Wind power variable pitch test system, method and equipment based on wireless sensor network

Abstract

The invention discloses a wind power variable pitch test system, method and equipment based on a wireless sensing network. The system comprises a multi-mode wireless sensing device, a quantum sensing integration device, an acoustic imaging analysis device, a data preprocessing device, a data processing device, an analysis device, a strategy generation device, a calibration device, an energy management device and a control device. The system integrates multiple advanced technologies such as multi-mode wireless sensing, quantum sensing and acoustic imaging, multi-dimensional parameter data of the wind power variable pitch system can be comprehensively and accurately collected and analyzed, and the quality and reliability of the data are effectively improved through the data preprocessing device and the data processing device. And the analysis device can predict the service life of the wind power variable pitch system by using the processed data, so that a scientific basis is provided for maintenance and management of a unit. The strategy generation device dynamically generates the test strategy according to the analysis data, and the test process is optimized. The calibration device ensures the accuracy and stability of the sensing device.

Description

Technical Field

[0001] The present invention belongs to the technical field of wind power pitch system testing, and particularly relates to a wind power pitch testing system, method and device based on a wireless sensor network. Background Art

[0002] Wind power generation is an important way to utilize renewable energy, and the pitch system is a key component in a wind turbine generator set, responsible for adjusting the windward angle of the wind turbine blades to adapt to different wind speed conditions, so as to optimize the power generation efficiency and ensure the safe operation of the unit. The performance of the pitch system directly affects the power generation efficiency, service life and safety of the wind turbine generator set. Therefore, it is of great significance to conduct comprehensive and accurate testing and monitoring on the pitch system. At present, for the testing of the pitch system, data collection is usually carried out by using a wired sensor network or a simple wireless sensor network, and data analysis and fault diagnosis are carried out through a central processing unit. This testing method has some limitations: First, the layout of the traditional wired sensor network is complex and it is difficult to cover all key measuring points; Second, although the simple wireless sensor network solves the wiring problem, there are often problems such as unstable data transmission and poor real-time performance. The specific reasons are as follows: Traditional wireless sensor networks (WSNs) usually adopt protocols such as Zigbee, Wi-Fi, LoRa, etc. In a complex wind farm environment, they may be affected by factors such as electromagnetic interference, signal attenuation, and wind turbine blade occlusion, resulting in unstable data transmission. The pitch system is located in the rotating part of the wind turbine unit, and the high-speed rotating blades will block the wireless signal transmission, causing some data packets to be lost or delayed.

[0003] When multiple wind turbines work simultaneously, wireless signal conflicts may occur, resulting in a decrease in data transmission efficiency. Due to the distributed layout of the wind farm, data synchronization between sensors is crucial. However, traditional wireless sensor network protocols (such as TDMA, CSMA) may lead to inaccurate data timestamps due to clock drift or different data collection rates of different nodes, affecting real-time performance.

[0004] In some low-power wireless networks (such as LoRaWAN), nodes adopt intermittent data transmission to save energy, but this will increase the data transmission delay and affect the real-time analysis ability of the system.

[0005] Sensors of the pitch system (such as vibration, acoustic, image sensors) may generate high-frequency and large-volume data, such as real-time vibration data or acoustic data, while the bandwidth of traditional wireless transmission protocols is limited and may not be able to meet the high-throughput requirements.

[0006] In a wind power generation environment, wireless signals may cause data packet loss due to factors such as wind speed changes and electromagnetic interference (such as thunderstorm weather), affecting data integrity.

[0007] Furthermore, the existing data processing methods mainly rely on preset thresholds and simple statistical analysis, making it difficult to cope with complex working conditions and potential failure modes. The specific reasons are as follows: Traditional pitch system monitoring mainly relies on fixed threshold judgment (such as excessive motor current, too high bearing temperature). If the data does not exceed the preset threshold, the system may not be able to detect potential failures.

[0008] Due to the complex operating environment of wind turbines, with large variations in working conditions such as wind speed, wind direction, temperature, and load, simple threshold settings cannot adapt dynamically, resulting in the inability to identify some early failures.

[0009] Statistical analysis methods (such as mean and variance) can only capture the overall trend of data, but it is difficult to identify complex non - linear failure modes, such as bearing micro - cracks and abnormal gear meshing.

