Method for calculating historical operation fatigue load spectrum of wind turbine generator

Abstract

The invention relates to data simulation of a wind turbine generator, in particular to a method for calculating a historical operation fatigue load spectrum of a wind turbine generator through SCADA historical operation data of the wind turbine generator, and aims to solve the problem that an existing wind turbine generator operation monitoring system is difficult to effectively record and process high-frequency load data under limited storage and calculation resources. The fatigue analysis is not sufficient, and the maintenance cost is increased. According to the method for calculating the historical operation fatigue load spectrum of the wind turbine generator set, the historical operation fatigue load spectrum of the wind turbine generator set is accurately calculated by using the historical operation data of SCADA and the working characteristics of the wind turbine generator set in combination with high-frequency load simulation and a rain flow statistical method, and the method can be used for design optimization, life prediction and maintenance strategy formulation.

Description

Technical Field

[0001] The invention relates to data simulation of a wind turbine generator set, and in particular to a method for calculating a historical operation fatigue load spectrum of a wind turbine generator set through SCADA (Supervisory Control and Data Acquisition) historical operation data of the wind turbine generator set. Background Art

[0002] As an important technology for the utilization of renewable energy, wind turbines are subject to a variety of complex loads during operation, especially cyclic loads. These cyclic loads come from air flow fluctuations, mechanical vibrations, and structural dynamic responses. The cumulative effect may lead to fatigue failure of key components of the unit, such as blades, transmission chains, and towers.

[0003] In the current wind turbine operation monitoring system, sensors are usually used to perform high-frequency sampling of the operating status and load. Due to the limitations of storage and data processing capabilities, most systems can only retain low-frequency data, such as the ten-minute average value. Therefore, it is difficult for existing technologies to effectively record and process high-frequency load data under limited storage and computing resources, which leads to the following major problems:

[0004] (1) Insufficient fatigue analysis: The existing low-frequency data cannot fully reflect the peak value and frequency distribution of high-frequency loads, which reduces the accuracy of fatigue life prediction;

[0005] (2) Increased maintenance costs: The inability to accurately assess component fatigue may lead to excessive maintenance or delayed repairs, increasing operating and maintenance costs. Summary of the invention

[0006] The purpose of the present invention is to solve the problem that the existing wind turbine operation monitoring system is difficult to effectively record and process high-frequency load data under limited storage and computing resources, resulting in insufficient fatigue analysis and increased maintenance costs, and to provide a method for calculating the historical operation fatigue load spectrum of a wind turbine.

[0007] In order to solve the deficiencies of the above-mentioned prior art, the present invention provides the following technical solutions:

[0008] A method for calculating the historical operation fatigue load spectrum of a wind turbine generator system, which is special in that it includes the following steps:

[0009] Step 1: Obtain wind turbine information and wind farm information;

[0010] Step 2: Obtain the SCADA historical operation data of the wind turbine, including power, wind speed, wind direction, pitch angle, rotor speed, and ambient temperature;

[0011] Step 3: Clean the SCADA historical operation data obtained in step 2 and delete invalid data;

[0012] Step 4: correct the air density according to the altitude in the wind farm information in step 1 and the ambient temperature in the SCADA historical operation data obtained in step 3, and use the corrected air density to calculate a standard power curve under standard air density;

[0013] Step 5: sort the SCADA historical operation data obtained in step 3 according to wind speed from small to large, and then divide the wind speed from small to large according to fixed intervals to obtain A wind speed intervals, A ≥ 1;

[0014] Step 6: Sort the power in the SCADA historical operation data obtained in step 3 corresponding to each wind speed interval from small to large, and divide it into B power intervals on average, where B≥1;

[0015] Step 7: Count the number of powers in each power interval in the SCADA historical operation data obtained in step 3 that fall within the power interval, and calculate the occurrence frequency of each power interval;

[0016] Step 8: Simulate and obtain high-frequency load data at different turbulence intensities at the average wind speed in each wind speed interval, and then calculate the average power of the wind turbine at each turbulence intensity according to the standard power curve obtained in step 4, to obtain multiple high-frequency load data groups with the same average wind speed and different average powers;

