Wind power climbing event detection method
By combining dynamic time warping and the rotating door algorithm, accurate detection of wind power ramp-up events in multi-wind turbine environments is achieved, solving the problems of high false alarm rate and poor adaptability in existing technologies, and improving the accuracy and robustness of detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-10
- Publication Date
- 2026-03-13
AI Technical Summary
Existing technologies struggle to accurately identify wind power ramp-up events, especially in multi-turbine environments where they suffer from high false alarm rates and poor adaptability.
By combining the Dynamic Time Warping (DTW) algorithm with the Rotating Door Algorithm (SDA), and through dynamic spatial weight calculation and adaptive dynamic door width adjustment, the alignment of multiple wind turbine power sequences and the extraction of key change points are achieved, generating the overall ramp-up event detection results of the wind farm.
It improves the detection accuracy of wind power ramp-up events, reduces false alarm and false negative rates, and enhances adaptability to complex operating conditions.
Smart Images
Figure CN121656641A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of wind power ramp detection technology, and in particular to a method for detecting wind power ramp events. Background Technology
[0002] With the accelerated global energy transition, wind power installed capacity continues to grow rapidly. The inherent volatility and uncertainty of its output pose serious challenges to the stable operation of the power system. Among these challenges, wind power ramp-up events—the phenomenon of rapid changes in wind farm output power within a short period—have become a critical issue affecting the safe and economical operation of the power grid. These events can trigger power fluctuations exceeding 20%-30% of the installed capacity within minutes, threatening not only system frequency stability and voltage quality but also forcing the grid to allocate substantial reserve capacity, significantly increasing operating costs.
[0003] Fixed threshold detection methods in related technologies are difficult to accurately identify complex and ever-changing climbing patterns, and have limitations such as poor adaptability and high false alarm rate, which urgently need to be improved. Summary of the Invention
[0004] This application provides a wind power ramp-up event detection method to solve the problems in the related technologies, such as the wind power ramp-up judgment threshold method being difficult to adapt to the dynamic operating conditions of wind farms and having poor identification accuracy, while the commonly used rotating door algorithm (SDA) can only process single time series, cannot align multi-wind turbine data, and is prone to misjudging normal fluctuations as ramp-up.
[0005] This application provides a method for detecting wind power ramp-up events, including the following steps: The wind power sequence of each wind turbine in the wind farm is obtained. The wind power sequence is dynamically weighted according to the wind turbine azimuth and upstream and downstream spatial positions. Based on the calculation results, a reference sequence is simulated using a dynamic weighting strategy. The dynamic time warping score of each wind turbine power sequence and the reference sequence is calculated to obtain the score that meets the preset optimal conditions. The timestamp of each wind turbine power sequence is adjusted according to the score that meets the preset optimal conditions to obtain the alignment sequence of each wind turbine. Based on the alignment sequence of each wind turbine, the adaptive dynamic rotating door threshold of each wind turbine is adjusted to obtain the dynamic door width of each wind turbine, and key change points exceeding the dynamic door width are extracted according to the deviation between power to generate a set of key change points for each wind turbine. Based on the set of key change points for each wind turbine, the power change rate within a sliding window is calculated. Based on the power change rate within the sliding window, when each wind turbine is detected to meet the preset ramp-up event determination conditions, a wind power ramp-up event for each wind turbine is determined. When the ratio of the wind turbines that have experienced the wind power ramp-up event to the total number of wind turbines in the wind farm reaches a preset threshold, an overall wind power ramp-up event for the wind farm is determined.
