Method and system for de-icing control of wind power blades

Abstract

The embodiment of the application provides a kind of wind power generation blade deicing control method and system, belong to wind power technology field.The method comprises: the operating state information of wind generating set is collected, and the operating state information is executed feature extraction, obtains operating state feature;Based on the operating state feature, call pre-trained deicing state prediction model, determine the deicing prediction level under current operating state;Based on the deicing prediction level, determine the control scheme of gas-thermal deicing device;The control scheme is executed, and the deicing prediction level of wind generating set is updated in real time in execution process, to execute the control scheme correction of corresponding thermal deicing device when deicing prediction level changes.The application scheme ensures that deicing process is more real-time and intelligent, effectively improve the response speed and energy consumption optimization level of deicing system, to improve the deicing efficiency of fan blade in complex environment and the operating stability of wind turbine generator set.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wind power technology, in particular to an ice removal control method for a wind power blade and an ice removal control system for a wind power blade. BACKGROUND

[0002] When a wind turbine operates in a cold environment, ice is likely to form on the surface of its blades, which changes the aerodynamic characteristics and reduces the power generation efficiency, and may even cause blade damage or turbine shutdown. The existing air-thermal ice removal technology mainly relies on air blowing equipment to deliver heated air to the inner cavity of the blade to melt the external ice layer through heat conduction. However, the traditional air-thermal ice removal scheme still has many problems, which affect its ice removal effect and energy efficiency performance.

[0003] Most existing air-thermal ice removal systems use a fixed start-stop strategy, relying on simple environmental parameter threshold to determine whether to start the ice removal device, which is difficult to adapt to complex and variable weather conditions. For example, simply relying on the threshold of temperature, humidity or wind speed to determine the ice removal start-stop may lead to early or late ice removal start, thereby affecting energy consumption control and ice removal efficiency. In addition, since the icing conditions of different parts of the blade may differ significantly, the existing technology fails to take fine control for different icing levels, often using full-power heating, which not only increases energy consumption, but also may cause local overheating, reducing the overall efficiency of the system.

[0004] The control mode of the air blower is relatively simple, usually using a fixed wind speed and constant heating mode, which fails to optimize and adjust for different icing levels and temperature distribution of different blade regions. Due to the complex shape of the blade, the hot air flow path may not be uniform, and the existing system cannot ensure that all regions are adequately heated, leading to insufficient ice melting or heat waste in local areas. At the same time, the operating state of the fan blade has an important influence on the ice removal process, such as blade angle, speed and other factors that will change the airflow distribution and heat transfer efficiency, and the existing control scheme fails to effectively combine the fan operating state for dynamic optimization.

[0005] Current ice removal strategies generally use a passive approach, i.e., starting the ice removal system only when the detected ice reaches a set threshold, without being able to actively prevent based on weather forecast data. When the weather conditions have icing risks, if the anti-icing mode can be entered in advance and preheating at a low power is appropriately performed, it will help to reduce ice formation and reduce subsequent ice removal energy consumption. However, the existing technology lacks effective multi-modal data fusion means, and cannot fully utilize the comprehensive information of weather forecast, real-time environmental parameters and ice monitoring data to optimize the ice removal strategy.

[0006] In summary, the current air-thermal deicing scheme still has great limitations in the intelligent degree of start-stop control, energy optimization, fan operation state adaptability, and prediction and prevention capability, and it is difficult to achieve an intelligent deicing control strategy that is efficient, accurate, and energy-saving. Therefore, an improved deicing control method is urgently needed to improve the overall deicing efficiency and energy consumption management level. SUMMARY

[0007] The purpose of the embodiments of the present application is to provide a deicing control method and system for wind power blades to at least solve the problem of great limitations of the current air-thermal deicing scheme in the intelligent degree of start-stop control, energy optimization, fan operation state adaptability, and prediction and prevention capability.

[0008] To achieve the above-mentioned purpose, the first aspect of the present application provides a deicing control method for wind power blades, the method comprising: collecting operation state information of a wind power generator set and performing feature extraction on the operation state information to obtain operation state features; calling a pre-trained deicing state prediction model based on the operation state features to determine a deicing prediction level under the current operation state; determining a control scheme of an air-thermal deicing device based on the deicing prediction level; executing the control scheme and updating the deicing prediction level of the wind power generator set in real time during the execution process to execute control scheme correction of the corresponding thermal deicing device when the deicing prediction level changes.

[0009] Optionally, the operation state information of the wind power generator set includes any one or more of meteorological prediction information, real-time environmental parameter information, and real-time icing thickness information of the wind power blade.

