Fan blade icing monitoring method, system, computer device, medium and product

Abstract

The application relates to the technical field of fan monitoring, and discloses a fan blade icing monitoring method, a system, computer equipment, a medium and a product, the method comprising the following steps: collecting multi-source data, the multi-source data comprising data of multiple sensors arranged at preset positions of fan blades, environmental meteorological data and fan operation data; fusing the multi-source data, and performing feature extraction on the fused data to obtain a comprehensive feature vector; monitoring the current fan blade icing condition based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information; and warning the fan blade icing state based on the size relationship between the icing condition information and a preset threshold. The application effectively solves the problem of low accuracy of fan blade icing monitoring in complex environments in the prior art by organically combining various advanced technical means such as multi-source data fusion, deep learning, icing condition information prediction and warning of the fan blade icing state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fan monitoring, in particular to a fan blade icing monitoring method, system, computer device, medium and product. BACKGROUND

[0002] In the field of fan blade icing monitoring, there are currently various monitoring methods and technical solutions. Common icing monitoring methods include visual-based monitoring methods, such as using a camera to take real-time photos of the blade and using image processing technology to determine the icing condition; and electrical parameter change-based monitoring methods, such as installing capacitors, resistors, and other sensors on the blade, which will change their electrical characteristics when icing occurs, thereby detecting icing.

[0003] In the field of fan blade icing monitoring, although there are various monitoring methods and technical solutions, the existing technology still has many significant shortcomings:

[0004] Visual-based monitoring methods: The method of using a camera to take real-time photos of the blade and determining the icing condition through image processing technology has extremely harsh requirements for environmental conditions. In rainy, foggy, night, or low-light conditions, the camera imaging quality will decrease significantly, the image will be blurred, and the contrast will be low, making it difficult for the image processing algorithm to accurately identify the icing area, and the misjudgment rate and omission rate will significantly increase. At the same time, complex weather conditions can also interfere with wireless signal transmission, causing monitoring data to be lost or delayed, seriously affecting the continuity and reliability of monitoring. In addition, as the installation height of the fan increases and the length of the blade gradually increases, the installation position and angle of the camera are limited, making it difficult to monitor the entire blade surface in all directions without dead angles, and local icing phenomena may not be detected in time.

[0005] Electrical parameter change-based monitoring methods: The method of installing capacitors, resistors, and other sensors on the blade to detect icing is susceptible to interference from various factors. In the early stages of icing, the ice layer is thin, and the influence on the electrical characteristics of capacitors, resistors, and other sensors is weak, and the sensor output signal changes slightly, making it difficult to accurately extract icing feature information from background noise, resulting in this method being insensitive to early icing and failing to achieve early warning of icing. Moreover, factors such as dirt, dust on the blade surface, and changes in environmental humidity can also cause changes in the capacitance and resistance values, which are coupled with the changes in electrical characteristics caused by icing, making it difficult to accurately distinguish and generating a large number of false positive signals. In addition, such sensors usually require complex wiring and power supply systems, which not only increase installation costs and maintenance workload, but also make the wiring vulnerable to damage in harsh natural environments, further reducing the stability and reliability of the system.

[0006] In summary, the existing fan blade icing monitoring method has deficiencies in accuracy, reliability, intelligence level, and adaptability to complex environments, and an advanced and efficient icing monitoring method is urgently needed to overcome these shortcomings. SUMMARY

[0007] Therefore, the present application provides a fan blade icing monitoring method, system, computer device, medium and product to solve the problem of low accuracy of fan blade icing monitoring in complex environments in the prior art.

[0008] In a first aspect, the present application provides a fan blade icing monitoring method, which comprises:

[0009] Collecting multi-source data, including data from multiple sensors arranged at predetermined positions on the fan blade, environmental meteorological data, and fan operation data;

[0010] Fusing the multi-source data and extracting features from the fused data to obtain a comprehensive feature vector;

[0011] Monitoring the current fan blade icing condition based on the comprehensive feature vector and a pre-set deep learning model to obtain icing condition information;

[0012] Based on the size relationship between the icing condition information and the pre-set threshold, the fan blade icing state is warned.

[0013] The fan blade icing monitoring method provided by the present application combines multiple advanced technical means such as multi-source data fusion, deep learning, icing condition information prediction, and warning of the fan blade icing state, and comprehensively improves the effect of the fan blade icing monitoring method, making it more accurate, intelligent, adaptable, and reliable in terms of fan blade icing monitoring and prevention and control, effectively solving the problem of low accuracy of fan blade icing monitoring in complex environments in the prior art.

[0014] In an optional embodiment, the data from the multiple sensors includes an output characteristic curve; before fusing the multi-source data, the fan blade icing monitoring method further comprises:

[0015] Real-time acquisition of the output characteristic curve of the multiple sensors, and comparison of the output characteristic curve with a standard characteristic curve to obtain a comparison error;

[0016] Based on the comparison error, the genetic algorithm is used to automatically find the optimal calibration parameter, and based on the optimal calibration parameter, the output characteristic curve of the multiple sensors is dynamically corrected to adaptively calibrate and compensate the data from the multiple sensors.

[0017] The application provides a fan blade icing monitoring method, which adopts an adaptive calibration strategy based on a genetic algorithm and an environmental compensation algorithm based on a physical model. By monitoring the output characteristics of the sensor in real time and comparing them with the pre-stored standard characteristic curve, the calibration parameters are dynamically optimized by using the genetic algorithm, and the measurement error of the sensor is compensated in real time. Meanwhile, the non-icing related strain and spectral change caused by environmental factors such as temperature and humidity can also be accurately compensated. This innovation ensures the measurement accuracy and stability of the fiber grating sensor in different operating environments and during long-term use, reduces the monitoring error caused by the sensor itself, improves the reliability of the entire monitoring system, and effectively solves the problems of insufficient monitoring performance adaptability and stability in the prior art.

[0018] In an optional embodiment, the multi-source data is fused, and the fused data is feature extracted to obtain a comprehensive feature vector, including:

[0019] The environmental meteorological data and the fan operation data are subjected to data cleaning and data standardization data preprocessing;

[0020] The multi-source data is fused, and the fused data is feature extracted to obtain a comprehensive feature vector, including:

[0021] The fused data is feature extracted to obtain a comprehensive feature vector.

[0022] The application provides a fan blade icing monitoring method, and the multi-source data includes data of multiple sensors arranged at preset positions of the fan blade, environmental meteorological data and fan operation data. A preset data fusion algorithm is used to form a more comprehensive and rich feature vector as the input of a deep learning model. This multi-source data fusion method enables the model to comprehensively evaluate the icing risk from multiple dimensions, overcoming the limitation of traditional methods that rely only on single strain and temperature information to represent the icing condition.

