Remote non-contact photo-thermal deicing method and system for fan blade surface

Abstract

The invention provides a remote non-contact photo-thermal deicing method for the surface of a fan blade, and the method comprises the steps: carrying out the ice condition data collection of the surface of the fan blade, and obtaining multi-modal data; performing feature recognition on the multi-modal data to obtain ice condition features; constructing a three-dimensional thickness distribution model by combining natural illumination factors based on the ice condition characteristics, and generating a deicing strategy; laser deicing is carried out according to the deicing strategy, and a laser beam is adjusted and controlled through a bee colony focusing algorithm; monitoring the temperature change of the coating on the surface of the fan blade in real time so as to adjust the laser power; and carrying out deicing rate detection on the surface of the de-iced fan blade, and if the deicing rate does not reach the standard, carrying out deicing operation again until the deicing rate reaches the standard. According to the method, through fusion of multispectral imaging, millimeter wave radar, a double-flow neural network and a dynamic environment modeling technology, accurate identification of ice conditions and adaptive laser deicing control are realized, and dependence on natural illumination is removed while energy consumption is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of stable and efficient operation and maintenance of wind power generation equipment, and in particular to a remote non-contact photothermal deicing method and system for the surface of wind turbine blades. Background Art

[0002] Wind energy is a clean, renewable energy source with vast reserves and widespread distribution. However, wind turbines built in high-altitude, cold regions often face severe icing problems. Ice accumulation on wind turbine blades can alter their aerodynamic performance, reducing power generation efficiency. In severe cases, it can cause unplanned downtime and blade breakage. Furthermore, ice falling from blades can pose a safety threat to nearby personnel and equipment. Therefore, addressing the issue of ice accumulation on wind turbine blades is crucial for the stable and efficient operation and maintenance of wind turbine equipment.

[0003] Active anti-icing technologies currently used on wind turbine blade surfaces, such as electric heating and hot air deicing, achieve deicing by heating the blade surface with external energy input. However, these technologies suffer from high deicing energy consumption, complex systems (requiring additional equipment such as heating, temperature control, and air supply), and expensive modification costs, making them difficult to implement in large-scale wind farms over the long term. Passive anti-icing technologies based on surface property regulation can achieve deicing without external energy input. However, once ice forms on the superhydrophobic coating, it is difficult to achieve rapid and efficient self-removal relying solely on the material's inherent properties. Although photothermal conversion materials can efficiently absorb sunlight and convert it into heat energy, preventing ice from forming on the blade surface or directly melting it, their deicing efficiency is highly dependent on natural lighting conditions. On cloudy days or at night, the light source is completely lost, resulting in insufficient energy absorption by the coating, slow ice melting, and even deicing failure. Therefore, it is crucial to design a remote, non-contact photothermal deicing method and system for wind turbine blade surfaces. Summary of the Invention

[0004] The purpose of the present invention is to provide a remote non-contact photothermal deicing method and system for the surface of wind turbine blades. By integrating multispectral imaging, millimeter-wave radar, dual-stream neural network and dynamic environment modeling technology, it can achieve accurate identification of ice conditions and adaptive laser deicing control, and reduce energy consumption while eliminating dependence on natural light.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A remote non-contact photothermal deicing method for a wind turbine blade surface comprises the following steps:

[0007] Using a multispectral camera and millimeter-wave radar to collect ice condition data on the surface of wind turbine blades, multimodal data is obtained. The multimodal data includes multispectral image data and radar echo data.

[0008] A two-stream convolutional neural network is used to identify features of multimodal data and obtain ice condition characteristics. Ice condition characteristics include ice type classification results and ice thickness.

[0009] Based on ice characteristics and combined with natural lighting factors, a three-dimensional thickness distribution model is constructed to generate de-icing strategies;

[0010] Perform laser de-icing operations according to the de-icing strategy and adjust and control the laser beam using a swarm focusing algorithm;

[0011] Monitor the coating temperature changes on the surface of the wind turbine blades in real time and adjust the laser power according to the coating temperature;

[0012] The deicing rate of the fan blade surface after deicing is tested. If the deicing rate does not reach the preset deicing standard, the deicing operation is repeated until the deicing rate reaches the preset deicing standard.

[0013] Optionally, ice condition data on the wind turbine blade surface is collected using a multispectral camera and millimeter-wave radar to obtain multimodal data, including:

[0014] The surface of the wind turbine blade is scanned by a multispectral camera in visible light and short-wave infrared bands to obtain raw image data;

[0015] Performing radiometric calibration on the original image data to obtain radiometric brightness image data;

[0016] Performing atmospheric scattering correction on the obtained radiometric brightness image data to obtain surface reflection image data;

[0017] The multi-band spatial registration operation of the surface reflectance image data is performed using the scale-invariant feature transformation algorithm to obtain multispectral image data.

