Large-scale wind turbine generator wind energy tracking control time history optimization method considering aging effect

Abstract

The invention discloses a large-scale wind turbine generator wind energy tracking control time history optimization method considering an aging effect. The method comprises the following steps: S100, quantitatively evaluating unit performance degradation, performing SCADA (Supervisory Control and Data Acquisition) parameter selection and data preprocessing, constructing a unit performance degradation quantitative evaluation model, and evaluating the aging state of a unit through field SCADA data; s300, optimizing the control time history of the unit, and judging the deterioration state of the unit through SCADA (Supervisory Control and Data Acquisition) data; and S200, maximum wind energy tracking control is carried out, a blade tip speed ratio-wind energy utilization coefficient curve under the aging state of the unit is drawn according to the evaluated aging state of the unit, the basic performance of the wind turbine generator is approximately described, and the maximum value of the wind energy utilization coefficient is obtained. The optimization method is provided for solving the problem that the wind power utilization coefficient of the wind turbine generator is reduced due to natural aging of the wind turbine generator, so that the power output of the wind turbine generator is directly influenced, the optimal operation condition parameters of the wind turbine generator in the aging state can be more accurately obtained, the operation condition is optimized, and the operation power generation efficiency is improved.

Description

Technical Field

[0001] The invention relates to wind power generation technology, and in particular provides a method for optimizing the time course of wind energy tracking control of a large wind turbine generator set taking into account aging effects. Background Art

[0002] Wind turbines are devices that convert wind energy into electrical energy. The ability to capture air kinetic energy is closely related to the aerodynamic performance of the unit. Therefore, when designing wind turbines, the analysis of structure and aerodynamic performance is the focus of attention. Some aerodynamic performance analysis methods, such as the classic blade element-momentum theory (BEM) and computational fluid dynamics (CFD), have received widespread attention. By combining aerodynamic design with control design, the power characteristic curve of the wind turbine can be designed as the basis for the unit operation control strategy [7]. With the increase in service life, the performance degradation caused by the natural aging of wind turbines is inevitable. This problem has become increasingly important as a large number of wind turbines enter their "middle and old" years. Research on wind turbine aging and its evaluation has attracted the attention of more and more researchers. In wind farms, describing the historical evolution of the aerodynamic performance of on-site units is a difficult problem related to the operation, maintenance and optimization of the units. First, because wind turbines are super-large mechanical structures, it is impossible to conduct aerodynamic flow field tests on site. Second, the on-site data is limited and there are no operating parameters that directly describe the aerodynamic performance.

[0003] For units in service on site, it is not realistic to add additional hardware test systems, but making full use of existing SCADA data is a feasible solution. SCADA data records many operating parameter data such as cabin wind speed, wind rotor (generator) speed, generator power, etc., which can indirectly reflect the aerodynamic performance of wind turbines. For example, the aerodynamic performance of wind turbines will affect their power coefficient, and the power coefficient can be calculated based on SCADA data, thus establishing a correlation between the aerodynamic performance of the unit and the SCADA parameters. However, SCADA data often cannot be used directly, and some zero values, null values ​​and abnormal values ​​interspersed in it need to be processed.

[0004] However, general data preprocessing cannot solve the problems of "randomness" and "bias" of wind speed. These two problems are the main difficulties that affect the reliable calculation of on-site power coefficient of wind turbines. The randomness of wind speed means that the wind speed is unstable and changes uncertainly, and it also means that the wind speed is mixed with high-frequency components. From the perspective of control theory, for a large inertia system such as a wind rotor, high-frequency wind speed will not affect its speed change. When performing power coefficient analysis, its influence should be suppressed as much as possible, otherwise it will lead to deviations in the power coefficient calculation. The deviation of wind speed means that the wind speed measured by the nacelle anemometer is smaller than the wind speed at the front of the wind rotor, and deviation compensation is required. Although some literature has carried out research on wind speed filtering, there is a lack of effective filtering methods that meet the actual characteristics of wind turbines. For example, the problem of how to determine the filter cutoff frequency (inertial system time constant) has not been effectively solved. Summary of the invention

[0005] Based on this, the present invention provides a method for optimizing the wind energy tracking control schedule of large wind turbines taking into account the aging effect, objectively assesses the natural aging condition of the wind turbine, and assesses the wind energy utilization coefficient of the wind turbine according to the aging condition, so as to more accurately obtain the optimal operating parameters of the wind turbine under the aging state, optimize the operating conditions, and improve the operating power generation efficiency.

[0006] In order to achieve the above object, the present invention provides a method for optimizing the wind energy tracking control time course of a large wind turbine taking into account the aging effect, comprising:

[0007] S100. Quantitative evaluation of unit performance degradation: select SCADA parameters for the unit and preprocess its data, build a quantitative evaluation model for unit performance degradation, and evaluate the aging status of the unit through on-site SCADA parameters;

[0008] S200. Optimize the unit control schedule: determine the unit degradation status through SCADA data, and update parameters according to the degradation status results, and determine the impact of the degradation degree in the speed-power control curve;

[0009] S300. Maximum wind energy tracking control: According to the assessed unit aging status, the tip speed ratio-wind energy utilization coefficient curve is drawn under the unit aging status to approximately describe the basic performance of the wind turbine unit, combined with the unit wind energy utilization coefficient C P The relationship between the blade tip speed ratio λ and the blade tip speed ratio is determined to be at the optimal value λ opt The maximum value of wind energy utilization coefficient is obtained.