[0010] Optionally, the feature extraction on the operation state information to obtain the operation state features comprises: pre-processing the operation state information; performing one-hot encoding on the pre-processed operation state information to obtain a numerical vector corresponding to each pre-processed operation state information; and performing feature fusion on the numerical vector of each pre-processed operation state information based on the AHP hierarchical analysis method to obtain a comprehensive feature vector as the operation state feature.

[0011] Optionally, the deicing prediction level includes a prevention level and a deicing level; the deicing level includes multiple levels, and the higher the level, the thicker the ice layer to be deiced.

[0012] Optionally, the air-thermal deicing device includes a heating module, a blower, and a blade angle adjustment device; and the determination of the control scheme of the air-thermal deicing device based on the deicing prediction level comprises: determining a target operation state of each air-thermal deicing device based on the deicing prediction level; and generating a corresponding control scheme based on the current operation state and the corresponding target operation state of each air-thermal deicing device.

[0013] Optionally, if the current deicing prediction level is the prevention level, the control scheme of the air-thermal deicing device is determined based on the deicing prediction level, including: predicting the icing time based on the operating state information of the wind turbine generator, obtaining the predicted icing time; determining the intervention time of each air-thermal deicing device based on the predicted icing time; generating an intermittent air supply operation scheme of each air-thermal deicing device based on the intervention time and a preset intermittent air supply timing table, as the control scheme of the air-thermal deicing device.

[0014] Optionally, if the current deicing prediction level is the deicing level, the control scheme of the air-thermal deicing device is determined based on the deicing prediction level, including: determining and predicting the icing thickness development based on the real-time icing thickness information of the wind turbine blade and the operating state information of the wind turbine generator; determining the thickness of the deicing layer based on the real-time icing thickness information and the prediction result of the icing thickness development; simulating the inverse development information of the icing thickness to the preset safety icing thickness value based on the thickness of the deicing layer, generating the target operating state of each air-thermal deicing device, and generating the corresponding control scheme based on the current operating state of each air-thermal deicing device and the corresponding target operating state.

[0015] Optionally, the deicing prediction level of the wind turbine generator is updated in real time during the execution process, so that the control scheme of the corresponding thermal deicing device is corrected when the deicing prediction level changes, including: collecting the operating state information of the wind turbine generator in real time during the execution process; extracting the features of the operating state information during the execution process to obtain the operating state features during the execution process; calling a pre-trained deicing state prediction model based on the operating state features during the execution process to determine the deicing prediction level under the current operating state; if the deicing prediction level under the current operating state is different from the deicing prediction level corresponding to the control scheme being executed, determining the control scheme of the air-thermal deicing device based on the deicing prediction level under the current operating state as a new control scheme, and executing the new control scheme.

[0016] The second aspect of the application provides a deicing control system for a wind turbine blade, the system comprising: a collection unit for collecting operating state information of a wind turbine generator and extracting features of the operating state information to obtain operating state features; a prediction unit for calling a pre-trained deicing state prediction model based on the operating state features to determine the deicing prediction level under the current operating state; a scheme generation unit for determining the control scheme of the air-thermal deicing device based on the deicing prediction level; and an execution unit for executing the control scheme and updating the deicing prediction level of the wind turbine generator in real time during the execution process, so that the control scheme of the corresponding thermal deicing device is corrected when the deicing prediction level changes.

[0017] In another aspect, the present application provides a computer readable storage medium, which stores instructions that, when executed on a computer, cause the computer to perform the above-mentioned ice removal control method for wind power generation blades.

[0018] Through the above technical solution, the present application scheme improves the accuracy of ice removal decision by collecting the running state information of the wind turbine generator set in real time, and performing feature extraction thereon to generate running state features. Using the pre-trained ice removal state prediction model, the method can intelligently evaluate the ice removal prediction level according to the current running state, avoiding the misjudgment or energy waste caused by the traditional method relying on a single threshold. Based on the prediction level, the strategy dynamically adjusts the control scheme of the air-thermal ice removal device to realize hierarchical ice removal and improve energy efficiency. At the same time, during the execution of the ice removal process, the system can monitor the state change of the wind turbine generator set in real time, update the ice removal prediction level, and correct the ice removal control scheme according to the latest situation, making it more adaptive. Through closed-loop control, the method ensures that the ice removal process is more real-time and intelligent, effectively improving the response speed and energy consumption optimization level of the ice removal system, thereby improving the ice removal efficiency of the wind turbine blade in complex environments and the operating stability of the wind turbine generator set.