[0023] In an optional embodiment, the current fan blade icing condition is monitored based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information, including:

[0024] The historical icing data corresponding to the multi-source data is obtained, and the historical icing data is input into the preset deep learning model for model training to obtain an icing trend prediction model;

[0025] The comprehensive feature vector is input into the icing trend prediction model to obtain the icing condition information; the icing condition information includes an icing position, an icing thickness, an icing range and an icing type.

[0026] In an optional embodiment, the icing trend prediction model comprises a feature change pattern learned when icing occurs on the fan blade and a correlation feature between the multi-source data; the comprehensive feature vector is input into the icing trend prediction model to obtain icing condition information, including:

[0027] When the comprehensive feature vector is input into the icing trend prediction model, the icing trend prediction model automatically judges the current icing condition of the fan blade based on the feature change pattern and the correlation feature between the multi-source data to obtain the icing condition information.

[0028] The fan blade icing monitoring method provided by the application can accurately predict the icing trend in the future period of time through the icing trend prediction module, can issue icing warning information in advance for hours or even days, provides a valuable time window for the anti-icing and de-icing decision of the fan, realizes the change from passive monitoring to active prevention, and lays a foundation for effectively solving the problem that the prior art lacks intelligent warning capability.

[0029] In an optional embodiment, the icing condition of the fan blade is warned based on the size relationship between the icing condition information and the preset threshold, including:

[0030] When the icing condition information reaches the preset threshold, a multi-channel warning signal is triggered to obtain warning information, and an anti-icing and / or de-icing decision suggestion is output based on the warning information and the preset strategy;

[0031] A response measure is made based on the anti-icing and / or de-icing decision suggestion.

[0032] The fan blade icing monitoring method provided by the application can quickly judge whether the icing reaches a dangerous degree by directly comparing the real-time monitored icing condition information with the preset threshold, avoid the warning lag caused by manual judgment delay or experience error, trigger the warning immediately when the icing condition reaches the threshold, and avoid the safety risk caused by icing in advance. The multi-channel warning signal means that the warning information can be delivered to relevant personnel or systems in multiple ways to ensure that the warning information can be received in time. The anti-icing and / or de-icing decision suggestion output based on the icing condition information and the preset strategy can provide individualized solutions in combination with specific icing conditions, realize the early prediction and intelligent warning of the icing trend, and provide strong support for the anti-icing and de-icing decision of the fan.

[0033] In a second aspect, the application provides a fan blade icing monitoring system, which comprises:

[0034] A sensor network is used to be fixedly installed at a preset position of the fan blade.

[0035] A multi-source data acquisition module is configured to acquire multi-source data, which includes data of multiple sensors arranged at preset positions of the fan blade, environmental meteorological data, and fan operation data;

[0036] An intelligent monitoring and analysis platform is configured to fuse the multi-source data, extract features from the fused data, and obtain a comprehensive feature vector; monitor the icing condition of the fan blade based on the comprehensive feature vector and a preset deep learning model, and obtain icing condition information;

[0037] A warning and decision support module is configured to warn the icing condition of the fan blade based on a size relationship between the icing condition information and a preset threshold.

[0038] In a third aspect, the present application provides a computer device, comprising a memory and a processor, the memory and the processor are communicatively connected with each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the fan blade icing monitoring method of the first aspect or any of the corresponding embodiments thereof.

[0039] In a fourth aspect, the present application provides a computer readable storage medium, which stores computer instructions, and the computer instructions are used to make a computer execute the fan blade icing monitoring method of the first aspect or any of the corresponding embodiments thereof.

[0040] In a fifth aspect, the present application provides a computer program product, which comprises computer instructions, and the computer instructions are used to make a computer execute the fan blade icing monitoring method of the first aspect or any of the corresponding embodiments thereof. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed in the specific embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0042] Figure 1 is a flowchart of a fan blade icing monitoring method according to an embodiment of the present application;

[0043] Figure 2 is a flowchart of another fan blade icing monitoring method according to an embodiment of the present application;

[0044] Figure 3 is a flowchart of still another fan blade icing monitoring method according to an embodiment of the present application;

[0045] Figure 4is a whole architecture diagram of a fan blade icing monitoring system according to an embodiment of the present application;

[0046] Figure 5 is an architecture diagram of a fiber grating sensor network according to an embodiment of the present application;

[0047] Figure 6 is an architecture diagram of a data acquisition and transmission module according to an embodiment of the present application;

[0048] Figure 7 is an architecture diagram of an intelligent monitoring and analysis platform according to an embodiment of the present application;

[0049] Figure 8 is an architecture diagram of a warning and decision support module according to an embodiment of the present application;

[0050] Figure 9 is a structural block diagram of a fan blade icing monitoring system according to an embodiment of the present application;

[0051] Figure 10 is a hardware structure schematic diagram of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0053] In the field of fan blade icing monitoring, the monitoring method based on fiber grating sensors is involved. Although the fiber grating sensor has many advantages, it still has obvious limitations in practical application. First, the fiber grating sensor mainly reflects the icing condition indirectly based on strain and temperature changes. However, the icing process is a complex multi-physical field coupling process. It is difficult to fully and accurately characterize the key parameters such as icing thickness, range, growth rate and ice layer type by relying only on strain and temperature information, resulting in inaccurate monitoring results and inability to provide sufficient decision basis for precise anti-icing and de-icing control of the fan. Second, this method lacks intelligent analysis capability for monitoring data, cannot deeply mine historical data to extract icing rules and trend information, and is difficult to realize early prediction and intelligent warning of icing. It can only conduct passive monitoring after icing occurs, and cannot meet the intelligent demand of fan blade icing monitoring.

[0054] In summary, the existing fan blade icing monitoring methods have deficiencies in accuracy, reliability, intelligence level, and adaptability to complex environments, and an advanced and efficient monitoring method is urgently needed to overcome these shortcomings and solve the following technical problems:

[0055] First, how to improve the accuracy of fan blade icing monitoring in complex environments, avoid misjudgment and missed judgment caused by environmental factors, and achieve accurate identification of icing conditions.

[0056] Second, how to enhance the intelligence level of the monitoring method to realize early prediction and intelligent warning of icing trends and overcome the limitations of traditional methods that can only conduct passive monitoring.

[0057] Third, how to improve the adaptability and stability of the monitoring method so that it can better adapt to different regions, different types of fan blades, and complex and variable environmental conditions, and reduce monitoring errors and system failure risks caused by sensor layout, environmental interference, and other factors.

[0058] Fourth, how to reduce monitoring costs and operational complexity to facilitate large-scale wind farm applications while ensuring that monitoring results can fully reflect key icing parameters and provide effective decision-making basis for fan ice prevention and removal control.

