Wind turbine blade icing segmentation detection system and method based on multi-modal data fusion

Abstract

The present application relates to the technical field of wind turbine blade state monitoring, and discloses a wind turbine blade icing subsection detection method based on multi-modal data fusion. Nodes containing flexible interdigital capacitors, temperature and humidity sensors and vibration sensors are deployed in the tip, middle and root subsections of the blade, and the dielectric constant change is perceived by using the edge electric field effect. A multi-modal cross-validation logic is used, combined with an environmental thermodynamic window to filter out dirt and water film interference, to accurately calculate the thickness of the ice layer. At the same time, PSD analysis of the vibration signal is introduced as a safety bottom logic to identify abnormalities caused by aerodynamic imbalance. The execution module implements differentiated deicing according to the subsection monitoring results, and uses the capacitive signal to capture the critical water film generation time. The present application solves the problems of high false positive rate, inability to monitor in subsections and large deicing energy consumption of the prior art, and has strong engineering applicability.

Description

Technical Field

[0001] This invention relates to the field of wind turbine blade condition monitoring technology, specifically to a wind turbine blade icing segment detection system and method based on multimodal data fusion. Background Technology

[0002] As a widely used clean and renewable energy source globally, wind power is experiencing strong growth. As of December 2025, the cumulative grid-connected wind power capacity in China reached 640 million kilowatts, with 120 million kilowatts of new installed capacity added in 2025. However, in cold, humid, or high-altitude areas, wind turbine blades are prone to icing. Blade icing not only alters the aerodynamic shape of the blades, leading to decreased aerodynamic efficiency and reduced power generation, but also disrupts the dynamic balance of the blades, causing abnormal vibrations, and in severe cases, even leading to major safety accidents such as turbine shutdown or blade breakage.

[0003] Existing wind turbine icing detection methods mostly rely on a single sensor, which has significant limitations: 1. High environmental interference: Ordinary dielectric sensors are easily affected by dirt and water film on the blade surface, which can cause false alarms; 2. Sensing lag: Temperature and humidity monitoring cannot reflect changes in the phase state of the surface ice layer in real time; 3. Inaccurate monitoring: The lack of segmented detection capabilities leads to extremely high energy consumption and low efficiency in de-icing.

[0004] Furthermore, existing monitoring methods often treat wind turbine blades as a whole, neglecting the differences in icing processes at different parts of the blade, such as the blade root, blade middle, and blade tip. Since the blade tip has the fastest linear velocity and the lowest temperature, it is usually the first and most severely iced part. If segmented and accurate monitoring cannot be achieved, it will lead to a "one-size-fits-all" de-icing strategy, resulting in insufficient local de-icing or overall energy waste.

[0005] Therefore, there is an urgent need for a detection method that can integrate the advantages of multiple sensors, has anti-interference capabilities, and can achieve precise segmented monitoring of blades, in order to solve the problems of high false alarm rate, low detection accuracy, and high de-icing energy consumption in existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a segmented detection system and method for wind turbine blade icing based on multimodal data fusion.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A segmented detection system for icing on wind turbine blades based on multimodal data fusion includes a control multimodal sensing module, a data acquisition and processing module, and an execution control module. The multimodal sensing modules are distributed across three sections of the wind turbine blade: the root, middle, and tip. The tip section is equipped with a composite sensing node integrating a vibration sensor, a dielectric sensor, and a temperature and humidity sensor. The middle and root sections are each equipped with at least one dielectric sensor. The data acquisition and processing module receives data from each section and executes multimodal cross-validation logic to output the corrected segmented ice thickness status. The execution control module drives the de-icing unit of the corresponding section based on the segmented ice thickness status.

[0008] Preferably, the dielectric sensor is a flexible interdigital capacitance sensor, which is attached to the blade surface and uses the edge electric field effect to sense changes in dielectric constant in order to capture phase transition signals in the early stage of icing.

[0009] Preferably, the multimodal cross-validation logic includes: When the ambient temperature is greater than 2℃ or the relative humidity is less than 60%, the fluctuation of the capacitance signal is identified as environmental interference and filtered out; when the ambient temperature and humidity meet the icing thermodynamic window, the capacitance-thickness fitting model is activated to calculate the ice layer thickness.

