Automatic identification method for wind power climbing event

Abstract

The invention relates to an automatic identification method for a wind power climbing event. The method specifically comprises the following steps: S1, acquiring configuration information and meteorological information of a tested wind power station; s2, acquiring wind power data of the tested wind power station, and establishing an association relationship with meteorological data information in the meteorological information; s3, acquiring feature information of a power climbing event of the tested wind power station, and constructing a power climbing kernel and a climbing section according to the feature information; and S4, screening a power climbing event. According to the method, the climbing event strictly conforming to the definition of the wind power climbing event is efficiently analyzed by using the greedy snake algorithm, and the phenomena of false detection and missing detection of the wind power climbing event are completely eradicated; and meanwhile, the wind power climbing event obviously associated with the weather information is extracted, and a basis is provided for carrying out forecasting service of the wind power climbing event in weather forecasting.

Description

Technical Field

[0001] The invention relates to an applied meteorological technology, in particular to an automatic identification method for wind power ramp-up events. Background Art

[0002] Due to its inherent randomness and volatility, wind power generation often shows short-term oscillations in output power, frequency and phase. This phenomenon will have a huge impact on the stability and economy of the power system, and accordingly will greatly reduce the reliability and power supply quality of the power system operation, and even cause serious power outages. The power ramp event of the wind power system refers to the drastic change in the active power output of the wind farm in a short period of time. Specifically, under the climatic conditions of surging wind speed, the wind power will increase significantly in a short period of time, and the increase may even exceed the backup capacity or backup regulation rate of the power system, thus forming an up-climbing event. Under severe meteorological conditions, when the wind speed is greater than the cut-out wind speed, the impeller of the wind turbine will automatically stop under the action of the safety protection device, which will cause the wind power to drop rapidly, resulting in a down-climbing event. When a down-climbing event occurs, with regard to current wind power technology, the backup power provided by other generators cannot make up for the power shortage caused by power fluctuations. This type of event is an extreme manifestation of the randomness and volatility of new energy power generation. When severe convective weather occurs, wind farms will proactively cut off most of the wind turbines connected to the power grid to provide artificial protection.

[0003] Generally, the greater the ramp amplitude and the shorter the duration, the more serious the harm of the ramp event. According to the severity classification, the ramp events of renewable energy power generation can be divided into mild, moderate and severe ramp events. How to smooth the fluctuation characteristics of the output power of large-scale wind farms, the first thing is to be able to effectively identify power ramp events, which is an urgent problem to be solved in improving the utilization rate of wind energy. Therefore, accurate identification of wind power ramp events is a primary prerequisite for smoothing the fluctuation of wind farm output power. Summary of the invention

[0004] The purpose of the present invention is to provide a method for automatically identifying wind power ramp events, so as to achieve accurate identification of wind power ramp events, thereby contributing to the safe operation and reasonable scheduling of the power system.

[0005] The object of the present invention is achieved in that:

[0006] A method for automatically identifying a wind power ramp event comprises the following steps:

[0007] S1. Obtain configuration information of the wind power station under test, including site area information, installed parameters, installed capacity and the section to which it belongs, as well as meteorological information of the area where the wind power station under test is located, including precipitation, temperature, air pressure, wind and humidity.

[0008] S2. Obtain wind power data of the wind power station under test, and preprocess the wind power data including deleting outliers and interpolating missing power data. Set up a numerical sequence for the processed wind power data, and establish a correlation with the meteorological data information in the meteorological historical information to obtain a data set X in the form of a matrix.

[0009] S3, obtaining characteristic information of the ramp event in the measured wind power station within a certain time range, wherein the characteristic information includes the power change threshold P threshold , time threshold T threshold and minimum climbing rate λ rate , and accordingly construct the power ramp core and ramp segment respectively.

[0010] S4, screening power ramp events, including the following sub-steps:

[0011] S4-1 Set power ramp event: at time threshold T threshold The difference between the maximum and minimum wind power is greater than or equal to the power change threshold P threshold , minimum climbing rate λ rate =P threshold / T threshold .

