A wind turbine blade icing monitoring system and method

Abstract

The present invention discloses a monitoring system and method for icing on wind turbine blades, which relates to the technical field of wind power generation. The system collects temperature, humidity, and vibration signals in real time through a multi-sensor module installed on the wind turbine blades, and uses an edge computing unit to fuse multi-parameter data to calculate the icing risk factor #imgabs0# of the blades. The calculation of the risk factor adopts a dynamic weighting model, which combines the normalized eigenvalues #imgabs1#, #imgabs2#, #imgabs3# to accurately evaluate the icing state of the blades. When the risk factor exceeds the set threshold, the system triggers an alarm and transmits the data to the cloud through a wireless communication module. The cloud platform predicts the future icing risk based on the long short-term memory neural network (LSTM) model, combined with historical data and real-time monitoring data. The present invention improves the accuracy and real-time performance of icing monitoring, is applicable to operation in cold regions and high-altitude wind farms, and provides a reliable guarantee for the safe and efficient operation of wind turbines.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular, to a monitoring system and method for icing on wind turbine blades. Background Art

[0002] As an important part of clean energy, wind power generation has been widely promoted globally in recent years. However, with the continuous increase in the installed capacity of wind power, the complexity of the operating environment of wind turbines has become increasingly apparent. Especially in cold regions and high-altitude areas, the icing problem on wind turbine blades has become an important factor affecting the efficiency and safety of wind power.

[0003] Influence of icing on wind turbine blades: When wind turbine blades operate in a cold environment, supercooled water droplets in the air hit the blade surface and freeze, easily forming ice. This phenomenon not only affects the normal operation of the wind turbine but may also pose the following serious threats to the safety and economy of the wind turbine:

[0004] Decrease in power generation efficiency: Ice formation on the blades changes the aerodynamic profile, resulting in flow separation and a decrease in lift, thereby significantly reducing the power generation efficiency. Research shows that in severe icing conditions, the power generation of the wind turbine may decrease by 20% - 50%.

[0005] Increase in mechanical load: Uneven icing on the blades disrupts the dynamic balance of the rotating system, increases the vibration amplitude, and causes excessive wear on core components such as the main shaft, gearbox, and bearings.

[0006] Safety hazards: During the rotation of the ice-covered blades, the ice may fall off due to centrifugal force, threatening the safety of nearby equipment and personnel. In addition, the structural strength of the blades may also be damaged due to excessive vibration caused by icing.

[0007] Existing icing monitoring technologies: For the icing problem on wind turbine blades, various technical solutions have been proposed to monitor the icing state of the blades, mainly including the following categories:

[0008] Meteorological data monitoring method: By monitoring meteorological parameters such as environmental temperature, humidity, and wind speed, and combining empirical formulas to predict whether the blades are iced. This method is simple to implement and has a low cost, but it can only indirectly infer the icing state and lacks an accurate reflection of the actual icing condition of the blades.

[0009] Visual monitoring method: Using cameras or infrared imaging devices to monitor the state changes on the blade surface. The visual method can clearly capture the icing situation on the blade surface in an ideal environment, but its monitoring effect will significantly decline in extreme weather conditions such as low light and heavy snowfall.

[0010] Vibration signal analysis method: By installing vibration sensors, analyze the changes in the vibration signals of the blades to determine whether icing occurs. This method is sensitive to the disruption of dynamic balance, can detect a certain degree of uneven icing, but has limited detection effect on incipient slight icing.

[0011] Surface temperature monitoring method: Use temperature sensors to directly measure the temperature of the blade surface, and combine with the ambient temperature to judge the possibility of icing. This method is easy to implement, but a single temperature parameter cannot reflect the complex icing state, and there may be false alarms or missed alarms.

[0012] Deficiencies of the prior art: Although the above methods can achieve the monitoring of blade icing to a certain extent, there are still many technical limitations, which are mainly manifested in the following aspects:

[0013] Single-parameter limitation: Most current monitoring methods rely on a single parameter (such as temperature or vibration signal) to judge icing, and do not comprehensively consider various physical characteristics of the blade surface, especially the icing state under the action of multiple variables.