[0059] The embodiment of the present application provides a fan blade icing monitoring method, which combines multiple advanced technical means such as multi-source data fusion, deep learning, icing condition information prediction, and early warning of fan blade icing state, and comprehensively improves the effect of the fan blade icing monitoring method, realizes high-precision, intelligent monitoring, early warning, and decision support of the fan blade icing condition, effectively solves the problems of low accuracy, insufficient intelligence, poor adaptability, and other problems of existing monitoring methods, and improves the operation safety and reliability of wind turbine generators in icing environments.

[0060] According to the embodiment of the present application, a fan blade icing monitoring method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0061] In this embodiment, a fan blade icing monitoring method is provided, which can be used in a fan blade icing monitoring system, as shown in Figure 4 The system includes a sensor network (such as a fiber Bragg grating sensor network), a data acquisition and transmission module, an intelligent monitoring and analysis platform, and an early warning and decision support module, Figure 1 is a flowchart of the fan blade icing monitoring method according to the embodiment of the present application, as shown in Figure 1As shown, the flow includes the following steps:

[0062] In step S101, multi-source data is collected, including data from multiple sensors arranged at predetermined positions on the fan blade, environmental meteorological data, and fan operation data.

[0063] Specifically, the multiple sensors of the embodiment are multiple fiber grating sensors, which form a fiber grating sensor network, the architecture diagram of which is as shown in Figure 5 As shown, multiple fiber grating sensors (FBG) are pasted or embedded inside the fan blade along the length direction and different cross-sectional positions of the blade. These sensors are uniformly distributed on the surface and key internal structure of the fan blade, such as the predetermined positions including the leading edge, trailing edge, pressure surface, and suction surface of the blade, which are prone to icing. Each fiber grating sensor has a unique identifier for distinguishing the measurement data at different positions, ensuring accurate correspondence between the data and the physical position.

[0064] The function and principle of the fiber grating sensor are as follows: the fiber grating sensor utilizes its sensitivity to changes in the surrounding physical quantities. When the fan blade is iced, the additional load caused by the ice layer causes the deformation of the fan blade, which in turn changes the grating period and effective refractive index of the fiber grating sensor, causing the wavelength of the reflected or transmitted light to shift. By measuring the wavelength shift, the strain data and temperature data of the fan blade can be obtained (the refractive index of the fiber grating changes with temperature, and for every 1℃ change in temperature, the wavelength offset is about 10pm, and the temperature can be calculated from the wavelength offset (which needs to be distinguished from the wavelength offset caused by strain, which will be compensated for later by an algorithm)). Compared with the traditional single arrangement method, the distributed arrangement of fiber grating sensors in this embodiment can more comprehensively capture the multi-point and multi-direction deformation characteristics caused by icing, providing a rich data source for subsequent accurate analysis of icing conditions. Among them, the strain data is the tensile / compressive strain (unit: με, microstrain) of the fan blade caused by the additional load of icing. The temperature data is the temperature (unit: ℃) of the surface and interior of the fan blade.

[0065] The multi-source data is collected by a data acquisition and transmission module, and the architecture diagram of the data acquisition and transmission module is as shown in Figure 6 As shown, the data acquisition and transmission module includes a data acquisition unit (such as a high-precision grating demodulator) and a data transmission link. The data acquisition unit is connected with the fiber grating sensors. When the data acquisition unit uses a grating demodulator, the grating demodulator collects the reflected spectrum signals of each fiber grating sensor in real time at a set sampling frequency (such as 1Hz-10000Hz adjustable), and converts the spectrum signals into digital signals. These digital signals contain accurate wavelength shift data, reflecting the strain, vibration, and temperature changes of the blade.

[0066] When the data collection of multiple fiber grating sensors is completed, the multiple fiber grating sensor data collected through the data transmission link is transmitted to the intelligent monitoring and analysis platform through a wired communication (such as optical fiber communication) or wireless communication (such as 4G / 5G, Wi-Fi, etc.) communication link. The system uses standard communication protocols such as TCP / IP to ensure the stability and reliability of data transmission. At the same time, encryption algorithms are used during data transmission to prevent data tampering or theft and ensure data security.

[0067] The data acquisition and transmission module realizes efficient acquisition and reliable transmission of multiple fiber grating sensor data, providing continuous, complete and accurate data basis for subsequent data analysis.

[0068] Collection of environmental meteorological data: The top of the fan tower drum (closest to the blade, with strong representative environmental parameters) or the wind farm meteorological station (if a single fan has no independent equipment, the data of the meteorological station within a 3km range can be called) can be used as the collection point, and integrated meteorological sensors (including temperature sensors, humidity sensors, anemometers, wind direction indicators, precipitation type detectors, etc.) are used for data collection. The data is connected to the fan monitoring system through wired mode. The collected environmental meteorological data includes environmental air temperature, relative humidity, wind speed and direction, and precipitation type (distinguishing between rain, snow, freezing rain, and rime (identified by infrared or microwave sensors)) and other data. The fan monitoring system transmits the environmental meteorological data to the intelligent monitoring and analysis platform.

[0069] Collection of fan operation data: Through the fan monitoring system (SCADA system, Supervisory Control And Data Acquisition, which is the abbreviation of Data Acquisition and Monitoring Control System), the system records the unit operation parameters in real time, and communicates with the monitoring system through internal bus (such as Modbus, Profibus). The collected fan operation data includes fan blade speed, fan power generation, pitch angle and nacelle vibration data. The fan monitoring system transmits the fan operation data to the intelligent monitoring and analysis platform.

[0070] Step S102, fuse multiple source data, and extract features from the fused data to obtain a comprehensive feature vector.

[0071] Specifically, the intelligent monitoring and analysis platform includes a multi-source data fusion module.

[0072] The multi-source data fusion module is used to fuse the strain data and temperature data of the fiber grating sensor and environmental data (temperature, humidity, wind speed, precipitation type, etc.) from a weather station and operation state data (rotation speed, power, etc.) provided by a fan monitoring system, and a preset data fusion algorithm such as Kalman filtering, principal component analysis (PCA), etc. is used to perform feature extraction, dimension reduction and fusion on the multi-source data to generate a comprehensive feature vector, which is used as an input of a deep learning model to improve the perception ability of the deep learning model to the icing condition.

[0073] In step S103, the current fan blade icing condition is monitored based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information.

[0074] Specifically, the intelligent monitoring and analysis platform further includes a deep learning model. The comprehensive feature vector is input into the trained preset deep learning model to predict the fan icing condition, and the icing condition information is obtained.

[0075] In step S104, the fan blade icing state is warned based on the size relationship between the icing condition information and a preset threshold.

