Wind turbine generator health state diagnosis system and method

Abstract

The invention relates to the technical field of wind power generation, in particular to a wind turbine generator health state diagnosis system and method. The system comprises a monitoring module used for acquiring multi-source data by using a preset sensor group; the calculation module is used for receiving the multi-source data and performing feature extraction operation on the multi-source data to obtain multi-source feature information; and the analysis module is used for fusing the multi-source feature information and evaluating the health state of the generator set based on a fusion result. Therefore, through the generator set health state diagnosis system, multi-sensor fusion, edge calculation and cloud intelligent analysis are integrated, the problem that the prior art is lack of system-level health state evaluation capability and intelligent diagnosis capability is solved, and real-time monitoring and early fault early warning of the health state of the whole wind turbine generator system are realized.

Description

Technical Field

[0001] This application relates to the field of wind power generation technology, and in particular to a health status diagnosis system and method for wind turbine generators. Background Technology

[0003] In related technologies, single-point monitoring systems (such as gearbox vibration monitoring and tower stress monitoring) are typically used to operate independently to achieve health monitoring of wind turbine units. However, this method uses a single sensor to monitor a specific component, which easily leads to "data silos," lacks system-level health status assessment capabilities, and lacks intelligent diagnostic capabilities, which urgently need to be addressed. Summary of the Invention

[0004] This application provides a wind turbine health status diagnosis system and method to solve the problems of the lack of system-level health status assessment capability and the lack of intelligent diagnosis capability in the prior art, so as to realize real-time monitoring of the health status of the entire wind turbine system and early fault warning.

[0005] To achieve the above objectives, a first aspect of this application proposes a wind turbine health status diagnosis system, comprising: a monitoring module, an edge computing module, and an analysis module, wherein...

[0006] The monitoring module is used to acquire multi-source data using a preset sensor group;

[0007] The computing module is used to receive the multi-source data and perform feature extraction operations on the multi-source data to obtain multi-source feature information;

[0008] The analysis module is used to fuse the multi-source feature information and evaluate the health status of the wind turbine based on the fusion results.

[0009] According to one embodiment of this application, the monitoring module includes: a support system monitoring unit, the support system monitoring unit comprising an ultrasonic sensor array and a fiber optic grating sensor network, wherein...

[0010] The ultrasonic sensor array is used to monitor the stress of the tower connecting bolts;

[0011] The fiber optic grating sensor network is used to monitor the stress on the tower body.

[0012] According to one embodiment of this application, the monitoring module further includes: a rotation system monitoring unit, the rotation system monitoring unit comprising a fiber optic grating array integrated inside the wind turbine blades and a triaxial vibration sensor deployed in the hub, wherein...

[0013] The fiber grating array is used to determine the damage status of the blade structure using a preset damage identification algorithm;

[0014] The triaxial vibration sensor is used to acquire vibration data of the wheel hub area, and the preset damage identification algorithm is used to analyze the vibration data to determine the abnormal conditions of the wheel hub area.

[0015] According to one embodiment of this application, the monitoring module further includes: a transmission system monitoring unit, which includes a high-frequency acceleration sensor, a broadband acoustic emission sensor, and an oil parameter monitoring subunit, wherein...

[0016] The high-frequency acceleration sensor is deployed at the gearbox input / output end and the spindle bearing;

[0017] The broadband acoustic emission sensor is mounted on the gearbox housing;

[0018] The oil parameter monitoring subunit includes a particle counter, a viscosity sensor, and a moisture sensor, used to monitor the physical and chemical parameters of lubricating oil or hydraulic oil in real time.

[0019] According to one embodiment of this application, the monitoring module further includes: a yaw system monitoring unit, which comprises a yaw gear wear monitoring subunit, a yaw bearing infrared imaging subunit, and a dual-mode positioning subunit, wherein...

[0020] The yaw gear wear monitoring subunit includes at least one of a high-definition industrial camera and an LED ring light source.

[0021] According to one embodiment of this application, the monitoring module further includes:

[0022] The brake system monitoring unit includes at least one of a fiber optic grating temperature sensor, a hydraulic line pressure sensor, a brake caliper displacement sensor, and an ultrasonic brake pad thickness sensor.

[0023] According to one embodiment of this application, the monitoring module further includes:

[0024] A cooling and lubrication system monitoring unit, comprising at least one of an online oil quality multi-parameter sensor, a turbine flow meter, an infrared thermocouple array, and an inductive particle counter.

[0025] According to one embodiment of this application, the monitoring module further includes:

[0026] The electronic control system monitoring unit includes at least one of a high-frequency current transformer, a distributed optical fiber temperature sensing subunit, and a combustible particulate matter sensor.