[0014] Inaccurate meteorological monitoring method: The prediction model based on ambient temperature and humidity is easily affected by the fluctuations of meteorological conditions and cannot truly reflect the icing condition on the blade surface.

[0015] Phenomena of missed alarms and false alarms: The recognition rate for slight icing or incipient icing is relatively low, and the monitoring effect for complex icing (such as slight disruption of dynamic balance but no obvious temperature anomaly) is not good.

[0016] Reaction lag: Some monitoring schemes rely on manual data analysis or meteorological prediction and it is difficult to respond immediately when icing occurs.

[0017] Limited data processing ability: Most current systems adopt simple threshold judgment or single-parameter analysis and do not utilize the multi-sensor fusion data to achieve fast calculation and real-time evaluation.

[0018] Difficulty in coping with complex environments: In alpine regions, the wind field environment is mostly strong wind, high humidity and low temperature. Visual devices are restricted by light, and vibration signals are easily interfered, resulting in unreliable monitoring results.

[0019] Weak anti-interference ability: Existing systems are easily interfered in a wind field with a complex electromagnetic environment, especially the signal attenuation of wireless communication devices is serious during long-distance transmission.

[0020] Lack of intelligent thawing determination: The current thawing control process is usually judged manually, lacking an automated thawing completion determination function based on real-time data, which may lead to over-operation or premature stop of the thawing device, thus affecting the operation efficiency and energy utilization rate.

[0021] To make up for the deficiencies of the existing technology, future blade icing monitoring systems should have the following capabilities:

[0022] Multi-parameter integrated monitoring: By combining multiple parameters such as temperature, humidity, and vibration signals, and using data fusion algorithms to comprehensively reflect the actual icing condition on the blade surface.

[0023] Intelligent prediction: Utilize big data and machine learning technologies to predict future icing risks based on historical data and real-time data, and improve the early warning ability of the system.

[0024] Real-time and reliability: Introduce an efficient edge computing unit to achieve rapid data processing and real-time judgment, and improve the transmission reliability through high-stability wireless communication technologies.

[0025] Automatic thawing determination: Develop a thawing completion determination algorithm based on the dynamic change of risk factors, and monitor the operation effect of the thawing device in real time to avoid energy waste.

[0026] Therefore, this application aims to design and provide a fan blade icing monitoring system and method to solve the above problems.

[0027] The purpose of the present invention is to provide a fan blade icing monitoring system and method for the deficiencies of the existing technology to solve the problems raised in the background technology. Summary of the Invention

[0028] To achieve the above object, the present invention provides the following technical solutions:

[0029] One aspect of the solution of the present invention provides a fan blade icing monitoring system, which includes:

[0030] A sensor module, installed at the leading edge, middle, and trailing edge positions of the fan blade, and a set of sensor groups are set at each position. The sensor group includes:

[0031] A surface temperature sensor for measuring the surface temperature of the blade , in degrees Celsius;

[0032] A capacitive humidity sensor for measuring the surface humidity of the blade , in relative humidity percentage;

[0033] A micro-vibration accelerometer for collecting the vibration signal of the blade , in decibels;

[0034] A data processing and edge computing unit: The data processing and edge computing unit communicates with the sensor module and is used for multi-parameter fusion processing of real-time data, and judges the blade icing risk factor through the following formula , that is:

[0035] ;

[0036] Among them, , , are weight coefficients, satisfying ;

[0037] , , are respectively the normalization functions of temperature, humidity and vibration characteristics;

[0038] When , the system determines that the blade is iced and triggers an alarm signal;

[0039] Wireless communication module: Adopting LoRa or 5G technology, it transmits real-time data to the cloud;

[0040] Cloud computing platform: Through regional meteorological data and historical data, combined with the long short-term memory neural network (LSTM) model, it predicts the icing risk factor within the next 30 minutes. The calculation formula is:

[0041] ;

[0042] Among them, , which is the current monitoring data vector;

[0043] Alarm and control module: When the blade icing risk factor or the predicted value exceeds the threshold , it triggers the blade heating device or the shutdown protection operation.