[0010] Furthermore, the data preprocessing of SCADA parameters includes:

[0011] S110. Supplement missing data and handle outliers: Find and supplement missing data to facilitate subsequent processing; judge and handle null values, zero values ​​and outliers that are obviously inconsistent with reality;

[0012] S120. High-frequency processing of temperature data: high-frequency processing of temperature data with a long sampling period to achieve the same sampling frequency as wind speed, rotation speed, power and other parameters, and put them in the same data packet with wind speed, rotation speed, power and other parameter data to facilitate subsequent analysis;

[0013] S130. Wind farm air density data supplement: Use the relationship ρ=f(T C ,h,RH), the instantaneous air density of the wind farm is obtained by numerical calculation of temperature, humidity and atmospheric pressure, where ρ is the air density, T C is the temperature in degrees Celsius, which is obtained through the SCADA system records; h is the altitude, which can be obtained through the technical data of the wind farm; RH is the relative humidity, which can be obtained through the meteorological website;

[0014] S140. Wind speed data deviation compensation and filtering: Compensate for wind speed deviation and eliminate high-frequency components mixed in the random changes of wind speed through filtering.

[0015] Furthermore, the quantitative evaluation model of unit performance degradation is constructed as follows:

[0016]

[0017] A wind energy tracking control time optimization method for large wind turbines taking into account aging effects is provided

[0018] 1. Use the power coefficient as the basic indicator for on-site analysis of the aerodynamic performance of wind turbine blades, fully consider the flow field characteristics of on-site wind speed data, and use data compensation and data filtering to overcome the influence of the two unfavorable factors of "randomness" and "bias" of SCADA wind speed data;

[0019] 2. Considering the relationship between wind speed and rotor speed as an inertial system, a data-driven time constant determination method is proposed. The system time constant is determined from the data itself, overcoming the inconvenience caused by the model-driven time constant determination method requiring precise unit parameters and without considering the influence of blade flexibility;

[0020] 3. Designed a calculation condition area division algorithm that conforms to the actual characteristics of the field data, selected the speed-power curve as the basis for judgment, and made full use of the stability and slow-changing advantages of these two parameters relative to wind speed. The data binning method and the "conservative" data comparison method were used to solve the "single-value" problem of scattered data and the problem of determining the interval critical value that needed to be solved when dividing the calculation condition area;

[0021] 4. The effectiveness of the "four-index" multi-angle observation algorithm is proposed. The calculation case analysis shows that the proposed method is reliable and effective, and the evolution characteristics of the aerodynamic performance of large wind turbines on site are discovered. The analysis results are consistent with the theoretical analysis and provide new ideas and basis for quantitative evaluation. This provides important knowledge support for the operation, maintenance and performance improvement of large wind turbines after they have been in service for a long time. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0023] Figure 1 It is a flowchart of an embodiment of a method for optimizing the time course of wind energy tracking control of a large wind turbine generator set taking into account the aging effect;

[0024] Figure 2 It is the logic block diagram of SCADA parameter selection and processing provided;

[0025] Figure 3 It is a flowchart of data preprocessing of the provided SCADA parameters;

[0026] Figure 4 It is the block diagram of the wind turbine timing optimization control provided;

[0027] Figure 5 is the wind turbine performance curve;

[0028] Figure 6 is the wind turbine speed-power curve;

[0029] Figure 7 High-frequency processing of temperature data;

[0030] Figure 8 It is the three-dimensional surface of the mean value of wind energy utilization coefficient from 2017 to 2020;

[0031] Fig. 9 Three-dimensional scatter plot of wind energy utilization coefficient based on wind speed and power;

[0032] Fig.10 It is the wind energy utilization factor and degradation degree from 2017 to 2020;

[0033] Fig.11 is the wind turbine performance curve;

[0034] Fig.12 It is the time course optimization performance analysis under stable wind speed conditions;

[0035] Fig.13 It is the power increase rate of the unit after process optimization;

[0036] Fig.14 It is the power characteristic of the unit under variable wind speed conditions. DETAILED DESCRIPTION

[0037] This application provides a method for optimizing the time course of wind energy tracking control of large wind turbines taking into account the aging effect, and studies the on-site historical evolution behavior of the aerodynamic performance of large wind turbines from the dimensions of description of the on-site aerodynamic performance of blades, preprocessing of on-site wind speed data, division of calculation conditions and analysis of calculation cases. The power coefficient is used as the basic indicator for on-site analysis of the aerodynamic performance of wind turbine blades, and the flow field characteristics of the on-site wind speed data are fully considered. Data compensation and data filtering are used to overcome the influence of the two unfavorable factors of "randomness" and "bias" of SCADA wind speed data. The system time constant is determined from the data itself, and a data-driven time constant determination method is proposed. A calculation condition area division algorithm that conforms to the actual characteristics of the on-site data is designed, and the effectiveness of the "four indicators" multi-angle observation algorithm is proposed.

[0038] (I) Maximum wind energy tracking control

[0039] According to the wind speed characteristics, wind turbines can be divided into two operating conditions: above the rated wind speed and below the rated wind speed. Below the rated wind speed, wind turbines are not limited in operation by power, while above the rated wind speed, the power capture and output need to be limited by adjusting the pitch angle. Below the rated wind speed, the main task of the wind turbine control system is to obtain the maximum wind energy capture efficiency as much as possible, and the control method is mainly to adjust the speed.