[0019] Other features and advantages of the present application will be described in detail in the following detailed description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0020] The accompanying drawings are included to provide a further understanding of the embodiments of the application, and constitute a part of the specification, and are used together with the following detailed description of the embodiments to explain the embodiments of the application, but do not constitute a limitation of the embodiments of the application. In the drawings:

[0021] Figure 1 is a step flow chart of the ice removal control method for wind power generation blades provided by an embodiment of the present application;

[0022] Figure 2 is a system structure diagram of the ice removal control system for wind power generation blades provided by an embodiment of the present application. DETAILED DESCRIPTION

[0023] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present application, and are not intended to limit the present application.

[0024] Figure 1 is a method flow chart of the ice removal control method for wind power generation blades provided by an embodiment of the present application. As shown in Figure 1 , the present application embodiment provides an ice removal control method for wind power generation blades, which comprises:

[0025] Step S10: Collecting the operating state information of the wind turbine generator set, and performing feature extraction on the operating state information to obtain operating state features.

[0026] Specifically, the operating state information of the wind turbine generator set includes any one or more of meteorological prediction information, real-time environmental parameter information, and real-time icing thickness information of the wind power blade.

[0027] Further, the feature extraction on the operating state information to obtain operating state features includes: performing preprocessing on the operating state information; performing one-hot encoding on the preprocessed operating state information to obtain a numerical vector corresponding to each preprocessed operating state information; and performing feature fusion on the numerical vector of each preprocessed operating state information based on the AHP hierarchical analysis method to obtain a comprehensive feature vector as the operating state feature.

[0028] In the embodiments of the present application, the operating state information of the wind turbine generator set is composed of multiple dimensions of data, including but not limited to:

[0029] 1) Meteorological prediction information: meteorological forecast data of the area where the wind turbine is located, including temperature, humidity, wind speed, air pressure, precipitation, cloud thickness, etc. in the future period of time. These data can help predict whether the wind turbine blade is at risk of icing.

[0030] 2) Real-time environmental parameter information: temperature, humidity, wind speed, wind direction, air pressure, etc. monitored by the environmental sensors around the wind turbine in real time, which can reflect the real-time condition of the environment where the wind turbine is located, and provide key basis for judging icing formation.

[0031] 3) Real-time icing thickness information of the wind power blade: using icing sensors, thermal imaging cameras or other detection means to measure the actual ice layer thickness formed on the blade surface to obtain the most direct icing state data.

[0032] The collection of these data can be carried out through meteorological stations, wind turbine sensors and external remote monitoring, to ensure the diversity and real-time nature of the data and improve the comprehensive judgment ability of the wind turbine operating state.

[0033] Further, after obtaining the operating state information of the fan, feature extraction needs to be performed on these data to form a numerical feature vector that can be used by the subsequent deicing prediction model. Due to the problems such as noise, incompleteness, and inconsistent format in the collected original data, data cleaning and preprocessing are needed. Linear interpolation, mean filling, or machine learning prediction methods are used to supplement missing data. Abnormal values are detected and removed using box plot analysis or the 3σ principle to prevent data from misleading decisions. Min-Max normalization or Z-score normalization is used to ensure that different data types have a uniform scale during feature extraction, improving computational stability.

[0034] In order to ensure the effective use of different types of data, the method performs One-Hot encoding on the processed operating state information, converting discrete category variables into calculable numerical vectors to avoid misleading due to the size of category values. One-Hot encoding is a commonly used data preprocessing method, mainly used to convert discrete category variables into numerical vectors to enable computers to more effectively understand and process these data. It is widely used in machine learning, deep learning, and natural language processing. In the original data, category variables are usually represented in the form of words or integers, such as "sunny, cloudy, rainy" or "low, medium, high" classification information. However, directly converting these categories into numerical values (such as 0, 1, 2) may introduce an incorrect order relationship, causing the model to mistakenly believe that there is a size or weight relationship between categories. Therefore, One-Hot encoding provides an unordered and equidistant conversion method, making different categories have the same weight and distance in numerical values. After One-Hot encoding, all data are converted into binary vectors of a uniform format, ensuring that data input into the model will not be affected by numerical size.

[0035] Further, since the fan operating state information comes from various sources, each data source has different degrees of influence on deicing decisions, so AHP is used to calculate the importance weights of different data sources and perform feature fusion to form the final comprehensive feature vector.