[0017] Step 9, matching each power interval and its occurrence frequency with the high-frequency load data set at the corresponding wind speed obtained in step 8 to obtain the load distribution corresponding to each power interval;

[0018] Step 10: Combining the load distribution corresponding to each power interval with the Weibull distribution parameters in the wind farm information in step 1 to obtain the historical operating load spectrum distribution of the wind turbine;

[0019] Step 11: Perform cyclic statistics on the historical operation load spectrum distribution of the wind turbine generator set by using the rain flow statistics method, and calculate the number of load cycles corresponding to each stress amplitude;

[0020] Step 12: According to the number of load cycles corresponding to each stress amplitude and the corresponding time distribution, the historical operation fatigue load spectrum of the wind turbine is obtained.

[0021] Furthermore, the step 8 is specifically as follows:

[0022] Step 8.1, calculate the average wind speed in each wind speed interval, and select different turbulence intensities in the wind speed interval as simulation conditions;

[0023] Step 8.2: For each turbulence intensity, use a simulation tool to use multiple wind seeds to generate different wind speed sequences, and perform simulations to obtain the corresponding high-frequency load data set;

[0024] Step 8.3, statistically analyzing the high-frequency load data set corresponding to all wind seeds in the wind speed range to obtain high-frequency load data at different powers in the wind speed range;

[0025] Step 8.4: For each wind speed sequence under each turbulence intensity, use the standard power curve obtained in step 4 to calculate the average power of the wind turbine under each turbulence intensity, and obtain multiple high-frequency load data sets with the same average wind speed and different average powers.

[0026] Furthermore, the step 9 is specifically as follows:

[0027] Step 9.1, matching each power interval with the high-frequency load data set at the corresponding wind speed obtained in step 8 to obtain a high-frequency load spectrum for each power interval;

[0028] Step 9.2: Use the occurrence frequency of each power interval obtained in step 7 as a weighting factor and assign it to the high-frequency load spectrum of the power interval to obtain the load distribution corresponding to each power interval.

[0029] Furthermore, the step 10 is specifically as follows:

[0030] Step 10.1: Calculate the probability of wind speed occurrence P(U) in each power interval based on the Weibull distribution parameters and Weibull distribution function f(U). l , U h ):

[0031]

[0032] Where U is the wind speed, U l , and are the minimum and maximum values ​​of wind speed respectively; k and c are shape parameters and scale parameters respectively;

[0033] Step 10.2: Use the wind speed occurrence probability P(U l , U h ) weighting the load distribution of the power interval to obtain a weighted load distribution;

[0034] Step 10.3: Summarize the weighted load distribution of all power intervals to obtain the historical operating load spectrum distribution of the wind turbine.

[0035] Furthermore, the step 11 is specifically as follows:

[0036] Step 11.1, arranging the load amplitudes in the historical operation load spectrum distribution into a load data sequence in chronological order;

[0037] Step 11.2, using the rainflow statistics method to identify the load cycles in the load data sequence;

[0038] Step 11.3, for each identified load cycle, calculate its stress amplitude, where the stress amplitude is defined as the difference between the maximum load and the minimum load in each load cycle;

[0039] Step 11.4, sort all identified stress amplitudes from small to large, and divide them into multiple amplitude intervals, and count the occurrence frequency of stress amplitudes in each amplitude interval, that is, the number of load cycles corresponding to each stress amplitude.

[0040] Furthermore, the step 12 is specifically as follows:

[0041] Step 12.1: The number of load cycles N under the i-th stress amplitude obtained in step 11 is i , and the corresponding time distribution T i Multiply them together to get the weighted fatigue cycle number N of the i-th stress amplitude i，total ;

[0042] Step 12.2: According to Miner's law, the weighted fatigue cycle number N of the i-th stress amplitude is combined i，total , calculate the total fatigue damage D of the wind turbine in historical operation, and then obtain the fatigue load spectrum of the wind turbine in historical operation;

[0043] The total fatigue damage D of the historical operation is as follows:

[0044]

[0045] Where n is the number of stress amplitudes; N f，i is the fatigue life for the ith stress amplitude.