[0018] Optionally, ice condition data on the surface of wind turbine blades is collected using a multispectral camera and millimeter-wave radar to obtain multimodal data, which also includes:

[0019] The millimeter-wave radar transmits electromagnetic wave signals to the surface of the wind turbine blade to obtain the original signal;

[0020] Performing coherent mixing operation on the transmitted electromagnetic wave signal and the original signal to obtain a difference frequency signal;

[0021] The difference frequency signal is decomposed into time and frequency by wavelet packet transform to obtain radar echo data.

[0022] Optionally, feature recognition is performed on the multimodal data using a two-stream convolutional neural network to obtain ice condition features, including:

[0023] The ice texture features of multimodal data are extracted through the first sub-network of the two-stream convolutional neural network;

[0024] The spectral absorption features of multimodal data are extracted through the second sub-network of the two-stream convolutional neural network;

[0025] Compare the ice layer texture features and spectral absorption features with the preset ice and snow state feature library to obtain the ice type classification results;

[0026] The ice thickness is calculated based on the time delay difference of the radar echo data.

[0027] Optionally, a three-dimensional thickness distribution model is constructed based on ice characteristics and combined with natural lighting factors, and a de-icing strategy is generated, including:

[0028] The irradiance sensor array performs multi-point sampling on the current ambient light intensity to obtain a real-time light distribution map;

[0029] Perform attenuation compensation on the real-time illumination distribution map through the atmospheric transmittance model to obtain the effective illumination intensity map;

[0030] Perform finite element meshing operations on the surface of the wind turbine blade to obtain a unit mesh;

[0031] The LSTM neural network is used to analyze the meteorological data provided by the weather station in time series to obtain the predicted value of ice growth rate;

[0032] The thickness prediction distribution of the unit grid is obtained according to the predicted values ​​of ice thickness and ice growth rate;

[0033] The three-dimensional thickness distribution model is obtained by performing three-dimensional reconstruction of the predicted thickness distribution through non-uniform rational B-spline surface fitting.

[0034] Laser action parameters are generated according to the three-dimensional thickness distribution model, and conditional constraints are imposed on the laser action parameters to obtain the deicing strategy.

[0035] Optionally, laser deicing operations are performed according to the deicing strategy, and the laser beam is adjusted and controlled by a swarm focusing algorithm, including:

[0036] The adaptive particle swarm optimization algorithm is used to plan the path of the unit grid and obtain the optimal focusing sequence of the main beam;

[0037] Monitor the ice layer's temperature in real time and identify points where the temperature drops sharply;

[0038] Generate an auxiliary beam according to the temperature drop point, and adjust the laser power of the auxiliary beam according to the real-time temperature;

[0039] The vibration spectrum of the wind blade surface is collected to perform vibration compensation and motion prediction on the wind blade surface to locate the position of the unit grid in real time.

[0040] Optionally, the coating temperature change on the surface of the wind turbine blade is monitored in real time, and the laser power is adjusted according to the coating temperature, including:

[0041] When the coating temperature is lower than a preset first temperature threshold, increasing the laser power to a reference power;

[0042] When the coating temperature reaches a first temperature threshold and is lower than a preset second temperature threshold, the laser power is maintained within a fluctuation range of the reference power and the laser power is fine-tuned according to the temperature gradient of the coating temperature;

[0043] When the coating temperature is higher than the second temperature threshold, the laser power is reduced to a preset safety power and the pulse mode is started.

[0044] Optionally, a deicing rate test is performed on the deiced fan blade surface. If the deicing rate does not meet a preset deicing standard, the deicing operation is repeated until the deicing rate meets the preset deicing standard, including:

[0045] Perform thermal radiation scanning on the surface of the wind turbine blades after deicing to obtain the residual ice temperature distribution map;

[0046] The dielectric constant of the de-iced wind turbine blade surface is scanned by millimeter-wave radar to obtain the complex dielectric constant distribution;

[0047] The residual ice volume is obtained by fusing the residual ice temperature distribution map and the complex dielectric constant distribution through DS evidence theory.

[0048] The de-icing rate is calculated based on the residual ice volume.

[0049] A remote non-contact optical thermal deicing system for wind turbine blade surfaces, comprising:

[0050] The data acquisition module is used to collect ice condition data on the surface of wind turbine blades using a multispectral camera and millimeter-wave radar to obtain multimodal data; the multimodal data includes multispectral image data and radar echo data;

[0051] The feature extraction module is used to identify features of multimodal data using a two-stream convolutional neural network to obtain ice condition characteristics. Ice condition characteristics include ice type classification results and ice thickness.