[0040] Since the pitch angle of wind turbines generally remains unchanged below the rated wind speed, it can be Figure 5 The "tip speed ratio-wind energy utilization coefficient" curve shown in the figure is used to approximate the basic performance of the wind turbine. As can be seen from the figure, the wind energy utilization coefficient C of the wind turbine P It is closely related to the tip speed ratio λ. When the tip speed ratio is at an optimal value λ opt The wind energy utilization coefficient reaches its maximum when . The wind energy utilization coefficient reflects the ability of wind turbines to capture energy from natural wind. From the existing wind turbine operation theory, we know that the mechanical power captured by wind turbines from wind can be written as

[0041]

[0042] Where ρ is the air density (kg / m 3 ), S is the swept area of ​​the wind wheel (m 2 ), v is the wind speed at the front edge of the wind rotor (m / s), and β is the pitch angle (°).

[0043] The tip speed ratio is used to indicate the state of the wind rotor at different wind speeds. Its value is the ratio of the circumferential speed of the blade tip to the wind speed. The expression is:

[0044]

[0045] In the formula, ω t is the wind rotor angular frequency (rad / s), and R is the wind rotor radius (m).

[0046] Will Figure 5 Medium opt and C Pmax Substituting into formula (1), the optimal wind energy capture expression below the rated wind speed can be obtained as follows:

[0047]

[0048] Then, substitute formula (2) into formula (3), and we have

[0049]

[0050] Rewrite the above formula as

[0051]

[0052] In the maximum wind energy tracking stage, the speed ω r With power P opt The above relationship should be satisfied, that is, when the tip speed ratio and wind energy utilization coefficient are kept at constant values, the power and speed are in a cubic relationship. Figure 6 This is a "speed-power" curve drawn based on the actual SCADA data of a 2MW wind turbine. It can be seen from the figure that the wind turbine can be divided into four areas during operation, namely transition area 1, maximum wind energy tracking area, transition area 2, and constant power area. It can be seen that the speed changes greatly in the maximum wind energy tracking area, and the wind turbine in this area is controlled by the given "speed-power" curve.

[0053] For newly installed wind turbines, the key to achieving the best energy capture performance is to set the "speed-power" curve in accordance with the actual characteristics of the wind turbine. For wind turbines that have already been in service, as the service time increases, the performance aging (deterioration) of the wind turbine is inevitable. The "speed-power" curve set initially will deviate from the actual characteristics of the wind turbine. If this deviation caused by the aging (deterioration) of the unit is ignored, the energy capture efficiency will inevitably be affected. Therefore, the author proposes the idea of ​​"time course optimization", that is, to adjust the "speed-power" curve in real time according to the dynamic performance degradation state of the wind turbine. However, during the service process of the wind turbine, the aging state is difficult to describe effectively, and there is no prior knowledge to refer to. The judgment of the aging state of the unit can only be based on the on-site SCADA data. Therefore, judging the aging state of the unit based on SCADA data is the prerequisite for "time course optimization" of performance.

[0054] (II) Quantitative evaluation of unit performance degradation

[0055] SCADA parameter selection and data preprocessing: The SCADA system records many parameters, but not all parameters need to be used as the basis for performance degradation evaluation. Figure 4 As shown, the whole process includes relevant parameter analysis, parameter data classification, supplementary parameter analysis, data preprocessing and other processes. The present application uses the wind energy utilization coefficient as an evaluation indicator of performance degradation caused by unit aging. Therefore, in the "relevant parameter analysis", only the SCADA parameters related to its calculation need to be considered, including hub speed, instantaneous wind speed, active power, ambient temperature, pitch angle and other parameters. On the other hand, although these parameters are related to the calculation of wind energy utilization coefficient, they can be divided into two categories, one is directly used for the calculation of wind energy utilization coefficient, and the other is used as observation values ​​for data extraction when calculating wind energy utilization coefficient. For example, the pitch angle parameter will not be used directly when calculating the wind energy utilization coefficient, but its value should be kept unchanged within the data interval used for extraction calculation, so it is a state monitoring parameter during calculation. This process is called "parameter data classification".

[0056] Data analysis found that air density is generally not directly recorded in SCADA data. If the dynamic change characteristics of air density are ignored, the accuracy of wind energy utilization coefficient calculation will inevitably be affected. Therefore, the process of "supplementary parameter analysis" mainly adds air density into the SCADA data table, which requires supplementary calculations through temperature, humidity, air pressure and other data. In addition, in some SCADA systems, temperature data is stored at the minute level, which is not convenient for batch data analysis and calculation. Therefore, it is also necessary to process minute-level data into second-level data at a high frequency.

[0057] In the process of wind turbine SCADA data collection and transmission, due to internal and external signal interference (noise), it is inevitable that data missing, array empty values, zero values, and abnormal values ​​will appear. Therefore, "data preprocessing" is an indispensable step in SCADA data analysis. For different data processing and analysis purposes, the data preprocessing strategies used will be different. The following mainly conducts data preprocessing in four steps:

[0058] Step 1: Missing data supplementation and outlier processing.

[0059] Through SCADA data analysis, it is found that the system will have a phenomenon of discontinuous data storage time during the data storage process, which is called "data missing". In order to facilitate subsequent processing, the missing data needs to be found and supplemented. Each set of data in the SCADA data not only contains state parameter information, but also contains sampling time series. Data storage is stored in packages with a certain time period as a unit. In the process of missing data judgment and search, first record the start time of the data packet, and then accumulate and compare them item by item according to the sampling time interval (usually one second). If the time item in the package is inconsistent with the accumulated time item, it indicates that a missing item has occurred. The supplementation algorithm is as follows:

[0060] X=(time,parameter1,parameter2,···)

[0061] i=start_time:1:end_time

[0062] if X.time≠i,

[0063] insert X.time=i;

[0064] X.parameters=0.