[0006] Optionally, in one embodiment of this application, the step of calculating the dynamic spatial weight of the wind power sequence based on the wind turbine azimuth and upstream and downstream spatial positions includes: The azimuth angles and upstream and downstream spatial positions of the wind turbines in the wind farm are determined, wherein the formula for calculating the azimuth angle of the wind turbines is:
[0007] in, i For the aforementioned fan, The origin of the coordinate system is the center of the wind farm or the location of the meteorological tower. Let be the plane rectangular coordinates of each wind turbine; Based on the azimuth angle of the wind turbines and their upstream and downstream spatial positions, the dynamic weight of each wind turbine at a preset time is calculated, wherein the formula for calculating the dynamic weight of each wind turbine at the preset time is:
[0008] in, Indicates wind turbine The straight-line distance from the origin Indicates the weighting coefficient. It is a tiny constant. The wind direction matching coefficient for the upstream wind turbine. This is the weight attenuation coefficient for downstream wind turbines.
[0009] Optionally, in one embodiment of this application, the step of simulating the reference sequence using a dynamic weighting strategy includes: The dynamically weighted reference power of each wind turbine is calculated based on the dynamic weights and the power of each turbine. A dynamic weighting strategy is then determined based on the dynamically weighted reference power, and the strategy is executed to simulate the reference sequence. The formula for calculating the dynamically weighted reference power is as follows:
[0010] in, For the generated sequence, t For a moment.
[0011] Optionally, in one embodiment of this application, the step of calculating the dynamic time warping score of each wind turbine power sequence and the reference sequence to obtain a score that satisfies a preset optimal condition, and adjusting the timestamps of each wind turbine power sequence according to the score that satisfies the preset optimal condition to obtain the alignment sequence of each wind turbine, includes: Calculate the similarity distance between the reference sequence and the target sequence at each time point, wherein the formula for calculating the similarity distance is:
[0012] in, For the reference sequence and the real sequence at time... With time The single-point Euclidean distance, For the reference sequence at time power, For the real sequence at time... The power; Based on the similarity distance, the globally optimal alignment path score between the real sequence and the reference sequence is calculated recursively, wherein the formula for calculating the globally optimal alignment path score is:
[0013] in, Indicates the sequence starts from the beginning. arrive The cumulative DTW distance; Based on the globally optimal alignment path score, the real sequence timestamp is adjusted to the reference sequence timestamp to obtain the aligned sequence. The formula for calculating the aligned sequence is as follows:
[0014] in, The power of the real sequence at that time.
[0015] Optionally, in one embodiment of this application, the formula for calculating the dynamic gate width is:
[0016] in, For a moment Dynamic door width, Based on the basic door width, For fluctuation coefficient, For the time mentioned The standard deviation of power within the nearby sliding window, The average power within this window. For correction factor, For the time mentioned DTW alignment error.
[0017] Optionally, in one embodiment of this application, the formula for calculating the deviation between the power values is:
[0018] in, The current starting index power, For the time mentioned The power of.
[0019] Optionally, in one embodiment of this application, the formula for calculating the power change rate within the window is:
[0020] in, Key Change Points Starting point, time window The rate of change of power within, For time windows The number of key change points contained within. This represents the power at the key change point at the end of the time window.
[0021] Optionally, in one embodiment of this application, the preset hill-climbing event determination condition is:
[0022] in, The intensity of the event.
[0023] Optionally, in one embodiment of this application, determining the wind power ramp-up event of each wind turbine when it is detected that each wind turbine meets the preset ramp-up event determination conditions includes: When it is detected that more than a preset proportion of wind turbines in the wind farm meet the positive or negative ramp event in the preset ramp event determination conditions within the same time period, it is determined that the wind turbines in the wind farm have experienced the positive or negative ramp event. Within two adjacent time windows, if a single wind turbine in the wind farm is detected to simultaneously satisfy both a positive and a negative ramp event in the preset ramp event determination conditions, it is determined that the single wind turbine in the wind farm has experienced the same ramp event.
[0024] Optionally, in one embodiment of this application, after determining the overall wind power ramp-up event of the wind farm, the method further includes: The overall wind power ramp-up events are detected to generate ramp-up event detection results; the detection results are then examined to determine the proportional relationships within the ramp-up events.