[0012] S4-2 In the detection of power ramp events, as long as the power change range is not less than the power change threshold P in a certain time interval, threshold , and the climbing rate is not less than the minimum climbing rate λ rate , that is, it is judged to constitute a power ramp event.

[0013] Furthermore, the data structure of the data set X is:

[0014]

[0015] The first column of the matrix is ​​the time point information T = (T 1 , T j ,…,T n ); the second column of the matrix is ​​the rainfall information associated with time T Y = (Y 1 ,Y j ,…,Y n ); The third column of the matrix is ​​the wind speed information V associated with time T = (V 1 ,V j ,…,V n); the fourth column of the matrix is ​​the associated air pressure information PA at time T = (PA 1 ,PA j ,…,PA n ); The fifth column of the matrix is ​​the temperature information associated with time T TEMP = (TEMP 1 ,TEMP j ,…,TEMP n ); The sixth column of the matrix is ​​the humidity information RH associated with time T = (RH 1 ,RH j ,…,RH n ); The seventh column of the matrix is ​​the wind power data associated with time T P = (P 1 ,P j ,…,P n ).

[0016] Furthermore, the power ramp kernel is constructed by using a first-order difference transformation matrix M to transform the wind power data P = (P 1 ,P j ,…,P n ) to obtain the first-order difference signal of wind power data:

[0017] P f =(p 1 f ,p j f ,…,p n f )

[0018] Among them, P f is the set of power ramp kernels on the entire wind power dataset; p j f is the wind power ramp kernel corresponding to time j.

[0019] Furthermore, the ramping section is constructed by setting the ramping section to the absolute value of the power ramping kernel |p j f |Continuously greater than or equal to the minimum climbing rate λ that constitutes a climbing event rate When the value of the power ramp kernel is positive, it is the rising slope of the power ramp; when the value of the power ramp kernel is negative, it is the falling slope of the power ramp; other areas are invalid areas of the power ramp event.

[0020] Furthermore, the specific construction process of the climbing section includes the following sub-steps:

[0021] S3-1 sets the climbing segment set to CS:

[0022] CS={cs i|i>=1,i<=n}

[0023] Among them, cs i is a single climbing segment, i represents the order value of the climbing segment in the set CS, and n represents the number of climbing segments contained in the set CS.

[0024] In each climbing section cs i It includes the start time, end time, absolute value of the climbing amplitude and the climbing direction.

[0025] S3-2 Create the first climbing segment cs 1 , traverse the set P of power ramp kernels f The trend value of each data point at a moment in the rate When the power ramps up, this moment is taken as the first ramping segment cs 1 The starting time; continue to traverse the set P backward f , when encountering the first one less than the minimum climbing rate λ rate When the power ramp core is used, the moment before this moment is taken as the first ramp segment cs 1 At the same time, the difference in wind power between the end time and the start time of the first climbing segment is calculated, and its absolute value is taken as the climbing amplitude, and recorded to the first climbing segment cs 1 Then take its numerical sign as the climbing direction, with “–” representing the descending slope and “+” representing the ascending slope; the first climbing slope cs 1 The data is put into the climbing segment set CS.

[0026] S3-3 Create the next climbing segment cs 2 , continue to traverse the set P of power ramp kernels f The trend value of each data point at a certain moment in the process is determined by referring to the method in sub-step S3-2 to locate the next climbing segment cs 2 The start and end time of the next climbing segment cs 2 The climbing amplitude and climbing direction, and then the next climbing segment cs 2 The data is put into the climbing segment set CS.

[0027] S3-4 repeats sub-step S3-3, and finally puts the data of all climbing segments into the climbing segment set CS.