[0027] According to one embodiment of this application, the analysis module includes:

[0028] A wind turbine digital twin unit, which is used to simulate the state of a wind turbine under different preset physical conditions;

[0029] A lifespan prediction unit, which integrates a preset neural network model, is used to predict the remaining lifespan of the wind turbine.

[0030] The fault knowledge graph unit is used to determine the fault type of the wind turbine and provide corresponding fault handling suggestions to the operator.

[0031] An alarm unit is provided, which is used to set an alarm threshold based on the real-time ambient temperature, and to trigger an alarm signal when the current operating temperature of the fan exceeds the alarm threshold, so as to notify the operator to take appropriate measures.

[0032] The wind turbine health status diagnosis system proposed in this application involves a monitoring module that acquires multi-source data using a pre-set sensor group; a computing module that receives the multi-source data and performs feature extraction on it to obtain multi-source feature information; and an analysis module that fuses the multi-source feature information and assesses the health status of the wind turbine based on the fusion result. Thus, this wind turbine health status diagnosis system integrates multi-sensor fusion, edge computing, and cloud-based intelligent analysis, solving the problems of existing technologies lacking system-level health status assessment capabilities and intelligent diagnostic capabilities, and achieving real-time monitoring and early fault warning of the entire wind turbine system's health status.

[0033] To achieve the above objectives, a second aspect of this application proposes a method for diagnosing the health status of a wind turbine generator, employing the wind turbine generator health status diagnosis system described in the first aspect embodiment. The method includes the following steps:

[0034] Acquire multi-source data and perform feature extraction on the multi-source data to obtain multi-source feature information;

[0035] The multi-source feature information is fused, and the health status of the wind turbine is evaluated based on the fusion result to obtain the evaluation result;

[0036] Based on the assessment results, predictive wind turbine maintenance recommendations are generated and then visualized.

[0037] The wind turbine health status diagnosis method proposed in the embodiments of this application integrates multi-sensor fusion, edge computing and cloud intelligent analysis through a wind turbine health status diagnosis system, which solves the problems of the lack of system-level health status assessment capability and the lack of intelligent diagnosis capability in the prior art, and realizes real-time monitoring and early fault warning of the entire wind turbine system health status.

[0038] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0039] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0040] Figure 1 This is a block diagram of a wind turbine health status diagnosis system according to an embodiment of this application;

[0041] Figure 2 This is a flowchart of a wind turbine health status diagnosis method provided according to an embodiment of this application. Detailed Implementation

[0042] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0043] The following describes the wind turbine health status diagnosis system and method according to the embodiments of this application with reference to the accompanying drawings. First, the wind turbine health status diagnosis system according to the embodiments of this application will be described with reference to the accompanying drawings.

[0044] Figure 1 This is a block diagram of a wind turbine health status diagnosis system according to an embodiment of this application.

[0045] For example, such as Figure 1 As shown, the wind turbine health status diagnosis system 10 includes a monitoring module 100, an edge computing module 200, and an analysis module 300. The monitoring module 100 is used to acquire multi-source data using a preset sensor group; the computing module 200 is used to receive multi-source data and perform feature extraction operations on the multi-source data to obtain multi-source feature information; the analysis module 300 is used to fuse the multi-source feature information and evaluate the health status of the wind turbine based on the fusion result.

[0046] Specifically, the monitoring module 100 is a multi-sensor fusion network that can achieve "holographic perception" of the wind turbine's operating status using a multi-sensor group. This module can align multi-source data to the microsecond level through spatiotemporal synchronization technology (PTP (Precision Time Protocol) + GPS (Global Positioning System) timing), providing high-fidelity raw signal input to the computing module 200. The computing module 200 is an edge computing node that can adopt industrial-grade FPGA (Field-Programmable Gate Array) technology. This high-performance hardware platform can be flexibly programmed and configured to adapt to different computing needs. On this node, a lightweight 1D-CNN (One-Dimensional Convolutional Neural Network) model runs. This model is specifically optimized for edge computing environments and can efficiently process data under resource constraints. Through the application of this model, the edge computing node has successfully achieved a data compression ratio of over 90%, greatly reducing the burden of data transmission and improving the overall efficiency of the system.

[0047] Furthermore, this edge computing node supports multiple industrial communication protocols, including the Modbus / TCP network communication protocol (an industrial communication protocol based on TCP / IP (Transmission Control Protocol / Internet Protocol), widely used for data exchange between devices) and the OPC UA (Open Platform Communications-Unified Architecture) protocol. Support for these protocols ensures seamless compatibility between the edge computing node and existing SCADA (Supervisory Control and Data Acquisition) systems, enabling smooth data flow and collaborative system operation. Through this compatible design, the edge computing node can not only efficiently process and compress data but also easily integrate into existing industrial control systems, providing strong technical support for industrial automation and intelligence.