[0046] Furthermore, it is characterized in that:

[0047] In step 1, the wind turbine information includes manufacturer, model, transmission chain type, rated power, rated wind speed, rated speed, gearbox ratio, blade length, tower height, standard power curve, and grid connection time;

[0048] The wind farm information includes the wind farm location, altitude, annual average temperature, annual average wind speed, average air density, Weibull distribution parameters, wind shear index, and longitude and latitude coordinates of the machine site.

[0049] Furthermore, it is characterized in that:

[0050] In step 3, the invalid data includes shutdown data, data with a pitch angle greater than 70 degrees, data exceeding a preset wind speed boundary, data with a power less than the power corresponding to the cut-in wind speed, and speed data not connected to the grid;

[0051] The preset wind speed boundary = [cut-in wind speed V in -0.5, cut out wind speed V out ].

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] (1) The present invention provides a method for calculating the historical operating fatigue load spectrum of a wind turbine. The method utilizes the SCADA historical operating data and the working characteristics of the wind turbine, combined with high-frequency load simulation and rain flow statistics, to accurately calculate the historical operating fatigue load spectrum of the wind turbine, which can be used for design optimization, life prediction and formulation of maintenance strategies. The present invention not only helps to improve the design optimization and life prediction of wind turbines, but also provides data support for reasonable maintenance strategies.

[0054] (2) The present invention combines low-frequency SCADA data (such as wind speed, power, etc.) with high-frequency load data generated by simulation, and effectively compensates for the limitations of data collection through intelligent algorithms and simulation methods. DETAILED DESCRIPTION

[0055] The present invention is further described below in conjunction with exemplary embodiments.

[0056] A method for calculating a wind turbine historical operation fatigue load spectrum comprises the following steps:

[0057] Step 1: Obtain wind turbine information and wind farm information;

[0058] The wind turbine information includes manufacturer, model, transmission chain type, rated power, rated wind speed, rated speed, gearbox ratio, blade length, tower height, standard power curve, and grid connection time;

[0059] The wind farm information includes the wind farm location, altitude, annual average temperature, annual average wind speed, average air density, Weibull distribution parameters (shape parameter k, scale parameter c), wind shear index, and longitude and latitude coordinates of the machine position;

[0060] Step 2: Obtain the SCADA historical operation data of the wind turbine, including power, wind speed, wind direction, pitch angle, rotor speed, and ambient temperature (ten-minute average);

[0061] Step 3, cleaning the SCADA historical operation data obtained in step 2, deleting invalid data, including shutdown data, data with a pitch angle greater than 70 degrees (when exceeding this angle, the pitch angle usually changes and does not represent a normal operating state), data exceeding a preset wind speed boundary, data with a power less than the power corresponding to the cut-in wind speed, and speed data not connected to the grid (generator speed is less than the grid-connected speed);

[0062] The preset wind speed boundary = [cut-in wind speed V in -0.5, cut out wind speed V out ], wind speeds below the cut-in wind speed usually do not produce effective power;

[0063] Step 4: Correct the air density according to the altitude of the wind farm information in step 1 and the ambient temperature in the SCADA historical operation data obtained in step 3, and use the corrected air density to calculate the standard air density (1.225kg / m 3 ) standard power curve under;

[0064] Step 5: sort the SCADA historical operation data obtained in step 3 according to wind speed from small to large, and then divide the wind speed from small to large according to a fixed interval of 1m / s to obtain A wind speed intervals, A≥1;

[0065] The wind speed range of this embodiment is shown in Table 1:

[0066] Table 1

[0067]

[0068]

[0069] Step 6: Sort the power in the SCADA historical operation data obtained in step 3 corresponding to each wind speed interval from small to large, and divide it into B power intervals on average, where B≥1;

[0070] Step 7: Count the number of powers in each power interval in the SCADA historical operation data obtained in step 3 that fall within the power interval, and calculate the occurrence frequency of each power interval;

[0071] Step 8: Simulate and obtain high-frequency load data at different turbulence intensities at the average wind speed in each wind speed interval, and then calculate the average power of the wind turbine at each turbulence intensity according to the standard power curve obtained in step 4, to obtain multiple high-frequency load data groups with the same average wind speed and different average powers;

[0072] Step 8.1, calculate the average wind speed in each wind speed interval, and select different turbulence intensities (low, medium, and high) in the wind speed interval as simulation conditions;

