Electric heating deicing system and method based on machine learning prediction and adaptive adjustment

Abstract

The invention discloses an electric heating deicing system and method based on machine learning prediction and adaptive adjustment, and belongs to the technical field of wind power generation. The system comprises a multi-physical sensor array and a dynamic heating execution module which are arranged on a blade. The multi-physical sensor array and the dynamic heating execution module are both connected with a CAN bus; the CAN bus is connected with an intelligent control module, a cloud collaboration platform and an edge computing node. The method comprises the following steps: acquiring historical operation data of an original unit, and preprocessing the historical operation data to obtain preprocessed data; the original unit historical operation data comprises environmental parameters and blade state monitoring data; analyzing the preprocessed data based on a machine learning algorithm, and establishing a blade icing prediction model; real-time environment parameters and blade state monitoring data are obtained, and the blade surface icing condition is predicted based on the blade icing prediction model; and determining the heating requirement of the blade based on the surface icing condition of the blade, and carrying out deicing.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wind power generation, and relates to an electric heating deicing system and method based on machine learning prediction and adaptive adjustment. BACKGROUND

[0002] In the field of wind power generation, wind turbine generators are exposed to complex and changeable natural environments for a long time, and when they operate in cold and humid areas, the problem of ice formation on the surface of blades has become a key technical bottleneck restricting the stable operation of wind turbine generators. Blade icing not only causes changes in aerodynamic shape, leading to a significant decrease in power generation efficiency, but more seriously, ice accumulation changes the mass distribution and stiffness characteristics of the blade, generating alternating stress during rotation and accelerating fatigue damage of the blade structure, significantly increasing operation and maintenance costs and safety risks. As the current mainstream active deicing solution, electric heating deicing technology melts ice layers by arranging heating elements on the surface of blades, and has the advantage of complete deicing compared with passive anti-icing solutions.

[0003] However, the traditional electric heating deicing system has the following technical defects: its control strategy is seriously lagging behind the actual icing state, and it mostly uses timed heating or simple threshold triggering mechanism, lacking intelligent sensing ability for the microclimate environment of the blade. This "blind control" mode causes the heating process to be out of sync with the actual icing process, resulting in a large amount of energy waste due to continuous heating during non-icing periods, and incomplete deicing due to insufficient heating power during the icing acceleration period. The deeper technical bottleneck is that the existing system does not have an icing prediction function and cannot predict icing risks in advance based on the trend of environmental parameters, always being in a passive response state. The heating control strategy is mostly open-loop control or simple PID adjustment, without establishing a dynamic mapping relationship between heating power and deicing effect, and over-heating or under-heating may occur under complex working conditions. This extensive control mode not only leads to high energy consumption of the system, but also accelerates the aging process of the blade material due to frequent and drastic temperature changes. SUMMARY

[0004] The purpose of the present application is to solve the technical problem that the control strategy of the blade electric heating in the prior art is seriously lagging behind the actual icing state, and to provide an electric heating deicing system and method based on machine learning prediction and adaptive adjustment.

[0005] In order to achieve the above purpose, the following technical solutions are adopted in the present application:

[0006] In a first aspect, the present application discloses an electric heating deicing system based on machine learning prediction and adaptive adjustment, comprising a multi-physical sensor array and a dynamic heating execution module arranged on a blade; the multi-physical sensor array and the dynamic heating execution module are both connected with a CAN bus; the CAN bus is connected with an intelligent control module, a cloud collaborative platform and an edge computing node.

[0007] Further improvements are that:

[0008] The intelligent control module comprises a machine learning prediction engine and an optimization controller; the machine learning prediction engine is used to receive the data processed by the edge computing node, generate a blade dynamic thermal field prediction model, and obtain blade predicted thermal field data; the optimization controller generates a multi-objective control parameter set based on the blade predicted thermal field data; and the cloud collaborative platform is integrated with a remote monitoring platform.

[0009] The machine learning prediction engine is an LSTM-GRU prediction engine, and the optimization controller is an NSGA-II optimization controller.