[0048] This edge computing node is deployed inside the wind turbine nacelle. Its main function is to process various types of data (multi-source data) collected during wind turbine operation in real time, and to perform feature extraction to obtain multi-source feature information. This deployment method ensures that data is processed promptly at its source, avoiding data transmission delays, improving data processing efficiency and accuracy, and providing strong technical support for the stable operation and intelligent maintenance of wind turbines. Feature extraction is the process of filtering out important information from raw data. For example, during wind turbine operation, sensors record a large amount of data. Through feature extraction, the parameters most relevant to the health status of the wind turbine can be identified, thereby optimizing maintenance strategies. The analysis module 300 is a cloud-based intelligent analysis platform. This module can transform "fragmented data" into "interpretable health knowledge." That is, it organically integrates diverse and multi-dimensional feature data collected by various sensors, and based on the comprehensive information obtained from the fusion (fusion result), it further conducts detailed analysis on various key performance indicators of wind turbine units, identifies potential health hazards, and quantifies their severity. Finally, it can output a detailed health status report, providing strong support for the operation and maintenance decisions of wind farms and ensuring the safe and stable operation of wind turbine units.

[0049] It should be noted that the monitoring module 100 may consist of a support system monitoring unit, a rotation system monitoring unit, a transmission system monitoring unit, a yaw system monitoring unit, a braking system monitoring unit, a cooling and lubrication system monitoring unit, and an electronic control system monitoring unit. These monitoring units are equipped with different types of sensors to monitor the operating status of each part of the wind turbine 10 in real time. For ease of understanding, these units will be described in detail below.

[0050] Optionally, in some embodiments, the monitoring module 100 includes: a support system monitoring unit, which includes an ultrasonic sensor array and a fiber Bragg grating sensor network, wherein the ultrasonic sensor array is used to monitor the stress of the tower connecting bolts; and the fiber Bragg grating sensor network is used to monitor the stress of the tower body.

[0051] Understandably, the tower is a crucial component of a wind turbine, typically a tall, cylindrical structure supporting the nacelle and blades. To ensure the safe operation of the wind turbine, the tower must withstand various loads (such as wind loads and gravity loads), making the reliability and stress state of its connecting components (such as bolts) critical. Connecting bolts are fasteners used to secure different parts of the tower together (such as connections between flanges). Due to the tower's long-term exposure to harsh environments (such as strong winds, vibrations, and temperature changes), connecting bolts may loosen, develop fatigue cracks, or experience stress concentration, affecting the overall stability of the tower. A fiber optic grating sensor is a sensor based on the principle of fiber Bragg gratings. It uses a grating structure with periodically varying refractive indices etched into an optical fiber to reflect light of specific wavelengths while transmitting other wavelengths. When external environmental factors (such as strain, temperature, and pressure) change, the wavelength of the reflected light shifts, enabling high-precision measurement of external physical quantities. Fiber Bragg grating sensor networks are distributed monitoring systems composed of multiple fiber Bragg grating sensors. They can acquire multi-point data over a wide area in real time and are suitable for health monitoring of large structures.

[0052] Specifically, the support system monitoring unit follows the principle of "point-surface combination and dynamic-static coordination," constructing a stress field monitoring system around the tower, the core of the support. That is, the support system monitoring unit can include an ultrasonic sensor array and a fiber optic grating sensor network. The ultrasonic sensor array is used for point stress capture; at the tower flange connection, a miniature piezoelectric ultrasonic sensor can be embedded in the end face of each high-strength bolt, forming an 8×8 array layout. The sensor sampling frequency is ≥1MHz. Utilizing the propagation characteristics of ultrasound in materials, the stress distribution inside the connecting bolts can be non-destructively assessed by analyzing changes in the echo signal. The fiber optic grating sensor network is used for surface stress sensing. A distributed fiber optic grating sensor chain is spirally laid along the tower height direction. For example, a measuring point is set every 2m (monitoring point spacing ≤2m), forming 120 strain detection nodes. The fiber core diameter can be 9μm, and the grating center wavelength can be 1550nm. The circumferential / axial strain field of the tower is analyzed in real time by measuring the wavelength drift.

[0053] Furthermore, the relevant data from the ultrasonic sensor array and the fiber optic grating sensor network can be spatiotemporally aligned and fused for analysis using the computing module 200. Specifically, the PTP protocol enables microsecond-level timescale unification, eliminating clock drift between ultrasonic waves (MHz sampling) and fiber optic gratings (kHz sampling). By establishing a three-dimensional coordinate system for the tower, bolt point stress can be mapped to the boundary conditions of the tower wall stress field, and Kalman filtering is used to eliminate measurement noise caused by structural vibration. The correlation feature between "bolt stress concentration factor and tower strain gradient" is extracted. Under normal operating conditions, the bolt stress concentration factor and tower strain gradient maintain a fixed ratio (e.g., 1:3). The system can calculate the average value (baseline) and fluctuation range (standard deviation) of the bolt stress concentration factor, as well as the average value (baseline) and fluctuation range (standard deviation) of the tower strain gradient, based on normal data from a past period (e.g., one year). When a bolt stress increase is observed at a certain time point, but the tower strain gradient does not increase proportionally, resulting in a ratio exceeding 1.5 times the historical baseline standard deviation, it can be determined that the tower structure stiffness has degraded.