[0052] Strategy generation module, which is used to build a three-dimensional thickness distribution model based on ice characteristics and natural lighting factors, and generate de-icing strategies;

[0053] The de-icing execution module is used to perform laser de-icing operations according to the de-icing strategy and adjust and control the laser beam through the swarm focusing algorithm;

[0054] Laser adjustment module, used to monitor the temperature change of the coating on the surface of the wind turbine blade in real time and adjust the laser power according to the coating temperature;

[0055] The detection module is used to detect the deicing rate of the fan blade surface after deicing. If the deicing rate does not reach the preset deicing standard, the deicing operation is repeated until the deicing rate reaches the preset deicing standard.

[0056] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: the remote non-contact photothermal deicing method for the surface of a wind turbine blade provided by the present invention includes: collecting ice condition data on the surface of the wind turbine blade through a multispectral camera and a millimeter-wave radar to obtain multimodal data; the multimodal data includes: multispectral image data and radar echo data; performing feature recognition on the multimodal data through a two-stream convolutional neural network to obtain ice condition characteristics; the ice condition characteristics include: ice type classification results and ice layer thickness; based on the ice condition characteristics, a three-dimensional thickness distribution model is constructed in combination with natural lighting factors, and a deicing strategy is generated; laser deicing operations are performed according to the deicing strategy, and the laser beam is adjusted and controlled by a swarm focusing algorithm; real-time monitoring of the coating temperature changes on the surface of the wind turbine blade, and adjusting the laser power according to the coating temperature; performing deicing rate detection on the surface of the wind turbine blade after deicing, and if the deicing rate does not reach the preset deicing standard, the deicing operation is re-performed until the deicing rate reaches the preset deicing standard. This method achieves accurate identification of ice conditions and adaptive laser de-icing control by integrating multispectral imaging, millimeter-wave radar, dual-stream neural network and dynamic environment modeling technology, and eliminates dependence on natural light while reducing energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. 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 any creative work.

[0058] Figure 1 This is a flow chart of the remote non-contact photothermal deicing method of the present invention;

[0059] Figure 2 This is a flow chart of the laser beam adjustment control process of the present invention. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0061] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] like Figure 1 As shown, the present invention provides a remote non-contact photothermal deicing method for the surface of a wind turbine blade, comprising the following steps:

[0063] Step 100: Collect ice condition data on the surface of a wind turbine blade using a multispectral camera and a millimeter-wave radar to obtain multimodal data; the multimodal data includes multispectral image data and radar echo data;

[0064] Specifically, the multispectral camera, equipped with dual-band sensors for visible light and shortwave infrared, scans the rotating blade surface line by line using a high-precision servo mechanism. The visible light band captures the contours and color characteristics of the ice layer and blade base, while the shortwave infrared band responds to the characteristic absorption peaks of ice / water molecules at 1520nm and 1950nm, thereby detecting wet ice hidden in the surface layer. The raw image data generated by the scan is a digital matrix containing spatial coordinates and band intensity information.

[0065] The raw image data is then radiometrically calibrated using a pre-established calibration parameter library to generate radiometric brightness image data. Dark current noise data acquired by the multispectral camera under a standard light source, nonlinear response curves for each band, and sensor gain coefficients are incorporated into the radiation transfer equation to convert the pixel grayscale values ​​of the raw image into physically dimensioned radiometric brightness values. This eliminates the effects of sensor noise and optical system vignetting, ensuring the physical consistency of subsequent ice condition inversion.

[0066] The radiance image is then corrected for atmospheric scattering. The MODTRAN atmospheric radiation transfer model is used to calculate the scattering and absorption of downward solar radiation by the atmosphere, as well as the superposition of radiation from the upward path. This creates a lookup table of atmospheric influence factors. This table is used to compensate the radiance image pixel by pixel, ultimately outputting surface reflectance image data that eliminates environmental interference such as clouds and haze, highlighting the spectral characteristics of the leaf surface.

[0067] Finally, the scale-invariant feature transform (SIFT) algorithm was used to spatially register the multi-band surface reflectance images. The SIFT algorithm extracts stable features, such as blade leading edge bolts or coating defects, from images of different bands as key points, calculates their gradient directional histogram descriptors, and uses the RANSAC algorithm to select matching point pairs. An affine transformation matrix is ​​then constructed based on the matching results. The shortwave infrared band images are then precisely aligned to the visible light band reference coordinate system, generating spatially consistent multispectral image data.

[0068] It is important to note that by combining shortwave infrared (SWIR) with characteristic absorption bands, it is possible to penetrate the surface to detect wet and stratified ice, and identify the internal ice structure. Radiometric calibration and atmospheric correction eliminate environmental noise and reduce ice reflectivity inversion errors. SIFT registration technology addresses motion blur and multi-band misalignment, improving the spatial registration accuracy of images.

