Blade icing monitoring and early warning method based on blade clearance data

Abstract

The invention provides a blade clearance data-based blade icing monitoring and early warning method, which comprises the following steps of: firstly, acquiring blade clearance data, temperature and humidity data and unit control parameters from a wind turbine generator, grouping the parameters by time, and constructing a data set; comparing the data set with the icing environment condition model; the data set conforming to the icing environment condition model enters a deviation dispersion overrun early warning model for calculation, and the clearance difference, namely the deviation dispersion, of three adjacent blades in the same impeller period is judged through a sliding window; and finally, entering a clearance oscillation over-limit early warning model for calculation. According to the method, through wind turbine generator blade clearance short-term time sequence characteristics, in combination with environmental factors and unit control parameters, the phenomenon of imbalance or clearance oscillation occurring on unit clearance is calculated, and therefore the blade icing recognition and early warning functions can be well achieved by rapidly recognizing the clearance abnormal characteristics.

Description

Technical Field

[0001] The invention provides a blade icing monitoring and early warning method based on blade clearance data, belonging to the technical field of health monitoring of wind turbine generator sets. Background Art

[0002] Icing of wind turbine blades is an important issue in the field of wind power generation. The meteorological conditions for blade icing mainly include ambient temperature, air humidity, wind speed, supercooled water droplets and condensation height. When the air temperature is below freezing, supercooled water droplets will freeze on the blade surface to form ice. In addition, meteorological factors such as wind speed, wind direction and air humidity will also affect the formation of ice. Factors such as the shape, material and surface coating of the blade will affect the adhesion and freezing process of water droplets on the blade surface. If there are rough or uneven areas on the blade surface, it is easy for water droplets to adhere and form ice. The geographical location, altitude and surrounding environment of the wind turbine will affect the meteorological exposure conditions of the blades of the turbine. In some areas, due to factors such as terrain and climate, turbine blades are more susceptible to ice.

[0003] Blade icing will have many negative effects on wind turbines: 1. Increased weight and resistance: Ice on the blades will increase the weight and resistance of the blades, resulting in reduced power generation efficiency of the wind turbine. 2. Changes in aerodynamic performance: Ice will change the aerodynamic performance of the blades, making the output power of the wind turbine unstable and affecting the stable operation of the power grid. 3. Increased mechanical wear: Ice on the blades will cause the mechanical parts of the wind turbine to bear greater loads, increase mechanical wear and damage, and increase maintenance costs. 4. Safety hazards: Frozen blades may no longer be balanced, causing unstable rotation, thus threatening the safety of wind turbines. Under extreme weather conditions, ice may lead to serious consequences such as blade breakage and unit collapse, posing a threat to people's lives and property.

[0004] At present, common blade icing diagnosis methods include icing sensor diagnosis method, wind power mismatch diagnosis method and SCADA multi-dimensional data comprehensive diagnosis method. When the icing sensor diagnosis method is used in the existing scheme, the icing of the icing sensor does not mean that the blade is actually frozen, resulting in problems such as high false alarm rate; the wind power mismatch diagnosis method mainly relies on the monitoring of output power, but the change of output power may be affected by many factors, such as wind speed fluctuations, mechanical failures and other factors; moreover, in the early stage of icing or when the degree of icing is light, this method may not be able to accurately diagnose the icing phenomenon. The SCADA multi-dimensional data comprehensive diagnosis method cannot achieve real-time blade diagnosis and protection functions. Summary of the invention

[0005] The patent of the present invention solves the deficiencies in the prior art and provides a blade icing monitoring and early warning method based on blade clearance data. In this method, the short-term time series characteristics of the wind turbine blade clearance are combined with environmental factors and unit control parameters. By calculating the imbalance or clearance oscillation phenomenon that occurs in the unit clearance, the abnormal clearance characteristics can be quickly identified, which can effectively play the role of blade icing identification and early warning.

