficient. This indicates the response time of wind speed changes to power fluctuations;

[0081] When the predicted abnormal value exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0082] Leveraging the characteristic that wind speed changes typically precede power anomalies, and combining this with identified response delays, predictions can be made before actual abnormal power occurs; not all data points require complex intensity assessments. This step quickly identifies high-risk periods, focusing analytical resources. It transforms passive alerts into proactive predictions, buying valuable time for regulation and response. The first-level threshold filtering significantly reduces the amount of data required for subsequent detailed assessments, improving the overall real-time performance of the method.

[0083] Specifically, the process of quantifying the degree of interaction between the mechanical inertial response and meteorological condition variables for the candidate event set, and generating corresponding power anomaly event intensity assessment indicators, includes:

[0084] For each potential power anomaly event in the event candidate set, multi-dimensional data of each potential power anomaly event is extracted, including time-series data of mechanical inertial response index, time-series data of meteorological condition variables, and time-series data of operating parameters;

[0085] Based on the aforementioned multi-dimensional data, the degree of interaction between mechanical inertial response and various meteorological condition variables is quantified, specifically including:

[0086] Key statistical features were extracted from the time series data of mechanical inertial response index and meteorological condition variables, respectively, to obtain the mechanical inertial response characteristics and meteorological condition characteristics.

[0087] Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of each meteorological condition, and obtain the interaction influence quantity y, which characterizes the degree of interaction between the mechanical inertial response and the meteorological condition variables, by weighted averaging the correlation coefficients.

[0088] By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value p of the historical normal pattern, the final power anomaly event intensity assessment index I is generated, and the generation formula is as follows:

[0089] ;

[0090] Where s represents the duration of the potential power anomaly event. , , This represents the weighting coefficient.

[0091] Specifically, the preprocessing includes: time alignment of the collected data, outlier removal and correction, missing value imputation, noise reduction, and standardization; the initial running dataset is a multi-dimensional time series matrix containing meteorological condition variables, power output, and operating parameters.

[0092] Since meteorological condition variable data and operational parameter data may originate from different clock sources, all received data entries are first unified to the same time base (e.g., UTC time). To address the issue of inconsistent sampling frequencies from different data sources, linear interpolation or nearest neighbor interpolation algorithms are used to resample all time series data to a unified, fixed time interval (e.g., 1 minute). Using this common time point as a reference, an aligned, equally spaced initial time series data framework is generated. Moving averages or filtering are then applied to the meteorological condition variable data and operational parameter data to reduce noise and minimize random fluctuations. Data of different dimensions and orders of magnitude are normalized to eliminate scale effects and facilitate subsequent model calculations.

[0093] Specifically, the preset classification rule is to set multiple threshold intervals based on the power anomaly event intensity assessment index and power ramp rate, with each threshold interval corresponding to a warning level; the warning levels include at least attention level, warning level, alarm level and emergency level.

[0094] Multiple consecutive threshold intervals are predefined, each interval corresponding to a warning level. For example:

[0095] I < A1: Attention level (minor anomaly, trend needs to be monitored).

[0096] A1 ≤ I < A2: Warning level (there is an abnormal risk, and monitoring needs to be strengthened).

[0097] A2 ≤ I < A3: Alert level (high probability of an anomaly, and countermeasures should be prepared).

[0098] I ≥ A3: Emergency level (serious abnormality is about to occur or has already occurred, requiring immediate action).

[0099] Meanwhile, the classification rules combine key operating parameters such as power ramp rate (power change rate per unit time) to set auxiliary thresholds to prevent "false alarms" where the intensity index is high but the power change is slow.

[0100] The intensity assessment index I for each event is compared with a preset threshold range to determine its warning level; different severity levels of events require different response strategies. Tiered warnings can prioritize the allocation of limited operation and maintenance and scheduling resources to the highest-risk events; each warning level corresponds to a standardized emergency plan and operating procedure, facilitating rapid decision-making by on-site personnel and the dispatch center.

[0101] Based on the determined warning level, a structured warning signal containing the following key information is automatically generated:

[0102] Event ID, time, location (wind turbine or wind farm number).

[0103] Warning level (e.g., alert level).

[0104] Overview of predicted abnormal power and expected duration.

[0105] Analysis of the main contributing factors (e.g., strong gusts combined with turbulence).