[0025] This application's embodiments organically combine algorithms with two different objectives: time series alignment and data compression. It proposes a wind power ramping event detection method based on DTW-SDA fusion. By simultaneously considering time alignment and spatial correlation, a reference sequence is established for matching to obtain the adaptive dynamic threshold of SDA. This solves the problems of poor detection effect caused by the fixed ramping event detection threshold and the asynchronous operation of multiple wind turbines, ultimately achieving accurate ramping detection of wind power in wind farms.
[0026] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description
[0027] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is an overall flowchart of a wind power ramping event detection method provided according to an embodiment of this application; Figure 2 This is a graph showing the results of a wind power ramping event detection method according to an embodiment of this application. Figure 3 This is a flowchart illustrating a wind power ramp-up event detection method according to an embodiment of this application. Detailed Implementation
[0028] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.
[0029] The following describes a wind power ramp-up event detection method according to an embodiment of this application, with reference to the accompanying drawings. As mentioned in the background art, wind power can experience sudden increases or decreases in a short period. However, due to the influence of wind speed turbulence and turbine layout, ramp-up events for different turbines exhibit time shifts, such as upstream turbines changing first and downstream turbines responding with a delay, as well as amplitude differences in wind speed during flow. Traditional wind power ramp-up judgment threshold methods are ill-suited to the dynamic operating conditions of wind farms, resulting in poor identification accuracy. The commonly used rotating door algorithm (SDA) can only process single time series and cannot align multi-turbine data, easily misjudging normal fluctuations as ramp-ups. This application provides a novel detection method that combines the pattern recognition capability of the dynamic time warping algorithm with the efficient compression characteristics of the rotating door algorithm. The Dynamic Rotating Compression (DSC) algorithm, by mining the spatiotemporal features and morphological patterns of the power sequence, achieves accurate identification and quantification of different types of ramp-up events. This solves the problems in related technologies, such as the wind power ramp determination threshold method being difficult to adapt to the dynamic operating conditions of wind farms and having poor identification accuracy, while the commonly used rotating door algorithm (SDA) can only process single time series, cannot align data from multiple wind turbines, and is prone to misjudging normal fluctuations as ramps.
[0030] Specifically, Figure 1 This is a flowchart illustrating a wind power ramp-up event detection method provided in an embodiment of this application.
[0031] like Figure 1 As shown, the wind power ramp-up event detection method includes the following steps: In step S101, the wind power sequence of each wind turbine in the wind farm is obtained. The wind power sequence is dynamically weighted according to the wind turbine azimuth and upstream and downstream spatial positions. Based on the calculation results, a reference sequence is simulated using a dynamic weighting strategy to calculate the dynamic time warping score of each wind turbine power sequence and the reference sequence, and obtain the score that meets the preset optimal conditions. The timestamps of each wind turbine power sequence are adjusted according to the score that meets the preset optimal conditions to obtain the alignment sequence of each wind turbine.
[0032] In actual implementation, the embodiments of this application include: Step 1. Establish a reference sequence that aligns with the wind turbine sequence.
[0033] A virtual power time series, called the reference series, is established to represent the overall dynamic behavior of the wind farm. The reference series is not the actual power of any individual turbine, but rather an idealized signal constructed by fusing information from the entire farm. It provides a unified time scale, allowing power changes of all turbines to be aligned and compared on this scale. This eliminates time offsets caused by spatial heterogeneity among the turbines and enhances the robustness of alignment. The wind power data for a given month from 48 turbines in a wind farm is obtained, with a time resolution of 10 minutes.
[0034] Step 1.1 Construct a wind power reference sequence. Instead of directly using data from a single wind turbine, a dynamically weighted reference sequence is generated. The reference sequence values are calculated based on the spatial location of each wind turbine. .
[0035] Step 1.1.1 Determine the azimuth angle of the wind turbine and its spatial position upstream and downstream.