[0065] In the above expression, start_time and end_time are the start time and end time of packet sampling respectively.

[0066] In addition, during the field data sensing collection and transmission process, due to the influence of various interferences, null values, zero values, and values ​​that are obviously inconsistent with reality will appear. These are called "abnormal values" and also need to be judged and processed. For zero values ​​and null values, the following algorithms can be used:

[0067] ifX.parameters is null

[0068] X.parameters = 0;

[0069] ifX i .parameter=0andX i-1.parameter>0&X i+1 .parameter>0

[0070] X i .parameter=(X i-1 .parameter+X i+1 .parameter) / 2.

[0071] Step 2: High-frequency processing of temperature data.

[0072] In the stored SCADA data, the sampling periods of different wind turbine parameters are different. The sampling period of SCADA data for parameters such as wind speed, rotation speed, and power is 1s, while that of temperature data is 60s. Of course, this is related to the settings of the SCADA system itself. For the convenience of subsequent analysis, the temperature data needs to be processed at a high frequency and placed in the same data packet with other parameter data. Considering that temperature is a slow variable, high-frequency processing can be simplified. The specific method is to evenly disperse the minute-level temperature data into second-level data, that is, it is assumed that the temperature data does not change within one minute, such as Figure 7 shown.

[0073] Step 3: Supplement wind farm air density data.

[0074] The wind energy utilization coefficient of a wind turbine is an important indicator for evaluating the aging status of the unit. According to formula (1), the wind energy utilization coefficient is related to factors such as wind speed, power, and air density. The real-time air density of a wind farm is generally not recorded in the SCADA system, but it can be obtained through numerical calculations of temperature, humidity, and atmospheric pressure.

[0075] ρ=f(T C ,h,RH) (6)

[0076] Where ρ is the air density, T C is the Celsius temperature, which can be obtained through the SCADA system record; h is the altitude, which can be obtained through the wind farm technical data; RH is the relative humidity, which can be found through the meteorological website. The specific calculation expression of formula (6) can be adopted as follows:

[0077]

[0078] Where, T K is the Kelvin temperature; P V is the water vapor pressure (Pa); R d is the specific gas constant of dry air (J / (kg·K)); R v is the specific gas constant of water vapor (J / (kg·K)); Po is the sea level pressure; g is the acceleration due to gravity, g=9.8m / s 2; M is the molar mass of air (kg / mol); R is the universal gas constant (N·m / [mol·K]).

[0079] Step 4: Wind speed data deviation compensation and filtering.

[0080] Among various SCADA parameters, wind speed needs to be paid special attention due to its random uncertainty. In addition, wind speed data is mostly measured by an anemometer installed in the nacelle (behind the wind rotor). Due to the spatial location, the wind speed measured by the anemometer is not the actual incoming wind speed. When the airflow passes through the wind rotor, part of the air kinetic energy is converted into mechanical energy, which causes the wind speed measured by the anemometer to deviate from the actual wind speed.

[0081] The random changes in wind speed (wind speed mixed with high-frequency components) can be processed by filtering, for example, using a sliding average filtering algorithm:

[0082]

[0083] In the formula, y i is the output of the filter, x i is the input of the filter, and n is the window length of the MAF filter.

[0084] For the problem of wind speed deviation compensation, the compensation expression is adopted:

[0085]

[0086] In the formula, v 1 is the wind speed after compensation, P is the generator output power, v 0 is the wind speed measured by the sensor, and S is the swept area of ​​the wind turbine rotor.

[0087] 2.2 Quantitative evaluation model for unit performance degradation

[0088] As the service life of wind turbines increases, the natural aging of the units will inevitably lead to performance degradation. Wind turbines are equipment that converts wind energy into electrical energy, and the wind energy utilization coefficient is the most important indicator for measuring the aging of the units. Therefore, this application designs a quantitative evaluation model for wind turbine performance degradation based on the wind energy utilization coefficient as follows. The model describes the entire process from SCADA data to the unit degradation (aging) degree.

[0089] In the model, x is a data set composed of SCADA system parameters. A set of data represents a sampling result of SCADA. i Represents a certain operating parameter of a wind turbine. X represents several groups of parameter data stored in a SCADA data packet. These data groups may be continuous in time or may be discontinuous in time during the data collection process. 1The new data packet is composed of several consecutive data packets. This is mainly because the SCADA system data group is stored in packets. The data of a packet is limited no matter how big it is (for example, 10 minutes or 1 day). It is necessary to integrate multiple data packets into a new data packet with a large enough data volume for analysis. For example, select the data volume of 10 consecutive days (N day =10). D 2 This is the data space after the relevant parameter selection and data preprocessing. In this data space, the system parameters unrelated to aging have been eliminated and the data have been processed, including the supplementation of relevant parameters that are not directly recorded, the supplementation of missing data, the high-frequency processing of data, and the compensation and filtering of data deviation. 3 It is in D 2 In the data space, several subspaces that meet the calculation conditions are selected based on wind speed. The data screening based on wind speed is mainly to ensure that the selected data are in the maximum wind energy tracking area, because the wind energy utilization coefficient in this area is theoretically constant, which is conducive to obtaining more reliable wind energy utilization coefficient calculation results. In addition, data screening also requires setting the data capacity of the subspace (for example, 3000 groups of data in each subspace). If the amount of data is too large, it is difficult to achieve the same amount of data in different years, and if the amount of data is too small, it is easy to cause large deviations in the calculation results. Data analysis found that the number of subspaces obtained according to the above method is large. In order to simplify the calculation, in D 3 In the data subspace of , randomly select several subspaces without data intersection and get D 4 ; Then, for D 4 In each subspace, the series C is obtained according to the calculation method of wind energy utilization coefficient. Pi It should be noted that although the wind energy utilization coefficient is theoretically constant in the maximum wind energy tracking area, the calculated C P The value changes dynamically, so for each subspace, the average value of wind energy utilization coefficient can be used To express the wind energy utilization coefficient in this range; using this method, in D 4 r subspaces can get r value, select the maximum value As the final maximum wind energy tracking area C of the year P According to the above method, the wind energy utilization coefficient of different years is calculated respectively and compared with the result of the selected base year, so as to obtain the unit degradation (aging) degree Γ which evolves by year.