[0054] Optionally, in some embodiments, the monitoring module 100 further includes: a rotation system monitoring unit, which includes a fiber optic grating array integrated inside the wind turbine blade and a triaxial vibration sensor deployed in the hub. The fiber optic grating array is used to determine the damage status of the blade structure using a preset damage identification algorithm; the triaxial vibration sensor is used to acquire vibration data at the hub and analyze the vibration data using the preset damage identification algorithm to determine any abnormalities at the hub.

[0055] Specifically, the rotation system monitoring unit is primarily responsible for accurately monitoring the real-time status of the wind turbine's rotation system. It consists of two main parts: First, a fiber optic grating array integrated inside the wind turbine blades. This array, embedded within the blades, provides a quasi-static strain field with millimeter-level resolution (≤10cm) in the spatial domain, enabling highly sensitive detection of blade stress and deformation through fiber optic technology. When the blades are subjected to loads (such as wind pressure or fatigue), the wavelength of the fiber optic grating array shifts with strain changes; the damage location can be pinpointed by demodulating the wavelength change. Second, a triaxial vibration sensor deployed at the hub center. This sensor monitors X / Y / Z-axis vibration, capturing a wideband dynamic response of 0.5-10kHz in the time-frequency domain, comprehensively capturing vibration data of the hub in three dimensions (characteristic frequencies of blade rotation, imbalance, bearing failure, etc.). These two parts work together to ensure the stability and safety of the rotation system under various operating conditions, providing solid data support for the optimized operation of the wind turbine. The frequency response range of 0.5-10kHz fully encompasses the blade passage frequency, shaft rotation frequency and its harmonics, gear meshing frequency, and bearing fault characteristic frequency. It can simultaneously capture two types of signals: low-frequency structural modes (shimmy, flapping) and high-frequency impacts (stripping, cracking), enabling broadband identification of early defects. The vibration sensor uses a three-dimensional orthogonal layout to acquire radial, tangential, and axial accelerations in the hub coordinate system. Through vector synthesis, the true direction and amplitude of the vibration can be reconstructed, avoiding directional coupling errors of single-axis sensors in the rotating reference frame.

[0056] Furthermore, fiber optic grating arrays can accurately locate the spatial coordinates of cracks (e.g., 37m from the blade root), vibration sensors can identify the fault excitation source (impact frequency of hub bearing = shaft rotation frequency × number of rollers), and using preset damage identification algorithms (e.g., wavelet packet energy entropy damage identification algorithm), the vibration signal is decomposed into different frequency bands through wavelet packet decomposition, detailed features are extracted, and the energy distribution of each frequency band is analyzed by energy entropy calculation. Sudden changes in entropy value usually correspond to structural damage or mechanical failure, ultimately achieving millimeter-level location, micrometer-level quantization, and hourly prediction of crack-impact coupling faults.

[0057] Optionally, in some embodiments, the monitoring module 100 further includes: a transmission system monitoring unit, which includes a high-frequency acceleration sensor, a broadband acoustic emission sensor, and an oil parameter monitoring subunit. The high-frequency acceleration sensor is deployed at the gearbox input / output end and the spindle bearing; the broadband acoustic emission sensor is installed in the gearbox housing; and the oil parameter monitoring subunit includes a particle counter, a viscosity sensor, and a moisture sensor for real-time monitoring of the physical and chemical parameters of the lubricating oil or hydraulic oil.