[0006] The technical solution adopted to achieve the above-mentioned purpose of the present invention is:

[0007] A blade icing monitoring and early warning method based on blade clearance data comprises the following steps:

[0008] (1) Collect blade clearance data and temperature and humidity data from the blade clearance monitoring system and temperature and humidity sensors installed on the wind turbine, and obtain unit control parameters from the unit control system, group the above parameters by time and construct a data set;

[0009] (2) Constructing an icing environment condition model through an empirical model, comparing the data set in the previous step with the icing environment condition model, and the data set that meets the icing environment condition model proceeds to the next stage for calculation, while the data set that does not meet the icing environment condition model terminates the calculation;

[0010] (3) The data set that meets the icing environment condition model is entered into the deviation dispersion over-limit warning model for calculation. The difference in the clearance of three adjacent blades in the same impeller cycle is determined by sliding window, that is, the deviation dispersion. The calculation formula of the deviation dispersion devi is as follows:

[0011]

[0012] Where i is the impeller rotation period, Ai, Bi, and Ci are the clearance values ​​of the three blades in the same impeller rotation period i. Furthermore, the sliding window deviation discreteness Devj is calculated as follows: Where Devj is the sliding window deviation discreteness of the jth impeller rotation cycle, and n is the window length; if the sliding window deviation discreteness Devj is greater than the set deviation discreteness over-limit warning threshold, the blade icing alarm is triggered and the calculation is terminated; if the sliding window deviation discreteness Devj is not greater than the set deviation discreteness over-limit warning threshold, the calculation continues to the next stage;

[0013] (4) The data set that is not greater than the deviation discreteness over-limit warning threshold enters the clearance oscillation over-limit warning model for calculation. The clearance oscillation over-limit warning model is preset with a clearance reference table, which is prepared by combining the clearance historical data of different models under different unit control parameters; the clearance reference value under the unit control parameters in the data set is obtained by querying the difference against the clearance reference table, and the clearance oscillation value osc is further calculated. i : Where cl i is the clearance value of any blade, CL (ωi,αi) is the speed ω in the clearance reference table i and pitch angle α i The clearance value at that time; then calculate the sliding window clearance oscillation value Osc j , the calculation formula is: If the sliding window clearance oscillation value Osc j If the clearance oscillation exceeds the set warning threshold, the blade icing alarm will be triggered.

[0014] Furthermore, the freezing environment condition model in step (2) includes freezing temperature conditions and freezing humidity conditions. When the temperature and humidity data in the data set simultaneously meet the freezing temperature conditions and the freezing humidity conditions, the next stage of calculation is entered.

[0015] Furthermore, the freezing temperature condition is -40 to 0°C, and the freezing humidity condition is that the air humidity is greater than 80%.

[0016] Furthermore, in step (3), the maximum value of the sliding window deviation discreteness under normal operating conditions of the wind turbine is collected and analyzed, and the maximum value is multiplied by a safety factor to set a deviation discreteness over-limit warning threshold.

[0017] Furthermore, the safety factor is 1.3.

[0018] Furthermore, in step (4), the maximum value of the sliding window clearance oscillation value under normal conditions of the unit is collected and analyzed, and the maximum value is multiplied by the safety factor to set the clearance oscillation over-limit warning threshold.

[0019] Furthermore, the safety factor is 1.3.

[0020] Furthermore, in step (4), a control parameter clearance reference table for different models is formulated based on historical data, historical data of clearance conditions at a preset time is collected, the relationship between speed, pitch angle and clearance is established, the speed and pitch angle are divided into intervals, and the average clearance in each bin is used as the reference value.

[0021] Furthermore, after the blade icing alarm is triggered in steps (3) and (4), the alarm information is uploaded to the wind turbine main control system, and the crew responds and handles it.

[0022] Compared with the prior art, the blade icing monitoring and early warning method based on blade clearance data provided by the present invention has the following advantages: 1. The blade icing monitoring and early warning method in this patent is based on the clearance time series characteristics combined with environmental factors and unit control parameters. Compared with the prior art, there is no need to install additional devices such as icing sensors, power sensors, etc. All parameters used are derived from the data monitored during the routine operation and maintenance of the wind turbine set. Therefore, the original data collection is relatively simple, which further reduces the cost.

[0023] 2. In addition to the deviation dispersion over-limit warning model, this patent also has a clearance oscillation over-limit warning model. When all three blades are frozen, the clearance deviations of the blades may not have obvious periodic differences due to symmetry, which will cause the warning ability of the deviation dispersion over-limit warning model to decrease. On this basis, the clearance oscillation over-limit warning model can conduct further analysis and warning for this situation to avoid missed reports. At the same time, the clearance oscillation over-limit warning model also eliminates the clearance changes caused by control parameters by constructing a clearance reference table for the model, making monitoring and warning more accurate.