[0044] Furthermore, the power spectrum of the vibration signal is calculated based on wavelet transform. The formula is:

[0045] ;

[0046] Among them, is the wavelet transform result of the vibration signal, is the frequency variable, is the time variable, .

[0047] Furthermore, the normalization function of the temperature is calculated through the following formula:

[0048] ;

[0049] Among them, and are the maximum and minimum values of the temperature monitoring range respectively.

[0050] Furthermore, the humidity normalization function is calculated by the following formula:

[0051] ;

[0052] where and are the maximum and minimum values of the humidity monitoring range respectively.

[0053] Furthermore, the vibration normalization function is calculated by the following formula:

[0054] ;

[0055] where and are the maximum and minimum values of the vibration characteristics respectively.

[0056] Furthermore, the weight coefficient , , is determined by the following optimization formula:

[0057] ,

[0058] ;

[0059] where is the th risk factor calculated from real-time data, is the real icing risk factor in historical data, is the number of samples, indicates that during the optimization process, the weights , , must satisfy the constraint that the sum of the weights is 1.

[0060] Furthermore, when the cloud computing platform optimizes the LSTM model, based on the following loss function :

[0061] ;

[0062] where is the icing risk factor predicted by the model, is the true value, is the predicted future time step.

[0063] Furthermore, the alarm and control module determines whether thawing is completed based on the following formula:

[0064] ;

[0065] in, Icing risk factor recalculated after blade thawing;

[0066] is the real-time temperature of the leaf surface after thawing, in degrees Celsius;

[0067] It is the real-time humidity of the leaf surface after thawing, in relative humidity percentage;

[0068] is the power spectrum amplitude of the vibration signal after the blade surface is thawed, in decibels;

[0069] Based on temperature Normalization function of ;

[0070] Based on humidity Normalization function of ;

[0071] Based on vibration signal Normalization function of ;

[0072] , , are the weight coefficients of temperature, humidity and vibration signals respectively, and satisfy ;

[0073] set up is the thawing completion threshold, when When the blades are judged to be thawed, the system stops heating or releases the alarm signal.

[0074] Furthermore, the wireless communication module adopts LoRa technology, specifically a module based on SemtechSX127x series or SX126x series chips, with a data transmission range of not less than 10 kilometers and anti-interference performance.

[0075] Another aspect of the present invention provides a method for monitoring icing of fan blades, which is implemented based on the above-mentioned fan blade icing monitoring system and includes the following steps:

[0076] The blade surface temperature is collected through the sensor module ,humidity and vibration signal ;

[0077] The data processing unit fuses the , , to calculate the risk factor ;

[0078] Let be the icing risk threshold. When , an alarm is triggered and uploaded to the cloud;

[0079] The cloud platform predicts the future risk factor ;

[0080] When the alarm signal is triggered, the control module activates the blade heating device or the shutdown protection operation.

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] 1. The technical solution of the present invention adopts a multi-parameter fusion algorithm, taking temperature, humidity, and vibration signals as the core monitoring indicators, and calculating the risk factor through dynamic weighting, which comprehensively reflects the icing condition on the blade surface. Compared with the traditional monitoring method relying on a single parameter, the solution of the present invention can accurately identify the icing states of slight icing, local icing, and complex environments through normalization and weighted optimization, significantly reducing the false alarm and missed alarm rates;

[0083] 2. The technical solution of the present invention introduces the LSTM model in the cloud platform, combines historical data and real-time monitoring data to predict the future icing risk factor ; This intelligent prediction mechanism breaks through the lag of traditional monitoring, can detect potential risks in advance, and provides an efficient early warning for the operation of the wind farm; at the same time, the multi-parameter weights are optimized through historical data , , , realizing the dynamic adjustment of the risk factor, further improving the accuracy and flexibility of icing assessment;

[0084] 3. The technical solution of the present invention realizes the automatic determination of the thawing completion state through real-time data monitoring after the operation of the thawing device, combined with the dynamically calculated risk factor , avoiding the disadvantages of manual intervention or traditional fixed-time control; the system of the present invention can accurately adjust the operation time of the thawing device according to the actual state of the blade, effectively saving energy, while reducing the loss caused by excessive operation of the equipment, and significantly improving the system energy efficiency and control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 FIG. is a system block diagram of a wind turbine blade icing monitoring system proposed by the present invention;

[0086] Figure 2 This is the flowchart of a method for monitoring ice formation on a wind turbine blade proposed by the present invention. Detailed implementation manners

[0087] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0088] The following will be combined with Figure 1 - Figure 2 multiple embodiments to describe the specific implementation manners of the present invention in detail.