[0076] Specifically, as shown in Figure 8 The warning and decision support module is used to warn the fan blade icing state based on the size relationship between the icing condition information and a preset threshold. When the predicted icing thickness or range is about to reach the set threshold (determined according to the fan blade design and operation safety standard), the system automatically triggers a warning signal.

[0077] The fan blade icing monitoring method provided in the embodiment combines multiple advanced technical means such as multi-source data fusion, deep learning, icing condition information prediction, and warning of the fan blade icing state, and comprehensively improves the effect of the fan blade icing monitoring method, so that more accurate, reliable and timely fan blade icing monitoring and prevention are realized in terms of accuracy, intelligence, adaptability and reliability, and the problem of low accuracy of the fan blade icing monitoring in the prior art in a complex environment is effectively solved.

[0078] In the embodiment, a fan blade icing monitoring method is provided, which can be used in a mobile terminal such as a mobile phone, a tablet computer, etc. (the execution subject is described in combination with the actual situation) of the above-mentioned fan blade icing monitoring system, as shown in Figure 4 The system includes a sensor network (for example, a fiber grating sensor network), a data acquisition and transmission module, an intelligent monitoring and analysis platform, and a warning and decision support module, Figure 2 is a flowchart of the fan blade icing monitoring method according to the embodiment of the application, as shown in Figure 2As shown, the flow includes the following steps:

[0079] Step S201, collect multi-source data, which includes data of arranging multiple sensors at a predetermined position of the fan blade, environmental meteorological data, and fan operation data. For details, please refer to Figure 1 Step S101 of the embodiment shown will not be described here.

[0080] Step S202, the data of multiple sensors includes an output characteristic curve; real-time acquisition of the output characteristic curve of multiple sensors, comparison of the output characteristic curve with the standard characteristic curve, and obtaining the comparison error; based on the comparison error, the genetic algorithm is used to automatically find the optimal calibration parameter, and the output characteristic curve of multiple sensors is dynamically corrected based on the optimal calibration parameter, to adaptively calibrate and compensate the data of multiple sensors.

[0081] Specifically, the architecture diagram of the intelligent monitoring and analysis platform is as shown in Figure 7 As shown, the hardware of the intelligent monitoring and analysis platform is composed of high-performance servers, equipped with multi-core CPUs, large-capacity memories, and high-speed storage devices, to meet the needs of complex data processing and model training.

[0082] In addition to containing deep learning models, the intelligent monitoring and analysis platform also includes an adaptive calibration and compensation module. In order to solve the measurement error problem of fiber grating sensors, the intelligent monitoring and analysis platform runs an adaptive calibration and compensation module based on a genetic algorithm, and the specific adaptive calibration and compensation process is as follows:

[0083] Real-time acquisition of the output characteristic curve of the fiber grating sensor, comparison with the preset standard characteristic curve, dynamic adjustment of the calibration parameter by using the genetic algorithm, compensation of the measurement error caused by the process difference of the fiber grating sensor, long-term use drift, and environmental changes. At the same time, according to the physical model, the non-icing related strain and spectral change caused by environmental factors are accurately compensated, to ensure that the data input into the deep learning model is accurate and reliable.

[0084] Step S203, fuse the multi-source data, and extract features from the fused data to obtain a comprehensive feature vector.

[0085] Specifically, the adaptive calibration and compensation module is connected with the multi-source data fusion module, and after the adaptive calibration and compensation of the fiber grating sensor is completed, the fiber grating sensor data after adaptive calibration and compensation is transmitted to the multi-source data fusion module.

[0086] The above step S203 includes:

[0087] Step S2031, data preprocessing of data cleaning and data standardization for environmental meteorological data and fan operation data.

[0088] Specifically, the data cleaning includes:

[0089] De-noising processing: for high-frequency noise in fiber grating sensor data (strain, temperature) caused by blade vibration and electromagnetic interference, wavelet transform de-noising (decompose the signal into different frequency components, and eliminate high-frequency components dominated by noise) or mean filtering (take the mean value in the sliding window, smooth the instantaneous fluctuation) is adopted; for sudden outliers in environmental meteorological data (such as wind speed, humidity) and fan operation data (such as speed, power) (such as jump caused by sensor instantaneous failure), 3σ principle (eliminate data deviating from the mean value by 3 times the standard deviation) or median filtering (replace outliers with window median, more resistant to interference).

[0090] Missing value filling: for missing values caused by data transmission interruption or sensor short-time failure, select the filling method according to the data type: linear interpolation (fit the intermediate value based on the data before and after the time) for time series continuous data (such as real-time strain of fiber grating, fan speed); adopt adjacent mean value filling (take the mean value of 1 hour data before and after the missing time).

[0091] Data standardization: the three types of data have large differences in magnitude (such as strain in με, temperature in ℃, and speed in r / min), which need to be unified in scale to avoid that a certain type of data has too high weight in fusion.

[0092] For strain and temperature data of fiber grating, Z-score standardization is adopted to convert it into normal distribution data with mean value of 0 and standard deviation of 1, which preserves the distribution characteristics of the data.

[0093] For environmental meteorological data and fan operation data, min-max normalization is adopted to map to the interval [0, 1], which is suitable for data with clear range (such as humidity 0-100%, speed 0-rated speed).

[0094] It should be noted that the collection frequencies of the three types of data may be different (such as fiber grating data sampling frequency 1-100Hz, environmental meteorological data 1 time / minute, and fan operation data 10Hz), which need to be synchronized through time stamp (take the time stamp of the highest frequency data as the reference, and interpolate or resample the low-frequency data), to ensure that the multi-source data at the same time can be matched (such as interpolating the humidity data 1 minute 1 time to 1Hz, aligning the time granularity with the fiber grating data).

[0095] Step S2032, adopt a preset fusion algorithm to fuse the multiple sensor data after adaptive calibration and error compensation, and the environmental meteorological data and fan operation data after data preprocessing.

[0096] Specifically, the calibrated fiber grating data (strain, temperature) and the pre-processed environmental meteorological data (temperature, humidity, wind speed, precipitation type), fan operation data (rotation speed, power) need to be integrated through a fusion algorithm, key features are extracted, and "low redundancy, high correlation" comprehensive information is generated. According to the characteristics of the data, two fusion paths are included:

[0097] Path 1: Kalman filter (dynamic data real-time fusion): suitable for scenarios that require real-time monitoring (such as the dynamic development process of blade icing), focusing on the fusion of time-series data (such as real-time strain of fiber grating, instantaneous changes in wind speed, dynamic adjustment of fan speed).