[0010] Existing monitoring systems often adopt rule - driven fault detection, that is, setting the characteristics of certain failure modes based on experience, but this method is difficult to cover all possible failures.

[0011] The failures of the pitch system have multi - modal data correlation (for example, abnormal vibration may be caused by various factors such as blade imbalance, motor failure, and bearing damage), but traditional data analysis methods are difficult to fuse data from different sensors, resulting in inaccurate root cause analysis of failures.

[0012] The performance of the pitch system of a wind turbine usually degrades on a time scale of several years, while traditional data analysis methods often focus on short - term data anomaly detection and are difficult to capture long - term degradation trends.

[0013] For example, the wear of bearings will cause the vibration characteristics to change gradually over time, but if only relying on short - term data analysis, it may not be able to detect in time that the health state of the equipment is deteriorating step by step.

[0014] Wind turbines operate in a complex wind field environment, where factors such as wind speed, wind direction, temperature, and humidity are constantly changing, making the operating state of the pitch system adjust dynamically.

[0015] Traditional monitoring systems are difficult to adapt to this time - varying non - linear working condition, resulting in different manifestations of the same failure mode under different working conditions and increasing the difficulty of fault diagnosis. Summary of the Invention

[0016] The purpose of the present invention is to provide a wind power pitch test system, method, and device based on a wireless sensor network to solve the technical defects of incomplete data acquisition coverage, unstable transmission, and insufficient analysis in the prior art.

[0017] To achieve the above - mentioned purpose, the present invention adopts the following technical solutions: In a first aspect, a wind turbine pitch testing system based on a wireless sensor network is provided, including: a multi-modal wireless sensing device, a quantum sensing integration device, an acoustic imaging analysis device, a data preprocessing device, a data processing device, an analysis device, a strategy generation device, a calibration device, an energy management device, and a control device. The multi-modal wireless sensing device, the quantum sensing integration device, the acoustic imaging analysis device, the data preprocessing device, the data processing device, the analysis device, the strategy generation device, the calibration device, and the energy management device are electrically connected to the control device; the multi-modal wireless sensing device, the quantum sensing integration device, and the acoustic imaging analysis device are signal-connected to the data preprocessing device, and the data preprocessing device, the analysis device, the strategy generation device, and the calibration device are all signal-connected to the data processing device; Among them, the multi-modal wireless sensing device is used to collect parameter data of the wind turbine pitch system; The data preprocessing device is used to preprocess the parameter data; The data processing device is used to analyze the preprocessed parameter data and obtain analysis data; The analysis device is used to receive the analysis data and predict the life of the wind turbine pitch system by using the analysis data; The strategy generation device is used to generate a test strategy according to the analysis data; The calibration device is used to calibrate the multi-modal wireless sensing device; The quantum sensing integration device is used to detect the magnetic field of the wind turbine pitch system; The acoustic imaging analysis device is used to capture and analyze the acoustic characteristics of the wind turbine pitch system; The energy management device is used to provide energy for the multi-modal wireless sensing device.

[0018] Further, the multi-modal wireless sensing device includes a strain sensor, an acceleration sensor, a pressure sensor, a temperature sensor, an angle encoder, an acoustic sensor array, and a quantum magnetometer; Among them, the strain sensor is arranged on the support structure of the pitch system bearing and the blade root, and is used to detect the mechanical stress borne by the blade during the pitch process; The acceleration sensor is arranged on the motor housing of the pitch system and the blade root, and is used to detect the vibration characteristics during the operation of the pitch system; The pressure sensor is arranged on the hydraulic pipelines, the inner cavity of the oil cylinder, and the blade surface of the hydraulic actuator and the electric actuator of the pitch system, and is used to monitor the pressure change; The temperature sensor is arranged on the drive motor, the oil cylinder, and the main bearing seat of the pitch system, and is used to monitor the temperature change; The angle encoder is disposed on the rotating shaft and the blade root of the pitch system for measuring the pitch angle of the blade; The acoustic sensor array is disposed on the pitch system housing, the main bearing and the motor for collecting the acoustic characteristics of the pitch system during operation; The quantum magnetometer is disposed in the bearing cavity and the motor stator of the pitch system for measuring the magnetic field change.