[0089] The implementation solution of the present invention provides a wind turbine blade ice formation monitoring system and method, aiming to accurately judge the ice formation state of the blade by real-time monitoring of the temperature, humidity and vibration signals of the wind turbine blade, combining a multi-parameter fusion algorithm and cloud intelligent prediction, and taking necessary protection measures.

[0090] I. Description of the overall system framework

[0091] The system includes the following modules:

[0092] Sensor module: Installed on the surface of the wind turbine blade, an integrated multi-parameter sensor is used (including a surface temperature sensor; a capacitive humidity sensor; a micro vibration accelerometer), and the surface state data of the blade is collected in real time. A slip ring is provided at the wind turbine shaft, and the sensor is connected to the data processing and edge computing unit through the slip ring;

[0093] Data processing and edge computing unit: Based on an embedded AI chip, deployed in the wind turbine nacelle, preprocessing and anomaly recognition of the sensor data are performed.

[0094] Wireless communication module: Transmits the data to the cloud through LoRa or 5G technology.

[0095] Cloud big data platform: Integrates meteorological data, and combines machine learning algorithms for ice formation prediction and alarm decision-making.

[0096] Alarm and control module: Triggers an alarm signal according to the calculation result, and starts the thawing device or shutdown protection.

[0097] II. System operation process

[0098] Data acquisition: The sensor module consists of the following components:

[0099] Surface temperature sensor: Collects the temperature data on the surface of the blade , with the unit of degrees Celsius (°C).

[0100] Capacitive humidity sensor: Collect humidity data on the leaf surface , with the unit of relative humidity percentage (%).

[0101] Micro-vibration accelerometer: Collect vibration signals of the leaf , with the unit of decibel (dB).

[0102] Vibration signal The characteristics of are calculated by wavelet transform to obtain the power spectrum , and the formula is as follows:

[0103] ;

[0104] where is the power spectrum of the vibration signal at frequency ; is the result of wavelet transform of the vibration signal.

[0105] Risk factor calculation: The data processing and edge computing unit normalizes the collected data and calculates the icing risk factor of the leaf .

[0106] The specific steps are as follows:

[0107] Temperature normalization:

[0108] ;

[0109] where is the result of temperature normalization; and are the maximum and minimum temperatures within the monitoring range respectively.

[0110] Humidity normalization:

[0111] ;

[0112] where is the result of humidity normalization; and are the maximum and minimum humidities respectively.

[0113] Vibration normalization:

[0114] ;

[0115] where is the result of vibration feature normalization; and are the maximum and minimum vibration features respectively.

[0116] Risk factor calculation: ;

[0117] in, , , is the multi-parameter weight coefficient, satisfying ; is the icing risk factor of the blade.

[0118] Determine whether it is frozen: If , the system determines that the blades are frozen and triggers an alarm signal.

[0119] Cloud prediction and optimization: The cloud computing platform optimizes weights based on historical data using the following formula:

[0120] ,

[0121] ;

[0122] In addition, the LSTM model is used to predict the icing risk factor within the next 30 minutes:

[0123] ;

[0124] in, , which is the current monitoring data.

[0125] Thawing completion judgment: After the thawing device is started, the system recalculates the risk factor using the following formula:

[0126] ;

[0127] when , the system determines that the thawing is completed and releases the alarm signal.

[0128] 3. Implementation Method

[0129] Example 1: Detailed calculation process for real-time monitoring of a single wind turbine

[0130] Background conditions: Wind farm environment: northern high-altitude cold area, winter temperature -20°C to 5°C, humidity range 10%-95%, vibration characteristic range 0-60dB.