[0098] Path 2: Principal Component Analysis (PCA, static feature dimension reduction fusion): suitable for batch data analysis scenarios (such as training models with historical data), focusing on the fusion of multi-dimensional, high-redundancy data (such as strain data from different locations fiber grating, environmental parameters for multiple hours, and multiple operating state indicators), the core is dimension reduction, that is, to retain the main information of the original data with a few principal components.

[0099] Step S2033, feature extraction is performed on the fused data to obtain a comprehensive feature vector.

[0100] Specifically, whether it is the real-time fusion result of Kalman filter or the dimension reduction feature matrix of PCA, it needs to be further integrated into a single sample feature vector as the input of the deep learning model.

[0101] The specific operation includes: for time series data (such as 100 Kalman filter results in 1 minute), statistical features (such as mean, variance, peak value, trend slope) are extracted, and time series information is converted into static features; for the principal components after PCA dimension reduction, they are directly arranged in dimension order; integrate all features (such as statistical features + principal component features) to form a fixed length vector, ensure the uniformity of dimensions, and meet the model input requirements.

[0102] Step S204, based on the comprehensive feature vector and the preset deep learning model, the current fan blade icing condition is monitored to obtain icing condition information. For details, please refer to Figure 1 the step S103 of the embodiment shown in the figure, which will not be repeated here.

[0103] Step S205, based on the size relationship between the icing condition information and the preset threshold, the fan blade icing state is warned. For details, please refer to Figure 1 the step S103 of the embodiment shown in the figure, which will not be repeated here.

[0104] The fan blade icing monitoring method provided by the embodiment adopts an adaptive calibration strategy based on a genetic algorithm and an environmental compensation algorithm based on a physical model. By monitoring the output characteristics of the sensor in real time and comparing them with the pre-stored standard characteristic curve, the calibration parameters are dynamically optimized using the genetic algorithm to compensate for the measurement error of the sensor in real time. At the same time, the non-icing related strain and spectral changes caused by environmental factors such as temperature and humidity can also be accurately compensated. This innovation ensures the measurement accuracy and stability of the fiber grating sensor in different operating environments and during long-term use, reduces the monitoring error caused by the sensor itself, improves the reliability of the entire monitoring system, and effectively solves the problems of insufficient monitoring performance adaptability and stability in the prior art. The multi-source data includes data of multiple sensors arranged at preset positions of the fan blade, environmental meteorological data, and fan operation data. Through a preset data fusion algorithm, a more comprehensive and rich feature vector is formed as the input of the deep learning model. This multi-source data fusion method enables the model to comprehensively evaluate the icing risk from multiple dimensions, overcoming the limitations of traditional methods that rely solely on single strain and temperature information to fully characterize the icing condition.

[0105] In the embodiment, a fan blade icing monitoring method is provided, which can be used in a fan blade icing monitoring system. Figure 4 As shown in the figure, the system includes a sensor network (such as a fiber grating sensor network), a data acquisition and transmission module, an intelligent monitoring and analysis platform, and a warning and decision support module. Figure 3 The flowchart of the fan blade icing monitoring method according to the embodiment of the application is shown in Figure 3 The flowchart includes the following steps:

[0106] Step S301, collect multi-source data, including data of multiple sensors arranged at preset positions of the fan blade, environmental meteorological data, and fan operation data. For details, see step S201 of the embodiment shown in Figure 2 which will not be repeated here.

[0107] Step S302, fuse the multi-source data and extract features from the fused data to obtain a comprehensive feature vector. For details, see step S203 of the embodiment shown in Figure 2 which will not be repeated here.

[0108] Step S303, monitor the current fan blade icing condition based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information.

[0109] Specifically, step S303 includes:

[0110] In step S3031, the historical icing data corresponding to the multi-source data is obtained, and the historical icing data is input into a preset deep learning model for model training to obtain an icing trend prediction model.

[0111] Specifically, the preset deep learning model can adopt a convolutional neural network (CNN) model.

[0112] The intelligent monitoring and analysis platform integrates a deep learning framework (such as TensorFlow or PyTorch), and trains a convolutional neural network (CNN) model by using a large amount of historical icing data (including fiber grating sensor data under different icing thickness, range, type and other scenarios, and corresponding environmental meteorological data and fan operation data). During the training process, the convolutional neural network structure (such as the number of layers and the number of neurons of the convolutional layer, the pooling layer and the fully connected layer), the optimizer parameters (such as the learning rate, the momentum, etc.) and the regularization strategy (such as the L1 / L2 regularization, the Dropout regularization technology, etc.) are adjusted, so that the convolutional neural network model can automatically learn the change mode of the fiber grating reflection spectrum features before and after the icing occurs, and the correlation features between the multi-source data, thereby realizing accurate identification and classification of the icing condition.

[0113] In the fan blade icing monitoring scene, the correlation features between the multi-source data (sensor data, environmental meteorological data and fan operation data) refer to the internal coordinated change law, the cause-effect relationship or the mutual confirmation relationship formed by different source data due to the physical mechanism of the icing process. These correlation features can reflect the dynamic process of the icing occurrence and development, and are the core basis for the deep learning model to judge the icing condition.

[0114] For example, the correlation between the sensor data (blade state) and the environmental meteorological data (external inducement) reflects the cause-effect relationship that the environmental conditions drive the icing formation and the icing leads to the change of the blade state. For another example, the correlation between the sensor data (blade physical state) and the fan operation data (device response) reflects the linkage relationship that the icing changes the blade characteristics and leads to the adaptive adjustment of the fan operation parameters; or the correlation between the environmental meteorological data (icing inducement) and the fan operation data (device response) reflects the law that the environmental conditions trigger the icing risk and the fan avoids the risk by changing the operation parameters.

[0115] In step S3032, the comprehensive feature vector is input into the icing trend prediction model to obtain icing condition information, and the icing condition information includes the icing position, the icing thickness, the icing range and the icing type.

[0116] In an optional embodiment, the icing trend prediction model comprises learned feature change patterns when icing occurs on the wind turbine blade and associated features between multi-source data; the step S3032 comprises:

[0117] Step a, when the comprehensive feature vector is input into the icing trend prediction model, the icing trend prediction model automatically judges the current icing condition of the wind turbine blade based on the feature change patterns and the associated features between multi-source data, and obtains icing condition information.

[0118] Specifically, the deep learning model analyzes the comprehensive feature vector to determine the icing condition of the current wind turbine blade. If the deep learning model identifies that icing occurs on the wind turbine blade, the location of the icing, the thickness of the icing, the range of the icing, the type of the icing and other icing condition information are displayed in real time on the monitoring interface.

[0119] Step S304, based on the size relationship between the icing condition information and the preset threshold, the icing state of the wind turbine blade is warned.