[0028] Furthermore, in the process of detecting the power ramp event in step S4-2, the ramp segments are merged using the snake method, and the specific merging method includes the following sub-steps:

[0029] S4-2-1 sets the set PS of all climbing events to:

[0030] PS = {ps i |i≥1,i≤n}

[0031] Among them, ps i is a single climbing event, which contains four parts of information: start time, end time, absolute value of climbing amplitude and climbing direction; i represents the order value of the climbing event in the set PS; n represents the total number of climbing events contained in the set PS;

[0032] S4-2-2 Setting the time domain extension caliper T ruler , set the caliper time length to the time domain value T in the wind power ramp event threshold ;

[0033] S4-2-3 First, extend the time domain extension caliper T backward for a certain climbing section that needs to be extended and merged later. ruler The first step is to eat up the longest climbing segment as much as possible in the time window. Secondly, if other climbing segments have been eaten up in the previous time window extension, try to extend it again, and the starting point is the end of the previous time window extension. Thirdly, add an extension time window, and so on, until no other climbing segments have been eaten up in a certain extension time window.

[0034] The present invention utilizes the snake algorithm to efficiently parse out the climbing events that strictly meet the definition of wind power climbing events, thereby eliminating the false detection and missed detection of wind power climbing events; at the same time, it is capable of extracting wind power climbing events that have a clear correlation with meteorological information, thus providing a basis for carrying out forecast services for wind power climbing events in future meteorological forecasts. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a minute-by-minute graph of wind power changes and ramp events at a certain wind power station from 12:00 to 18:00 on May 3, 2022. DETAILED DESCRIPTION

[0036] The present invention will be further described below in conjunction with the accompanying drawings.

[0037] The automatic identification method of wind power ramping events of the present invention comprises the following steps:

[0038] S1. Obtain configuration information of the wind power station under test, including site area information, installed parameters, installed capacity and the section to which it belongs, as well as meteorological information of the area where the wind power station under test is located, including precipitation, temperature, air pressure, wind and humidity.

[0039] S2. Obtain wind power data of the wind power station under test, and preprocess the wind power data, including deleting abnormal values ​​and interpolating missing power data. Set up a numerical sequence for the processed wind power data, and establish a correlation with the meteorological data information in the meteorological historical information to obtain a data set X in the form of a matrix. The data structure of data set X is:

[0040]

[0041] The first column of the matrix is ​​the time point information T = (T 1 , T j ,…,T n ); the second column of the matrix is ​​the rainfall information associated with time T Y = (Y 1 ,Y j ,…,Y n ); The third column of the matrix is ​​the wind speed information V associated with time T = (V 1 ,V j ,…,V n ); the fourth column of the matrix is ​​the associated air pressure information PA at time T = (PA 1 ,PA j ,…,PA n ); The fifth column of the matrix is ​​the temperature information associated with time T TEMP = (TEMP 1 ,TEMP j ,…,TEMP n ); The sixth column of the matrix is ​​the humidity information RH associated with time T = (RH 1 ,RH j ,…,RH n ); The seventh column of the matrix is ​​the wind power data associated with time T P = (P 1 ,P j ,…,P n ).

[0042] S3, obtaining characteristic information of the ramp event of the wind power station under test, the characteristic information including the power change threshold P threshold , time threshold T threshold and minimum climbing rate λ rate , and accordingly construct the power ramp core and ramp segment respectively.

[0043] The power ramp kernel is constructed by using the first-order difference transformation matrix M to transform the wind power data P = (P 1 ,P j ,…,P n ) is transformed to obtain the first-order difference signal of wind power data. The first-order difference transformation matrix M is:

[0044]

[0045] The minute-by-minute wind power value of the wind farm under test in May 2022 is P = (P 1 ,P j ,…,P n ) performs first-order difference signal transformation to obtain a new signal sequence:

[0046] P f =(p 1 f ,p j f ,…,p n f )

[0047] Among them, P f is the set of power ramp kernels on the entire wind power dataset; p j f is the wind power ramp kernel corresponding to time j.

[0048] The ramp segment is constructed by setting the ramp segment to the absolute value of the power ramp kernel |p j f |Continuously greater than or equal to the minimum climbing rate λ that constitutes a climbing event rate The interval of power ramp kernel is positive, which is the rising slope of power ramp; when the value of power ramp kernel is negative, it is the falling slope of power ramp; other areas are invalid areas of power ramp events. It is impossible to form a ramp event in the invalid area. Since the ramp event is a low-probability event, most of the data areas on the wind power data set are eliminated, which greatly saves the processing time for detecting wind power ramp events and improves efficiency.