[0010] The multi-physical sensor array adopts an orthogonal layout and comprises a three-dimensional temperature field sensor group, a humidity sensor group, a flow field pressure sensor array, and a structure vibration sensor array; the three-dimensional temperature field sensor group comprises an infrared sensor and a thermocouple sensor.

[0011] The chordwise arrangement spacing of the multi-physical sensor array is 12-18 cm in the blade root area and 22-28 cm in the blade tip area; the multi-physical sensor array comprises a fiber grating and a microwave dielectric sensor; the wavelength resolution of the fiber grating is 1 pm; and the working frequency of the microwave dielectric sensor is 0.5-2.5 GHz, and the ice thickness resolution is 0.1 mm.

[0012] In a second aspect, the application discloses an electric heating deicing method based on machine learning prediction and adaptive adjustment based on the above system, comprising:

[0013] Original unit historical operation data are acquired, and preprocessed to obtain preprocessed data; the original unit historical operation data comprise environmental parameters and blade state monitoring data;

[0014] The preprocessed data are analyzed based on a machine learning algorithm, and a blade icing prediction model is established;

[0015] Real-time environmental parameters and blade state monitoring data are acquired, and a blade surface icing condition is predicted based on the blade icing prediction model;

[0016] The blade heating demand is determined based on the blade surface icing condition, and deicing is performed.

[0017] After deicing is performed based on the blade surface icing condition, the blade heating demand is determined, and the method further comprises:

[0018] The heating effect and energy consumption are determined based on the real-time environmental parameters and blade state monitoring data, and the heating strategy is optimized.

[0019] The heating effect and energy consumption are determined based on real-time environmental parameters and blade state monitoring data, and the heating strategy is optimized, and the specific heating strategy is:

[0020] By adjusting the time function of the heating power P(t), the total energy consumption of the entire deicing period t0 to the termination time t f is minimized; the temperature rise rate is limited to not more than 5℃ / min, so as to avoid thermal fatigue damage of the blade material due to excessive temperature difference; the remaining power of the energy storage system is maintained to be not less than 20%, so as to prevent the deicing process from being interrupted due to power depletion;

[0021]

[0022] Wherein, P(t) represents the heating power function; represents the cumulative minimum energy consumption of the electric heating element from the initial time t0 to the termination time t f ; T(x, t) represents the blade surface temperature field, and the temperature value of any position x of the blade surface at time t; T melt represents the ice layer melting temperature threshold; represents the temperature change rate; SOC represents the battery state of charge.

[0023] The icing situation of the blade surface includes icing speed, icing thickness and icing position information; the blade heating demand includes heating power, heating time and heating area.

[0024] 10. The machine learning-based prediction and adaptive adjustment electric heating deicing method according to claim 7, wherein the machine learning algorithm comprises a neural network algorithm, a random forest algorithm or a support vector machine algorithm.

[0025] Compared with the prior art, the present application has the following beneficial effects:

[0026] The application discloses an electric heating deicing system based on machine learning prediction and adaptive adjustment. Through the synergistic effect of a multi-physical sensor array and a dynamic heating execution module, combined with the architecture design of a CAN bus, an intelligent control module, a cloud collaborative platform and an edge computing node, the multi-physical sensor array is used to collect blade surface and environmental parameters (such as temperature, humidity, wind speed, etc.) in real time, thereby providing high-precision data input for the intelligent control module. A prediction model constructed in combination with a machine learning algorithm can predict the blade icing trend in advance, thereby realizing the technical leap from passive response to active prediction. This prediction mechanism enables the system to formulate a targeted heating strategy at the initial icing stage, thereby avoiding energy waste in the traditional timed heating or threshold triggering mode and significantly improving the accuracy of deicing operation. The dynamic heating execution module realizes the dynamic adaptation of the heating strategy through three-dimensional optimization of power regulation, heating time control and region selection based on the prediction results and real-time monitoring data. The system can distribute differentiated heating power according to the icing severity of different positions of the blade, thereby avoiding the energy consumption surge caused by global heating. The CAN bus is used to realize low-delay communication (≤10ms) of the sensor array, the execution module and the control module, thereby ensuring the real-time interaction of state data and control instructions. The edge computing node completes data preprocessing and preliminary decision-making locally, thereby reducing the cloud computing pressure and shortening the system response cycle. The cloud collaborative platform realizes remote model updating and system state monitoring through digital twinning technology, thereby forming a double closed-loop control architecture of "edge-cloud collaboration", and improving the adaptability and stability of the system under complex working conditions.