[0036] Set the center of the wind farm or the location of the meteorological tower as the origin of the coordinate system. Obtain information for each wind turbine. Cartesian coordinates Time obtained from the wind measurement tower at the wind farm Real-time wind direction .
[0037] The process involves dynamically calculating the spatial weight of the wind power sequence based on the azimuth angle of the wind turbines and their upstream and downstream spatial locations. This includes determining the azimuth angle of the wind turbines and their upstream and downstream spatial locations within the wind farm. The formula for calculating the azimuth angle of the wind turbines is as follows: (1) in, i For the fan, The origin of the coordinate system is the center of the wind farm or the location of the meteorological tower. Let each wind turbine be represented by a Cartesian coordinate system. The upstream and downstream determination logic is: when the wind turbine The difference between the azimuth and the real-time wind direction satisfies And wind turbine blade yaw angle deviation ( When the fan is running, Upwind is the location of the wind turbine, while downwind is the location of the downstream wind turbine.
[0038] Step 1.1.2 Calculation of dynamic spatial weights.
[0039] Based on the azimuth angle of the wind turbine and the spatial location of the upstream and downstream areas, this application calculates the distance of each wind turbine from the origin according to the upstream and downstream relationships. At any moment Dynamic weights This ensures that upstream wind turbines have a higher weight, and that upstream wind turbines closer to the origin have a higher weight. The dynamic weight calculation formula for each wind turbine at a preset time is as follows: (2) in, Indicates wind turbine The straight-line distance from the origin, in meters. The closer the distance, the higher the weighting coefficient. The larger; It is a tiny constant; The wind direction matching coefficient for the upstream wind turbine, when and When the fan is directly upstream, the coefficient is 1. The wind direction is crosswind, and the coefficient is 0. This is the weight attenuation coefficient for downstream wind turbines, ensuring that the weight of downstream wind turbines is lower than that of upstream wind turbines.
[0040] Step 1.1.3 Generate a reference sequence using a dynamic weighting strategy.
[0041] Multiply the normalized power of each wind turbine by its corresponding dynamic weight, and sum them to obtain the time step. The weighted reference power is used to replace the original average total power. This application can calculate the dynamic weighted reference power of each wind turbine based on the dynamic weights and the turbine's power, determine the dynamic weighting strategy based on the dynamic weighted reference power, and execute the dynamic weighting strategy to simulate a reference sequence. The formula for calculating the dynamic weighted reference power is: (3) The generated sequence It can prioritize reflecting the power change trend of upstream wind turbines, better aligning with the spatial heterogeneity characteristics of wind power ramping events where upstream responses are earlier and downstream responses are later, avoiding interference from ineffective upstream wind turbines, and providing a more accurate reference benchmark for subsequent DTW alignment. Figure 3 The process of calculating and obtaining the reference sequence is shown.
[0042] Step 1.2 DTW multi-fan timing alignment optimization.
[0043] The core of DTW is to find the optimal alignment path through dynamic programming in order to eliminate the time offset of multiple wind turbine ramping events. The key formulas include two parts: the DTW distance matrix and dynamic programming recursion.
[0044] The process involves calculating the dynamic time warping score (DTW) between the power sequence of each wind turbine and the reference sequence to obtain a score that meets a preset optimal condition. The timestamps of each wind turbine's power sequence are then adjusted based on these scores to obtain an aligned sequence for each turbine. This includes: First, calculating the similarity distance between the reference and target sequences at each time point. The formula for calculating the similarity distance is: (4) in, For the reference sequence and the real sequence at time... With time The single-point Euclidean distance, For the reference sequence at time power, For the true sequence at time... The power.