[0073] Step 8.2: For each turbulence intensity, use a simulation tool (such as DNV Bladed / FAST / OpenFAST) to generate different wind speed sequences using multiple wind seeds, and perform simulations to obtain the corresponding high-frequency load data set;

[0074] Step 8.3, statistically analyzing the high-frequency load data set corresponding to all wind seeds in the wind speed interval to obtain the high-frequency load data at different powers in the wind speed interval, the high-frequency load data being the instantaneous load fluctuation caused by wind speed fluctuation during the operation of the wind turbine, including load amplitude and load change frequency;

[0075] Step 8.4, for each wind speed sequence under each turbulence intensity, use the standard power curve obtained in step 4 to calculate the average power of the wind turbine under each turbulence intensity, and obtain multiple high-frequency load data sets with the same average wind speed and different average powers;

[0076] Step 9, matching each power interval and its occurrence frequency with the high-frequency load data set at the corresponding wind speed obtained in step 8 to obtain the load distribution corresponding to each power interval;

[0077] Step 9.1, matching each power interval with the high-frequency load data set at the corresponding wind speed obtained in step 8 to obtain a high-frequency load spectrum for each power interval;

[0078] Step 9.2, using the occurrence frequency of each power interval obtained in step 7 as a weighting factor, assigning it to the high-frequency load spectrum of the power interval, and obtaining the load distribution corresponding to each power interval;

[0079] Step 10: Combining the load distribution corresponding to each power interval with the Weibull distribution parameters in the wind farm information in step 1 to obtain the historical operating load spectrum distribution of the wind turbine;

[0080] Step 10.1: Calculate the probability of wind speed occurrence P(U) in each power interval based on the Weibull distribution parameters and Weibull distribution function f(U). l , U h ):

[0081]

[0082] Where U is the wind speed, U l , and are the minimum and maximum values ​​of wind speed respectively; k and c are shape parameters and scale parameters respectively;

[0083] Step 10.2: Use the wind speed occurrence probability P(U l , U h ) weighting the load distribution of the power interval to obtain a weighted load distribution;

[0084] Step 10.3, summarizing the weighted load distribution of all power intervals to obtain the historical operation load spectrum distribution of the wind turbine, which can reflect the load fluctuation of the wind turbine in different power intervals;

[0085] Step 11: Perform cyclic statistics on the historical operating load spectrum distribution of the wind turbine obtained in step 10 by using the rain flow statistics method, and calculate the number of load cycles corresponding to each stress amplitude;

[0086] Step 11.1, sorting the load amplitudes (torque, stress, etc.) in the historical operation load spectrum distribution into a load data sequence in chronological order;

[0087] Step 11.2, using the rainflow statistics method to identify the load cycles in the load data sequence;

[0088] For each load, check whether it is an extreme value (maximum or minimum); when the load reaches a maximum value and starts to decrease, it is marked as an upward point; when the load value reaches a minimum value and starts to increase, it is marked as a downward point;

[0089] By scanning these rising and falling points, it is determined whether they form a complete load cycle; a complete load cycle includes a rising and a falling process;

[0090] Step 11.3, for each identified load cycle, calculate its stress amplitude, where the stress amplitude is defined as the difference between the maximum load and the minimum load in each load cycle;

[0091] Step 11.4, sort all identified stress amplitudes from small to large and divide them into multiple amplitude intervals, and count the occurrence frequency of stress amplitudes in each amplitude interval, that is, the number of load cycles corresponding to each stress amplitude;

[0092] Step 12: according to the number of load cycles corresponding to each stress amplitude and the corresponding time distribution, obtain the historical operation fatigue load spectrum of the wind turbine;

[0093] Step 12.1: The number of load cycles N under the i-th stress amplitude obtained in step 11 is i , and the corresponding time distribution T i Multiply them together to get the weighted fatigue cycle number N of the i-th stress amplitude i，total ;

[0094] Step 12.2: According to Miner's Rule, the weighted fatigue cycle number N of the i-th stress amplitude is combined i，total, calculate the total fatigue damage D of the wind turbine in historical operation, and then obtain the fatigue load spectrum of the wind turbine in historical operation, that is, the distribution of different stress amplitudes and the number of load cycles borne by the wind turbine in historical operation;