[0019] Further, the data processing device includes a data access layer, a data processing layer, a model service layer and a storage layer, and the data access layer, the data processing layer, the model service layer and the storage layer are sequentially connected by signals; Among them, the data access layer is used to receive the parameter data transmitted by the data preprocessing device; The data processing layer is used to further process the parameter data; The model service layer is used to deploy and manage the analysis model; The storage layer is used to store the original data of the parameter data and the processing results of the parameter data.

[0020] Further, the analysis device includes a pitch system performance prediction model, a fault diagnosis and classification model, and a life prediction model, and the pitch system performance prediction model, the fault diagnosis and classification model, and the life prediction model are connected to the data processing device by signals.

[0021] Further, the energy management device includes a piezoelectric energy harvesting component, a wireless charging component and a sleep component, and the piezoelectric energy harvesting component, the wireless charging component and the sleep component are electrically connected to the control device; Among them, the piezoelectric energy harvesting component is used to convert the mechanical vibration of the pitch system into electrical energy; The wireless charging component is used to provide electrical energy; The sleep component is used to adjust the working mode of the multimodal wireless sensing device.

[0022] In a second aspect, a wind power pitch test method based on a wireless sensor network is provided. The method is carried out by using the wind power pitch test system based on a wireless sensor network as described above, and includes: Using the multimodal wireless sensing device to collect the parameter data of the key parts in the wind power pitch system; Based on the data preprocessing device, the parameter data is subjected to real-time processing and anomaly detection; Through the data processing device, in-depth analysis is carried out on the parameter data after real-time processing; Relying on the analysis device and combining the analysis data, the life of the wind power pitch system is predicted; According to the life prediction result, the test strategy is dynamically adjusted.

[0023] Furthermore, the data preprocessing device performs real-time processing and anomaly detection on the parameter data, specifically including: Performing real-time anomaly detection on the parameter data using the Isolation Forest algorithm; Adopting wavelet transform combined with an adaptive quantization method to compress the parameter data after anomaly detection; Extracting time-domain, frequency-domain, and time-frequency-domain features from the compressed parameter data.

[0024] Furthermore, during the real-time anomaly detection process, specifically: Detecting mechanical anomalies in the pitch system to determine whether there is overload stress, abnormal vibration, or mechanical shock in the pitch system; Detecting the current, voltage, and magnetic field data of the pitch motor to identify overcurrent, undervoltage, and magnetic field drift fault signals; Detecting the temperature of the pitch system to determine whether there is abnormal temperature rise in the pitch system; Detecting the pressure data of the pitch system to determine whether there are faults such as hydraulic oil leakage, blockage, or actuator jamming; Analyzing the acoustic data of the pitch system to detect whether there is abnormal noise; Detecting the magnetic field data of the pitch system to identify magnetic saturation, magnetic offset, or winding faults.

[0025] In a third aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned wind power pitch test method based on a wireless sensor network are implemented.

[0026] In a fourth aspect, a computer program product is provided, including computer instructions, and the computer instructions direct a computing device to perform operations corresponding to the above-mentioned wind power pitch test method based on a wireless sensor network.

[0027] Compared with the prior art, the present invention has the following beneficial effects: 1. The system integrates a variety of advanced technologies such as multi-modal wireless sensing, quantum sensing, and acoustic imaging, and can comprehensively and accurately collect and analyze multi-dimensional parameter data of the wind turbine pitch system. Through data preprocessing and data processing devices, the quality and reliability of the data are effectively improved. The analysis device can use the processed data to predict the lifespan of the wind turbine pitch system, providing a scientific basis for the maintenance and management of the unit. The strategy generation device dynamically generates test strategies according to the analysis data, optimizing the test process. The calibration device ensures the accuracy and stability of the sensing device. The energy management device provides continuous and reliable energy support for the multi-modal wireless sensing device, ensuring the long-term stable operation of the system. Overall, the system improves the test efficiency and accuracy of the wind turbine pitch system, ensuring the safe and reliable operation of the unit.