[0131] Current sensor data collection: temperature ;humidity ; Vibration power spectrum .

[0132] Calculation process:

[0133] Temperature normalization calculation:

[0134] ;

[0135] Set the temperature range , substitute the data:

[0136] ;

[0137] Humidity normalization calculation:

[0138] ;

[0139] Set the humidity range , substitute the data:

[0140] ;

[0141] Vibration normalization calculation:

[0142] ;

[0143] Set the vibration range , substitute the data:

[0144] ;

[0145] Risk factor calculation:

[0146] ;

[0147] Set the weights , , , substitute the normalization results:

[0148] ;

[0149] Determine whether icing occurs: Set the icing risk threshold . Because , the system determines that the blade is iced.

[0150] Example 2: Status monitoring after blade thawing

[0151] Background conditions: After the heating device operates for 20 minutes, when thawing is completed, the sensor collects data: Temperature ; Humidity ; Vibration power spectrum .

[0152] Calculation process:

[0153] Temperature normalization calculation:

[0154] ;

[0155] Substitute the temperature range and data:

[0156] ;

[0157] Humidity normalization calculation:

[0158] ;

[0159] Substitute the humidity range and data:

[0160] .

[0161] Example 3: Calculation process for large-scale monitoring of the wind field:

[0162] Background conditions: Wind field scale: It includes 50 wind turbines, each with 3 blades, for a total of 150 monitoring points.

[0163] Real-time risk factor for each blade: Wind turbine 1: , , ; Wind turbine 2: , , .

[0164] Calculation process:

[0165] Calculation of the risk factor for a single wind turbine:

[0166] ;

[0167] Risk factor for wind turbine 1:

[0168] ;

[0169] Risk factor for wind turbine 2:

[0170] ;

[0171] Determine whether to alarm: Set the alarm threshold for the wind turbine ;

[0172] Wind turbine 1: , trigger the alarm signal;

[0173] Wind turbine 2: , no alarm is triggered.

[0174] Example 4: Application of optimizing weights based on historical data:

[0175] Background conditions: During the past month of operation of a certain wind turbine, the true values of the risk factors collected and the system calculated values

[0176] , , ;

[0177] System calculated value , , .

[0178] The system uses historical data to optimize the risk factor weights , , .

[0179] Calculation process:

[0180] Optimization objective: The system aims to minimize the squared error:

[0181] ,

[0182] ;

[0183] Set the initial weights: , , .

[0184] Error calculation:

[0185] For each sample the squared error:

[0186] Error ;

[0187] Substitute the data: Error ;

[0188] Optimization process: The system adjusts the weights by the gradient descent method , , , updated to:

[0189] , , ;

[0190] Optimization effect: Calculate the risk factor using the new weights , the error is reduced to 0.0015, and the optimization is completed.

[0191] Example 5: Application of thawing completion judgment:

[0192] Background conditions: The ice state of the fan blade triggers the thawing device. After running for 20 minutes, the thawing device stops, and the sensor collects the data after thawing:

[0193] ;

[0194] ;

[0195] .

[0196] The system needs to recalculate the risk factor after the thawing is completed .

[0197] Calculation process: Temperature normalization: ;

[0198] Set temperature range , substitute:

[0199] ;

[0200] Humidity normalization: The formula is as follows:

[0201] ;

[0202] Humidity range , substitute:

[0203] ;

[0204] Vibration Normalization:

[0205] ;

[0206] Set vibration range , substitute:

[0207] ;

[0208] Calculation of risk factor after thawing:

[0209] ;

[0210] Set weight , , , after substituting:

[0211] .

[0212] Determine whether thawing is complete: set a thawing completion threshold.

[0213] Because the system determines that thawing is complete, it stops the heating device and releases the alarm signal.