[0090]

[0091] Where v is D 2The wind speed parameter in the data space is the wind speed after filtering and compensation; P i Yes D 4 Active power data recorded in subspace, ρ i Yes D 4 Air density data in subspace (supplementary value), v i Yes D 4 Wind speed data in the subspace; s is the number of parameters retained after data selection, k is the length of sample data; C pi Based on D 4 The dynamic wind energy utilization coefficient obtained by subspace is is the average value of wind energy utilization coefficient in the sub-interval; C p is the maximum wind energy utilization coefficient among the randomly selected r sub-intervals; C pb is the wind energy utilization factor of the base (historical) year, C pc is the wind energy utilization factor for the assessment year.

[0092] (III) Unit control schedule optimization

[0093] In the variable speed permanent magnet wind turbine, the torque equation of the permanent magnet generator is:

[0094]

[0095] Where, T e is the electromagnetic torque of the generator; p is the number of pole pairs; ψ f is the permanent magnet flux; L d , L q are the AC and DC axis inductances of the permanent magnet generator, i d 、i q is the AC and DC axis current. If the rotor flux ψ f If the d-axis current component is always zero, the electromagnetic torque is proportional to the q-axis current. d = 0, the mechanical dynamic model of the wind turbine can be written as

[0096]

[0097] In the formula, J t is the moment of inertia; k g K is the speed ratio of the gearbox; t is the friction coefficient; T r is the aerodynamic torque. Formula (12) can be rewritten as

[0098]

[0099] From formula (13), we can see that changing the generator quadrature axis current i qThe speed of the wind turbine can be adjusted. Select the first-order dynamic speed tracking error and get the expression of the generator quadrature axis current as

[0100]

[0101] In the formula, a 0 is a constant; ε is the dynamic speed tracking error, and its expression is

[0102] ε=ω tref -ω t (15)

[0103] ω t is the wind wheel speed, which can be measured in real time using a high-precision speed sensor. The speed command ω tref It can be obtained by setting the "speed-power" curve. Formula (5) gives the relationship between the speed and power in the maximum wind energy tracking area.

[0104] The method provided in this application takes the power coefficient as the basic indicator for on-site analysis of the aerodynamic performance of wind turbine blades, fully considers the flow field characteristics of on-site wind speed data, and uses data compensation and data filtering to overcome the influence of the two unfavorable factors of "randomness" and "bias" of SCADA wind speed data.

[0105] Assume that after a certain period of service, the aerodynamic degradation degree of the wind turbine is Γ 0 , corresponding to the maximum wind energy utilization coefficient C of the optimal blade tip speed ratio Pmax becomes

[0106]

[0107] Then the speed command considering the aging effect can be written as

[0108]

[0109] Considering the aerodynamic torque T in formula (14) r The real-time value is difficult to measure, and the calculation of the quadrature axis current is only considered from the dynamic speed tracking error ε, that is,

[0110]

[0111] In the formula, K p , K i are the proportional coefficient and integral coefficient respectively; ε(k) is the dynamic speed tracking error at the kth sampling moment.

[0112] According to the above analysis, the wind turbine timing optimization control mode is drawn as follows: Figure 4First, the degradation state of the unit is determined by SCADA data, and then the parameters are updated according to the degradation state results, that is, the influence of the degradation degree is considered in the "speed-power" control curve according to formula (17).

[0113] The relationship between wind speed and rotor speed is regarded as an inertial system, and a data-driven time constant determination method is proposed. The system time constant is determined from the data itself, overcoming the inconvenience of the model-driven time constant determination method, which requires precise unit parameters and does not need to consider the influence of blade flexibility.

[0114] At the same time, an algorithm for dividing the calculation working condition area that conforms to the actual characteristics of the field data is designed. The speed-power curve is selected as the basis for judgment, making full use of the stability and slow-changing advantages of these two parameters relative to wind speed. The data binning method and the "conservative" data comparison method are used to solve the "single-value" problem of scattered data and the problem of determining the interval critical value that need to be solved when dividing the calculation working condition area.