[0036] In one possible implementation, through expert experience and historical data analysis, the relative importance of each data source is set, including the weight of meteorological prediction information (α), the weight of real-time environmental parameter information (β), and the weight of blade icing thickness information (γ). So that α + β + γ = 1, the relative importance of each feature is calculated using pairwise comparison matrix, and consistency check (CR<0.1) is performed to ensure the rationality of the weights. The comprehensive feature vector after weight fusion can more accurately reflect the real operating state of the fan and eliminate the false positives that may be caused by a single data source.

[0037] Based on the scheme, combined with weather prediction, real-time environment monitoring and blade icing monitoring, the AHP is used to calculate the importance of features, so that the final feature vector is more reasonable, and the adaptability of the model to complex weather conditions is improved. Through data preprocessing + One-Hot encoding, the data format is unified, the noise influence is reduced, the input stability of the deep learning model is improved, and the numerical size is avoided to mislead the classification decision. The AHP is used to calculate the weighted feature vector, compared with the traditional simple weighting method, different data sources can be more scientifically integrated, the accuracy of deicing prediction is improved, and the start and stop of the deicing device is more intelligent and energy-saving. Because the blade icing condition and external environment change can be accurately calculated, more efficient deicing control can be realized, energy waste caused by excessive heating is avoided, and the risk of blade damage caused by excessive ice thickness is reduced.

[0038] Step S20: calling a pre-trained deicing state prediction model based on the operating state features to determine the deicing prediction level under the current operating state.

[0039] Specifically, the scheme adopts a deep convolutional neural network (ResNet18) as the core algorithm of the deicing state prediction model to realize accurate identification and intelligent decision of the operating state of the wind turbine generator. ResNet18, as a deep residual network, can efficiently extract features of input data and identify complex patterns of wind turbine blade icing through a hierarchical learning structure. In the scheme, the input data is a comprehensive feature vector generated by feature extraction, encoding and feature fusion of the wind turbine's weather prediction information, real-time environmental parameter information and icing thickness information, which can fully reflect the current operating state of the wind turbine.

[0040] Further, the deep convolutional neural network has deep feature extraction capability through multiple convolution layers, pooling layers and residual connections. ResNet18 can effectively alleviate the gradient vanishing and gradient explosion problems in deep networks by introducing a residual structure, so that the model can learn more deep features. In the scheme, the comprehensive feature vector is input into the model, first through multiple convolution operations to extract local features, and then through residual blocks for deep feature fusion to ensure that the complex patterns of wind turbine operating state can be fully captured. Finally, the fully connected layer of the network maps the extracted high-dimensional features to the classification output space, and uses Softmax for multi-classification prediction to determine the current deicing prediction level.

[0041] Preferably, the deicing prediction level includes a prevention level and a deicing level; the deicing level includes multiple levels, and the higher the level, the thicker the corresponding layer to be deiced.

[0042] In the embodiments of the present application, the deicing prediction grading system of the present application includes a prevention grade and a deicing grade, to realize accurate identification and grading control of the icing state of the wind turbine blade. The prevention grade is mainly used to determine whether to take preventive anti-icing operation in low power mode in advance, and the deicing grade is further subdivided into multiple grades, according to the different thickness of the ice layer on the blade surface, to determine the corresponding deicing intensity and control strategy. The higher the grade, the more serious the icing of the blade, the thicker the ice layer to be removed, and the higher power and longer time deicing measures need to be taken.

[0043] In actual application, the prevention grade is mainly based on meteorological prediction information and real-time environmental parameter information to determine whether there is icing risk. For example, when the future temperature is close to the freezing point, the humidity is high, and the wind speed is low, the low-power anti-icing mode is entered in advance, and the intermittent air supply or small-scale heating is used to reduce the possibility of ice layer formation and reduce the energy consumption of subsequent large-scale deicing. At this time, high-power deicing is not directly started, but preventive measures are taken to reduce ice layer accumulation and improve energy utilization efficiency.

[0044] The deicing grade is dynamically adjusted according to real-time ice thickness information to ensure accurate matching of different degrees of icing. Specifically, the deicing grade can be divided into light deicing, medium deicing, heavy deicing and the like. For example, when the ice thickness on the blade surface is thin (such as 0.5mm-1mm), the light deicing mode is entered, and low wind speed and short time heating are used for deicing to avoid energy waste; when the ice thickness reaches 2mm or more, the medium deicing mode is entered, the heating intensity is increased, and the fan angle is adjusted to optimize the heat distribution; if the ice thickness further increases (such as 3mm or more), the heavy deicing mode is entered, and full power is run to ensure that the ice layer is completely melted by high-temperature air blowing and long-time heating to prevent ice blocks from falling and causing damage to the blade or unit.