[0024] 3. In this patent, the data is filtered through the icing environment condition model to improve the accuracy of the model and reduce the amount of calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A system flow chart of a blade icing monitoring and early warning method based on blade clearance data provided by the present invention;

[0026] Figure 2 A schematic diagram of a control parameter clearance reference table established in the present invention;

[0027] Figure 3 It is a calculation diagram of the deviation dispersion over-limit warning model in the embodiment (normal period);

[0028] Figure 4 It is a calculation diagram of the deviation discreteness over-limit warning model in the embodiment (starting to freeze);

[0029] Figure 5 It is a calculation diagram of the deviation dispersion over-limit warning model in the embodiment (partial blade icing);

[0030] Figure 6 This is a calculation diagram of the deviation dispersion over-limit warning model in the embodiment (3 blades are frozen);

[0031] Figure 7It is a calculation diagram of the clearance oscillation over-limit warning model in the embodiment (normal period);

[0032] Figure 8 It is a calculation diagram of the clearance oscillation over-limit warning model in the embodiment (starting to freeze);

[0033] Fig. 9 It is a calculation diagram of the clearance oscillation over-limit warning model in the embodiment (partial blade icing);

[0034] Fig.10 This is a calculation diagram of the clearance oscillation over-limit warning model in the embodiment (3 blades are frozen). DETAILED DESCRIPTION

[0035] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments, but the protection scope of the present invention is not limited to the following embodiments.

[0036] The system flow chart of the blade icing monitoring and early warning method based on blade clearance data provided in the present invention is as follows: Figure 1 As shown, the steps are as follows:

[0037] (1) Collect blade clearance data, temperature and humidity data from the blade clearance monitoring system and temperature and humidity sensors installed on the wind turbine, and obtain unit control parameters from the unit control system. The above parameters are grouped by time and a data set is constructed; the unit control parameters mainly include impeller speed and pitch angle.

[0038] (2) Construct an icing environment condition model through an empirical model, compare the data set in the previous step with the icing environment condition model, and the data set that meets the icing environment condition model enters the next stage for calculation, and the data set that does not meet the icing environment condition model terminates the calculation. The icing environment condition model includes icing temperature conditions and icing humidity conditions. When the temperature and humidity data in the data set meet both the icing temperature conditions and the icing humidity conditions, the data set enters the next stage for calculation. The icing temperature condition is -40 to 0°C, and the icing humidity condition is that the air humidity is greater than 80%.

[0039] In this patent, the temperature and humidity that must be met for the blades of the unit to freeze are limited through empirical models to reduce false alarms of icing caused by special operating conditions. Freezing occurs at -40 to 0°C, with the highest frequency between -10 and -2°C. Freezing will only occur below 0°C, and the ice layer will basically melt when the temperature is above 3°C. When the temperature is too low, the probability of freezing will become smaller, and the liquid water in the air will directly condense into ice crystals and fall. The higher the water content, the greater the amount of freezing, and the ice type changes from frost ice to clear ice. In addition, the diameter of the water droplets is related to the area of ​​the impact surface and the shape of the ice. Air humidity is related to the humid regional climate and temperature. If the temperature is too low, the water droplets will directly condense into ice crystals and fall, so the humidity will not be too high.

[0040] (3) Blade icing is a changing process, usually starting from 1-2 blades to 3 blades. During this process, there will be obvious differences between the iced blades and the non-iced blades, resulting in differences in aerodynamic characteristics and blade deformation, which will be reflected in the obvious differences in the clearance of the three blades. The deviation dispersion over-limit warning model uses a sliding window to judge the difference in the clearance of three adjacent blades in the same impeller cycle, which is the deviation dispersion mentioned in this patent.

[0041] The data set that meets the icing environment condition model is entered into the deviation dispersion over-limit warning model for calculation. The difference in clearance between three adjacent blades in the same impeller cycle is determined by sliding window, i.e., deviation dispersion. The calculation formula of deviation dispersion devi is as follows:

[0042]

[0043] Where i is the impeller rotation period, Ai, Bi, and Ci are the clearance values ​​of the three blades in the same impeller rotation period i. Furthermore, the sliding window deviation discreteness Devj is calculated as follows: Where Devj is the sliding window deviation discreteness of the jth impeller rotation period, and n is the window length; if the sliding window deviation discreteness Devj is greater than the set deviation discreteness over-limit warning threshold, the blade icing alarm is triggered and the calculation is terminated; if the sliding window deviation discreteness Devj is not greater than the set deviation discreteness over-limit warning threshold, the calculation continues to the next stage.