[0106] Recommended measures (generated according to the corresponding contingency plan based on the level).

[0107] Warning signals are sent in real time to wind farm operators, equipment maintenance teams, and the power grid dispatch center through various means such as audible and visual alarms, monitoring system pop-ups, SMS, email, and API interface push.

[0108] A wind power generation anomaly prediction and early warning system under extreme weather conditions, the system comprising a data acquisition module, a response module, an anomaly prediction module, an evaluation module, and an early warning module;

[0109] The acquisition module is used to collect and preprocess meteorological condition variables and operating parameters of wind power generation devices under extreme weather conditions in real time, and construct an initial operating dataset containing time series data. The preprocessing includes: time alignment of the collected data, outlier removal and correction, missing value imputation, noise reduction, and standardization. The initial operating dataset is a multi-dimensional time series matrix containing meteorological condition variables, power output, and operating parameters.

[0110] The response module is used to extract the dynamic correlation between wind speed change and power output based on the initial running dataset, determine the response delay time of wind speed change to power fluctuation, and obtain the wind speed change trend and mechanical inertial response index.

[0111] The anomaly prediction module is used to predict power anomalies based on the response delay time and operating parameters. When the prediction result exceeds a first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0112] The evaluation module is used to quantify the degree of interaction between the mechanical inertial response and meteorological condition variables for the event candidate set, and generate corresponding power anomaly event intensity evaluation indicators.

[0113] The early warning module is used to determine the early warning level based on the intensity assessment index and a preset classification rule; and to trigger and output an early warning signal based on the early warning level.

[0114] The preset classification rule is to set multiple threshold intervals based on the power anomaly event intensity assessment index and power ramp rate, with each threshold interval corresponding to a warning level; the warning levels include at least attention level, warning level, alarm level and emergency level.

[0115] Specifically, the response module includes: extracting the short-term wind speed change rate based on meteorological condition variables to obtain the wind speed change trend;

[0116] Based on the operating parameters of the wind power generation device, obtain the power with timestamps;

[0117] Based on the time series data in the preliminary operational dataset, extract the time series features of wind speed changes and power output;

[0118] Based on the time series characteristics of the wind speed change and the power output, the correlation coefficients between wind speed change and power output under different lag times are calculated.

[0119] Based on the correlation coefficient between wind speed change and power output at different lag times, the lag time with the largest correlation coefficient is selected as the response delay time of wind speed change to power fluctuation, thus obtaining the mechanical inertial response index.

[0120] Specifically, the anomaly prediction module includes:

[0121] Based on the response delay time and operating parameters, the abnormal predicted value of power is calculated using the following formula:

[0122] ;

[0123] in, This represents the predicted abnormal power value at time t, where T represents the length of the time series. This represents the weight coefficient at time i. Indicates delay time The wind speed value after that, This represents the reference wind speed, and β represents the attenuation coefficient. This indicates the response time of wind speed changes to power fluctuations;

[0124] When the predicted abnormal value exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set.

[0125] Specifically, the evaluation module includes;

[0126] For each potential power anomaly event in the event candidate set, multi-dimensional data of each potential power anomaly event is extracted, including time-series data of mechanical inertial response index, time-series data of meteorological condition variables, and time-series data of operating parameters;

[0127] Based on the aforementioned multi-dimensional data, the degree of interaction between mechanical inertial response and various meteorological condition variables is quantified, specifically including:

[0128] Key statistical features were extracted from the time series data of mechanical inertial response index and meteorological condition variables, respectively, to obtain the mechanical inertial response characteristics and meteorological condition characteristics.

[0129] Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of each meteorological condition, and obtain the interaction influence quantity y, which characterizes the degree of interaction between the mechanical inertial response and the meteorological condition variables, by weighted averaging the correlation coefficients.

[0130] By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value p of the historical normal pattern, the final power anomaly event intensity assessment index I is generated, and the generation formula is as follows:

[0131] ;

[0132] Where s represents the duration of the potential power anomaly event. , , This represents the weighting coefficient.

[0133] It should be noted that, in this document, relational terms such as "first" and "second" are used only 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 process, method, article, or apparatus.