[0045] Based on the present application, compared with the traditional simple start-stop control mode, more accurate energy consumption management can be realized, unnecessary high-power deicing operation can be avoided, and the influence of icing on the running efficiency of the fan can be reduced. In addition, due to the introduction of the prevention grade, corresponding measures can be taken when the ice layer has not yet accumulated seriously, the demand for subsequent high-power deicing is reduced, and the running safety of the fan and the overall energy efficiency of deicing are further improved. In summary, the deicing prediction grading system realizes intelligent grading control, making the deicing strategy more flexible and efficient, which helps to improve the long-term stability of the wind turbine unit in complex environmental conditions.

[0046] Step S30: determining a control scheme of the air-thermal deicing device based on the deicing prediction grade.

[0047] Specifically, the air heating deicing device comprises a heating module, a blower and a blade angle adjusting device; the control scheme of the air heating deicing device is determined based on the deicing prediction level, comprising: determining the target operating state of each air heating deicing device based on the deicing prediction level; generating a corresponding control scheme based on the current operating state and the corresponding target operating state of each air heating deicing device.

[0048] In the embodiments of the present application, the heating module, the blower and the blade angle adjusting device are precisely controlled according to different deicing prediction levels to adapt to different icing conditions, thereby improving deicing efficiency, reducing energy consumption and ensuring the safe operation of the fan blades. Based on the difference between the current operating state and the corresponding target operating state of each air heating deicing device, a specific control scheme is generated. During the control process, the temperature of the fan blades, the deicing progress and the changes in environmental conditions are detected in real time, and if the actual deicing progress does not achieve the expected effect, the heating temperature, the wind speed or the blade angle will be automatically adjusted to ensure that the deicing process is carried out efficiently. For example, if it is detected that the local ice layer has not completely melted, the control can adjust the blade direction to expose the area to a higher temperature airflow, or appropriately extend the deicing time to ensure that the ice layer is completely removed.

[0049] Further, if the current deicing prediction level is the prevention level, the control scheme of the air heating deicing device based on the deicing prediction level comprises: predicting the icing time based on the operating state information of the wind turbine generator, obtaining the predicted icing time; determining the intervention time of each air heating deicing device based on the predicted icing time; generating an intermittent air supply operation scheme of each air heating deicing device based on the intervention time and a preset intermittent air supply timing table as the control scheme of the air heating deicing device.

[0050] In the embodiments of the present application, when the current deicing prediction level is in the prevention level, it is necessary to first predict the icing time based on the operating state information of the wind turbine generator (including weather prediction, real-time environmental parameters, blade temperature and humidity, wind speed, etc.). Specifically, a method based on physical modeling is used to predict the time point when icing may occur in the future by combining historical data and real-time weather data. The icing time prediction mainly depends on multiple variables such as environmental temperature, air humidity, wind speed, air pressure, precipitation, blade surface temperature, etc. Through multivariate regression analysis or deep learning algorithm (such as LSTM), the formation time of ice layer is predicted. For example, when the weather data shows that the temperature will drop below the freezing point in the next 30 minutes, and the air humidity is higher than 80%, the wind speed is low (lower than 3m / s, reducing the self-cleaning ability of the blade), it can be predicted that the blade will start icing after 25-40 minutes, and the anti-icing measures will be deployed in advance according to this prediction result.

[0051] Furthermore, after obtaining the predicted icing time, the gas-thermal de-icing device needs to be activated in advance to reduce the probability of icing formation and decrease the energy consumption of subsequent de-icing. For example, if the predicted icing time is 30 minutes, the device may enter a low-power prevention mode 10-1 minutes earlier. The temperature difference between the current blade temperature and the predicted icing threshold also plays a role; if the blade surface temperature is low, the anti-icing mode needs to be activated earlier. Combining wind speed and ambient temperature, if the wind speed is low and heat remains on the blade surface for a longer time, the activation time should be appropriately delayed; if the wind speed is high and heat is easily lost, activation should be earlier. The goal of optimizing the activation time is to minimize energy consumption while ensuring anti-icing effectiveness, using intelligent algorithms to dynamically calculate the optimal activation point.