[0214] The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A wind turbine blade icing monitoring system, characterized in that, The system includes: A sensor module, which is installed at the leading edge, middle and trailing edge positions of the fan blade. One set of sensor groups is set at each position. The sensor group includes: A surface temperature sensor, which is used to measure the surface temperature T of the blade, and the unit is degree Celsius; A capacitive humidity sensor, which is used to measure the surface humidity H of the blade, and the unit is relative humidity percentage; A micro vibration accelerometer, which is used to collect the blade vibration signal V, and the unit is decibel; A data processing and edge computing unit: The data processing and edge computing unit communicates with the sensor module and is used to perform multi-parameter fusion processing on real-time data. The icing risk factor R of the blade is judged through the following formula, that is: R = w1·f T (T) + w2·f H (H) + w3·f V (V); Where, w1, w2, and w3 are weight coefficients, and w1 + w2 + w3 = 1; f T (T), f H (H), f V (V) are the normalization functions of temperature, humidity and vibration characteristics respectively; When R≥R threshold the system determines that the blade is iced and triggers an alarm signal; A wireless communication module: Adopting LoRa or 5G technology, it transmits real-time data to the cloud; A cloud computing platform: Through regional meteorological data and historical data, combined with a long short-term memory neural network model, it predicts the icing risk factor within the next 30 minutes. The calculation formula is: Among them, X t = [T, H, V], which is the current monitoring data vector; Alarm and control module: When the blade icing risk factor R or the predicted value exceeds the threshold value R threshold it triggers the blade heating device or the shutdown protection operation; The power spectrum P(f) of the vibration signal V is calculated based on wavelet transform, and the formula is: Where, W(f,t) is the wavelet transform result of the vibration signal, f is the frequency variable, t is the time variable, and P(f) = V; The weight coefficients w1, w2, and w3 are determined through the following optimization formula: s.t. w1 + w2 + w3 = 1; Among them, R i is the i-th risk factor calculated according to real-time data, and R actual，i is the real icing risk factor in historical data. N is the number of samples. s.t. w1 + w2 + w3 = 1 means that in the optimization process, the weights w1, w2, and w3 must satisfy the constraint that the sum of the weights is 1; When optimizing the LSTM model on the cloud computing platform, based on the following loss function Among them, is the icing risk factor predicted by the model, R t+j is the true value, and M is the predicted future time step; The alarm and control module judges the completion of thawing based on the following formula: R after = w1·f T (T after ) + w2·f H (H after ) + w3·f V (V after ); Among them, R after is the icing risk factor recalculated after the blade is thawed; T after is the real-time temperature after the leaf surface thaws, in degrees Celsius; H after is the real-time humidity of the blade surface after thawing, in units of relative humidity percentage; V after is the power spectral amplitude of the vibration signal after the blade surface thaws, with the unit of decibel; f T (T after ) is a normalization function based on the temperature T after ; f H (H after ) is a normalization function based on the humidity H after ; f V (V after ) is a normalization function based on the vibration signal V after ; w1, w2, and w3 are the weight coefficients of the temperature, humidity, and vibration signals respectively, and w1 + w2 + w3 = 1; Setting R threshold，1 is the thawing completion threshold, when R after <R threshold，1 When the blades are judged to be thawed, the system stops heating or releases the alarm signal; The wireless communication module adopts LoRa technology, specifically a module based on Semtech SX127x series or SX126x series chips, with a data transmission range of not less than 10 kilometers and anti-interference performance; the normalization function f T (T) is calculated by the following formula: where T max and T min are the maximum and minimum values of the temperature monitoring range, respectively; The normalization function f H (H) is calculated by the following formula: Among them, H max and H min are the maximum and minimum values of the humidity monitoring range respectively; the normalization function f V (V) is calculated by the following formula: Among them, V max and V min are the maximum and minimum values of the vibration characteristics, respectively.

2. A method for monitoring ice formation on a wind turbine blade, which is implemented based on the wind turbine blade ice formation monitoring system described in claim 1, characterized in that, It includes the following steps: Collect the surface temperature T, humidity H, and vibration signal V of the blade through the sensor module; The data processing unit fuses the T, H, and V to calculate the risk factor R; Let R threshold be the icing risk threshold. When R ≥ R threshold , an alarm is triggered and uploaded to the cloud; The cloud platform predicts future risk factors When the alarm signal is triggered, the control module starts the blade heating device or the shutdown protection operation.

Images