[0134] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting and warning of abnormal wind power generation under extreme weather conditions, characterized in that, The method includes the following steps: Real-time acquisition and preprocessing of meteorological condition variables and operating parameters of wind power generation devices under extreme weather conditions to construct an initial operating dataset containing time series data; Based on the initial operating dataset, the dynamic correlation between wind speed change and power output is extracted to obtain the wind speed change trend. Combined with the operating parameters of the wind power generation device, the response delay time of wind speed change to power fluctuation is determined as a mechanical inertial response index. Based on the response delay time and operating parameters, power anomaly prediction is performed, and the anomaly prediction value of power is calculated using the following formula: in, This represents the predicted abnormal power value at time t, where T represents the length of the time series. This represents the weight coefficient at time i. Indicates delay time The wind speed value after that, This represents the reference wind speed, and β represents the attenuation coefficient. This indicates the response time of wind speed changes to power fluctuations; When the predicted abnormal value exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set; For the event candidate set, the degree of interaction between the mechanical inertial response index and meteorological condition variables is quantified to generate a corresponding power anomaly event intensity assessment index; key statistical features are extracted from the time series data of the mechanical inertial response index and the time series data of the meteorological condition variables to obtain mechanical inertial response features and meteorological condition features. Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of various meteorological conditions, and obtain the interaction influence quantity, which characterizes the degree of interaction between the mechanical inertial response index and the meteorological condition variables, by weighted averaging the correlation coefficients. By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value between historical normal patterns, the final power anomaly event intensity assessment index is generated. Based on the intensity assessment indicators and the preset classification rules, the warning level is determined; Based on the aforementioned warning level, a warning signal is triggered and output.

2. The method for predicting and warning of wind power generation anomalies under extreme weather conditions according to claim 1, characterized in that, The specific implementation process of extracting the dynamic correlation between wind speed changes and power output based on the initial operating dataset, obtaining the wind speed change trend, and determining the response delay time of wind speed changes to power fluctuations as a mechanical inertial response index, in conjunction with the operating parameters of the wind power generation device, includes: Based on meteorological condition variables, the short-term wind speed change rate is extracted to obtain the wind speed change trend; Based on the operating parameters of the wind power generation device, obtain the power with timestamps; Based on the time series data in the initial running dataset, extract the time series features of wind speed changes and power output; Based on the time series characteristics of the wind speed change and the time series characteristics of the power output, the correlation coefficient between wind speed change and power output under different lag times is calculated. Based on the correlation coefficient between wind speed change and power output at different lag times, the lag time with the largest correlation coefficient is selected as the response delay time of wind speed change to power fluctuation, thus obtaining the mechanical inertial response index.

3. The method for predicting and warning of wind power generation anomalies under extreme weather conditions according to claim 2, characterized in that, The specific implementation process of quantifying the degree of interaction between the mechanical inertial response index and meteorological condition variables for the candidate event set, and generating corresponding power anomaly event intensity assessment indexes, includes: For each potential power anomaly event in the event candidate set, multi-dimensional data of each potential power anomaly event is extracted, including time-series data of mechanical inertial response index, time-series data of meteorological condition variables, and time-series data of operating parameters; Based on the aforementioned multi-dimensional data, the degree of interaction between the mechanical inertial response index and various meteorological condition variables is quantified, specifically including: Key statistical features were extracted from the time series data of mechanical inertial response index and meteorological condition variables, respectively, to obtain the mechanical inertial response characteristics and meteorological condition characteristics. Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of various meteorological conditions, and obtain the interaction influence quantity y, which characterizes the degree of interaction between the mechanical inertial response index and the meteorological condition variables, by weighted averaging the correlation coefficients. By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value p between historical normal patterns, the final power anomaly event intensity assessment index I is generated. The generation formula is as follows: ； Where s represents the duration of the potential power anomaly event. , , This represents the weighting coefficient.

4. The method for predicting and warning of wind power generation anomalies under extreme weather conditions according to claim 1, characterized in that, The preprocessing includes: time alignment of the collected data, outlier removal and correction, missing value imputation, noise reduction and standardization; the initial running dataset is a multi-dimensional time series matrix containing meteorological condition variables, power output and operating parameters.

5. The method for predicting and warning of abnormal wind power generation under extreme weather conditions according to claim 1, characterized in that, The preset classification rule sets multiple threshold intervals based on the power anomaly event intensity assessment index and power ramp rate, with each threshold interval corresponding to a warning level; the warning levels include at least attention level, warning level, alarm level and emergency level.