[0052] Furthermore, after determining the intervention time, a specific intermittent air supply operation plan needs to be formulated to prevent ice formation under low energy consumption. A series of optional intermittent air supply plans are stored, and the optimal plan is selected based on different meteorological conditions and blade status. Low-power air supply (e.g., air supply for 2 minutes every 10 minutes) can avoid rapid heat loss and improve thermal efficiency. Zoned air supply, that is, prioritizing heating the blade roots and then extending to the middle and tip of the blades, improves the overall anti-icing effect. Adjusting the wind speed: if the ambient wind speed is high, a high-temperature, low-wind-speed mode is used to prevent excessive heat loss; if the wind speed is low, a low-temperature, high-wind-speed mode is used to ensure that hot air evenly covers the blade surface.

[0053] Preferably, if the current de-icing prediction level is the de-icing level, the step of determining the control scheme of the gas-thermal de-icing device based on the de-icing prediction level includes: determining the real-time icing thickness information of the wind turbine blades based on the operating status information of the wind turbine generator set, and predicting the development of icing thickness; determining the thickness of the layer to be de-iced based on the real-time icing thickness information and the prediction result of icing thickness development; simulating the reverse development information of icing thickness towards the preset safe icing thickness based on the thickness of the layer to be de-iced, generating the target operating state of each gas-thermal de-icing device, and generating a corresponding control scheme based on the current operating state of each gas-thermal de-icing device and the corresponding target operating state.

[0054] In this embodiment of the invention, under the de-icing level, it is first necessary to accurately measure the current ice thickness on the blades and predict the future development trend of the ice layer. This process includes the following technical steps:

[0055] 1) Real-time data acquisition: Using icing sensors, thermal imaging cameras or ultrasonic thickness gauges, the icing thickness data on the blade surface is acquired in real time to ensure measurement accuracy.

[0056] 2) Environmental Factor Analysis: Simultaneously monitor meteorological conditions (temperature, humidity, wind speed, precipitation), and combine them with the operating parameters of the wind turbine (speed, blade angle, power output, etc.) to construct a comprehensive environmental impact model in order to accurately predict the formation rate of icing.

[0057] 3) Icing Thickness Trend Prediction: Based on historical data and machine learning models, the potential growth rate of ice under current meteorological conditions is calculated, and the future icing thickness of leaves is predicted without external intervention. For example, if the current icing thickness on the leaves is 2.5 mm, and it is predicted that it may increase to 3.5 mm in the next 30 minutes without intervention, the required ice thickness is calculated, and the de-icing strategy is adjusted accordingly.

[0058] Furthermore, after obtaining real-time icing thickness data and predicted icing trends, it is necessary to calculate the thickness of the layer to be de-iced, i.e., how much ice needs to be removed to restore the blades to a safe operating state. A preset safe icing thickness (e.g., 0.5mm or no icing) is set, and the difference between the current icing thickness and the safe thickness is calculated. If the current ice thickness is significantly higher than the safe value (e.g., above 3.5mm), a full-power de-icing strategy must be adopted; if the thickness slightly exceeds the safe threshold (e.g., 1.5mm), low-power or intermittent de-icing should be used to reduce energy consumption. Based on the ice distribution in different areas of the blades, the de-icing process is regionally optimized, i.e., heavily iced areas are heated, while lighter iced areas are heated with lower power to reduce unnecessary energy consumption.

[0059] Furthermore, after determining the thickness of the ice layer to be de-iced, it is necessary to calculate the target operating state of the pneumatic de-icing device to ensure that the ice thickness can be effectively reduced to a safe range. This process includes:

[0060] 1) Simulate the reverse development process of ice thickness: Based on the thermodynamic simulation model, calculate the melting rate of ice layer under different heating power and wind speed conditions, and thus obtain the optimal heating strategy.

[0061] 2) Adjust the heating module: If the ice layer to be de-iced is thick (e.g., >3mm), the heating module will be set to high temperature mode to provide stronger heat input; if the ice layer is thin (e.g., 1~2mm), a lower temperature will be used for melting to reduce unnecessary energy consumption.

[0062] 3) Adjust the blower speed: Adjust the blower speed according to the power setting of the heating module to optimize the circulation path of hot air and improve heating efficiency. For example, the high-speed mode can quickly diffuse hot air to the entire surface of the blades, while the low-speed mode is suitable for more precise local de-icing.

[0063] 4) Adjusting the blade angle: If the wind turbine is in a low wind speed environment, fine-tune the pitch angle of the blades to ensure that the blades receive hot air at the optimal angle, ensuring uniformity of deicing and improving heat transfer efficiency.