[0045] Based on the similarity distance, the globally optimal alignment path score between the real sequence and the reference sequence is calculated recursively. The formula for calculating the globally optimal alignment path score is as follows: (5) in, Indicates the sequence starts from the beginning. arrive The cumulative DTW distance. Boundary conditions are: Based on the optimal path ( (where the path length is used), the timestamps of the actual sequence are adjusted to the timestamps of the reference sequence. That is, the timestamps of the power sequences of each wind turbine are adjusted according to the best score to obtain the aligned sequence. The formula for calculating the aligned sequence is: (6) In step S102, based on the alignment sequence of each wind turbine, the adaptive dynamic rotating door threshold of each wind turbine is adjusted to obtain the dynamic door width of each wind turbine, and key change points exceeding the dynamic door width are extracted according to the deviation between power to generate a set of key change points for each wind turbine.
[0046] The embodiments of this application may include: Step 2. Extract key change points from the alignment sequence of each wind turbine.
[0047] Step 2.1 Adjust the adaptive dynamic rotating door threshold.
[0048] The dynamic gate width is adjusted in real time based on the DTW alignment result, solving the problem that a fixed threshold is not suitable for dynamic changes. The following dynamic gate width calculation formula is introduced: (7) in, For a moment Dynamic gate width; The base gate width is obtained by calibration using historical data. This is the volatility coefficient; For a moment The standard deviation of power within the nearby sliding window, This represents the average power within the window. This is a correction factor; For a moment The larger the DTW alignment error, the smaller the gate width to improve sensitivity.
[0049] Step 2.2. Extract key change points using SDA dynamic compression.
[0050] Data exceeding the dynamic gate width is identified using SDA; key changes are preserved. This applies to the aligned sequence. , This represents the number of data points. Starting from the current index. power Begin by calculating subsequent points. With line segment The deviation, where the formula for calculating the deviation between power values is: (8) in, The current starting index power, For a moment The power of.
[0051] Criteria for identifying key change points: If , Within a dynamic revolving door, this is considered normal fluctuation and is not retained; if , Once outside the revolving door, mark it as a key change point and update the starting point to [value]. Then recalculate the subsequent deviations. The final output is a set of key change points. , The number of key change points. Timestamps of key change points This corresponds to the power.
[0052] To address the issues of fixed thresholds for current ramp event detection and poor detection performance caused by asynchronous operation of multiple wind turbines, this application organically combines Time Series Alignment (DTW) and Data Compression (SDA) algorithms to develop a wind power ramp event detection method based on DTW-SDA fusion, namely Dynamic Rotation Compression (DSC).
[0053] In step S103, based on the set of key change points for each wind turbine, the power change rate within the sliding window is calculated. Based on the power change rate within the sliding window, when each wind turbine is detected to meet the preset ramp-up event judgment conditions, the wind power ramp-up event of each wind turbine is determined. When the ratio of wind turbines that have experienced wind power ramp-up events to the total number of wind turbines in the wind farm reaches a preset threshold, the overall wind power ramp-up event of the wind farm is determined.
[0054] In actual implementation, the embodiments of this application may include: Step 3: Wind farm ramp-up event detection.
[0055] Step 3.1 Determine the formula for judging wind power ramp-up events.
[0056] The climbing event is defined by the rate of change of power during the climbing event, and the set of key change points is used. Using sliding windows Calculate the rate of change of power within the window, where the formula for calculating the rate of change of power within the window is: (9) in, Key Change Points Starting point, time window The rate of change of power within; For time windows The number of key change points contained within; This represents the power at the key change point at the end of the time window.
[0057] Step 3.2 Determining the wind farm ramp-up event.
[0058] For a single wind turbine power, the ramp-up event determination condition is set as follows: (10) in, The intensity of the event.
[0059] Specifically, when each wind turbine is detected to meet the preset ramp-up event judgment conditions, the wind power ramp-up event of each wind turbine is determined, including: when more than a preset proportion of wind turbines in the wind farm are detected to meet the positive ramp-up event or negative ramp-up event in the preset ramp-up event judgment conditions within the same time period, the wind turbine of the wind farm is determined to have experienced a positive ramp-up event or a negative ramp-up event; within two adjacent time windows, when a single wind turbine in the wind farm is detected to simultaneously meet the positive ramp-up event and the negative ramp-up event in the preset ramp-up event judgment conditions, the single wind turbine of the wind farm is determined to have experienced the same ramp-up event.