[0058] Understandably, the transmission system monitoring unit utilizes multi-sensor collaborative monitoring and intelligent analysis technology to achieve high-precision, real-time monitoring and early fault warning for key components such as gearboxes and spindle bearings. Specifically, the transmission system monitoring unit includes a high-frequency accelerometer, a broadband acoustic emission sensor, and an oil parameter monitoring subunit. The high-frequency accelerometer is deployed at the gearbox input / output ends and the spindle bearings, with a sampling frequency ≥40kHz, capable of capturing high-frequency vibration signals such as gear meshing and bearing rolling. For example, when pitting or tooth breakage occurs in the gears, the high-frequency accelerometer can detect characteristic frequencies (such as the sidebands of the meshing frequency). Combined with envelope demodulation technology, it can effectively separate the modulation components in the fault signal, thereby identifying early micro-damage. The broadband acoustic emission sensor is installed in the gearbox housing, covering a frequency band of 100-500kHz, and is used to monitor stress wave signals generated by gears and bearings during operation. Acoustic emission technology is extremely sensitive to localized dynamic defects such as crack propagation and frictional wear. For example, when microcracks appear in the bearing raceway, the acoustic emission signal will exhibit sudden waveform characteristics, and the fault location can be determined through time-frequency analysis. The oil parameter monitoring subunit includes a particle counter, viscosity sensor, and moisture sensor, which monitor the physical and chemical state of the lubricating oil or hydraulic oil in real time. The particle counter can detect metal particles larger than 5μm in the oil, and combined with abrasive morphology analysis, it can determine the wear type (such as cutting wear, fatigue wear). For example, the detection of a large number of plate-like ferromagnetic particles may indicate fatigue spalling on the gear surface. The viscosity sensor can monitor changes in oil viscosity; an abnormally high viscosity may indicate oil oxidation or contamination. The moisture sensor can detect the water content in the oil; excessive moisture will accelerate bearing corrosion.

[0059] Furthermore, by utilizing a fault diagnosis model based on envelope demodulation and deep residual networks, low-frequency fault features are extracted from high-frequency vibration signals (for example, for local gear faults, rotational frequency and its harmonic components will appear in the envelope spectrum). End-to-end fault classification and fault severity assessment are performed using multi-sensor data (vibration, acoustic emission, oil parameters), thereby improving the operational reliability and economy of wind turbine units.

[0060] Optionally, in some embodiments, the monitoring module 100 further includes: a yaw system monitoring unit, which includes a yaw gear ring wear monitoring subunit, a yaw bearing infrared imaging subunit, and a dual-mode positioning subunit, wherein the yaw gear ring wear monitoring subunit includes at least one of a high-definition industrial camera and an LED ring light source.

[0061] Understandably, the yaw system is a critical subsystem in a wind turbine, adjusting the nacelle's orientation to align with the wind direction and ensure maximum wind energy capture. Its core components (such as the yaw gear ring and yaw bearing) are subjected to complex alternating loads over long periods, making them prone to wear, deformation, or lubrication failure, leading to decreased yaw accuracy or even jamming. This yaw system monitoring unit can employ multimodal sensing technology and high-precision positioning to achieve comprehensive health monitoring of the yaw mechanism.

[0062] Specifically, the yaw system monitoring unit includes a yaw gear wear monitoring subunit, a yaw bearing infrared imaging subunit, and a dual-mode positioning subunit. The yaw gear wear monitoring subunit is a machine vision-based yaw gear wear detection system, including a high-definition industrial camera (resolution ≥ 0.1 mm) and an LED ring light source, ensuring clear, shadow-free imaging of the gear surface. By periodically capturing images of the gear meshing area and comparing them with the initial state, the wear rate can be calculated. For example, a tooth root crack will appear as a sudden change in image grayscale value. The yaw bearing infrared imaging subunit is a yaw bearing infrared thermal imager with a temperature detection accuracy of ±1℃ and a scanning frequency ≥ 5Hz, capable of capturing abnormal bearing temperature rises in real time. Compared to traditional temperature sensors, the thermal imager can globally monitor the bearing surface temperature field, avoiding missed detections. The dual-mode positioning subunit is a BeiDou / GPS dual-mode positioning and orientation module with an azimuth measurement accuracy of ±0.1°, providing real-time feedback on yaw angle errors in the cabin. The use of dual-mode positioning ensures data reliability under extreme weather conditions (such as thunderstorms).

[0063] Optionally, in some embodiments, the monitoring module 100 further includes a brake system monitoring unit, which includes at least one of a fiber optic grating temperature sensor, a hydraulic line pressure sensor, a brake caliper displacement sensor, and an ultrasonic brake pad thickness sensor.

[0064] Understandably, braking systems can be used to achieve reliable braking when wind turbines are overspeeding, malfunctioning, or undergoing maintenance.

[0065] Specifically, the brake system monitoring unit includes a fiber optic temperature sensor, a hydraulic line pressure sensor, a brake caliper displacement sensor, and an ultrasonic brake pad thickness sensor. The fiber optic temperature sensor (range 0-600℃) is directly embedded inside the brake disc to monitor its temperature. Compared to traditional thermocouples, the fiber optic temperature sensor is resistant to electromagnetic interference and supports multi-point distributed measurement. The hydraulic line pressure sensor (range 0-25MPa, accuracy 0.1%) provides real-time feedback on hydraulic system pressure fluctuations. The brake caliper displacement sensor (resolution 0.01mm) monitors the brake caliper piston displacement and, combined with pressure data, calculates the braking delay time (normal operation should be <200ms). The ultrasonic brake pad thickness sensor (measurement accuracy ±0.1mm) enables non-contact wear monitoring, dynamically scanning the entire circumference of the brake pad to generate a 3D wear cloud map.