[0044] Usually, the maximum value of the sliding window deviation dispersion under normal working conditions of the wind turbine is collected and analyzed, and the maximum value is multiplied by a safety factor to set a deviation dispersion over-limit warning threshold, and the safety factor is 1.3.

[0045] (4) When all three blades are frozen, due to symmetry, there is no obvious periodic difference in the deviation of the blade clearance. The warning capability of the deviation dispersion over-limit warning model decreases. However, due to the instability of the aerodynamic characteristics, the blade clearance will show obvious oscillation, reducing the stability and safety of the unit operation.

[0046] First, establish a reference clearance table. The preparation work requires the development of a control parameter clearance reference table for different models based on historical data, that is, to collect a sufficiently long period of historical data on clearance conditions, establish the relationship between speed, pitch angle and clearance, divide the speed and pitch angle into bins at appropriate intervals, and use the average clearance in each bin as the reference value, which reflects the clearance value applied by the unit under stable wind conditions. A control parameter clearance reference table can be established by covering the vast majority of unit operation conditions, such as Figure 2 As shown, in the subsequent calculation process, the clearance reference value can be obtained by interpolation in the table based on the current rotation speed and pitch angle.

[0047] The data set that is not greater than the deviation discreteness over-limit warning threshold in step (3) enters the clearance oscillation over-limit warning model for calculation. The clearance oscillation over-limit warning model is preset with a clearance reference table, which is prepared by combining the clearance historical data of different models under different unit control parameters; the clearance reference table is queried according to the unit control parameters in the data set to obtain the clearance reference value under this unit control parameter, and the clearance oscillation value osc is further calculated. i : In the formula, cl i is the ith clearance value, CL (ωi,αi) is the speed ω in the clearance reference table i and pitch angle α i The clearance value at that time; then calculate the sliding window clearance oscillation value Osc j , the calculation formula is: If the sliding window clearance oscillation value Osc j If the clearance oscillation exceeds the set warning threshold, the blade icing alarm will be triggered.

[0048] Usually, the maximum value of the sliding window clearance oscillation value under normal conditions of the unit is collected and analyzed, and the maximum value is multiplied by a safety factor to set a clearance oscillation excessive warning threshold, wherein the safety factor is 1.3.

[0049] After the blade icing alarm is triggered, the alarm information is uploaded to the wind turbine main control system, and the crew responds and handles it.

[0050] The method provided by the present invention is demonstrated below in conjunction with specific application examples.

[0051] Example 1

[0052] The data of a known icing unit were selected for analysis, including four groups of data for comparative analysis, including normal period, beginning of icing, partial blade icing, and 3 blade icing, to test the application effects of the deviation discreteness over-limit warning model and the clearance oscillation over-limit warning model.

[0053] The application effect of the deviation dispersion over-limit warning model is as follows Figure 3-Figure 6 As shown in the figure, the deviation discreteness of the three blades in the normal period does not exceed 0.04, the warning threshold is set to 0.06, with a certain margin, and the window length is 30. It can be seen from the figure that the beginning of icing, partial blade icing, and 3 blade icing are basically greater than the warning threshold, which can play a good warning effect. Among them, in the beginning of icing samples, the blades are obviously frozen after 5 impeller cycles, which can play an obvious alarm effect. It can also be seen that in the 3 blade icing samples, the margin between the warning value and the measured value is small, and there may be a risk of underreporting.

[0054] The application effect of the clearance oscillation over-limit warning model is as follows Figure 7-10 As shown in the figure, the deviation dispersion of the three blades in the normal period does not exceed 0.56, the warning value is set to 0.7, with a certain margin, and the window length is 90. It can be seen from the figure that the initial icing, partial icing of blades, and icing of three blades are basically greater than the warning value, which can play a good warning effect. Among them, in the initial icing sample, the blades are obviously frozen about 17 impeller cycles (51 clearances), which can play an obvious alarm effect. It can also be seen that in the 3-blade icing sample, the warning rate is 100%.