[0069] More specifically, the millimeter-wave radar uses a frequency-modulated continuous wave (FMCW) system with a center frequency of 94 GHz. A voltage-controlled oscillator (VCO) generates a linear frequency-modulated (LFM) signal at the radar transmitter. After amplification by a power amplifier, the signal is directed onto the blade surface via a conical lens antenna. The electromagnetic wave reflects at the ice-blade and ice-air interfaces, and the receiving array antenna captures the original echo signal, which contains phase and amplitude information. A dual-balanced diode structure then coherently mixes the transmitted and received signals. The transmitted signal serves as the local oscillator reference signal and is nonlinearly superimposed with the original echo signal. Due to blade motion, the echo has a Doppler shift, resulting in a difference frequency signal containing range information. This difference frequency signal is then subjected to full subtree decomposition using the Daubechies-5 wavelet basis, adaptively selecting a decomposition level of 6 to 8 layers, to generate the radar echo data.

[0070] It's important to note that non-contact millimeter-wave detection has strong penetration and is not limited by blade coating type. Coherent mixing converts distance information into easily processable frequency-domain signals, thereby improving the signal-to-noise ratio. Wavelet packet transform overcomes the time-frequency resolution limitations of FFT, reducing the false positive rate in ice identification.

[0071] Step 200: Perform feature recognition on the multimodal data using a two-stream convolutional neural network to obtain ice condition features; the ice condition features include: ice type classification results and ice thickness;

[0072] Specifically, a heterogeneous two-stream convolutional neural network was first constructed, consisting of two parallel subnetworks. The first subnetwork uses a ResNet-50 backbone with dilated convolutions. Its input is spatially registered multispectral image data of size 512×512×6, encompassing three bands: visible light, RGB, and shortwave infrared. The first subnetwork uses a shallow 7×7 convolution kernel to capture local texture features such as ice cracks and bubbles in the image. A mid-layer dilated convolution with a dilation rate of 2 then expands the receptive field to identify regions such as frost crystal clusters and wet ice seepage zones. Finally, global average pooling is performed to output a 1024-dimensional high-semantic feature vector to characterize the texture of the ice on the leaf surface.

[0073] The second subnetwork is a fully 1D convolutional structure, taking as input a spectral reflectance curve that has undergone dimension reduction. This subnetwork consists of four convolutional blocks. The first layer uses a wide convolution kernel with a width of 15, which is used to extract the broad absorption valleys of ice / water molecules at 1420-1520nm and 1900-2050nm. Subsequent layers use a stepwise reduction in convolution kernel width, with kernel widths of 7, 3, and 1, respectively, to gradually focus on the characteristic bands of ice-type sensitivity, such as the reflectance mutation point of transparent ice at 1310nm, and ultimately output the spectral absorption characteristics.

[0074] The spectral feature vector is then projected into the texture feature space through a learnable affine transformation matrix, and the feature weights of the two sub-network output features are obtained respectively using a gated fusion mechanism, and the two features are fused using a sigmoid function. The fused joint feature is then compared with a preset ice and snow state feature library, which contains template vectors of typical ice types and frost accumulation. The matching score is calculated by cosine similarity, and the highest-scoring category is the ice type classification result. In some embodiments, the identified joint feature is white granular or feathery, and the classification result is rime; the feature with a smooth and hard surface and a glassy appearance is identified as rime; the feature with a hard outer layer and a loose inner layer is identified as mixed rime; the feature with a needle-shaped or scaly shape is identified as frost; and the classification results of other irregular ice coverings such as wet snow, fan-shaped ice, and ice ridges are also included.

[0075] Finally, based on the radar echo data, the time delay difference sequence generated by wavelet packet transform is used to select the echo delay of the ice-coating interface and the ice-air interface, and the equation Calculate the ice thickness , where c is the speed of light, Δt is the echo delay, and εᵣ is the dielectric constant of the ice layer, which is dynamically selected based on the ice type classification results, such as 3.17 for clear ice and 5.3-7.1 for wet ice.

[0076] It's important to note that traditional single-modal detection methods, such as infrared thermal imagers, cannot distinguish visually similar ice types. However, the dual-stream network proposed in this paper can simultaneously capture both spatial texture and spectral absorption fingerprints, improving the accuracy of mixed ice identification. Dynamic dielectric constant correction is also used to couple radar time-delay data with optical ice type classification results, reducing ice thickness measurement errors.