[0049] The specific construction process of the climbing section includes the following sub-steps:

[0050] S3-1 sets the climbing segment set CS as:

[0051] CS={cs i |i≥1,i≤n}

[0052] Among them, cs i is a single climbing segment, i represents the order value of the climbing segment in the set CS, and n represents the number of climbing segments contained in the set CS. i It includes the start time, end time, absolute value of the climbing amplitude and the climbing direction.

[0053] S3-2 Create the first climbing segment cs 1 , traverse the set P of power ramp kernels f The trend value of each data point at a moment in the rateWhen the power ramps up, this moment is taken as the first ramping segment cs 1 The starting time; continue to traverse the set P backward f , when encountering the first one less than the minimum climbing rate λ rate When the power ramp core is used, the moment before this moment is taken as the first ramp segment cs 1 At the same time, the difference in wind power between the end time and the start time of the first climbing segment is calculated, and its absolute value is taken as the climbing amplitude, and recorded to the first climbing segment cs 1 Then take its numerical sign as the climbing direction, with “–” representing the descending slope and “+” representing the ascending slope; the first climbing slope cs 1 The data is put into the climbing segment set CS.

[0054] S3-3 Create the next climbing segment cs 2 , continue to traverse the set P of power ramp kernels f The trend value of each data point at a certain moment in the process is determined by referring to the method in sub-step S3-2 to locate the next climbing segment cs 2 The start and end time of the next climbing segment cs 2 The climbing amplitude and climbing direction, and then the next climbing segment cs 2 The data is put into the climbing segment set CS.

[0055] S3-4 repeats sub-step S3-3, and finally puts the data of all climbing segments into the climbing segment set CS.

[0056] S4, screening power ramp events, including the following sub-steps:

[0057] S4-1 Set power ramp event: at time threshold T threshold The difference between the maximum and minimum wind power is greater than or equal to the power change threshold P threshold , minimum climbing rate λ rate =P threshold / T threshold .

[0058] S4-2 In the detection of power ramp events, as long as the power change range is not less than the power change threshold P in a certain time interval, threshold , and the climbing rate is not less than the minimum climbing rate λ rate , that is, it is judged to constitute a power ramp event.

[0059] In the previous step, a set of all ramp segments CS on the wind power data set is constructed. The power value changes in the ramp segments contained in it are absolutely monotonic, and the absolute values ​​of the slopes are greater than or equal to the minimum ramp rate λ in the definition of the ramp event. rate, the climbing segment set CS only includes monotonically rising climbing segments and monotonically falling climbing segments, excluding the climbing intervals that will not constitute a climbing event. This is because when the absolute value of the slope of a single climbing kernel is less than the minimum climbing rate λ rate When the power change amplitude is greater than or equal to the power change threshold P in the definition of wind power ramp event, no matter how long the accumulated time is, threshold , will not constitute a wind power ramp event.

[0060] Based on the snake algorithm, all the climbing segments in the climbing segment set CS are extended to the maximum extent possible to eat up as many mergible climbing segments as possible. The specific merging method includes the following sub-steps:

[0061] S4-2-1 sets the set PS of all climbing events to:

[0062] PS = {ps i |i≥1,i≤n}

[0063] Among them, ps i is a single climbing event, which contains four pieces of information: start time, end time, absolute value of climbing amplitude and climbing direction; i represents the order value of the climbing event in the set PS; n represents the total number of climbing events contained in the set PS.

[0064] S4-2-2 Setting the time domain extension caliper T ruler , set the caliper time length to the time domain value T in the wind power ramp event threshold Since a single climbing segment cs in the climbing segment set CS i It may not meet the definition of a ramp event, that is, the wind power ramp amplitude may not reach the threshold P threshold , but in the wind power set P = (P 1 ,P j ,…,P n ), by extending a certain time window backward and merging other climbing segments in the time window, it is possible to meet the requirements of the climbing event definition.