[0028] 2. The multi-modal wireless sensing device integrates various types of sensors and can monitor the operating status of the wind turbine pitch system in all directions and from multiple angles. The settings of strain sensors, acceleration sensors, and pressure sensors can monitor the mechanical stress, vibration characteristics, and pressure changes endured by the pitch system in real time, providing important data for evaluating the structural integrity and operating stability of the system. The temperature sensor can monitor the temperature changes of key components, promptly detect abnormal conditions such as overheating, and prevent equipment damage. The angle encoder accurately measures the blade pitch angle, ensuring the precise control of the pitch system. The acoustic sensor array and the acoustic imaging analysis device cooperate to deeply capture and analyze the acoustic characteristics of the pitch system, providing strong support for fault diagnosis. The addition of the quantum magnetometer enables the system to measure magnetic field changes, further expanding the dimension and depth of monitoring. Overall, this multi-modal sensing device improves the comprehensiveness and accuracy of the monitoring of the wind turbine pitch system, providing strong guarantee for the safe and efficient operation of the system.

[0029] 3. The data processing device realizes the efficient and orderly processing of data through hierarchical design. The data access layer is responsible for receiving the preprocessed parameter data, ensuring the accurate input of the data; the data processing layer further processes the parameter data, extracts useful information, and provides a basis for subsequent analysis; the model service layer deploys and manages analysis models, enabling the system to flexibly use various algorithms and models to deeply analyze the data, improving the accuracy and efficiency of the analysis; the storage layer is responsible for storing the original data and processing results, providing convenience for data traceability and subsequent analysis. This hierarchical design improves the flexibility and scalability of data processing, enabling the system to better adapt to different monitoring requirements and changes, and providing strong data support for the operation and maintenance of the wind turbine pitch system.

[0030] 4. By deploying the pitch system performance prediction model, fault diagnosis and classification model, and life prediction model, the analysis device can comprehensively and deeply analyze the operating status of the wind power pitch system. These models are based on a large amount of data and advanced algorithms, enabling more accurate prediction of system performance, diagnosis of fault types, and prediction of system life, providing reliable decision-making support for maintenance personnel. The model service layer in the analysis device can flexibly deploy and manage various analysis models, enabling the system to automatically select the appropriate model for analysis according to actual needs. This intelligent analysis method improves the adaptability and flexibility of the system, enabling it to better cope with complex and changing operating environments.

[0031] 5. By integrating piezoelectric energy harvesting components, wireless charging components, and sleep components, the energy management device achieves multiple beneficial effects such as self-power supply, optimized energy management, and improved system reliability, providing strong guarantee for the long-term stable operation of the wind power pitch test system.

[0032] 6. Using the multi-modal wireless sensing device, it is possible to comprehensively and accurately collect parameter data of key parts in the wind power pitch system. This multi-dimensional data collection method provides a rich and reliable data basis for subsequent analysis and prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0034] Figure 1 Schematic diagram of the wind power pitch test system based on wireless sensor network provided by the present invention; Figure 2 Flowchart of the wind power pitch test method based on wireless sensor network provided by the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. The components of the embodiments of the present invention usually described and illustrated in the drawings here can be arranged and designed in various different configurations.

[0035] Accordingly, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0036] It should be noted that like reference numerals and letters denote like items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0037] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper", "lower", "horizontal", "inner", etc. are used to indicate the orientation or positional relationship, it is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the inventive product is usually placed during use. This is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present invention. In addition, terms such as "first", "second", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0038] In addition, if the term "horizontal" appears, it does not mean that the component is required to be absolutely horizontal, but it can be slightly inclined. For example, "horizontal" only means that its direction is more horizontal relative to "vertical", and does not mean that the structure must be completely horizontal, but it can be slightly inclined.

[0039] In the description of the embodiments of the present invention, it should also be noted that unless otherwise clearly defined and limited, if terms such as "set", "installed", "connected", "coupled" are used, they should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0040] Wind power generation is an important way to utilize renewable energy. The pitch system is a key component in wind turbines, responsible for adjusting the windward angle of the wind turbine blades to adapt to different wind speed conditions, thereby optimizing the power generation efficiency and ensuring the safe operation of the unit. The performance of the pitch system directly affects the power generation efficiency, lifespan, and safety of wind turbines. Therefore, it is of great significance to conduct comprehensive and accurate testing and monitoring of the pitch system. Currently, for the testing of the pitch system, wired sensor networks or simple wireless sensor networks are usually used for data acquisition, and data analysis and fault diagnosis are carried out through a central processing unit. This testing method has some limitations: First, the layout of traditional wired sensor networks is complex and it is difficult to cover all key measurement points; Second, although simple wireless sensor networks solve the wiring problem, they often have problems such as unstable data transmission and poor real-time performance. The specific reasons are as follows: Traditional wireless sensor networks (WSNs) usually adopt protocols such as Zigbee, Wi-Fi, LoRa, etc. In a complex wind farm environment, they may be affected by factors such as electromagnetic interference, signal attenuation, and wind turbine blade occlusion, resulting in unstable data transmission. The pitch system is located in the rotating part of the wind turbine, and the high-speed rotating blades will block the wireless signal transmission, causing some data packets to be lost or delayed.