[0077] Step 300: Based on ice characteristics and combined with natural lighting factors, a three-dimensional thickness distribution model is constructed, and a de-icing strategy is generated;

[0078] Specifically, the irradiance sensor array first collects ambient light field data, combines GPS positioning with astronomical algorithms to calculate the solar azimuth, and generates a real-time light distribution map with a spatial resolution of 1m×1m. Then, based on the air pressure, humidity, and aerosol concentration data from the weather station, combined with the atmospheric optical depth at 550nm obtained by the CE-318 sun photometer, the real-time light distribution map is converted into a map of the effective light intensity received on the leaf surface using a solar spectrum model. The solar spectrum model is expressed as follows: ,in is the effective irradiance, is direct radiation, is the scattered radiation, is the atmospheric transmittance, is the solar zenith angle, is the blade surface inclination angle.

[0079] An adaptive quadtree algorithm was then used to perform finite element meshing on the wind turbine blade surface, resulting in a curvature-sensitive triangular mesh with a mesh size of 2cm×2cm at the leading edge and 5cm×5cm at the trailing edge. Ice growth trends were then predicted using meteorological data such as temperature, humidity, wind speed, and liquid water content provided by the weather station using an LSTM neural network. This neural network, which employs a three-layer gated structure with 128 hidden units and was trained using historical observations of 2,000 icing events, ultimately outputs a predicted ice growth rate for each grid cell over the next 10 minutes, expressed as: ,in Ice growth prediction, is the local collection coefficient of the blade, is the wind speed, is the liquid water content, is the temperature-dependent phase transition function, For humidity.

[0080] The current ice thickness is then added to the predicted ice growth rate to predict the thickness distribution for each cell. Non-uniform rational B-splines (NURBS) are then used to insert thickness control points at the cell vertices. The influence factors of these control points are set based on the confidence level of the radar measurement. Regions with a signal-to-noise ratio (SNR) greater than 20dB are weighted to 1, while regions with a SNR less than 10dB are weighted to 0.3. A three-dimensional thickness distribution model is then derived by solving the tensor product equation of the NURBS basis functions.

[0081] Finally, the laser energy density is calculated for the three-dimensional thickness distribution model. The calculation formula is: ,in is the energy density, is the density of the ice layer, is the latent heat of fusion, which in some embodiments is 334 kJ / kg, is the current grid temperature, is the coating absorption rate, is the maximum laser power. Adding conditional constraints, such as the coating safety temperature must not exceed 80°C, ultimately yields deicing strategies for different unit grids.

[0082] It should be noted that the integration of real-time atmospheric transmission correction and LSTM time series prediction improves the accuracy of ice thickness prediction. NURBS surface fitting overcomes the limitations of grid discretization and significantly reduces reconstruction errors.

[0083] Step 400: Perform laser de-icing operation according to the de-icing strategy and adjust and control the laser beam through the swarm focusing algorithm; the specific steps are as follows: Figure 2 As shown, including:

[0084] Step 401: performing path planning on the unit grid using an adaptive particle swarm optimization algorithm to obtain an optimal focusing sequence for the main beam;

[0085] Specifically, the main beam path is planned based on the adaptive particle swarm optimization algorithm, and the particle swarm state is initialized with the unit grid divided by the three-dimensional thickness distribution model as the target. The fitness function expression of the algorithm is: ,in 、 and is the weight coefficient, is the ratio of theoretical melting energy consumption to actual input laser energy, is the ratio of the scanned grid to the total ice layer grid, and finally the optimal focusing sequence of the grid when the main beam performs de-icing operation is obtained.

[0086] Step 402: Monitor the temperature of the ice layer in real time and identify the temperature drop point;

[0087] Specifically, the temperature field of the ice layer is monitored in real time by an infrared thermal imager, a temperature distribution thermogram is generated at a frame rate of 10 Hz, and the temperature drop points are detected by the Sobel gradient operator.

[0088] Step 403: generating an auxiliary beam according to the temperature drop point, and adjusting the laser power of the auxiliary beam according to the real-time temperature;

[0089] Specifically, 2-4 independently controllable auxiliary beams are derived, and the auxiliary beams are positioned to the target area according to the coordinates of the temperature drop point. The laser power of the auxiliary beams is adjusted according to the real-time temperature. The adjustment formula is:

[0090] ;

[0091] in 8W / ℃, 0.5W / (℃·s), is the benchmark melting temperature corresponding to the current ice thickness, is the current temperature.

[0092] Step 404: Vibration compensation and motion prediction are performed on the surface of the wind turbine blade using the collected vibration spectrum of the surface of the wind turbine blade, so as to locate the position of the unit grid in real time.

[0093] Specifically, the blade vibration spectrum is generated based on the first-order flapping and second-order shimmy frequencies extracted from the radar echo data through FFT transformation. Using the ARIMA time series model, the blade deformation for the next 0.5 seconds is predicted based on the blade vibration data from the previous 5 seconds. The target unit grid is then tracked in real time based on the predicted deformation, ensuring that the grid center is continuously locked.