[0010] Preferably, the data acquisition and processing module is equipped with vibration triggering logic, which extracts the inherent frequency change characteristics by performing power spectral density analysis on the vibration signal; when the aerodynamic polarization signal induced by icing is captured, the module bypasses the thickness threshold judgment logic and directly triggers the highest priority emergency de-icing procedure.

[0011] Preferably, the method further includes the following steps: Capacitance signals, ambient temperature and humidity, and blade vibration data are collected in segments at the tip, middle, and root of the wind turbine blades. Real-time temperature and humidity data are thermodynamically matched with capacitance signals to filter out interference signals and calculate the segmented ice layer thickness. Vibration signal features are extracted to identify changes in mass distribution and aerodynamic anomalies caused by icing. Differentiated de-icing strategies are output based on the segmented icing status, and the critical water film formation moment is captured by the recovery of capacitance signals to assist in de-icing.

[0012] Compared with existing technologies, this invention provides a segmented detection method for icing on wind turbine blades based on multimodal data fusion, which has the following advantages: 1. This low-energy collaborative control system and method for predicting and dynamically de-icing wind turbine blades utilizes a flexible interdigital capacitive sensor to detect changes in the dielectric constant of the blade surface based on the edge electric field effect, supplemented by temperature and humidity sensors and vibration sensors to acquire environmental and dynamic characteristics. The data acquisition and processing module uses multimodal cross-validation logic to thermodynamically match real-time temperature and humidity data with the capacitance signal, filtering out interference signals and accurately calculating the ice layer thickness. The execution control module then drives the de-icing unit to perform precise de-icing or anti-icing operations based on the segmented icing status output by the processing module.

[0013] 2. This wind turbine blade icing prediction and dynamic de-icing low-energy collaborative control system and method utilizes interdigitated capacitance sensors to detect the differences in dielectric constants of ice, water, and air, enabling very early icing warnings. It also effectively avoids misjudgments caused by surface dirt or water films by using temperature and humidity windows. During the execution phase, vibration characteristics are introduced as a safety fallback logic; once abnormal vibration caused by aerodynamic imbalance is detected, emergency de-icing is immediately triggered. Furthermore, it possesses precise phase change feedback capabilities, able to capture the critical water film formation moment at the blade-ice interface by monitoring the rise in capacitance signals, thereby guiding the execution module to use centrifugal force to assist in de-icing, significantly improving energy efficiency.

[0014] 3. This wind turbine blade icing prediction and dynamic de-icing low-energy collaborative control system and method adopts a segmented monitoring strategy. By independently sensing and controlling the icing differences at the blade tip (0.9R), blade middle (0.5R), and blade root (0.2R), it effectively solves the energy waste problem caused by the one-size-fits-all approach in traditional detection schemes. It can be integrated into the intelligent monitoring system of newly manufactured wind turbine blades, and can also be applied to the digital transformation of existing wind farm blades through flexible sensor bonding technology, which has strong engineering applicability. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the segmented deployment of wind turbine blade sensors in an embodiment of the present invention.

[0016] Figure 2 This is a flowchart illustrating the logic of multimodal data cross-validation in an embodiment of the present invention.

[0017] Figure 3 This is a schematic diagram of the on-site experimental operation in an embodiment of the present invention.

[0018] Figure 4 This is a graph showing experimental data of the interdigital capacitive sensor in an embodiment of the present invention.

[0019] Figure 5 This is a PSD image of the vibration sensor blade tip (0.9R) in the X, Y, and Z directions in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] As described in the background section, there are shortcomings in the existing technology. In order to solve the above-mentioned technical problems, this application proposes a wind turbine blade icing segment detection system and method based on multimodal data fusion.

[0022] A segmented detection method for icing on wind turbine blades based on multimodal data fusion includes a multimodal sensing module, a data acquisition and processing module, and an execution control module integrated into different sections of the wind turbine blade. This allows for precise sensing of the blade's icing status and refined de-icing control. Furthermore, the deployment density of sensing nodes can be flexibly adjusted according to different wind turbine models.

[0023] Specifically, the multimodal sensing module adopts a distributed deployment architecture. In a preferred embodiment, a full-modal node including interdigital capacitance, temperature and humidity, and vibration sensors is deployed at the blade tip (0.9R), while at least interdigital capacitance sensors are deployed at the blade middle (0.5R) and blade root (0.2R). (See attached diagram.) Figure 3-4 As shown, in this embodiment of the invention, the interdigital capacitance sensor is attached to the surface of the blade and uses the edge electric field effect generated by it to sense the change in dielectric constant. Since there is a significant difference in the dielectric constant between ice and water, the sensor can capture the phase transition signal in the early stage of icing in real time. Under experimental conditions, its effective monitoring range can cover ice thickness of 0-2 mm and above.