[0041] When multiple wind turbines work simultaneously, wireless signal conflicts may occur, leading to a decrease in data transmission efficiency. Due to the distributed layout of wind farms, data synchronization between sensors is crucial. However, traditional wireless sensor network protocols (such as TDMA, CSMA) may result in inaccurate data timestamps due to clock drift or different data acquisition rates of different nodes, affecting real-time performance.

[0042] In some low-power wireless networks (such as LoRaWAN), nodes use intermittent data transmission to save energy, but this will increase the data transmission delay and affect the real-time analysis ability of the system.

[0043] Sensors in the pitch system (such as vibration, acoustic, image sensors) may generate high-frequency and large-volume data, such as real-time vibration data or acoustic data. However, the bandwidth of traditional wireless transmission protocols is limited and may not be able to meet the high-throughput requirements.

[0044] In the wind power generation environment, wireless signals may cause data packet loss due to factors such as wind speed changes and electromagnetic interference (such as thunderstorm weather), affecting data integrity.

[0045] Furthermore, existing data processing methods mainly rely on preset thresholds and simple statistical analysis, and it is difficult to cope with complex working condition changes and potential fault modes. The specific reasons are as follows: Traditional pitch system monitoring mainly relies on fixed threshold judgment (such as excessive motor current, too high bearing temperature). If the data does not exceed the preset threshold, the system may not be able to detect potential faults.

[0046] Due to the complex operating environment of wind turbines, with large variations in operating conditions such as wind speed, wind direction, temperature, and load, simple threshold settings cannot adapt dynamically, resulting in some early faults not being recognized.

[0047] Statistical analysis methods (such as mean, variance) can only capture the overall trend of data, but it is difficult to identify complex non-linear fault patterns, such as bearing micro-cracks, abnormal gear meshing, etc.

[0048] Existing monitoring systems often adopt rule-driven fault detection, that is, setting the characteristics of certain fault patterns based on experience, but this method is difficult to cover all possible faults.

[0049] The faults of the pitch system have multi-modal data correlation (for example, abnormal vibration may be caused by various factors such as blade imbalance, motor failure, bearing damage, etc.), but traditional data analysis methods are difficult to fuse data from different sensors, resulting in inaccurate root cause analysis of faults.

[0050] The performance of the pitch system of a wind turbine usually degrades on a time scale of several years, while traditional data analysis methods often focus on short-term data anomaly detection and are difficult to capture long-term degradation trends.

[0051] For example, the wear of the bearing will cause the vibration characteristics to change gradually over time, but if only relying on short-term data analysis, it may not be able to detect in time that the health state of the equipment is deteriorating step by step.

[0052] Wind turbines operate in a complex wind field environment, with factors such as wind speed, wind direction, temperature, and humidity constantly changing, making the operating state of the pitch system adjust dynamically.

[0053] Traditional monitoring systems are difficult to adapt to this time-varying non-linear operating condition, resulting in different manifestations of the same fault pattern under different operating conditions, increasing the difficulty of fault diagnosis.