[0064] Further, based on the current operating state of each air-thermal deicing device and the calculated target operating state, a specific deicing control scheme is finally generated. When the ice thickness is far above the safety threshold (e.g., above 3.5mm), the highest power deicing mode is entered, all heating modules operate at full power, the blower maintains the highest wind speed, and the blade angle is adjusted to the optimal position to ensure rapid ice melting. If the ice thickness is relatively high but not at the limit (e.g., 2mm~3mm), a gradual deicing strategy is adopted, first preheating at medium power, then gradually increasing the power to prevent overheating and waste of heat in a short period of time. If the ice thickness varies greatly in different blade regions, a zoned heating strategy is adopted, i.e., focusing on heating areas with thicker ice, while areas with thinner ice are in low power mode to optimize energy consumption. When the ice thickness is thin (e.g., 1mm~2mm), an intermittent air supply and short-time heating strategy is adopted to complete the deicing process with minimal power consumption and prevent blade overheating.

[0065] Step S40: Execute the control scheme and update the deicing prediction level of the wind turbine generator set in real time during execution to perform control scheme correction of the corresponding thermal deicing device when the deicing prediction level changes.

[0066] Specifically, during execution, real-time collection of operating state information during execution of the wind turbine generator set is performed; feature extraction is performed on the operating state information during execution to obtain operating state features during execution; a pre-trained deicing state prediction model is called based on the operating state features during execution to determine the deicing prediction level under the current operating state; if the deicing prediction level under the current operating state is different from the deicing prediction level corresponding to the control scheme being executed, the control scheme of the air-thermal deicing device is determined based on the deicing prediction level under the current operating state as a new control scheme, and the new control scheme is executed.

[0067] In the embodiments of the present application, during the deicing process of the wind turbine generator set, the external environmental conditions and the ice thickness of the blades will change constantly, so the deicing control scheme needs to have the ability to adjust in real time. During the execution of the control scheme, the operating state information of the wind turbine is continuously collected, including environmental temperature, humidity, wind speed, blade surface temperature, ice thickness, and operating state of the deicing device. In order to ensure the effectiveness of the data, feature extraction is performed on these information to remove abnormal data, and a deep learning model is used to analyze the operating state features to evaluate the current deicing prediction level.

[0068] Further, during the de-icing process, the pre-trained de-icing state prediction model is called continuously to update the de-icing prediction level according to the real-time monitoring data. If the prediction level changes, for example, the de-icing effect is not as expected, or the environmental conditions change suddenly to cause the ice layer to grow faster, the optimal de-icing control strategy is recalculated immediately. For example, when the ice layer has not reached a safe thickness, the heating power is increased, the wind speed is increased, or the blade angle is adjusted to optimize the heat flow distribution; on the contrary, if it is detected that the ice layer has been basically melted, the power is reduced, and the operation time of the air blower is reduced to save energy.

[0069] Figure 2 is a system structure diagram of a de-icing control system of a wind power blade provided by an embodiment of the present application. As shown in Figure 2 the embodiment of the present application provides a de-icing control system of a wind power blade, the system comprising: a collection unit for collecting the operating state information of a wind turbine generator set and performing feature extraction on the operating state information to obtain operating state features; a prediction unit for calling a pre-trained de-icing state prediction model based on the operating state features to determine the de-icing prediction level under the current operating state; a scheme generation unit for determining the control scheme of the air-thermal de-icing device based on the de-icing prediction level; and an execution unit for executing the control scheme and updating the de-icing prediction level of the wind turbine generator set in real time during the execution to execute the control scheme correction of the corresponding thermal de-icing device when the de-icing prediction level changes.

[0070] The embodiment of the present application also provides a computer readable storage medium, which stores instructions, and the instructions make the computer execute the de-icing control method of the wind power blade when the computer runs.

[0071] Those skilled in the art can understand that all or part of the steps in the method for implementing the above-mentioned embodiments can be completed by programs instructing related hardware, the programs are stored in a storage medium, and the programs include a plurality of instructions for making a single-chip microcomputer, a chip or a processor execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk and various storage medium capable of storing program codes.

[0072] The optional embodiments of the present application are described in detail above with reference to the drawings, but the embodiments of the present application are not limited to the specific details in the above-described embodiments. Within the technical concept of the embodiments of the present application, various simple modifications can be made to the technical solutions of the embodiments of the present application, and these simple modifications all belong to the protection scope of the embodiments of the present application. In addition, it should be noted that, in the above-described specific embodiments, various specific technical features can be combined in any appropriate manner without contradiction. In order to avoid unnecessary repetition, various possible combinations are not described again in the embodiments of the present application.

[0073] In addition, various different embodiments of the present application can also be combined in any appropriate manner, as long as it does not deviate from the idea of the embodiments of the present application, and it should also be considered as disclosed by the embodiments of the present application.