[0120] Specifically, the architecture of the warning and decision support module is as shown in Figure 8 The step S304 comprises:

[0121] Step S3041, when the icing condition information reaches the preset threshold, a multi-channel warning signal is triggered, warning information is obtained, and anti-icing and / or de-icing decision suggestions are output based on the warning information and the preset strategy.

[0122] Specifically, the warning mechanism comprises: the icing trend prediction model based on the long short term memory (Long Short Term Memory, LSTM for short) model runs on the intelligent detection and analysis platform. The LSTM model learns the time sequence law of icing development by using historical icing monitoring data, environmental meteorological data and wind turbine operation data. When the icing thickness or range is predicted to reach the set threshold (such as determined according to the wind turbine blade design and operation safety standard), the system automatically triggers a warning signal. The warning signal is sent to the wind farm operation and maintenance personnel through various ways (such as short message, email, sound and light alarm, etc.), and a prominent warning prompt information is popped up on the monitoring interface, displaying the predicted development trend of the icing, the possible affected blade position and other key information.

[0123] Step S3042, a response measure is made based on the anti-icing and / or de-icing decision suggestion.

[0124] Specifically, the decision support function: the intelligent detection and analysis platform provides decision suggestions for the operation and maintenance personnel according to the early warning information and the current fan operation state, combined with the anti-icing and / or de-icing decision suggestions (such as adjusting the fan speed, starting the blade heating system, changing the pitch angle, etc.). The operation and maintenance personnel can view the detailed decision scheme through the monitoring interface, including operation steps, expected effect, resource demand and other information, to assist them to quickly and accurately make response measures, effectively prevent the damage of icing to the fan blades, and ensure the safe and stable operation of the wind turbine generator. The early warning and decision support module realizes the early prediction and intelligent warning of the icing trend, and provides strong support for the anti-icing and de-icing decision of the fan.

[0125] The fan blade icing monitoring method provided by the embodiment can realize accurate prediction of the icing trend in a future period of time, can issue an icing early warning information in advance for hours or even days, provides a valuable time window for the anti-icing and de-icing decision of the fan, realizes the change from passive monitoring to active prevention, and lays a foundation for effectively solving the problem of lack of intelligent early warning capability in the prior art. By directly comparing the real-time monitored icing condition information with the preset threshold, it can be quickly judged whether the icing reaches a dangerous degree, the early warning lag caused by manual judgment delay or experience error is avoided, and the early warning is triggered immediately when the icing condition reaches the threshold, so that the safety risk caused by icing can be avoided in advance. The multi-channel early warning signal means that the early warning information can be delivered to the relevant personnel or system in multiple ways to ensure that the early warning information can be received in time. The anti-icing and / or de-icing decision suggestions output based on the icing condition information and the preset strategy can provide personalized schemes combined with the specific icing condition, realize the early prediction and intelligent warning of the icing trend, and provide strong support for the anti-icing and de-icing decision of the fan.

[0126] As one or more specific application embodiments of the embodiment of the application, the application is combined with Figure 4 The fan blade icing monitoring method provided by the application is further described in detail as follows.

[0127] 1. Fan blade icing monitoring system initialization: after power-on, the fiber grating sensor network, the data acquisition and transmission module, the intelligent monitoring and analysis platform and the early warning and decision support module perform self-checking and initialization configuration, establish communication connection, and prepare to start monitoring work.

[0128] 2. Data acquisition and transmission: the fiber grating sensor senses the strain and temperature change of the fan blade in real time, the data acquisition unit collects the spectrum signal of the fiber grating sensor through the grating demodulator and converts it into a digital signal, which is sent to the intelligent monitoring and analysis platform through the data transmission link. At the same time, environmental meteorological data and fan operation data are collected and transmitted to the intelligent monitoring and analysis platform.

[0129] 3. Data processing and analysis: After receiving the fiber Bragg grating sensor data, environmental meteorological data, and wind turbine operation data, the intelligent monitoring and analysis platform first performs preprocessing (such as filtering, denoising, etc.), then fuses the processed fiber Bragg grating sensor data with environmental meteorological data and wind turbine operation data to form a comprehensive feature vector. The deep learning model analyzes the feature vector to determine the current icing condition of the blade. At the same time, the self-adaptive calibration and compensation module performs real-time calibration and compensation on the fiber Bragg grating sensor data to ensure the accuracy of the analysis results.

[0130] 4. Icing monitoring and early warning: If the deep learning model identifies icing, the system displays the location, thickness, and range of the icing on the monitoring interface in real time. The LSTM icing trend prediction model predicts the icing development trend based on the current icing condition and environmental data. When the predicted icing reaches the warning threshold, the warning and decision support module immediately issues a warning signal and provides corresponding decision recommendations.

[0131] 5. Operation and response: After receiving the warning signal, the operation personnel take appropriate anti-icing and de-icing measures based on the decision recommendations provided by the system, and feed back the processing results to the system to update the icing state information and continue monitoring.

[0132] The fan blade icing monitoring method provided by the embodiment realizes high-precision, intelligent monitoring, early warning, and decision support for the icing condition of the fan blade through the cooperative work of the above modules, effectively solves the problems of low accuracy, insufficient intelligence, and poor adaptability of existing monitoring methods, and improves the operation safety and reliability of the wind turbine generator set in icing environments.

[0133] The fan blade icing monitoring method provided by the embodiment effectively solves many problems existing in existing fan blade icing monitoring methods through a series of innovative technical means, and achieves remarkable technical effects:

[0134] 1) Improve monitoring accuracy: The embodiment introduces a deep learning algorithm (convolutional neural network, CNN) to analyze and process the data collected by the fiber Bragg grating sensor. Since the CNN model is trained with a large amount of historical data, it can automatically learn the change pattern of the fiber Bragg grating reflectance spectrum before and after icing occurs. Compared with the traditional method of judging strain changes based on fixed thresholds, it can capture more subtle and complex spectral feature changes. Therefore, the application can accurately identify and classify the early stage of icing, significantly improve the accuracy and sensitivity of monitoring, and effectively solve the problems of misjudgment and missed judgment in existing technologies.

[0135] 2) Enhance the intelligent level: the embodiment of the application constructs a feature extraction module of multi-source data fusion, integrates the data of fiber grating sensor, environmental meteorological data and fan operation data. Through a specific data fusion algorithm, a more comprehensive and rich feature vector is formed as the input of the deep learning model. This multi-source data fusion method enables the model to comprehensively evaluate the icing risk from multiple dimensions, overcoming the limitations of traditional methods that rely only on single strain and temperature information to fully characterize the icing condition. At the same time, the application develops an icing trend prediction module based on long short-term memory network (LSTM), and the powerful time series modeling capability of the LSTM model can realize accurate prediction of the icing trend in the future period, and provide valuable time window for the anti-icing and de-icing decision of the fan, realize the change from passive monitoring to active prevention, effectively solve the problem of lack of intelligent warning capability in the prior art.