[0027] Further, by setting different chordwise spacing in the root area (12-18 cm) and tip area (22-28 cm), the strain characteristics of different areas of the blade are accurately matched. The root area, as the area with the largest bending moment, uses a dense arrangement spacing of 12-18 cm to capture the complex aeroelastic deformation of the blade root; the tip area, due to the large amplitude of flapping and edgewise motion, uses a spacing of 22-28 cm to ensure monitoring coverage while avoiding signal coupling interference caused by excessive sensor density. This gradient layout deeply couples the sensor network with the mechanical characteristics of the blade, improving the spatial resolution of ice monitoring. The composite monitoring architecture of fiber Bragg grating (FBG) and microwave dielectric sensor forms complementary monitoring capabilities. Based on the wavelength coding principle, 1 pm wavelength resolution can sense 0.1με-level micro-strain on the blade surface, accurately capturing the modal changes of the blade caused by ice layer attachment, providing nanoscale deformation monitoring capability for early warning of icing. The 0.5-2.5 GHz operating frequency band covers the ice layer dielectric constant sensitive frequency band, and the 0.1 mm ice thickness resolution can quantify the ice layer accumulation process. Its non-contact measurement characteristics avoid interference with the aerodynamic performance of the blade. The 1 pm wavelength resolution of the fiber Bragg grating corresponds to a temperature sensitivity of 0.1℃ / pm, combined with the 0.1 mm ice thickness detection threshold of the microwave sensor, a dual-dimensional monitoring system of "micro-strain-dielectric properties" is constructed. This system can detect as little as 0.01 mm of initial ice crystal attachment, improving the monitoring accuracy by two orders of magnitude compared to traditional icing sensors (usually ≥0.5 mm resolution), providing earlier icing feature input for machine learning models. The intrinsic electromagnetic interference prevention characteristics of the fiber optic sensor combined with the all-weather working capability of the microwave sensor make the system maintain monitoring stability in complex weather conditions such as strong electromagnetic environments (e.g., thunderstorm weather), rain and fog weather, etc. The selection of the 0.5-2.5 GHz frequency band of the microwave sensor effectively avoids the signal attenuation problem caused by water film coverage, ensuring continuous monitoring of the ice-water phase change process.

[0028] The application discloses an electric heating deicing method based on machine learning prediction and adaptive adjustment, and creatively introduces a machine learning algorithm into a blade icing prediction field, and breaks through the limitation of traditional threshold judgment. Through construction of a high-precision prediction model, time and space correlation characteristics of environmental parameters and blade state monitoring data are deeply mined, and accurate description of a blade icing process is realized. The model can capture weak signs in an early icing stage, provides prospective guidance for deicing decision, and significantly improves the timeliness and pertinence of deicing operation. A closed-loop control system of prediction-decision-execution is constructed, and dynamic optimization of a heating strategy is realized. Based on an icing trend output by the prediction model, combined with real-time state monitoring data of the blade, the system can intelligently generate a differentiated heating scheme, including power adjustment, heating area selection and heating time length control. The adaptive adjustment mechanism ensures that heating energy is accurately put into key areas, avoids energy waste caused by global heating in a traditional scheme, and prevents material damage caused by local overheating. Through collaborative work of an edge computing node and a cloud platform, organic combination of local real-time response and global strategy optimization is realized. A light-weight algorithm model is deployed on the edge side, ensuring fast decision-making ability in a harsh environment; and a digital twin system is constructed on the cloud side, supporting complex strategy verification and model continuous evolution. The dual-mode architecture guarantees system response speed and endows it with self-learning evolution ability, and significantly improves operation intelligent level. An integrated remote monitoring platform provides a panoramic state perception interface for operation personnel, and visually displays blade temperature field, stress field and icing distribution through three-dimensional visualization technology. An intelligent decision support system automatically generates optimization suggestions and pushes early warning information based on prediction results and equipment states, so that artificial intervention is changed from "passive response" to "active decision", and operation efficiency and decision quality are greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0030] Figure 1 It is an architecture diagram of an electric heating deicing system based on machine learning prediction and adaptive adjustment in the embodiments of the present application.