[0066] Furthermore, this braking system monitoring unit integrates a braking performance evaluation algorithm based on pressure-displacement time delay analysis. This algorithm establishes a time-series relationship model between hydraulic pressure and brake caliper displacement, and uses time delay correlation analysis to determine the system state. Under normal conditions, pressure rise and displacement are synchronized (correlation coefficient > 0.95). Under abnormal conditions, the time delay increases or the correlation decreases (e.g., in one case, the correlation coefficient < 0.7, diagnosed as hydraulic line blockage). Compared to threshold alarms, this algorithm can identify early performance degradation.

[0067] Optionally, in some embodiments, the monitoring module 100 further includes a cooling and lubrication system monitoring unit, which includes at least one of an online oil quality multi-parameter sensor, a turbine flow meter, an infrared thermocouple array, and an inductive particle counter.

[0068] Understandably, the condition of the cooling and lubrication system directly affects the lifespan of critical components such as gearboxes and bearings. Traditional oil monitoring relies on periodic sampling and laboratory analysis, which suffers from significant delays and high costs. This cooling and lubrication system monitoring unit, through multi-parameter real-time sensing and intelligent wear diagnosis technology, achieves comprehensive online monitoring and accurate assessment of oil condition.

[0069] Specifically, the cooling and lubrication system monitoring unit includes an online multi-parameter oil quality sensor, a turbine flow meter, an infrared thermocouple array, and an inductive particle counter. The online multi-parameter oil quality sensor integrates dielectric constant, viscosity, and moisture measurements onto a single probe using MEMS (MicroElectroMechanical Systems) technology. The dielectric constant reflects the degree of oil oxidation (normal value is 2.0-2.5, >3.0 indicates severe degradation), and the moisture detection sensitivity reaches 0.01%, far exceeding the traditional Karl Fischer method. The turbine flow meter (accuracy ±1%) is used for real-time monitoring of lubricating oil flow. The infrared thermocouple array can simultaneously acquire multi-channel temperatures, monitoring the temperature of various parts of the gearbox with a resolution of 0.1℃. The inductive particle counter uses a high-frequency alternating magnetic field principle to distinguish between ferromagnetic and non-ferromagnetic particles and can detect metal particles >5μm, improving sensitivity by 10 times compared to traditional optical technology.

[0070] Furthermore, the cooling and lubrication system monitoring unit integrates a wear type classification model based on abrasive particle morphology recognition. This model uses microscopic images to acquire seven key characteristic parameters of abrasive particles (aspect ratio, edge serration, etc.), and employs a convolutional neural network to classify wear types, such as cutting wear (long strip-shaped particles), fatigue wear (flaky particles), and corrosion wear (spherical particles). This provides a comprehensive protective barrier for the wind turbine cooling and lubrication system.

[0071] Optionally, in some embodiments, the monitoring module 100 further includes an electronic control system monitoring unit, which includes at least one of a high-frequency current transformer, a distributed optical fiber temperature sensing subunit, and a combustible particulate matter sensor.

[0072] Understandably, the reliability of the electrical control system is directly related to the overall operational safety of the machine. Traditional monitoring methods struggle to detect early, latent faults in electrical equipment. This electrical control system's monitoring unit, through an innovative combination of high-frequency sensing and intelligent analysis, achieves accurate prediction and rapid location of potential risks within the electrical control system.

[0073] Specifically, the monitoring unit of the electrical control system includes a high-frequency current transformer, a distributed fiber optic temperature sensing subunit, and a combustible particulate matter sensor. The high-frequency current transformer monitors partial discharge at the converter output, employing a wide-bandwidth (100-500MHz) Rogowski coil sensor to capture nanosecond-level discharge pulses. The distributed fiber optic temperature sensing subunit has a spatial resolution of 0.5m and a temperature accuracy of ±0.5℃, and is laid along the entire collector line to achieve "line-to-surface" monitoring, enabling simultaneous detection of temperature and strain. The combustible particulate matter sensor utilizes laser scattering principles, with a detection range of 100nm-20μm, distinguishing particle types (metal, carbon, dust) and triggering an alarm when the particle concentration reaches a preset threshold.

[0074] Furthermore, the monitoring unit of the electronic control system integrates a fault location algorithm based on spatiotemporal graph convolutional networks. By using the fault location algorithm based on spatiotemporal graph convolutional networks, the physical topology of the electronic control system (such as the connection relationship of converters, cables, and terminals) is modeled through graph structure. Sensor nodes are abstracted as vertices in the graph, and electrical connections are used as edges. The dynamic change sequence of parameters such as partial discharge, temperature, and particulate matter is analyzed simultaneously to capture the fault evolution law.