[0065] S4-2-3 First, extend the time domain extension caliper T backward for a certain climbing section that needs to be extended and merged later. ruler The first step is to eat up the longest climbing segment as much as possible in the time window. Secondly, if other climbing segments have been eaten up in the previous time window extension, try to extend it again, and the starting point is the end of the previous time window extension. Thirdly, add an extension time window, and so on, until no other climbing segments have been eaten up in a certain extension time window.

[0066] The idea of ​​merging climbing segments by the snake method is to extend all climbing segments in the climbing segment set CS to the maximum extent and eat up as many merging climbing segments as possible. This merging process is based on two aspects. First, a climbing segment that needs to be extended and merged is extended backward by T. ruler A time window of length, in which the longest climbing segment is eaten as much as possible. Then, as long as other climbing segments have been eaten in the previous time window extension, another extension will be tried. The starting point is the end after the last extension, and another extension time window is added, and so on, until no other climbing segments are eaten in a certain extension time window.

[0067] The specific operation method of step S4-2-3 includes the following sub-steps:

[0068] S4-2-3-1 Create a new climbing event ps m ; Among them, ps represents the climbing event, m represents the position of the climbing event in the set PS, and the initial position is 1.

[0069] S4-2-3-2 Take out a climbing segment cs to be extended from the climbing segment set CS j ; where cs represents the climbing segment, j represents the position of the climbing segment in the climbing segment set CS, and the initial value of j is 1; let the climbing segment to be extended cs j The start time is t 0 , the end time is t 1 The absolute value of the climbing amplitude is Δp 1 , climbing direction is d 1 , assuming that the wind power is at the start time t 0 The value of p 0 , at the end time t 1 The value of p 1 ; If the climbing segment set CS is traversed, the merging process ends; otherwise, execute step S4-2-3-3.

[0070] S4-2-3-3 takes the next climbing segment cs from the climbing segment set CS again i ; where cs represents a climbing segment, i represents the position of the climbing segment in the climbing segment set CS, and the initial value of i is j+1. i The start time is t 2 , the end time is t 3 , assuming that the wind power is at the start time t 2 The value of p 2 , at the end time t 3 The value of p 3 ; If a new climbing section cs can be obtained i , and t 2 –t1 ≤T ruler , then execute step S4-2-3-4; otherwise, detect |p 1 –p 0 |≥P threshold and|p 1 –p 0 |≥λ rate Are they satisfied at the same time? If they are satisfied at the same time, then the climbing section cs j Information copied to the climbing event ps m , and add the climbing event ps m Go to the climbing event set PS; then set j to the position j+1, add 1 to the value of m, and repeat step S4-2-3-1.

[0071] S4-2-3-4 traverses the wind power data set P at the start time t 1 To end time t 2 The data at each time point between i |i≥t 1 ,i <t 2 ), if the climbing section cs j is the ascending climbing section, and p i ≤p 0 , or climbing section cs j is the descending climbing section, and p i ≥p 0 , then let j increase by 1 and go to step S4-2-3-2. In other cases, continue to the next step S4-2-3-5.

[0072] S4-2-3-5 will be the climbing section cs i Try to merge into the climbing segment cs j In: If p 3 –p 0 ≥P threshold , and (p 3 –p 0 ) / (t3–t0)≥λ rate , then use the climbing section cs i Extended climbing section cs j , about to climb the slope cs j The end position is located in the climbing section cs i The end position of the , sets the value of j to the value of i, and updates the start time t 1 and the wind power p at the corresponding start time 1 value, repeat step S4-2-3-3.

[0073] Through the actual detection of a wind power station, the ramp events of the wind farm in May 2022 were finally obtained, as shown in Table 1. The minute-by-minute wind power changes and ramp events of the wind power station from 12:00 to 18:00 on May 3, 2022 are as follows: Figure 1 shown.