[0031] Figure 2 It is a high-precision environmental monitoring process schematic diagram in the electric heating deicing method based on machine learning prediction and adaptive adjustment in the embodiments of the present application.

[0032] Figure 3 It is a machine learning prediction process schematic diagram in the electric heating deicing method based on machine learning prediction and adaptive adjustment in the embodiments of the present application.

[0033] Figure 4 Figure 1 is a schematic diagram of an adaptive dynamic adjustment process in an embodiment of the present application for a method of electric heating deicing based on machine learning prediction and adaptive adjustment.

[0034] Figure 5 Figure 2 is a dynamic optimization control diagram in an embodiment of the present application for a method of electric heating deicing based on machine learning prediction and adaptive adjustment. DETAILED DESCRIPTION

[0035] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions 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 some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0036] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0037] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0038] In the description of the embodiments of the present application, it should be noted that if the terms "upper", "lower", "horizontal", "inner" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the present application is usually placed, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the indicated device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" and the like are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0039] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly inclined. For example, "horizontal" only means that its direction is relatively more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly inclined.

[0040] In the description of the embodiments of the present application, it also needs to be explained that, unless otherwise explicitly specified and limited, if the terms "arrange", "install", "connect", "connect" appear, they should be understood in a broad sense, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0041] The present application will be further described in detail below with reference to the accompanying drawings:

[0042] Referring to Figure 1 , the present application discloses an electric heating deicing system based on machine learning prediction and adaptive adjustment, comprising a multi-physical sensor array and a dynamic heating execution module arranged on the blade; the multi-physical sensor array and the dynamic heating execution module are both connected with a CAN bus; the CAN bus is connected with an intelligent control module, a cloud collaborative platform and an edge computing node. The intelligent control module comprises a machine learning prediction engine and an optimization controller; the machine learning prediction engine is used to receive the data processed by the edge computing node, generate a blade dynamic thermal field prediction model, and obtain blade predicted thermal field data; the optimization controller generates a multi-objective control parameter set based on the blade predicted thermal field data; the cloud collaborative platform is integrated with a remote monitoring platform. The machine learning prediction engine is an LSTM-GRU prediction engine; the optimization controller is an NSGA-II optimization controller.

[0043] Referring to Figure 2 , the multi-physical sensor array adopts an orthogonal layout, comprising a three-dimensional temperature field sensor group, a humidity sensor group, a flow field pressure sensor array and a structure vibration sensor array; the three-dimensional temperature field sensor group comprises an infrared sensor and a thermocouple sensor. Alternatively, the chordwise arrangement spacing of the multi-physical sensor array is 12-18 cm in the blade root area and 22-28 cm in the blade tip area; the multi-physical sensor array comprises a fiber grating and a microwave dielectric sensor; the wavelength resolution of the fiber grating is 1 pm; the working frequency of the microwave dielectric sensor is 0.5-2.5 GHz, and the ice thickness resolution is 0.1 mm. The system uses a variety of high-precision sensors to monitor the temperature, humidity, wind speed, wind direction and other key parameters of the blade surface and the surrounding environment in real time. These sensors are carefully arranged in the key parts of the blade and the surrounding environment to ensure the accuracy and comprehensiveness of the data. Through real-time monitoring, the system can obtain the real-time condition of the blade icing, providing high-quality data input for subsequent machine learning prediction.