[0075] Optionally, in some embodiments, the analysis module 300 further includes: a wind turbine digital twin unit, a life prediction unit, a fault knowledge graph unit, and an alarm unit. The wind turbine digital twin unit is used to simulate the state of the wind turbine under different preset physical conditions; the life prediction unit integrates a preset neural network model to predict the remaining service life of the wind turbine; the fault knowledge graph unit is used to determine the fault type of the wind turbine and provide corresponding fault handling suggestions to the operator; the alarm unit is used to set an alarm threshold based on the real-time ambient temperature and trigger an alarm signal when the current operating temperature of the wind turbine exceeds the alarm threshold to notify the operator to take appropriate measures.

[0076] Understandably, the analytics module 300 is responsible for integrating multi-source data, in-depth analysis, predictive maintenance decision support, and visualization. This platform integrates advanced technologies such as digital twins, AI (Artificial Intelligence) predictive models, knowledge graphs, and adaptive optimization to provide intelligent operation and maintenance management for wind turbine units throughout their entire lifecycle.

[0077] Specifically, the wind turbine digital twin unit (wind turbine digital twin) can construct a 1:1 virtual wind turbine model based on CAD (Computer-Aided Design), SCADA, and sensor data. It encompasses multi-physics coupled simulations of aerodynamics, structure, thermal, and electromagnetic fields, simulating the wind turbine's state under different preset physical conditions. The unit can update operational data every 5 minutes to ensure consistency between the digital twin and the physical turbine. This wind turbine digital twin unit can predict blade load distribution under typhoon conditions, optimize yaw strategies, and simulate gearbox wear processes by adjusting parameters to verify the effectiveness of maintenance plans. The life prediction unit integrates a remaining service life prediction model based on LTSM (Long Short-Term Memory) (i.e., a preset neural network model). The model's input can be 15-dimensional time-series data such as vibration amplitude, oil metal content, and bearing temperature, and its output can be the current health index and remaining life of wind turbine components (such as main shaft bearings and gearboxes). The fault knowledge graph unit constructs a knowledge system where node types can be fault modes (e.g., "bearing spalling"), symptoms ("high-frequency vibration"), and solutions ("grease replacement"). Relationship types can be causal relationships ("insulation aging → partial discharge") and handling relationships ("filter cleaning → oil temperature reduction"), etc. This fault knowledge graph unit can receive abnormal features reported by edge nodes (e.g., a 9.6kHz component in the vibration spectrum), match the closest fault path in the graph, thereby determining the fault type of the wind turbine and providing operators with corresponding fault handling suggestions. The alarm unit is an adaptive alarm threshold adjustment module that can intelligently adjust alarm thresholds according to different environmental conditions and operating states. For example, for every 10°C increase in temperature, the bearing vibration alarm threshold can be automatically increased by 5%; when the power generation is >80% of the rated value, the gearbox oil temperature threshold can be increased by 2°C, etc. This alarm unit can monitor the ambient temperature in real time and dynamically set the alarm threshold according to temperature changes. When the current operating temperature of the fan exceeds the set alarm threshold, the system will immediately trigger an alarm signal to remind the operator to take necessary measures, such as adjusting the equipment operating parameters or performing equipment maintenance, thereby ensuring the stable operation of the equipment and extending its service life.

[0078] The wind turbine health status diagnosis system proposed in this application acquires multi-source data through a monitoring module using a preset sensor group; a computing module receives the multi-source data and performs feature extraction operations on the multi-source data to obtain multi-source feature information; an analysis module fuses the multi-source feature information and evaluates the health status of the generator set based on the fusion result. Therefore, this generator set health status diagnosis system integrates multi-sensor fusion, edge computing, and cloud-based intelligent analysis, solving the problems of existing technologies lacking system-level health status assessment capabilities and intelligent diagnostic capabilities, and realizing real-time monitoring and early fault warning of the entire wind turbine system's health status.

[0079] Next, the method for diagnosing the health status of wind turbines according to embodiments of this application is described with reference to the accompanying drawings.

[0080] Figure 2 This is a flowchart of a wind turbine health status diagnosis method according to an embodiment of this application.

[0081] like Figure 2 As shown, the health status diagnosis method for this wind turbine adopts... Figure 1 The wind turbine health status diagnosis system of the embodiment includes the following steps:

[0082] In step S201, multi-source data is acquired, and feature extraction is performed on the multi-source data to obtain multi-source feature information.

[0083] In step S202, the multi-source feature information is fused, and the health status of the wind turbine is evaluated based on the fusion result to obtain the evaluation result.

[0084] In step S203, based on the evaluation results, predictive wind turbine maintenance recommendations are generated and visualized.