[0095] The total fatigue damage D of the historical operation is as follows:

[0096]

[0097] Where n is the number of stress amplitudes; N f，i is the fatigue life of the ith stress amplitude;

[0098] Step 13: Use the historical operating fatigue load spectrum of the wind turbine for design optimization, life prediction and maintenance strategy formulation;

[0099] Step 13.1, design optimization;

[0100] Determine the load amplitude, number of cycles and accumulated fatigue damage that have the greatest impact on the components of the wind turbine (such as blades, towers, transmission chains, etc.) based on the historical operating fatigue load spectrum of the wind turbine, and optimize the structural design so that the unit can operate safely under the maximum load and reduce fatigue damage;

[0101] By analyzing the frequency of occurrence of each load amplitude in the fatigue load spectrum of the wind turbine historical operation, the components that are subjected to large load fluctuations during operation are determined, so that appropriate materials and structures can be selected to improve the fatigue resistance of the components and extend their service life;

[0102] Step 13.2, lifespan prediction;

[0103] By calculating the cumulative fatigue damage under different stress amplitudes and combining it with Miners' law, the fatigue life of wind turbine components is predicted; then, the remaining life of the wind turbine is estimated based on the number of cycles and corresponding damage values ​​under each stress amplitude in the historical operating fatigue load spectrum of the wind turbine;

[0104] Step 13.3: Maintenance strategy formulation;

[0105] Based on the historical operating fatigue load spectrum of the wind turbine, the fatigue damage of each component of the wind turbine is determined. Combined with the prediction results of the total fatigue damage, the maintenance and replacement cycle is arranged to avoid excessive maintenance or delayed maintenance, thus ensuring the long-term stable operation of the unit.

Claims

1. A method for calculating the historical operating fatigue load spectrum of a wind turbine, characterized in that: The following steps are involved: Step 1: Obtain wind turbine information and wind farm information; Step 2: Obtain the SCADA historical operation data of the wind turbine, including power, wind speed, wind direction, pitch angle, rotor speed, and ambient temperature; Step 3: Clean the SCADA historical operation data obtained in step 2 and delete invalid data; Step 4: correct the air density according to the altitude in the wind farm information in step 1 and the ambient temperature in the SCADA historical operation data obtained in step 3, and use the corrected air density to calculate a standard power curve under standard air density; Step 5: sort the SCADA historical operation data obtained in step 3 according to wind speed from small to large, and then divide the wind speed from small to large according to fixed intervals to obtain A wind speed intervals, A ≥ 1; Step 6: Sort the power in the SCADA historical operation data obtained in step 3 corresponding to each wind speed interval from small to large, and divide it into B power intervals on average, where B≥1; Step 7: Count the number of powers in each power interval in the SCADA historical operation data obtained in step 3 that fall within the power interval, and calculate the occurrence frequency of each power interval; Step 8: Simulate and obtain high-frequency load data at different turbulence intensities at the average wind speed in each wind speed interval, and then calculate the average power of the wind turbine at each turbulence intensity according to the standard power curve obtained in step 4, to obtain multiple high-frequency load data groups with the same average wind speed and different average powers; Step 9, matching each power interval and its occurrence frequency with the high-frequency load data set at the corresponding wind speed obtained in step 8 to obtain the load distribution corresponding to each power interval; Step 10: Combining the load distribution corresponding to each power interval with the Weibull distribution parameters in the wind farm information in step 1 to obtain the historical operating load spectrum distribution of the wind turbine; Step 11: Perform cyclic statistics on the historical operation load spectrum distribution of the wind turbine generator set by using the rain flow statistics method, and calculate the number of load cycles corresponding to each stress amplitude; Step 12: According to the number of load cycles corresponding to each stress amplitude and the corresponding time distribution, the historical operation fatigue load spectrum of the wind turbine is obtained.