[0044] The application discloses an electric heating deicing system based on machine learning prediction and adaptive adjustment, which realizes real-time collection of blade surface and environmental parameters (such as temperature, humidity, wind speed, etc.) through a multi-physical sensor array in cooperation with a dynamic heating execution module, and provides high-precision data input for an intelligent control module in combination with a CAN bus, an intelligent control module, a cloud collaborative platform and an edge computing node architecture design. A prediction model constructed in combination with a machine learning algorithm can predict the blade icing trend in advance, realizing a technical leap from passive response to active prediction. This prediction mechanism enables the system to formulate a targeted heating strategy at the initial icing stage, avoiding energy waste in the traditional timed heating or threshold triggering mode, and significantly improving the accuracy of deicing operation. The dynamic heating execution module realizes dynamic adaptation of the heating strategy through three-dimensional optimization of power regulation, heating time control and region selection based on the prediction results and real-time monitoring data. The system can allocate differentiated heating power according to the icing severity of different positions of the blade, avoiding the energy consumption surge caused by global heating. The CAN bus is used to realize low-delay communication (≤10ms) of the sensor array, the execution module and the control module, ensuring real-time interaction of state data and control instructions. The edge computing node completes data preprocessing and preliminary decision-making locally, reducing the cloud computing pressure and shortening the system response cycle. The cloud collaborative platform realizes remote model updating and system state monitoring through digital twinning technology, forming a double closed-loop control architecture of "edge-cloud collaboration", and improving the adaptability and stability of the system under complex working conditions.

[0045] Referring to Figure 3 、 Figure 4 and Figure 5 , the application discloses an electric heating deicing method based on machine learning prediction and adaptive adjustment, comprising:

[0046] Step one, obtaining original unit historical operation data, and preprocessing the original unit historical operation data to obtain preprocessed data; the original unit historical operation data includes environmental parameters and blade state monitoring data;

[0047] Step two, analyzing the preprocessed data based on a machine learning algorithm, and establishing a blade icing prediction model; the machine learning algorithm includes a neural network algorithm, a random forest algorithm or a support vector machine algorithm. The system analyzes historical data and real-time monitoring data by using a machine learning algorithm to construct a blade icing prediction model. The model can predict the icing condition of the blade in the future, including icing speed, icing thickness, icing position, etc., and the time point and intensity of deicing demand. Through machine learning prediction, the system can formulate a deicing strategy in advance to avoid excessive heating and energy waste.

[0048] Step three, obtain real-time environmental parameters and blade state monitoring data, and predict the icing condition of the blade surface based on the blade icing prediction model; the icing condition of the blade surface includes icing speed, icing thickness and icing position information;

[0049] Step four, determine the blade heating requirement based on the icing condition of the blade surface, and perform de-icing. The blade heating requirement includes heating power, heating time and heating area.

[0050] Step five, determine the heating effect and energy consumption based on real-time environmental parameters and blade state monitoring data, and optimize the heating strategy. Based on the prediction results of machine learning, the system can adaptively adjust the heating strategy of the electric heating element. This includes the adjustment of heating power, the control of heating time and the selection of heating area, etc. The system can automatically determine the optimal heating scheme according to the prediction results and real-time monitoring data, to ensure ice melting at the best time and in the best way. At the same time, the system can continuously optimize the heating strategy according to the de-icing effect and energy consumption feedback, to realize adaptive dynamic adjustment and improve de-icing efficiency and energy efficiency.

[0051] By adjusting the time function of heating power P(t), the total energy consumption from the de-icing period t0 to the end time t f is minimized; the temperature rise rate is limited to not more than 5℃ / min to avoid thermal fatigue damage of the blade material due to large temperature difference; the remaining energy of the energy storage system is maintained to be not less than 20% to prevent the de-icing process from being interrupted due to energy depletion;

[0052]

[0053] Solution process: 1. Initialize the population: generate an initial solution set based on historical strategies; 2. Non-dominated sorting: calculate the Pareto front level; 3. Genetic operation: crossover rate 0.8, mutation rate 0.02; 4. Elite retention: retain the top 10% optimal individuals.