[0115] (IV) Calculation case analysis

[0116] Analysis of unit degradation based on SCADA data: A 2MW doubly-fed wind turbine in a mountain wind farm in the south was selected as an example for analysis. The power regulation mode of this type of wind turbine is variable speed and variable pitch, with a design life of 20 years, a rated power of 2100kW, a cut-in wind speed of 3m / s, and a cut-out wind speed of 20m / s. The SCADA data of a wind turbine for 10 consecutive days from 2017 to March 2020 were extracted, and parameter selection (supplementation) and data preprocessing were performed according to the method mentioned in 2.1. The wind speed range [5.0m / s, 9.5m / s] was used as the condition (it can be known from the design technical parameters and operation data of the wind turbine that the wind speed range is within the maximum wind energy tracking area). Continuous data were selected with a sampling interval of 1s and a data length of 3000 points. Among the 10-day data from 2017 to March 2020, there were 18156 groups, 51791 groups, 10940 groups, and 45671 groups of data that met the selection conditions, respectively. From the above groups, 8 groups (each year) of data with no overlap and a fixed blade angle were randomly selected to ensure that the selected wind turbine operation data were under the maximum wind energy tracking area. The selected data group numbers r = 1, 2, 3, ..., 8. Taking groups as units, the wind energy utilization coefficient calculation method of formula (10) is used to obtain the series C at the sampling time: Pi Value and average value of wind energy utilization coefficient of each group Take 8 groups per year The maximum value among the average values ​​is taken as the wind energy utilization coefficient C in the maximum wind energy tracking area in that year. P .

[0117] Figure 8 , Fig. 9As shown in the figure, the mean wind energy utilization coefficient calculated by randomly selecting 8 groups of data (32 groups in total) from 2017 to 2020 according to formula (10) is given. Three-dimensional surface graph. From the graph, we can see that the average wind energy utilization coefficient calculated by selecting 32 data groups The range is [0.332, 0.442], where the maximum value is in 2017, the wind utilization coefficient is 0.442, the minimum value is in 2017, the wind energy utilization coefficient is 0.332, and the 8 The fluctuation of the value is smaller than that of the other three years, and the fluctuation is the largest in 2018. According to formula (10), 8 The maximum average value among the values ​​is taken as the wind energy utilization coefficient C in the maximum wind energy tracking area in that year. P , which were 0.442, 0.420, 0.397 and 0.392 in 2017-2020 respectively.

[0118] In order to further analyze the degradation characteristics of the unit as the service time increases, Fig. 9 The three-dimensional scatter point relationship between wind speed (after compensation) and active power and wind energy utilization coefficient of the data group with the maximum mean value of wind energy utilization coefficient from 2017 to 2020 is given. It can be seen that the blade angle values ​​in the four groups of data from 2017 to 2020 are 0.02°, 0.03°, 0.02°, and -0.01° respectively. The blade angle values ​​in 2017 and 2019 are the same, which is different from the blade angle values ​​of the two groups of data in 2017 and 2020, but they are all fixed values, indicating that the wind turbine pitch system does not have pitch change in this operating condition area, that is, the operating conditions of the data group with the maximum mean value of wind energy utilization coefficient selected are all within the maximum wind power tracking area (MPPT). From the four-year dynamic Cp values ​​(3000) represented by the three-dimensional scatter plot, it can be seen that in 2019, the dynamic Cp values ​​were relatively concentrated, mainly concentrated in the range of [0.281, 0.530]. The dynamic Cp values ​​in 2020 were relatively dispersed, ranging from [0.251, 0.593], which was higher than the dynamic Cp values ​​of the other three groups of data. p The value is wider, and its variation value is 0.342, which is related to the change in the aerodynamic performance of the unit. Obviously, the larger the variation range, the more obvious the decline in aerodynamic performance. At the same time, among the four sets of data, the wind speed range in 2018 is [5.02m / s, 7.75m / s], which is the most concentrated, and the variation (fluctuation) is the smallest, indicating that the wind speed is the most stable compared with the other three sets of data. The wind speed range in 2019 is [5.01m / s, 8.85m / s], with the largest variation (fluctuation). The year with the smallest change in active power is 2017, with a variation range of [286kW, 1000kW] and a variation value of 714kW. The year with the largest change is 2019, with a variation range of [313kW, 1262kW], and a variation value of 949kW.

[0119] Fig.10 The wind energy utilization coefficient of the maximum wind energy tracking area in four years and the characteristic curve of the degradation degree of the historical years (2018-2020) compared with 2017 (base year) are given. p The values ​​are 0.442, 0.420, 0.397, and 0.392 respectively. It can be seen from the curve that the C p The value has an overall downward trend. According to the definition of degradation degree in formula (10), it can be calculated that the degradation degree (Γ) from 2017 to 2020 is 0, 5.1%, 10.2%, and 10.9%, respectively. The degradation degree value increases year by year with the increase of service years, which also indicates that the aerodynamic performance of the unit decreases year by year, which is consistent with the Figure 8 The degradation degree represented by the given unit degradation degree curve shows an overall upward trend, which is consistent with the above-mentioned aging assessment method.

[0120] 4.2 Timing Optimization Performance Analysis

[0121] In order to obtain the effect of time-optimized control of the unit, a numerical analysis method is used to first establish the aerodynamic characteristics analysis model of the wind turbine parameterized control as follows:

[0122]

[0123] In the formula, c i (i=1, 2, 3, 4, 5, 6) and k j (j=1, 2, 3) are setting coefficients, which are determined according to the specific characteristics of the wind turbine. The above calculation model is corrected in combination with the characteristics of the wind turbine. Given c1=0.312, c2=116, c3=0.4, c4=5, c5=15, c6=0.00912, k0=3, k1=0.08, k2=2.2, k3=0.035, the performance curve of the wind turbine is as follows: Fig.11 As shown in the figure, when the pitch angle is 0°, the optimal tip speed ratio λ opt is 8.09, and the corresponding C Pmax It is 0.421.