[0136] 3) Improve adaptability and stability: the application develops a self-adaptive calibration and compensation module, adopts a self-adaptive calibration strategy based on genetic algorithm and an environmental compensation algorithm based on physical model. By monitoring the output characteristics of the sensor in real time and comparing with the pre-stored standard characteristic curve, the genetic algorithm is used to dynamically optimize the calibration parameters, and the measurement error of the sensor is compensated in real time. At the same time, the non-icing related strain and spectrum change caused by temperature, humidity and other environmental factors are accurately compensated. This innovation ensures the measurement accuracy and stability of the fiber grating sensor in different operating environments and during long-term use, reduces the monitoring error caused by the sensor itself, improves the reliability of the whole monitoring system, and effectively solves the problem of insufficient adaptability and stability of the monitoring system in the prior art.

[0137] 4) Reduce system complexity and cost: the application uses a deep learning model to intelligently analyze the data of the fiber grating sensor, reducing the dependence on complex signal demodulation equipment and large-scale optical fiber transmission network. At the same time, the introduction of multi-source data fusion and self-adaptive calibration and compensation mechanism optimizes the overall performance of the method, reduces the system complexity and maintenance difficulty. This makes the monitoring system of the application more easily applied in large-scale wind farms, effectively reduces the monitoring cost, and ensures that the monitoring result can fully reflect the key parameters of icing, providing effective decision basis for the anti-icing and de-icing control of the fan.

[0138] In summary, the application combines various advanced technical means such as deep learning, multi-source data fusion, self-adaptive calibration and compensation, and time series prediction, to comprehensively improve the detection performance of the fan blade icing monitoring method based on fiber grating sensor, greatly improving the accuracy, intelligence, adaptability and reliability, effectively solving the problems and shortcomings of the prior art.

[0139] A fan blade icing monitoring system is also provided in the present embodiment, which is used to implement the above-mentioned embodiments and preferred embodiments, and will not be described again. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware, or a combination of software and hardware is also possible and contemplated.

[0140] The present embodiment provides a fan blade icing monitoring system, as shown in Figure 9 , comprising:

[0141] A sensor network 901 is fixedly installed at a predetermined position of a fan blade. A fiber grating sensor network is used in the present embodiment. Figure 4

[0142] A multi-source data acquisition module 902 is used to acquire multi-source data, which includes data of multiple sensors arranged at a predetermined position of a fan blade, environmental meteorological data, and fan operation data. The multi-source data acquisition module includes a data acquisition and transmission module. Figure 4

[0143] An intelligent monitoring and analysis platform 903 is used to fuse the multi-source data and extract features from the fused data to obtain a comprehensive feature vector; based on the comprehensive feature vector and a preset deep learning model, the current fan blade icing condition is monitored to obtain icing condition information.

[0144] An early warning and decision support module 904 is used to early warn the fan blade icing condition based on the size relationship between the icing condition information and a preset threshold. The early warning and decision support module is consistent with that in the present embodiment. Figure 4

[0145] In some optional embodiments, the data of multiple sensors includes an output characteristic curve; the fan blade icing monitoring system further comprises:

[0146] An adaptive calibration and compensation module is used to acquire the output characteristic curve of multiple sensors in real time, compare the output characteristic curve with a standard characteristic curve to obtain a comparison error, automatically find optimal calibration parameters based on the comparison error using a genetic algorithm, and dynamically correct the output characteristic curve of multiple sensors based on the optimal calibration parameters to adaptively calibrate and compensate the data of multiple sensors. The adaptive calibration and compensation module is consistent with that in the present embodiment. Figure 7

[0147] In some optional embodiments, the intelligent monitoring and analysis platform 903 comprises:

[0148] A data preprocessing module is used to perform data cleaning and data standardization data preprocessing on the environmental meteorological data and fan operation data.​​​​

[0149] a multi-source data fusion module configured to fuse the plurality of sensor data after adaptive calibration and error compensation, the environmental and meteorological data after data preprocessing, and the wind turbine operation data using a preset fusion algorithm.

[0150] a feature extraction module configured to extract features from the fused data to obtain a comprehensive feature vector.

[0151] In some optional embodiments, the intelligent monitoring and analysis platform 903 further comprises:

[0152] a deep learning model training module configured to obtain historical icing data corresponding to the multi-source data, and input the historical icing data into a preset deep learning model for model training to obtain an icing trend prediction model.

[0153] an icing trend prediction module configured to input the comprehensive feature vector into the icing trend prediction model to obtain icing condition information, wherein the icing condition information comprises an icing location, an icing thickness, an icing range, and an icing type.

[0154] In some optional embodiments, the icing trend prediction model comprises learned feature change patterns when icing occurs on the wind turbine blades and associated features between the multi-source data, and the icing trend prediction module comprises:

[0155] an icing trend prediction unit configured to, when the comprehensive feature vector is input into the icing trend prediction model, cause the icing trend prediction model to automatically determine a current icing condition of the wind turbine blades based on the feature change patterns and the associated features between the multi-source data to obtain the icing condition information.

[0156] In some optional embodiments, the early warning and decision support module 904 comprises:

[0157] an early warning mechanism unit configured to, when the icing condition information reaches a preset threshold, trigger a multi-channel early warning signal to obtain early warning information, and output an anti-icing and / or de-icing decision suggestion based on the early warning information and a preset strategy.

[0158] a decision support unit configured to take a response measure based on the anti-icing and / or de-icing decision suggestion.

[0159] Further function descriptions of the above-mentioned modules and units are the same as those of the corresponding embodiments described above, and will not be repeated here.

[0160] In this embodiment, the wind turbine blade icing monitoring system is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit), a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0161] This invention also provides a computer device having the above-described features. Figure 9 The wind turbine blade icing monitoring system shown.

[0162] Please see Figure 10 , Figure 10 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 10 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 10 Take a processor 10 as an example.

[0163] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0164] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0165] The memory 20 can include a program storage area and a data storage area, where the program storage area can store an operating system, application programs required for at least one function, and the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some alternative embodiments, the memory 20 can optionally include a memory disposed remotely from the processor 10, which can be connected to the computer device through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0166] The memory 20 can include a volatile memory, such as a random access memory, and can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid state disk, and can also include a combination of the above-mentioned kinds of memories.

[0167] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30, and the output device 40 can be connected through a bus or other means, Figure 10 The connection through the bus is taken as an example.

[0168] The input device 30 can receive inputted digital or character information, and generate key signal inputs related to the user settings and function controls of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touchpad, a pointing stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), a tactile feedback device (e.g., a vibration motor), etc. The display device includes, but is not limited to, a liquid crystal display, a light-emitting diode, a display, and a plasma display. In some alternative embodiments, the display device can be a touch screen.