[0054] wherein P(t) represents the heating power function; represents the cumulative minimum energy consumption of the electric heating element from the initial time t0 to the end time t f ; T(x, t) represents the blade surface temperature field, the temperature value of any position x on the blade surface at time t; T melt represents the ice layer melting temperature threshold; represents the temperature change rate; SOC represents the state of charge of the battery.

[0055] The system integrates a remote monitoring platform, realizing real-time monitoring and data analysis of the system running state. Operation and maintenance personnel can view real-time data, prediction results, heating strategies and other information through the remote monitoring platform, and understand the working state of the system in a timely manner. In addition, the platform also provides intelligent decision support function, which can provide decision suggestions for operation and maintenance personnel according to data analysis and prediction results, improve operation and maintenance efficiency and accuracy.

[0056] The application discloses an electric heating deicing method based on machine learning prediction and adaptive adjustment, which introduces machine learning algorithm into the field of blade icing prediction for the first time, breaking through the limitation of traditional threshold judgment. By constructing a high-precision prediction model, the spatio-temporal correlation characteristics of environmental parameters and blade state monitoring data are deeply mined to realize accurate description of the blade icing process. The model can capture the early signs of icing and provide forward-looking guidance for deicing decisions, significantly improving the timeliness and pertinence of deicing operations. The system constructs a closed-loop control system of prediction-decision-execution to realize dynamic optimization of heating strategies. Based on the icing trend output by the prediction model and combined with real-time blade state monitoring data, the system can intelligently generate differentiated heating schemes, including power adjustment, heating area selection and heating duration control. This adaptive adjustment mechanism ensures that heating energy is accurately delivered to key areas, avoiding energy waste caused by global heating in traditional schemes, while preventing material damage caused by local overheating. Through the collaborative work of edge computing nodes and cloud platforms, local real-time response and global strategy optimization are organically combined. Lightweight algorithm models are deployed on the edge side to ensure fast decision-making ability in harsh environments; the cloud builds a digital twin system to support complex strategy verification and model evolution. This dual-mode architecture not only ensures system response speed, but also endows it with self-learning evolution ability, significantly improving the level of operation intelligence. The integrated remote monitoring platform provides a panoramic state perception interface for operation and maintenance personnel, visually displaying blade temperature field, stress field and icing distribution through three-dimensional visualization technology. The intelligent decision support system automatically generates optimization suggestions and pushes warning information based on prediction results and device status, enabling artificial intervention to change from "passive response" to "active decision", significantly improving operation efficiency and decision quality.

[0057] The working principle of the application is as follows:

[0058] Precise prediction based on machine learning: The application first applies machine learning algorithm to the field of wind turbine blade deicing, realizes accurate prediction of blade icing conditions and deicing needs through the construction of a high-precision prediction model, and improves deicing efficiency and energy efficiency.

[0059] Adaptive dynamic adjustment technology: The system can automatically adjust the heating strategy according to the prediction results and real-time data, and continuously optimize the performance. This adaptive dynamic adjustment technology enables the system to flexibly respond to environmental changes, ensuring that ice melting is done at the best time and in the best way.

[0060] Remote monitoring and intelligent decision support: The integrated remote monitoring platform and intelligent decision support system provide an intuitive operation interface and decision basis for operation and maintenance personnel, improving operation and maintenance efficiency and accuracy. At the same time, the platform can also realize data analysis and fault warning, and timely discover and handle potential problems.

[0061] The above is only the preferred embodiment of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An electric heating deicing system based on machine learning prediction and adaptive regulation, characterized in that: It includes a multi-physical sensor array and a dynamic heating execution module arranged on the blade; the multi-physical sensor array and the dynamic heating execution module are both connected to a CAN bus; the CAN bus is connected to an intelligent control module, a cloud collaboration platform and an edge computing node.