6. A wind power generation anomaly prediction and early warning system under extreme weather conditions, characterized in that, The system includes a data acquisition module, a response module, an anomaly prediction module, an evaluation module, and an early warning module. The acquisition module is used to collect and preprocess meteorological condition variables and operating parameters of wind power generation devices under extreme weather conditions in real time, and construct an initial operating dataset containing time series data. The preprocessing includes: time alignment of the collected data, outlier removal and correction, missing value imputation, noise reduction, and standardization. The initial operating dataset is a multi-dimensional time series matrix containing meteorological condition variables, power output, and operating parameters. The response module is used to extract the dynamic correlation between wind speed change and power output based on the initial operating dataset, obtain the wind speed change trend, and determine the response delay time of wind speed change to power fluctuation in combination with the operating parameters of the wind power generation device, as a mechanical inertial response index. The anomaly prediction module is used to predict power anomalies based on the response delay time and operating parameters, and calculate the anomaly prediction value of the power using the following formula: in, This represents the predicted abnormal power value at time t, where T represents the length of the time series. This represents the weight coefficient at time i. Indicates delay time The wind speed value after that, This represents the reference wind speed, and β represents the attenuation coefficient. This indicates the response time of wind speed changes to power fluctuations; When the predicted abnormal value exceeds the first threshold, the corresponding time period is marked as a potential power anomaly event, forming an event candidate set; The evaluation module is used to quantify the degree of interaction between the mechanical inertial response index and meteorological condition variables for the event candidate set, and generate a corresponding power anomaly event intensity evaluation index; and to extract key statistical features from the time series data of the mechanical inertial response index and the time series data of the meteorological condition variables to obtain mechanical inertial response features and meteorological condition features respectively. Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of various meteorological conditions, and obtain the interaction influence quantity, which characterizes the degree of interaction between the mechanical inertial response index and the meteorological condition variables, by weighted averaging the correlation coefficients. By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value between historical normal patterns, the final power anomaly event intensity assessment index is generated. The early warning module is used to determine the early warning level based on the intensity assessment index and a preset classification rule; and to trigger and output an early warning signal based on the early warning level. The preset classification rule is to set multiple threshold intervals based on the power anomaly event intensity assessment index and power ramp rate, with each threshold interval corresponding to a warning level; the warning levels include at least attention level, warning level, alarm level and emergency level.

7. The wind power generation anomaly prediction and early warning system under extreme weather conditions according to claim 6, characterized in that, The response module includes: extracting the short-term wind speed change rate based on meteorological condition variables to obtain the wind speed change trend; Based on the operating parameters of the wind power generation device, obtain the power with timestamps; Based on the time series data in the initial running dataset, extract the time series features of wind speed changes and power output; Based on the time series characteristics of the wind speed change and the time series characteristics of the power output, the correlation coefficient between wind speed change and power output under different lag times is calculated. Based on the correlation coefficient between wind speed change and power output at different lag times, the lag time with the largest correlation coefficient is selected as the response delay time of wind speed change to power fluctuation, thus obtaining the mechanical inertial response index.

8. The wind power generation anomaly prediction and early warning system under extreme weather conditions according to claim 7, characterized in that: The evaluation module includes: For each potential power anomaly event in the event candidate set, multi-dimensional data of each potential power anomaly event is extracted, including time-series data of mechanical inertial response index, time-series data of meteorological condition variables, and time-series data of operating parameters; Based on the aforementioned multi-dimensional data, the degree of interaction between the mechanical inertial response index and various meteorological condition variables is quantified, specifically including: Key statistical features were extracted from the time series data of mechanical inertial response index and meteorological condition variables, respectively, to obtain the mechanical inertial response characteristics and meteorological condition characteristics. Calculate the correlation coefficient between the mechanical inertial response characteristics and the characteristics of various meteorological conditions, and obtain the interaction influence quantity y, which characterizes the degree of interaction between the mechanical inertial response index and the meteorological condition variables, by weighted averaging the correlation coefficients. By combining the correlation coefficient between actual wind speed changes and power output during the corresponding time period and the offset value p between historical normal patterns, the final power anomaly event intensity assessment index I is generated. The generation formula is as follows: ； Where s represents the duration of the potential power anomaly event. , , This represents the weighting coefficient.

Images