Claims

1. A method for controlling the de-icing of wind turbine blades, characterized in that, The method includes: Collect the operating status information of the wind turbine generator set, and perform feature extraction on the operating status information to obtain operating status features; Based on the aforementioned operational status characteristics, a pre-trained de-icing status prediction model is invoked to determine the de-icing prediction level for the current operational status; wherein, The de-icing prediction level includes a prevention level and a de-icing level; the de-icing level includes multiple levels, and the higher the level, the thicker the corresponding de-icing layer. The control scheme for the gas-thermal de-icing device is determined based on the aforementioned de-icing prediction level; wherein... The gas-thermal de-icing device includes: a heating module, a blower, and a blade angle adjustment device; the control scheme for determining the gas-thermal de-icing device based on the de-icing prediction level includes: determining the target operating state of each gas-thermal de-icing device based on the de-icing prediction level; and generating a corresponding control scheme based on the current operating state of each gas-thermal de-icing device and the corresponding target operating state. If the current de-icing prediction level is the prevention level, the step of determining the control scheme of the gas-thermal de-icing device based on the de-icing prediction level includes: predicting the icing time based on the operating status information of the wind turbine generator set to obtain the predicted icing time; determining the intervention time of each gas-thermal de-icing device based on the predicted icing time; and generating an intermittent air supply operation scheme for each gas-thermal de-icing device based on the intervention time and a preset intermittent air supply sequence table, which serves as the control scheme for the gas-thermal de-icing device. If the current de-icing prediction level is de-icing level, the step of determining the control scheme of the gas-thermal de-icing device based on the de-icing prediction level includes: determining the real-time icing thickness information of the wind turbine blades based on the operating status information of the wind turbine generator set, and predicting the development of icing thickness; determining the thickness of the layer to be de-iced based on the real-time icing thickness information and the prediction results of icing thickness development; simulating the reverse development information of icing thickness towards the preset safe icing thickness based on the thickness of the layer to be de-iced, generating the target operating state of each gas-thermal de-icing device, and generating a corresponding control scheme based on the current operating state of each gas-thermal de-icing device and the corresponding target operating state. The control scheme is executed, and the de-icing prediction level of the wind turbine generator is updated in real time during the execution process, so that the control scheme of the corresponding gas thermal de-icing device is corrected when the de-icing prediction level changes.

2. The method according to claim 1, characterized in that, The operating status information of the wind turbine generator set includes: Any one or more of the following: meteorological forecast information, real-time environmental parameter information, and real-time icing thickness information of wind turbine blades.

3. The method according to claim 1, characterized in that, The step of performing feature extraction on the running status information to obtain running status features includes: Preprocessing is performed on the aforementioned operating status information; One-hot encoding is performed on the preprocessed running status information to obtain the numerical vector of each preprocessed running status information. Based on the Analytic Hierarchy Process (AHP), feature fusion is performed on the numerical vectors of each preprocessed running status information to obtain a comprehensive feature vector, which serves as the running status feature.

4. The method according to claim 1, characterized in that, During execution, the de-icing prediction level of the wind turbine generator is updated in real time, so that the control scheme of the corresponding gas-thermal de-icing device is corrected when the de-icing prediction level changes, including: During the execution process, the operating status information of the wind turbine generator is collected in real time; The execution status information during the execution process is extracted to obtain the execution status features; Based on the running state characteristics during the execution process, a pre-trained de-icing state prediction model is invoked to determine the de-icing prediction level under the current running state. If the de-icing prediction level under the current operating state is different from the de-icing prediction level corresponding to the control scheme being executed, then the control scheme of the gas-thermal de-icing device is determined based on the de-icing prediction level under the current operating state, and the new control scheme is executed.

5. A de-icing control system for wind turbine blades, characterized in that, The system is applied to the de-icing control method for wind turbine blades according to any one of claims 1-4, and the system comprises: The acquisition unit is used to acquire the operating status information of the wind turbine generator set and perform feature extraction on the operating status information to obtain operating status features; The prediction unit is used to call a pre-trained de-icing state prediction model based on the operating state characteristics to determine the de-icing prediction level under the current operating state. The scheme generation unit is used to determine the control scheme of the gas-thermal de-icing device based on the de-icing prediction level. An execution unit is used to execute the control scheme and update the de-icing prediction level of the wind turbine generator in real time during the execution process, so as to correct the control scheme of the corresponding gas thermal de-icing device when the de-icing prediction level changes.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the de-icing control method for wind turbine blades as described in any one of claims 1-4.

Images