2. The electric heating deicing system based on machine learning prediction and adaptive regulation according to claim 1 is characterized in that: The intelligent control module includes a machine learning prediction engine and an optimization controller; the machine learning prediction engine is used to receive data processed by the edge computing node, generate a blade dynamic thermal field prediction model, and obtain blade predicted thermal field data; the optimization controller generates a multi-objective control parameter set based on the blade predicted thermal field data; the cloud-based collaborative platform is integrated with a remote monitoring platform.

3. The electric heating deicing system based on machine learning prediction and adaptive regulation according to claim 2 is characterized in that: The machine learning prediction engine is an LSTM-GRU prediction engine; the optimization controller is an NSGA-II optimization controller.

4. The electric heating deicing system based on machine learning prediction and adaptive regulation according to claim 1 is characterized in that: The multi-physical sensor array adopts an orthogonal layout, including a three-dimensional temperature field sensor group, a humidity sensor group, a flow field pressure sensor array and a structural vibration sensor array; the three-dimensional temperature field sensor group includes an infrared sensor and a thermocouple sensor.

5. The electric heating deicing system based on machine learning prediction and adaptive regulation according to claim 1 is characterized in that: The chord-wise spacing of the multi-physics sensor array is 12-18 cm in the blade root area and 22-28 cm in the blade tip area. The multi-physics sensor array includes fiber grating (FBG) and microwave dielectric sensors. The wavelength resolution of the fiber grating is 1 pm. The operating frequency of the microwave dielectric sensor is 0.5-2.5 GHz, and the ice thickness resolution is 0.1 mm.

6. An electric heating deicing method based on machine learning prediction and adaptive regulation based on the system according to any one of claims 1 to 5, characterized in that: include: Acquire original historical operation data of the unit and preprocess it to obtain preprocessed data; the original historical operation data of the unit includes environmental parameters and blade status monitoring data; Analyzing the preprocessed data based on a machine learning algorithm to establish a blade icing prediction model; Acquire real-time environmental parameters and blade status monitoring data, and predict blade surface icing conditions based on the blade icing prediction model; The blade heating requirement is determined based on the ice coverage condition on the blade surface, and de-icing is performed.

7. The electric heating deicing method based on machine learning prediction and adaptive regulation according to claim 6 is characterized in that: Determining blade heating requirements based on the ice coverage on the blade surface, and performing deicing further includes: Determine heating effect and energy consumption based on real-time environmental parameters and blade status monitoring data, and optimize heating strategy.

8. The electric heating deicing method based on machine learning prediction and adaptive regulation according to claim 7 is characterized in that: The method of determining the heating effect and energy consumption based on real-time environmental parameters and blade status monitoring data and optimizing the heating strategy is as follows: By adjusting the time function of the heating power P(t), the entire deicing cycle t0 to the end time t f Minimize total energy consumption; limit the temperature rise rate to no more than 5°C / minute to avoid thermal fatigue damage to the blade material due to excessive temperature differences; maintain the remaining power of the energy storage system at no less than 20% to prevent the de-icing process from being interrupted due to power depletion; Where P(t) represents the heating power function; Represents the time from the initial time t0 to the end time t f The minimum energy consumption of the electric heating element is accumulated; T(x,t) represents the temperature field on the blade surface, which is the temperature value at any position x on the blade surface at time t; T melt Indicates the ice melting temperature threshold; Indicates the rate of temperature change; SOC indicates the battery state of charge.

9. The electric heating deicing method based on machine learning prediction and adaptive regulation according to claim 7, characterized in that: The blade icing condition includes icing speed, icing thickness and icing position information; the blade heating requirement includes heating power, heating time and heating area.

10. The electric heating deicing method based on machine learning prediction and adaptive regulation according to claim 7, characterized in that: The machine learning algorithm includes a neural network algorithm, a random forest algorithm or a support vector machine algorithm.

Images