[0094] It should be noted that the optimal de-icing path design was achieved through an adaptive particle swarm optimization algorithm, eliminating the traditional fixed scanning path. Combined with the rapid response of the auxiliary beam, this significantly improved de-icing efficiency. Furthermore, real-time motion compensation during the de-icing process, using the vibration of the rotating blades, reduced positioning errors.

[0095] Step 500: monitoring the temperature change of the coating on the surface of the wind turbine blade in real time, and adjusting the laser power according to the coating temperature;

[0096] Specifically, when the coating temperature is lower than a preset first temperature threshold, the laser power is increased to the reference power; when the coating temperature reaches the first temperature threshold and is lower than a preset second temperature threshold, the laser power is maintained within the fluctuation range of the reference power and fine-tuned according to the temperature gradient of the coating temperature. The expression of the adjustment process is: ,in is the current laser power, is a temperature gradient; when the coating temperature is higher than the second temperature threshold, the laser power is reduced to a preset safety power and the pulse mode is started.

[0097] Step 600: Perform a de-icing rate test on the de-icing surface of the wind turbine blades. If the de-icing rate does not reach the preset de-icing standard, perform the de-icing operation again until the de-icing rate reaches the preset de-icing standard.

[0098] Specifically, a thermal radiation scan of the de-iced wind turbine blade surface is performed to obtain a residual ice temperature distribution map. A high-resolution infrared thermal imager is used to scan the entire blade surface in the 8-14μm infrared band at a frame rate of 15Hz. Due to the significant difference in thermal conductivity between residual ice and the blade coating, the residual ice area exhibits a different temperature signature in the thermal image than the surrounding coating. The residual ice area has a lower temperature and a more uniform distribution, while the temperature of the ice-free coating area fluctuates more significantly due to the influence of the environment and residual laser heat. During the scanning process, the thermal image data is spatially aligned with the pre-de-icing 3D model of the blade, mapping it to a 3D coordinate system on the blade surface to generate a residual ice temperature distribution map containing spatial coordinates and temperature values.

[0099] Next, a millimeter-wave radar is used to scan the dielectric constant of the de-iced blade surface to obtain the complex dielectric constant distribution. The radar performs a spiral scan of the blade surface with an angular resolution of 0.5°. After the transmitted electromagnetic wave is reflected by the blade surface, the receiver extracts the amplitude and phase information of the echo signal through coherent demodulation. This information is then combined with the phase change formula for electromagnetic wave propagation in different media to calculate the complex dielectric constant at each scanning point. Areas with a real part of the complex dielectric constant within the range of 3.0-3.3 and an imaginary part less than 0.1 are marked as candidate areas for residual ice. These areas are then aligned with the spatial coordinate system obtained from the thermal radiation scan to generate a complex dielectric constant distribution map, where each pixel contains the corresponding complex dielectric constant parameter.

[0100] The DS evidence theory was then used to fuse the residual ice temperature distribution and complex permittivity distribution to determine the residual ice volume. Both maps were divided into identical 1 cm × 1 cm grid cells, with each grid cell used as an identification target. For the temperature distribution, the average temperature within the grid cell was ≤ -2°C, serving as evidence 1. The basic probability allocation (BPA) for the "residual ice presence" hypothesis was calculated. Lower temperatures and greater temperature differences from the surrounding environment resulted in higher BPA values. For the complex permittivity distribution, the real part of the complex permittivity was 3.0-3.3 and the imaginary part ≤ 0.1, serving as evidence 2. The BPA was similarly calculated. The closer the permittivity was to pure ice, the higher the BPA value. The BPAs of the two pieces of evidence were then fused using the DS synthesis rule. When the confidence level of the fusion was ≥ 0.7, the grid cell was identified as a residual ice area. The total volume of residual ice was obtained by counting the number of grid cells with residual ice, combining the grid area with the average residual ice thickness inferred from the three-dimensional thickness distribution model.

[0101] Finally, the ratio of the difference between the total ice volume before deicing and the residual ice volume to the total ice volume before deicing is taken as the deicing rate.

[0102] The present invention also provides a remote non-contact optical thermal deicing system for the surface of a wind turbine blade, comprising:

[0103] The data acquisition module is used to collect ice condition data on the surface of wind turbine blades using a multispectral camera and millimeter-wave radar to obtain multimodal data; the multimodal data includes multispectral image data and radar echo data;

[0104] The feature extraction module is used to identify features of multimodal data using a two-stream convolutional neural network to obtain ice condition characteristics. Ice condition characteristics include ice type classification results and ice thickness.