[0060] In this application, for the entire wind turbine unit of the wind farm, at least 70% of the turbines must simultaneously experience a positive or negative ramp event before it is determined to be a positive or negative ramp event, in order to avoid false alarms caused by turbine abnormalities. Figure 2 The results of the hill-climbing event detection in this application are displayed.
[0061] If two adjacent time windows both meet the above conditions, they are merged into the same climbing event, and the event boundary is output. With event intensity .
[0062] After identifying the overall wind power ramp-up event of the wind farm, the process also includes: detecting the overall wind power ramp-up event to generate detection results of the ramp-up event; and verifying the detection results to determine the proportional relationship in the ramp-up event.
[0063] This application allows for the verification of test results to calculate precision and recall.
[0064] Precision represents the proportion of climbing events detected by the model that actually occurred. (11) Recall represents the proportion of actual, and precisely detected, climbing events out of all actual climbing events. (12) Of the 20 real-world ramping events detected that month, this application detected 19, of which 94.7% were genuine. The detection results for ramping events are shown in Table 1. The precision of this method is 15.8 percentage points higher than that of traditional SDA, indicating that the dynamic threshold effectively reduces false positives; the recall is 15 percentage points higher than that of traditional SDA, indicating that the SDA compression and retention of key inflection points combined with DTW alignment verification in this application can effectively reduce false negatives.
[0065] Table 1 shows the results of the detection method, as illustrated in Table 1.
[0066] Table 1
[0067] Furthermore, Figure 3 This includes reference sequence simulation, SDA dynamic compression extraction of key change points, and hill-climbing event determination.
[0068] This application organically combines algorithms with two different objectives: time series alignment and data compression. It proposes a wind power ramping event detection method based on DTW-SDA fusion. By simultaneously considering time alignment and spatial correlation, a reference sequence is established for matching to obtain the adaptive dynamic threshold of SDA. This solves the problems of poor detection effect caused by the fixed ramping event detection threshold and the asynchronous operation of multiple wind turbines, and finally achieves accurate ramping detection of wind power in wind farms.
[0069] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0070] 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, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0071] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.
Claims
1. A method for detecting wind power ramp-up events, characterized in that, Includes the following steps: The wind power sequence of each wind turbine in the wind farm is obtained. The wind power sequence is dynamically weighted according to the wind turbine azimuth and upstream and downstream spatial positions. Based on the calculation results, a reference sequence is simulated using a dynamic weighting strategy. The dynamic time warping score of each wind turbine power sequence and the reference sequence is calculated to obtain the score that meets the preset optimal conditions. The timestamp of each wind turbine power sequence is adjusted according to the score that meets the preset optimal conditions to obtain the alignment sequence of each wind turbine. Based on the alignment sequence of each wind turbine, the adaptive dynamic rotating door threshold of each wind turbine is adjusted to obtain the dynamic door width of each wind turbine, and key change points exceeding the dynamic door width are extracted according to the deviation between power to generate a set of key change points for each wind turbine. Based on the set of key change points for each wind turbine, the power change rate within a sliding window is calculated. Based on the power change rate within the sliding window, when each wind turbine is detected to meet the preset ramp-up event determination conditions, a wind power ramp-up event for each wind turbine is determined. When the ratio of the wind turbines that have experienced the wind power ramp-up event to the total number of wind turbines in the wind farm reaches a preset threshold, an overall wind power ramp-up event for the wind farm is determined.