[0124] exist Figure 4 In the control model, after the wind turbine speed command is obtained by equation (17), the generator quadrature axis current control command is given according to equation (18). From the perspective of response time, the time required for electrical and control is much smaller than that for mechanical systems. The tracking effect of wind turbine speed mainly depends on the inertial characteristics of the wind rotor speed. In order to speed up the analysis, the control system, electrical system and wind rotor inertial system are combined and simplified into a first-order system for speed tracking during simulation.

[0125]

[0126] Where τ is the time constant of the inertial system, which can be estimated by equation (21).

[0127]

[0128] Where, T rated is the rated wind wheel torque; J r is the moment of inertia of the wind wheel; ω rated is the rated speed of the wind wheel, ω 0 is the initial speed of the wind wheel.

[0129] The wind turbine rotor radius is set to 46.7m, the cut-in wind speed is set to 3m / s, and the cut-out wind speed is set to 25m / s; the wind rotor inertia time constant is set to 10s. Fig.12 The performance analysis of time course optimization under stable wind speed conditions is given. Fig.12 (a) is a scatter plot of power variation of wind turbines under different wind speeds and different degradation conditions; Fig.12 (b)-(f) are the power change comparisons before and after the schedule optimization under different degradation conditions when the wind speed is 6m / s, 7m / s, 8m / s, 9m / s, and 10m / s, respectively. In the figure, according to the definition of degradation in formula (10), combined with the possible situations that may occur when the wind turbine is in service for a long time, various situations such as degradation of 5%, 10%, 15%, 20%, 25%, and 30% are considered. Fig.12 (a) does not consider the impact of schedule optimization. From the change of power scatter points in the figure, it can be seen that as the unit degradation increases, the power output gradually decreases. The greater the wind speed, the greater the power reduction. For example, at a wind speed of 6m / s and a degradation degree of 5%, the power drops from 381.2kW to 361.9kW; at a degradation degree of 30%, the power drops to 260.4kW. At a wind speed of 10m / s and a degradation degree of 5%, the power drops from 1765.5kW to 1676.5kW; at a degradation degree of 30%, the power drops to 1206.3kW.

[0130] from Fig.12 From (b)-(f), it can be seen that the time optimization after the aerodynamic performance of the wind turbine is degraded can improve the power output to a certain extent. Relatively speaking, the greater the wind speed, the greater the power increase. For example, when the degradation degree is 5% and the wind speed is 6m / s, the power increases from 361.9kW to 362.1kW; when the wind speed is 10m / s, the power increases from 1676.5kW to 1677.2kW.

[0131] The power increase rate is defined as

[0132]

[0133] Where P DP is the power when the performance of the wind turbine is degraded; O It is the power after the unit is optimized in time schedule.

[0134] Although under the same degradation condition, the greater the wind speed, the more power can be improved by schedule optimization, but the improvement rate remains basically unchanged.

[0135] Fig.13 The power increase rate obtained by optimizing the schedule of wind turbines under different degradation degrees and wind speed conditions is given. The increase ratio is different under different degradation conditions. Under 5% degradation conditions, schedule optimization can increase power output by about 0.05%; under 10% degradation conditions, the power output is increased by about 0.19%; under 15% degradation conditions, the power output is increased by about 0.48%; under 20% degradation conditions, the power output is increased by about 0.92%; under 25% degradation conditions, the power output is increased by about 1.55%; under 30% degradation conditions, the power output is increased by about 2.46%. Because it is in the maximum wind energy tracking area, the power output increase ratio is mainly related to the degradation degree, and is less affected by the wind speed. From 6m / s to 10m / s, as long as the degradation degree remains unchanged, the power increase ratio after schedule optimization remains basically unchanged.

[0136] The above analysis is based on a constant wind speed. In fact, the wind speed on site varies with time. Therefore, a real-time wind speed of a certain wind farm is taken as the input for the simulation analysis.

[0137] The wind speed in the figure varies between 4.2m / s and 9.2m / s, which is in the maximum wind energy tracking area of ​​the wind turbine. Fig.14The power curves of wind turbines when the degradation degree Γ is 5%, 15%, and 30% are given in the figure, including the power change curves before and after the schedule optimization. It is obvious that under the same wind speed input conditions, the greater the degradation degree, the smaller the output power. When the unit does not experience performance degradation (i.e., Γ is 0), the power change range is 301.3kW-883.5kW. When Γ is 5%, the power change range before the schedule optimization is 287kW-837.4kW. In comparison, the reduction is 4.7% {(287kW-301.3kW) / 301.3kW*100%} when the power is minimum, and 5.2% when the power is maximum. The power reduction fluctuates around 5%, mainly due to the introduction of a first-order system considering the inertia of the wind turbine. The power change range after the schedule optimization is 286.5kW-839.4kW. When the degradation degree Γ is 15%, the power variation range is 257.6kW-742.7kW, and after the schedule optimization, the power variation range is 256.1kW-751kW; when Γ is 30%, the power variation range is 211kW-594.2kW, and after the schedule optimization, the power variation range is 210.9kW-618.5kW. Compared with the variation range before optimization, it can be seen that after the schedule optimization, the power is not greater than that before optimization at all times. The main factor is the influence of the large inertia of the wind rotor. The power variation characteristics are not only related to the current wind speed, but also to the wind speed variation characteristics over a period of time.

[0138] The effectiveness of the "four-index" multi-angle observation algorithm is proposed. The calculation case analysis shows that the proposed method is reliable and effective, and the evolution characteristics of the aerodynamic performance of large wind turbines on site are discovered. The analysis results are consistent with the theoretical analysis and provide new ideas and basis for quantitative evaluation. This provides important knowledge support for the operation, maintenance and performance improvement of large wind turbines after they have been in service for a long time.