[0085] It should be noted that the foregoing explanation of the wind turbine health status diagnosis system embodiment also applies to the wind turbine health status diagnosis method of this embodiment, and will not be repeated here.

[0086] The wind turbine health status diagnosis method proposed in the embodiments of this application integrates multi-sensor fusion, edge computing and cloud intelligent analysis through a wind turbine health status diagnosis system, which solves the problems of the lack of system-level health status assessment capability and the lack of intelligent diagnosis capability in the prior art, and realizes real-time monitoring and early fault warning of the entire wind turbine system health status.

[0087] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. Throughout the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0088] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0089] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A wind turbine health status diagnosis system, characterized in that, include: The module consists of a monitoring module, an edge computing module, and an analysis module. The monitoring module is used to acquire multi-source data using a preset sensor group; The computing module is used to receive the multi-source data and perform feature extraction operations on the multi-source data to obtain multi-source feature information; The analysis module is used to fuse the multi-source feature information and evaluate the health status of the wind turbine based on the fusion results.

2. The wind turbine health status diagnosis system according to claim 1, characterized in that, The monitoring module includes: a support system monitoring unit, which comprises an ultrasonic sensor array and a fiber optic grating sensor network, wherein... The ultrasonic sensor array is used to monitor the stress of the tower connecting bolts; The fiber optic grating sensor network is used to monitor the stress on the tower body.

3. The wind turbine health status diagnosis system according to claim 2, characterized in that, The monitoring module further includes: a rotation system monitoring unit, which comprises a fiber optic grating array integrated inside the wind turbine blades and a triaxial vibration sensor deployed in the hub. The fiber grating array is used to determine the damage status of the blade structure using a preset damage identification algorithm; The triaxial vibration sensor is used to acquire vibration data of the wheel hub area, and the preset damage identification algorithm is used to analyze the vibration data to determine the abnormal conditions of the wheel hub area.

4. The wind turbine health status diagnosis system according to claim 3, characterized in that, The monitoring module further includes: a transmission system monitoring unit, which comprises a high-frequency acceleration sensor, a broadband acoustic emission sensor, and an oil parameter monitoring subunit, wherein... The high-frequency acceleration sensor is deployed at the gearbox input / output end and the spindle bearing; The broadband acoustic emission sensor is mounted on the gearbox housing; The oil parameter monitoring subunit includes a particle counter, a viscosity sensor, and a moisture sensor, used to monitor the physical and chemical parameters of lubricating oil or hydraulic oil in real time.

5. The wind turbine health status diagnosis system according to claim 4, characterized in that, The monitoring module further includes: a yaw system monitoring unit, which comprises a yaw gear wear monitoring subunit, a yaw bearing infrared imaging subunit, and a dual-mode positioning subunit, wherein... The yaw gear wear monitoring subunit includes at least one of a high-definition industrial camera and an LED ring light source.

6. The wind turbine health status diagnosis system according to claim 5, characterized in that, The monitoring module also includes: The brake system monitoring unit includes at least one of a fiber optic grating temperature sensor, a hydraulic line pressure sensor, a brake caliper displacement sensor, and an ultrasonic brake pad thickness sensor.

7. The wind turbine health status diagnosis system according to claim 6, characterized in that, The monitoring module also includes: A cooling and lubrication system monitoring unit, comprising at least one of an online oil quality multi-parameter sensor, a turbine flow meter, an infrared thermocouple array, and an inductive particle counter.

8. The wind turbine health status diagnosis system according to claim 7, characterized in that, The monitoring module also includes: The electronic control system monitoring unit includes at least one of a high-frequency current transformer, a distributed optical fiber temperature sensing subunit, and a combustible particulate matter sensor.

9. The wind turbine health status diagnosis system according to claim 8, characterized in that, The analysis module includes: A wind turbine digital twin unit, which is used to simulate the state of a wind turbine under different preset physical conditions; A lifespan prediction unit, which integrates a preset neural network model, is used to predict the remaining lifespan of the wind turbine. The fault knowledge graph unit is used to determine the fault type of the wind turbine and provide corresponding fault handling suggestions to the operator. An alarm unit is provided, which is used to set an alarm threshold based on the real-time ambient temperature, and to trigger an alarm signal when the current operating temperature of the fan exceeds the alarm threshold, so as to notify the operator to take appropriate measures.

10. A method for diagnosing the health status of a wind turbine generator, characterized in that, The wind turbine health status diagnosis system as described in any one of claims 1-9 is used, wherein the method includes the following steps: Acquire multi-source data and perform feature extraction on the multi-source data to obtain multi-source feature information; The multi-source feature information is fused, and the health status of the wind turbine is evaluated based on the fusion result to obtain the evaluation result; Based on the assessment results, predictive wind turbine maintenance recommendations are generated and then visualized.

Images