[0074] Table 1: Power ramp events of a wind power station in May 2022

[0075]

[0076]

Claims

1. A method for automatically identifying wind power ramp events, characterized in that: The following steps are involved: S1. Obtain configuration information of the wind power station under test, including site area information, installed parameters, installed capacity and the section to which it belongs, as well as meteorological information of the area where the wind power station under test is located, including precipitation, temperature, air pressure, wind and humidity; S2, obtaining wind power data of the wind power station under test, and preprocessing the wind power data including deleting abnormal values ​​and interpolating missing power data, establishing a numerical sequence for the processed wind power data, and establishing a correlation relationship with the meteorological data information in the meteorological information, to obtain a data set X expressed in a matrix form; S3, obtaining characteristic information of the power ramp event of the wind power station under test, wherein the characteristic information includes the power change threshold P threshold , time threshold T threshold and minimum climbing rate λ rate , and accordingly construct the power ramp core and ramp segment respectively; S4, screening power ramp events, including the following sub-steps: S4-1 Set power ramp event: at time threshold T threshold The difference between the maximum and minimum wind power is greater than or equal to the power change threshold P threshold , minimum climbing rate λ rate =P threshold / T threshold ; S4-2 In the detection of power ramp events, as long as the power change range is not less than the power change threshold P in a certain time interval, threshold , and the climbing rate is not less than the minimum climbing rate λ rate , that is, it is judged to constitute a power ramp event.

2. The method for automatically identifying wind power ramp events according to claim 1, characterized in that: The data structure of the data set X is: The first column of the matrix is ​​the time point information T = (T1, T j ,…,T n ); the second column of the matrix is ​​the rainfall information associated with time T Y = (Y1,Y j ,…,Y n ); The third column of the matrix is ​​the wind speed information associated with time T V = (V1, V j ,…,V n ); the fourth column of the matrix is ​​the associated air pressure information PA at time T = (PA1, PA j ,…,PA n ); The fifth column of the matrix is ​​the temperature information associated with time T TEMP = (TEMP1, TEMP j ,…,TEMP n ); The sixth column of the matrix is ​​the humidity information associated with time T RH = (RH1, RH j ,…,RH n ); The seventh column of the matrix is ​​the wind power data associated with time T P = (P1, P j ,…,P n ).

3. The method for automatically identifying wind power ramp events according to claim 1, characterized in that: The power ramp kernel is constructed by using a first-order difference transformation matrix M to transform the wind power data P = (P1, P j ,…,P n ) to obtain the first-order difference signal of wind power data: P f =(p1 f ,p j f ,…,p n f ) Among them, P f is the set of power ramp kernels on the entire wind power dataset; p j f is the wind power ramp kernel corresponding to time j.

4. The method for automatically identifying wind power ramp events according to claim 3 is characterized in that: The ramp segment is constructed by setting the ramp segment to the absolute value of the power ramp kernel |p j f |Continuously greater than or equal to the minimum climbing rate λ that constitutes a climbing event rate When the value of the power ramp kernel is positive, it is the rising slope of the power ramp; when the value of the power ramp kernel is negative, it is the falling slope of the power ramp; other areas are invalid areas of the power ramp event.

5. The method for automatically identifying wind power ramp events according to claim 4 is characterized in that: The specific construction process of the climbing section includes the following sub-steps: S3-1 sets the climbing segment set CS as: CS={cs i |i≥1,i≤n} Among them, cs i is a single climbing segment, i represents the order value of the climbing segment in the set CS, and n represents the number of climbing segments contained in the set CS; in each climbing segment cs i It includes the start time, end time, absolute value of the climbing amplitude and the climbing direction; S3-2 creates the first ramp segment cs1 and traverses the set P of power ramp kernels f The trend value of each data point at a moment in the rate When the power ramp kernel is used, this moment is taken as the starting moment of the first ramp segment cs1; continue to traverse the set P backward f , when encountering the first one less than the minimum climbing rate λ rate When the power ramp kernel is used, the moment before this moment is taken as the end moment of the first ramp segment cs1, and the difference in wind power between the end moment and the start moment of the first ramp segment is calculated, and its absolute value is taken as the ramp amplitude, which is recorded in the first ramp segment cs1; then its numerical sign is taken as the ramp direction, with "–" representing the descending ramp segment and "+" representing the ascending ramp segment; the data of the first ramp segment cs1 is put into the ramp segment set CS; S3-3 creates the next ramp segment cs2 and continues to traverse the set P of power ramp kernels f According to the trend value of each data point at a certain moment in the process, refer to the method of sub-step S3-2 to locate the start time and end time of the next climbing segment cs2, calculate the climbing amplitude and climbing direction of the next climbing segment cs2, and then put the data of the next climbing segment cs2 into the climbing segment set CS; S3-4 repeats sub-step S3-3, and finally puts the data of all climbing segments into the climbing segment set CS.