[0139] The above description is only an exemplary embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing the wind energy tracking control time course of a large wind turbine taking into account the aging effect, characterized in that: include: S100. Quantitative evaluation of unit performance degradation: Carry out SCADA parameter selection and data preprocessing of the unit, build a quantitative evaluation model for unit performance degradation, and evaluate the aging status of the unit through on-site SCADA parameters; S200. Optimize unit control schedule: Determine the unit degradation status through SCADA data, update parameters based on the degradation status results, and determine the impact of the degradation degree in the speed-power control curve; S300. Maximum wind energy tracking control: According to the assessed aging status of the unit, the tip speed ratio-wind energy utilization coefficient curve under the aging status of the unit is drawn to approximately describe the basic performance of the wind turbine unit. Combined with the wind energy utilization coefficient C of the unit P The relationship between the blade tip speed ratio λ and the blade tip speed ratio is determined to be at the optimal value λ opt The maximum value of wind energy utilization coefficient is obtained.

2. The method for optimizing the wind energy tracking control time course of a large wind turbine taking into account the aging effect according to claim 1 is characterized in that: Make SCADA parameter selections, including: When analyzing relevant parameters, SCADA parameters related to calculation are selected as evaluation indicators of performance degradation caused by unit aging. The selected SCADA parameters include hub speed, instantaneous wind speed, active power, ambient temperature and pitch angle.

3. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 2 is characterized in that: Categorize SCADA parameters, including: The selected parameter data are divided into two categories: one category is directly used for calculating the wind energy utilization coefficient; the other category is used as observation values ​​for data extraction when calculating the wind energy utilization coefficient.

4. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 1, characterized in that: Data preprocessing of SCADA parameters includes: S110. Supplement missing data and handle outliers: Find and supplement missing data to facilitate subsequent processing; judge and handle null values, zero values ​​and outliers that are obviously inconsistent with reality; S120. High-frequency processing of temperature data: high-frequency processing of temperature data with a long sampling period to achieve the same sampling frequency as wind speed, rotation speed, power and other parameters, and put them in the same data packet with wind speed, rotation speed, power and other parameter data to facilitate subsequent analysis; S130. Wind farm air density data supplement: Use the relationship ρ=f(T C ,h,RH), the instantaneous air density of the wind farm is obtained by numerical calculation of temperature, humidity and atmospheric pressure, where ρ is the air density, T C is the temperature in degrees Celsius, which is obtained through the SCADA system records; h is the altitude, which can be obtained through the technical data of the wind farm; RH is the relative humidity, which can be obtained through the meteorological website; S140. Wind speed data deviation compensation and filtering: Compensate for wind speed deviation and eliminate high-frequency components mixed in the random changes of wind speed through filtering.

5. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 4, characterized in that: In the process of high-frequency processing of temperature data, based on the fact that temperature is a slow variable, the minute-level temperature data is evenly dispersed into second-level data, and the temperature data remains unchanged within one minute.

6. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 4, characterized in that: For the random changes in wind speed, the high-frequency components in the wind speed are processed by filtering, and the following formula is used for processing: In the formula, y i is the output of the filter, x i is the input of the filter, and n is the window length of the MAF filter.

7. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 4, characterized in that: The following formula is used to compensate for the deviation of wind speed: Where v1 is the compensated wind speed, P is the generator output power, v0 is the wind speed measured by the sensor, and S is the swept area of ​​the wind turbine rotor.

8. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 7, characterized in that: In the maximum wind energy tracking stage, the tip speed ratio and wind energy utilization coefficient are kept constant, and the speed ω r With power P opt Satisfaction relationship: In the formula, C Pmax is the maximum wind energy utilization factor obtained from the wind turbine performance curve, λ opt is the corresponding blade tip reduction ratio obtained from the wind turbine performance curve.

9. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 8, characterized in that: Based on the speed-power curve drawn by SCADA parameters, the operation process of the wind turbine is divided into four areas: transition zone 1, maximum wind energy tracking zone, transition zone 2 and constant power zone. It is determined that a large change in speed occurs in the maximum wind energy tracking zone, and the wind turbine in this area is controlled by a given speed-power curve.

10. The method for optimizing the wind energy tracking control time course of a large wind turbine generator system taking into account the aging effect according to claim 1, characterized in that: The quantitative evaluation model of unit performance degradation is constructed as follows: Where x is a data group composed of SCADA system parameters, X represents several groups of parameter data stored in a SCADA data packet, D1 is a new data packet composed of several consecutive data packets, D2 is the data space after relevant parameter selection and data preprocessing, D3 is a subspace that meets the calculation conditions in the D2 data space with wind speed as the selection condition, D4 is a subspace randomly selected from the D3 data subspace without data intersection, v is the wind speed parameter in the D2 data space, which is the wind speed after filtering and compensation; P i is the active power data recorded in the D4 subspace, ρ i is the air density data in the D4 subspace, v i is the wind speed data in the D4 subspace; s is the number of parameters retained after data selection, k is the sample data length; C pi is the dynamic wind energy utilization coefficient obtained based on the D4 subspace, is the average value of wind energy utilization coefficient in the sub-interval; C p is the maximum wind energy utilization coefficient among the randomly selected r sub-intervals; C p b is the wind energy utilization factor in the base year, C pc is the wind energy utilization factor for the assessment year.

Images