Claims

1. A blade icing monitoring and early warning method based on blade clearance data, characterized in that The following steps are involved: (1) Collect blade clearance data and temperature and humidity data from the blade clearance monitoring system and temperature and humidity sensors installed on the wind turbine, and obtain unit control parameters from the unit control system, group the above parameters by time and construct a data set; (2) Constructing an icing environment condition model through an empirical model, comparing the data set in the previous step with the icing environment condition model, and the data set that meets the icing environment condition model proceeds to the next stage for calculation, while the data set that does not meet the icing environment condition model terminates the calculation; (3) The data set that meets the icing environment condition model is entered into the deviation dispersion over-limit warning model for calculation. The difference in the clearance of three adjacent blades in the same impeller cycle is determined by sliding window, that is, the deviation dispersion. The calculation formula of the deviation dispersion devi is as follows: Where i is the impeller rotation period, Ai, Bi, and Ci are the clearance values ​​of the three blades in the same impeller rotation period i. Furthermore, the sliding window deviation discreteness Devj is calculated as follows: Where Devj is the sliding window deviation discreteness of the jth impeller rotation cycle, and n is the window length; if the sliding window deviation discreteness Devj is greater than the set deviation discreteness over-limit warning threshold, the blade icing alarm is triggered and the calculation is terminated; if the sliding window deviation discreteness Devj is not greater than the set deviation discreteness over-limit warning threshold, the calculation continues to the next stage; (4) The data set that is not greater than the deviation discreteness over-limit warning threshold enters the clearance oscillation over-limit warning model for calculation. The clearance oscillation over-limit warning model is preset with a clearance reference table, which is prepared by combining the clearance historical data of different models under different unit control parameters; the clearance reference value under the unit control parameters in the data set is obtained by querying the difference against the clearance reference table, and the clearance oscillation value osc is further calculated. i : In the formula, cl i is the clearance value of any blade, CL (ωi,αi) is the speed ω in the clearance reference table i and pitch angle α i The clearance value at that time; then calculate the sliding window clearance oscillation value Osc j , the calculation formula is: If the sliding window clearance oscillation value Osc j If the clearance oscillation exceeds the set warning threshold, the blade icing alarm will be triggered.

2. The blade icing monitoring and early warning method based on blade clearance data according to claim 1 is characterized in that: The freezing environment condition model described in step (2) includes freezing temperature conditions and freezing humidity conditions. When the temperature and humidity data in the data set simultaneously meet the freezing temperature conditions and the freezing humidity conditions, the next stage of calculation is entered.

3. The blade icing monitoring and early warning method based on blade clearance data according to claim 2 is characterized in that: The freezing temperature condition is -40 to 0°C, and the freezing humidity condition is that the air humidity is greater than 80%.

4. The blade icing monitoring and early warning method based on blade clearance data according to claim 1 is characterized in that: In step (3), the maximum value of the sliding window deviation discreteness under normal operating conditions of the wind turbine is collected and analyzed, and the maximum value is multiplied by the safety factor to set the deviation discreteness over-limit warning threshold.

5. The blade icing monitoring and early warning method based on blade clearance data according to claim 4 is characterized in that: The safety factor is 1.

3.

6. The blade icing monitoring and early warning method based on blade clearance data according to claim 1 is characterized in that: In step (4), the maximum value of the sliding window clearance oscillation value under normal conditions of the unit is collected and analyzed, and the maximum value is multiplied by the safety factor to set the clearance oscillation over-limit warning threshold.

7. The blade icing monitoring and early warning method based on blade clearance data according to claim 6 is characterized in that: The safety factor is 1.

3.

8. The blade icing monitoring and early warning method based on blade clearance data according to claim 1 is characterized in that: In step (4), a control parameter clearance reference table for different models is formulated based on historical data, the clearance condition historical data of a preset time is collected, the relationship between the speed, pitch angle and clearance is established, the speed and pitch angle are divided into intervals, and the average clearance in each bin is used as the reference value.

9. The blade icing monitoring and early warning method based on blade clearance data according to claim 1 is characterized in that: After the blade icing alarm is triggered in steps (3) and (4), the alarm information is uploaded to the wind turbine main control system, and the crew responds and handles it.

Images