6. The method for automatically identifying wind power ramp events according to claim 5, characterized in that: In the process of detecting the power ramp event in step S4-2, the ramp segments are merged using the snake method. The specific merging method includes the following sub-steps: S4-2-1 sets the set PS of all climbing events to: PS={ps i |i≥1,i≤n} in, ps i is a single climbing event, which contains four parts of information: start time, end time, absolute value of climbing amplitude and climbing direction; i represents the order value of the climbing event in the set PS; n represents the total number of climbing events contained in the set PS; S4-2-2 Setting the time domain extension caliper T ruler , set the caliper time length to the time domain value T in the wind power ramp event threshold ; S4-2-3 First, extend the time domain extension caliper T backward for a certain climbing section that needs to be extended and merged later. ruler The first step is to eat up the longest climbing segment as much as possible in the time window. Secondly, if other climbing segments have been eaten up in the previous time window extension, try to extend it again, and the starting point is the end of the previous time window extension. Thirdly, add an extension time window, and so on, until no other climbing segments have been eaten up in a certain extension time window.

7. The method for automatically identifying wind power ramp events according to claim 6, characterized in that: The specific operation method of step S4-2-3 includes the following sub-steps: S4-2-3-1 Create a new climbing event ps m ; Among them, ps represents the climbing event, m represents the position of the climbing event in the set PS, and the initial position is 1; S4-2-3-2 Take out a climbing segment cs to be extended from the climbing segment set CS j ; where cs represents the climbing segment, j represents the position of the climbing segment in the climbing segment set CS, and the initial value of j is 1; let the climbing segment to be extended cs j The start time is t0, the end time is t1, the absolute value of the climbing amplitude is Δp1, the climbing direction is d1, and the value of the wind power at the start time t0 is p0, and the value at the end time t1 is p1; if the climbing segment set CS is traversed, the merging process ends; otherwise, execute step S4-2-3-3; S4-2-3-3 takes the next climbing segment cs from the climbing segment set CS again i ; where cs represents a climbing segment, i represents the position of the climbing segment in the climbing segment set CS, and the initial value of i is j+1. i The start time is t2, the end time is t3, the wind power value at the start time t2 is p2, and the wind power value at the end time t3 is p3; if a new climbing section cs can be obtained i , and t2–t1≤T ruler , then execute step S4-2-3-4; otherwise, detect |p1–p0|≥P threshold and |p1–p0|≥λ rate Are they satisfied at the same time? If they are satisfied at the same time, then the climbing section cs j Information copied to the climbing event ps m , and add the climbing event ps m Go to the climbing event set PS; then set j to the position j+1, increase the value of m by 1, and repeat step S4-2-3-1; S4-2-3-4 traverses the data (p i |i≥t1, i<t2) of each time point in the wind power dataset P from the start time t1 to the end time t2. If the ramp segment cs j is an upward ramp segment and p i ≤p0, or the ramp segment cs j is a downward ramp segment and p i ≥p0, then increment j by 1 and go to step S4-2-3-2. Otherwise, continue to execute the next step S4-2-3-5; S4-2-3-5 will be the climbing section cs i Try to merge into the climbing segment cs j Middle: If p3–p0 ≥ P threshold , and (p3–p0) / (t3–t0)≥λ rate , then use the climbing section cs i Extended climbing section cs j , about to climb the slope cs j The end position is located in the climbing section cs i The end position of the grid is set, the value of j is set to the value of i, and the start time t1 and the value of the wind power p1 corresponding to the start time are updated, and step S4-2-3-3 is repeated.

Images