2. The method according to claim 1, characterized in that, The dynamic spatial weighting calculation of the wind power sequence based on the wind turbine azimuth and upstream and downstream spatial locations includes: The azimuth angles and upstream and downstream spatial positions of the wind turbines in the wind farm are determined, wherein the formula for calculating the azimuth angle of the wind turbines is: in, i For the aforementioned fan, The origin of the coordinate system is the center of the wind farm or the location of the meteorological tower. Let be the plane rectangular coordinates of each wind turbine; Based on the azimuth angle of the wind turbines and their upstream and downstream spatial positions, the dynamic weight of each wind turbine at a preset time is calculated, wherein the formula for calculating the dynamic weight of each wind turbine at the preset time is: in, Indicates wind turbine The straight-line distance from the origin Indicates the weighting coefficient. It is a tiny constant. The wind direction matching coefficient for the upstream wind turbine. This is the weight attenuation coefficient for downstream wind turbines.
3. The method according to claim 2, characterized in that, The process of simulating the reference sequence using a dynamic weighting strategy includes: The dynamically weighted reference power of each wind turbine is calculated based on the dynamic weights and the power of each turbine. A dynamic weighting strategy is then determined based on the dynamically weighted reference power, and the strategy is executed to simulate the reference sequence. The formula for calculating the dynamically weighted reference power is as follows: in, For the generated sequence, t For a moment.
4. The method according to claim 1, characterized in that, The calculation of the dynamic time warping score of each wind turbine power sequence and the reference sequence to obtain a score that meets a preset optimal condition, and the adjustment of the timestamps of each wind turbine power sequence based on the score that meets the preset optimal condition to obtain the alignment sequence of each wind turbine, includes: Calculate the similarity distance between the reference sequence and the target sequence at each time point, wherein the formula for calculating the similarity distance is: in, For the reference sequence and the real sequence at time... With time The single-point Euclidean distance, For the reference sequence at time power, For the real sequence at time... The power; Based on the similarity distance, the globally optimal alignment path score between the real sequence and the reference sequence is calculated recursively, wherein the formula for calculating the globally optimal alignment path score is: in, Indicates the sequence starts from the beginning. arrive The cumulative DTW distance; Based on the globally optimal alignment path score, the real sequence timestamp is adjusted to the reference sequence timestamp to obtain the aligned sequence. The formula for calculating the aligned sequence is as follows: in, The power of the real sequence at that time.
5. The method according to claim 1, characterized in that, The formula for calculating the dynamic gate width is: in, For a moment Dynamic door width, Based on the basic door width, For fluctuation coefficient, For the time mentioned The standard deviation of power within the nearby sliding window, The average power within this window. For correction factor, For the time mentioned DTW alignment error.
6. The method according to claim 5, characterized in that, The formula for calculating the deviation between the power values is: in, The current starting index power, For the time mentioned The power of.
7. The method according to claim 1, characterized in that, The formula for calculating the power change rate within the window is: in, Key Change Points Starting point, time window The rate of change of power within, For time windows The number of key change points contained within. This represents the power at the key change point at the end of the time window.
8. The method according to claim 1, characterized in that, The preset criteria for determining the hill-climbing event are as follows: in, The intensity of the event.
9. The method according to claim 1, characterized in that, The step of determining the wind power ramp-up event for each wind turbine when it is detected that each wind turbine meets the preset ramp-up event determination conditions includes: When it is detected that more than a preset proportion of wind turbines in the wind farm meet the positive or negative ramp event in the preset ramp event determination conditions within the same time period, it is determined that the wind turbines in the wind farm have experienced the positive or negative ramp event. Within two adjacent time windows, if a single wind turbine in the wind farm is detected to simultaneously satisfy both a positive and a negative ramp event in the preset ramp event determination conditions, it is determined that the single wind turbine in the wind farm has experienced the same ramp event.
10. The method according to claim 1, characterized in that, After determining the overall wind power ramp-up event of the wind farm, the process also includes: The overall wind power ramp-up events are detected to generate detection results for the ramp-up events; The test results are examined to determine the proportional relationship in the hill-climbing events.