2. The method for calculating the historical operation fatigue load spectrum of a wind turbine according to claim 1, characterized in that: The step 8 is specifically as follows: Step 8.1, calculate the average wind speed in each wind speed interval, and select different turbulence intensities in the wind speed interval as simulation conditions; Step 8.2: For each turbulence intensity, use a simulation tool to use multiple wind seeds to generate different wind speed sequences, and perform simulations to obtain the corresponding high-frequency load data set; Step 8.3, statistically analyzing the high-frequency load data set corresponding to all wind seeds in the wind speed range to obtain high-frequency load data at different powers in the wind speed range; Step 8.4: For each wind speed sequence under each turbulence intensity, use the standard power curve obtained in step 4 to calculate the average power of the wind turbine under each turbulence intensity, and obtain multiple high-frequency load data sets with the same average wind speed and different average powers.

3. The method for calculating the historical operation fatigue load spectrum of a wind turbine according to claim 2, characterized in that: The step 9 is specifically as follows: Step 9.1, matching each power interval with the high-frequency load data set at the corresponding wind speed obtained in step 8 to obtain a high-frequency load spectrum for each power interval; Step 9.2: Use the occurrence frequency of each power interval obtained in step 7 as a weighting factor and assign it to the high-frequency load spectrum of the power interval to obtain the load distribution corresponding to each power interval.

4. The method for calculating the historical operation fatigue load spectrum of a wind turbine according to claim 3, characterized in that: The step 10 is specifically as follows: Step 10.1: Calculate the probability of wind speed occurrence P(U) in each power interval based on the Weibull distribution parameters and Weibull distribution function f(U). l , U h ): Where U is the wind speed, U l , and are the minimum and maximum values ​​of wind speed respectively; k and c are shape parameters and scale parameters respectively; Step 10.2: Use the probability of wind speed occurrence P(U l , U h ) weighting the load distribution of the power interval to obtain a weighted load distribution; Step 10.3: Summarize the weighted load distribution of all power intervals to obtain the historical operating load spectrum distribution of the wind turbine.

5. The method for calculating the historical operation fatigue load spectrum of a wind turbine according to claim 4, characterized in that: The step 11 is specifically as follows: Step 11.1, arranging the load amplitudes in the historical operation load spectrum distribution into a load data sequence in chronological order; Step 11.2, using the rainflow statistics method to identify the load cycles in the load data sequence; Step 11.3, for each identified load cycle, calculate its stress amplitude, where the stress amplitude is defined as the difference between the maximum load and the minimum load in each load cycle; Step 11.4, sort all identified stress amplitudes from small to large, and divide them into multiple amplitude intervals, and count the occurrence frequency of stress amplitudes in each amplitude interval, that is, the number of load cycles corresponding to each stress amplitude.

6. A method for calculating the historical operation fatigue load spectrum of a wind turbine according to claim 5, characterized in that: The step 12 is specifically as follows: Step 12.1: The number of load cycles N under the i-th stress amplitude obtained in step 11 is i , and the corresponding time distribution T i Multiply them together to get the weighted fatigue cycle number N of the i-th stress amplitude i，total ; Step 12.2: According to Miner's law, the weighted fatigue cycle number N of the i-th stress amplitude is combined i，total , calculate the total fatigue damage D of the wind turbine in historical operation, and then obtain the fatigue load spectrum of the wind turbine in historical operation; The total fatigue damage D of the historical operation is as follows: Where n is the number of stress amplitudes; N f，i is the fatigue life for the ith stress amplitude.

7. A method for calculating the historical operating fatigue load spectrum of a wind turbine according to any one of claims 1 to 6, characterized in that: In step 1, the wind turbine information includes manufacturer, model, transmission chain type, rated power, rated wind speed, rated speed, gearbox ratio, blade length, tower height, standard power curve, and grid connection time; The wind farm information includes the wind farm location, altitude, annual average temperature, annual average wind speed, average air density, Weibull distribution parameters, wind shear index, and longitude and latitude coordinates of the machine site.

8. A method for calculating the historical operation fatigue load spectrum of a wind turbine according to any one of claims 1 to 6, characterized in that: In step 3, the invalid data includes shutdown data, data with a pitch angle greater than 70 degrees, data exceeding a preset wind speed boundary, data with a power less than the power corresponding to the cut-in wind speed, and speed data not connected to the grid; The preset wind speed boundary = [cut-in wind speed V in -0.5, cut out wind speed V out ].

Images