[0024] The data acquisition and processing module consists of a high-performance microprocessor and a multimodal fusion algorithm unit. This module receives data from the sensor and executes multimodal cross-validation logic: when temperature and humidity data fall within a preset non-icing meteorological range (e.g., temperature greater than 2℃ or humidity less than 60%), it automatically identifies and filters out capacitance signal drift caused by surface dirt or water film, determining it as environmental interference; when environmental conditions meet the icing thermodynamic window, it activates the capacitance-thickness fitting model to accurately pinpoint the icing moment. (See attached...) Figure 5 As shown in the figure, the dominant frequency distribution of the blade under normal operating conditions is clearly displayed. Simultaneously, this module performs power spectral density (PSD) analysis on the vibration signal to extract the first / second natural frequency variation characteristics of the blade, thereby identifying changes in mass distribution caused by icing. Once the aerodynamic polarization signal induced by icing is captured, the thickness threshold judgment logic is bypassed, and the highest priority emergency de-icing procedure is executed.

[0025] The execution control module includes a communication gateway and a de-icing drive unit. Based on the segmented icing status output by the processing module, this module executes a graded de-icing strategy. When the interdigital capacitance sensor detects a critical water film of 0.1 mm forming at the blade-ice interface (characterized by a rapid and characteristic increase in capacitance), the control module determines that adhesion has failed, immediately cuts off the heating current, and utilizes the centrifugal force generated by the blade rotation to achieve mechanical de-icing. This synergistic mode of thermal melting combined with centrifugal stripping minimizes de-icing energy consumption.

[0026] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A segmented detection system for icing on wind turbine blades based on multimodal data fusion, comprising a control multimodal sensing module, a data acquisition and processing module, and an execution control module, characterized in that: The multimodal sensing modules are distributed across three sections of the wind turbine blade: the root, middle, and tip. The tip section is equipped with a composite sensing node integrating a vibration sensor, a dielectric sensor, and a temperature and humidity sensor. The middle and root sections are each equipped with at least one dielectric sensor. The data acquisition and processing module receives data from each section and executes multimodal cross-validation logic to output the corrected segmented ice thickness status. The execution control module drives the de-icing unit of the corresponding section based on the segmented ice thickness status.

2. The wind turbine blade icing segment detection system based on multimodal data fusion according to claim 1, characterized in that: The dielectric sensor is a flexible interdigital capacitance sensor that is attached to the blade surface. It uses the edge electric field effect to sense changes in the dielectric constant in order to capture the phase transition signal in the early stage of icing.

3. The wind turbine blade icing segment detection system based on multimodal data fusion according to claim 1, characterized in that, The multimodal cross-validation logic includes: When the ambient temperature is greater than 2℃ or the relative humidity is less than 60%, the fluctuation of the capacitance signal is identified as environmental interference and filtered out; when the ambient temperature and humidity meet the icing thermodynamic window, the capacitance-thickness fitting model is activated to calculate the ice layer thickness.

4. The wind turbine blade icing segment detection system based on multimodal data fusion according to claim 1, characterized in that: The data acquisition and processing module is equipped with vibration triggering logic, which extracts the inherent frequency change characteristics by performing power spectral density analysis on the vibration signal; when the aerodynamic polarization signal induced by icing is captured, the module bypasses the thickness threshold judgment logic and directly triggers the highest priority emergency de-icing procedure.

5. The method for segmented detection of icing on wind turbine blades based on multimodal data fusion according to claim 1, characterized in that, It also includes the following steps: Capacitance signals, ambient temperature and humidity, and blade vibration data are collected in segments at the tip, middle, and root of the wind turbine blades. Real-time temperature and humidity data are thermodynamically matched with capacitance signals to filter out interference signals and calculate the segmented ice layer thickness. Vibration signal features are extracted to identify changes in mass distribution and aerodynamic anomalies caused by icing. Differentiated de-icing strategies are output based on the segmented icing status, and the critical water film formation moment is captured by the recovery of capacitance signals to assist in de-icing.

Images