[0169] The embodiments of the present application further provide a computer readable storage medium, and the method according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded to a local storage medium through network, so that the method described herein can be processed by such software on a storage medium using a general purpose computer, a special purpose processor, or programmable or special hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid state disk, etc. Further, the storage medium can also include a combination of the above-mentioned memories. It can be understood that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component that can store or receive software or computer code, when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0170] Part of the present application can be applied as a computer program product, for example, computer program instructions, when executed by a computer, the operation of the computer can invoke or provide the method and / or technical solutions according to the present application. Those skilled in the art should understand that the form of computer program instructions in computer readable medium includes but is not limited to source file, executable file, installation package file, etc. Correspondingly, the way of computer program instructions executed by computer includes but is not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Here, the computer readable medium can be any available computer readable storage medium or communication medium accessible to the computer.

[0171] Although the embodiments of the present application are described in conjunction with the accompanying drawings, various modifications and changes can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and changes fall within the scope defined by the appended claims.

Claims

1. A method of monitoring icing on a wind turbine blade, characterized by, The method comprises: Collecting multi-source data, the multi-source data comprising data of multiple sensors arranged at preset positions of the fan blade, environmental meteorological data, and fan operation data; Fusing the multi-source data and extracting features from the fused data to obtain a comprehensive feature vector; Monitoring the icing condition of the current fan blade based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information; Based on the size relationship between the icing condition information and a preset threshold, the icing state of the fan blade is warned; The data of the multiple sensors comprises an output characteristic curve; before the multi-source data is fused, the method further comprises: Real-time collection of the output characteristic curve of the multiple sensors, comparison of the output characteristic curve with a standard characteristic curve, and obtaining of a comparison error; Based on the comparison error, an optimal calibration parameter is automatically found by using a genetic algorithm, and the output characteristic curve of the multiple sensors is dynamically corrected based on the optimal calibration parameter to adaptively calibrate and compensate the data of the multiple sensors; The fusing of the multi-source data and the feature extraction from the fused data to obtain the comprehensive feature vector comprises: Data preprocessing of the environmental meteorological data and the fan operation data, including data cleaning and data standardization; Fusion of the multiple sensor data after adaptive calibration and error compensation, the environmental meteorological data after data preprocessing, and the fan operation data by using a preset fusion algorithm; Feature extraction from the fused data to obtain the comprehensive feature vector; The warning of the icing state of the fan blade based on the size relationship between the icing condition information and a preset threshold comprises: When the icing condition information reaches the preset threshold, a multi-channel warning signal is triggered to obtain warning information, and an anti-icing and / or de-icing decision suggestion is output based on the warning information and a preset strategy; the warning mechanism comprises: an icing trend prediction model based on a long short-term memory network model running on an intelligent detection and analysis platform; the long short-term memory network model learns the timing law of icing development by using historical icing monitoring data, environmental meteorological data, and fan operation data; when the predicted icing thickness or range reaches the set threshold, the system automatically triggers a warning signal; Response measures are taken based on the anti-icing and / or de-icing decision suggestion.

2. The method of claim 1, wherein, The monitoring of the icing condition of the current fan blade based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information comprises: Obtaining historical icing data corresponding to the multi-source data, and inputting the historical icing data into the preset deep learning model for model training to obtain an icing trend prediction model; Inputting the comprehensive feature vector into the icing trend prediction model to obtain the icing condition information; the icing condition information comprises icing position, icing thickness, icing range, and icing type.

3. The method of claim 2, wherein, The icing trend prediction model comprises learned feature change patterns and associated features between the multi-source data when the icing of the fan blade occurs; Inputting the comprehensive feature vector into the icing trend prediction model to obtain the icing condition information comprises: When the comprehensive feature vector is input into the icing trend prediction model, the icing trend prediction model is caused to automatically determine the current wind turbine blade icing condition based on the feature change mode and the associated features between the multi-source data, and icing condition information is obtained.

4. A wind turbine blade ice monitoring system characterized by, The system comprises: a sensor network for fixed installation at the preset positions of the wind turbine blades; a multi-source data acquisition module for acquiring multi-source data, the multi-source data comprising data of a plurality of sensors arranged at the preset positions of the wind turbine blades, environmental meteorological data and wind turbine operation data; an intelligent monitoring and analysis platform for fusing the multi-source data and extracting features from the fused data to obtain a comprehensive feature vector, and for monitoring the current wind turbine blade icing condition based on the comprehensive feature vector and a preset deep learning model to obtain icing condition information; an early warning and decision support module for early warning of the wind turbine blade icing condition based on the size relationship between the icing condition information and a preset threshold value; the data of the plurality of sensors comprises output characteristic curves; and the method further comprises: real-time acquisition of the output characteristic curves of the plurality of sensors, and comparison of the output characteristic curves with standard characteristic curves to obtain comparison errors; automatic finding of optimal calibration parameters based on the comparison errors using a genetic algorithm, and dynamic correction of the output characteristic curves of the plurality of sensors based on the optimal calibration parameters to adaptively calibrate and compensate errors in the data of the plurality of sensors; the fusing of the multi-source data and the extraction of features from the fused data to obtain a comprehensive feature vector comprises: data preprocessing of the environmental meteorological data and the wind turbine operation data, including data cleaning and data standardization; fusion of the adaptively calibrated and error-compensated data of the plurality of sensors, the data-preprocessed environmental meteorological data and the wind turbine operation data using a preset fusion algorithm; feature extraction from the fused data to obtain a comprehensive feature vector; the early warning of the wind turbine blade icing condition based on the size relationship between the icing condition information and a preset threshold value comprises: when the icing condition information reaches the preset threshold value, a multi-channel early warning signal is triggered to obtain early warning information, and an anti-icing and / or de-icing decision suggestion is output based on the early warning information and a preset strategy; the early warning mechanism comprises: an icing trend prediction model based on a long short-term memory network model is run on the intelligent detection and analysis platform; the long short-term memory network model learns the timing law of icing development using historical icing monitoring data, environmental meteorological data and wind turbine operation data; when it is predicted that the icing thickness or range will soon reach the set threshold value, the system automatically triggers an early warning signal; a response measure is taken based on the anti-icing and / or de-icing decision suggestion.

5. A computer device, comprising: The system comprises: a memory and a processor, which are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the wind turbine blade icing monitoring method of any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing a computer to execute the fan blade icing monitoring method of any one of claims 1 to 3.

7. A computer program product, characterised in that, The computer readable storage medium stores computer instructions for causing a computer to execute the fan blade icing monitoring method of any one of claims 1 to 3.

Images