[0105] Strategy generation module, which is used to build a three-dimensional thickness distribution model based on ice characteristics and natural lighting factors, and generate de-icing strategies;

[0106] The de-icing execution module is used to perform laser de-icing operations according to the de-icing strategy and adjust and control the laser beam through the swarm focusing algorithm;

[0107] Laser adjustment module, used to monitor the temperature change of the coating on the surface of the wind turbine blade in real time and adjust the laser power according to the coating temperature;

[0108] The detection module is used to detect the deicing rate of the fan blade surface after deicing. If the deicing rate does not reach the preset deicing standard, the deicing operation is repeated until the deicing rate reaches the preset deicing standard.

[0109] In some embodiments, except for the laser emission device, all other detection and identification related equipment can be carried on the drone for long-distance non-contact detection. The remote non-contact optical thermal deicing system can also switch to different working modes according to different weather conditions. There are three working modes, and the switching process is as follows:

[0110] Mode 1 (Photothermal Coating + Sunlight): On sunny days, the super-hydrophobic photothermal coating applied to the wind turbine blades converts sunlight into heat. The remote de-icing system monitors the ice's shedding and adhesion in real time. If ice melts and falls off, efficient, energy-saving de-icing is achieved.

[0111] Mode 2 (Photothermal Coating + Sunlight + Laser Assist): Even under sunlight, surface ice remains difficult to melt and adheres to the surface. In this case, a fixed laser is activated to assist with sunlight. With the assistance of the fixed laser, the coating's photothermal conversion generates sufficient heat to melt the surface ice.

[0112] Mode 3 (Photothermal Coating + Laser): On cloudy days with no sunlight, the photothermal coating applied to the wind turbine blades completely loses its photothermal deicing effect. Therefore, to achieve the desired deicing effect, a fixed laser can be used as a simulated light source. The photothermal conversion of the coating then generates sufficient heat to melt the ice on the surface.

[0113] The beneficial effects of the present invention are as follows:

[0114] 1) The system integrates multi-spectral imaging and millimeter-wave radar to collect multimodal data, and uses a two-stream convolutional neural network to extract ice texture and spectral absorption characteristics. The results are then compared with a pre-set ice and snow state feature library to obtain ice type classification results. The ice thickness is calculated based on the time delay difference of the radar echo data, improving the accuracy of ice condition identification.

[0115] 2) A de-icing strategy is generated by constructing a three-dimensional thickness distribution model based on ice characteristics and natural lighting factors. A swarm focusing algorithm is used to adjust the laser beam, and laser power is adjusted by real-time monitoring of coating temperature changes. This achieves adaptive de-icing control, avoids energy waste, and reduces energy consumption while ensuring de-icing effectiveness.

[0116] 3) It does not rely on natural light. On cloudy days, at night, or in conditions with no or insufficient light, laser can be used as an energy source to achieve de-icing, solving the problem of traditional photothermal conversion materials relying on natural light.

[0117] 4) The main beam and auxiliary beam are combined. The main beam operates in the optimal focusing sequence, and the auxiliary beam dynamically reinforces the temperature drop point. It can also compensate for blade vibration and predict motion, thereby improving de-icing efficiency.

[0118] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0119] The present invention uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A remote non-contact photothermal deicing method for the surface of a wind turbine blade, characterized in that: The steps include: Collecting ice condition data on the surface of wind turbine blades using a multispectral camera and millimeter-wave radar to obtain multimodal data; the multimodal data includes: multispectral image data and radar echo data; Performing feature recognition on the multimodal data through a two-stream convolutional neural network to obtain ice condition features; the ice condition features include: ice type classification results and ice layer thickness; Based on the ice characteristics and combined with natural lighting factors, a three-dimensional thickness distribution model is constructed to generate a de-icing strategy; Performing laser deicing operations according to the deicing strategy and adjusting and controlling the laser beam using a swarm focusing algorithm; Real-time monitoring of the coating temperature change on the surface of the wind turbine blade, and adjusting the laser power according to the coating temperature; The deicing rate of the fan blade surface after deicing is detected. If the deicing rate does not reach the preset deicing standard, the deicing operation is repeated until the deicing rate reaches the preset deicing standard.

2. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 1, characterized in that: Ice condition data on wind turbine blade surfaces is collected using a multispectral camera and millimeter-wave radar, generating multimodal data including: Scanning the surface of the wind turbine blade in visible light and short-wave infrared bands using the multispectral camera to obtain original image data; Performing a radiometric calibration operation on the original image data to obtain radiometric brightness image data; performing an atmospheric scattering correction operation on the obtained radiometric brightness image data to obtain surface reflection image data; The multi-spectral image data is obtained by performing a multi-band spatial registration operation on the surface reflection image data using a scale-invariant feature transformation algorithm.

3. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 2, characterized in that: Ice condition data on wind turbine blade surfaces is collected using a multispectral camera and millimeter-wave radar, generating multimodal data including: The millimeter-wave radar emits an electromagnetic wave signal to the surface of the wind turbine blade to obtain an original signal; performing a coherent mixing operation on the transmitted electromagnetic wave signal and the original signal to obtain a difference frequency signal; The radar echo data is obtained by performing a time-frequency decomposition operation on the difference frequency signal through wavelet packet transformation.

4. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 1, characterized in that: The multimodal data is subjected to feature recognition through a two-stream convolutional neural network to obtain ice condition features, including: Extracting ice texture features of the multimodal data through the first sub-network of the two-stream convolutional neural network; Extracting spectral absorption features of the multimodal data through the second sub-network of the two-stream convolutional neural network; Comparing the ice layer texture features and the spectral absorption features with a preset ice and snow state feature library to obtain an ice type classification result; The ice thickness is calculated based on the time delay difference of the radar echo data.

5. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 1, characterized in that: Based on the ice characteristics and combined with natural lighting factors, a three-dimensional thickness distribution model is constructed, and a de-icing strategy is generated, including: The irradiance sensor array performs multi-point sampling on the current ambient light intensity to obtain a real-time light distribution map; Performing an attenuation compensation operation on the real-time illumination distribution map through an atmospheric transmittance model to obtain an effective illumination intensity map; Performing a finite element meshing operation on the surface of the wind turbine blade to obtain a unit mesh; The LSTM neural network is used to analyze the meteorological data provided by the weather station in time series to obtain the predicted value of ice growth rate; Obtaining a predicted thickness distribution of the unit grid according to the ice layer thickness and the ice layer growth rate predicted value; Performing three-dimensional reconstruction of the predicted thickness distribution by non-uniform rational B-spline surface fitting to obtain the three-dimensional thickness distribution model; Laser action parameters are generated according to the three-dimensional thickness distribution model, and conditional constraints are imposed on the laser action parameters to obtain the deicing strategy.

6. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 5, characterized in that: Performing laser deicing operations according to the deicing strategy and adjusting and controlling the laser beam using a swarm focusing algorithm include: Performing path planning on the unit grid by using an adaptive particle swarm optimization algorithm to obtain an optimal focusing sequence of the main beam; Monitor the ice layer's temperature in real time and identify points where the temperature drops sharply; generating an auxiliary beam according to the temperature drop point, and adjusting the laser power of the auxiliary beam according to the real-time temperature; Vibration compensation and motion prediction are performed on the surface of the wind turbine blade using the collected vibration spectrum of the surface of the wind turbine blade, so as to locate the position of the unit grid in real time.

7. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 1, characterized in that: Real-time monitoring of the coating temperature change on the surface of the wind turbine blade and adjusting the laser power according to the coating temperature include: When the coating temperature is lower than a preset first temperature threshold, increasing the laser power to a reference power; When the coating temperature reaches the first temperature threshold and is lower than a preset second temperature threshold, maintaining the laser power within the fluctuation range of the reference power and fine-tuning the laser power according to the temperature gradient of the coating temperature; When the coating temperature is higher than the second temperature threshold, the laser power is reduced to a preset safety power and the pulse mode is started.

8. The remote non-contact photothermal deicing method for wind turbine blade surfaces according to claim 1, characterized in that: Performing a deicing rate detection on the surface of the wind turbine blade after deicing, and if the deicing rate does not reach a preset deicing standard, re-performing the deicing operation until the deicing rate reaches the preset deicing standard, including: Performing a thermal radiation scan on the surface of the wind turbine blade after deicing to obtain a residual ice temperature distribution map; Scanning the dielectric constant of the de-iced wind turbine blade surface using the millimeter-wave radar to obtain a complex dielectric constant distribution; The residual ice volume is obtained by fusing the residual ice temperature distribution map and the complex dielectric constant distribution through DS evidence theory; The de-icing rate is calculated according to the residual ice volume.

9. A remote non-contact thermal deicing system for wind turbine blade surfaces, characterized in that: include: The data acquisition module is used to collect ice condition data on the surface of wind turbine blades using a multispectral camera and millimeter-wave radar to obtain multimodal data; The multimodal data includes: multispectral image data and radar echo data; a feature extraction module for performing feature recognition on the multimodal data using a two-stream convolutional neural network to obtain ice condition features; the ice condition features include: ice type classification results and ice layer thickness; A strategy generation module is used to construct a three-dimensional thickness distribution model based on the ice characteristics and natural lighting factors, and generate a de-icing strategy; a deicing execution module, configured to perform laser deicing operations according to the deicing strategy and adjust and control the laser beam using a swarm focusing algorithm; A laser adjustment module, used to monitor the temperature change of the coating on the surface of the wind turbine blade in real time and adjust the laser power according to the coating temperature; The detection module is used to detect the deicing rate of the fan blade surface after deicing. If the deicing rate does not reach the preset deicing standard, the deicing operation is repeated until the deicing rate reaches the preset deicing standard.

Images