[0054] The following further describes the present invention in detail with reference to the accompanying drawings: In a first aspect, an embodiment of the present invention provides a wind power pitch testing system based on a wireless sensor network, comprising: a multimodal wireless sensing device, a quantum sensing integration device, an acoustic imaging analysis device, a data preprocessing device, a data processing device, an analysis device, a strategy generation device, a calibration device, an energy management device, and a control device. The multimodal wireless sensing device, the quantum sensing integration device, the acoustic imaging analysis device, the data preprocessing device, the data processing device, the analysis device, the strategy generation device, the calibration device, and the energy management device are electrically connected to the control device; the multimodal wireless sensing device, the quantum sensing integration device, and the acoustic imaging analysis device are signal-connected to the data preprocessing device, and the data preprocessing device, the analysis device, the strategy generation device, and the calibration device are all signal-connected to the data processing device. Among them, the multimodal wireless sensing device is used to collect parameter data of the wind power pitch system; the data preprocessing device is used to preprocess the parameter data; the data processing device is used to analyze the preprocessed parameter data and obtain analysis data; the analysis device is used to receive the analysis data and predict the life of the wind power pitch system using the analysis data; the strategy generation device is used to generate a test strategy according to the analysis data; the calibration device is used to calibrate the multimodal wireless sensing device; the quantum sensing integration device is used to detect the magnetic field of the wind power pitch system; the acoustic imaging analysis device is used to capture and analyze the acoustic characteristics of the wind power pitch system; the energy management device is used to provide energy for the multimodal wireless sensing device. Through the integration of advanced technologies such as multimodal wireless sensor networks, edge computing, artificial intelligence analysis, quantum sensing, and acoustic imaging, this system realizes comprehensive, real-time, and high-precision monitoring and analysis of the wind power pitch system, significantly improves the coverage and accuracy of data collection, and solves the problem of unstable data transmission in traditional testing methods. The introduction of edge computing greatly improves the real-time performance of the system, while the analysis device enhances the depth and accuracy of data analysis. The application of the strategy generation device and the calibration device makes the testing process more flexible and efficient, while ensuring the accuracy of long-term measurements. In addition, the energy management device ensures the continuous and stable operation of the entire testing system. Generally speaking, this system not only improves the testing efficiency and reliability of the wind power pitch system, but also provides strong technical support for the performance optimization, fault warning, and life prediction of wind turbines, thus contributing to the improvement of the overall efficiency and economy of wind power generation.

[0055] Furthermore, the multimodal wireless sensing device includes a strain sensor, an acceleration sensor, a pressure sensor, a temperature sensor, an angle encoder, an acoustic sensor array, and a quantum magnetometer. Among them, the strain sensor is arranged on the support structure of the pitch system bearing and the blade root area to detect the mechanical stress borne by the blade during the pitching process and evaluate the health status of the structure at the connection part between the blade and the bearing. The acceleration sensor is arranged on the pitch system motor housing and the blade root to detect the vibration characteristics during the operation of the pitch system and analyze possible mechanical shocks, abnormal vibrations, or fatigue damage conditions. The pressure sensor is arranged on the hydraulic pipelines, cylinder cavities of the hydraulic and electric actuators of the pitch system, and the key pneumatic areas on the blade surface to monitor the pressure changes in the hydraulic system, judge the working state of the actuator, and simultaneously monitor the influence of aerodynamic force on the blade. The temperature sensor is arranged on the drive motor, hydraulic cylinder of the hydraulic system, and main bearing seat of the pitch system to monitor the temperature changes of key components, evaluate the motor heat load, lubricating oil state, and bearing temperature rise to prevent overheating faults. The angle encoder is arranged at the rotating shaft and blade root position of the pitch system to measure the blade pitch angle, accurately measure the blade pitch angle, and ensure the adjustment accuracy of the blade and the accuracy of the control strategy. The acoustic sensor array is arranged on the pitch system housing, main bearing, and motor to collect the acoustic characteristics during the operation of the pitch system and detect abnormal noises for fault diagnosis and early warning. The quantum magnetometer is arranged in the bearing cavity and motor stator of the pitch system to measure the magnetic field changes, detect motor winding aging, bearing wear, and magnetic field anomalies for non-contact health monitoring. In application, the above multiple sensors can be combined to form a multimodal wireless sensor network. This network uses the TSCH protocol based on the IEEE 802.15.4e standard for communication. This protocol improves the reliability and real-time performance of communication through channel hopping and time slot synchronization mechanisms. The specific implementation includes: Channel hopping: Use a pseudo-random sequence to hop among 16 available channels to improve the anti-interference ability.

[0056] Time slot synchronization: Adopt an improved adaptive time synchronization algorithm to control the time synchronization error between nodes within ±10 μs.

[0057] Network topology: Adopt a dynamic mesh topology to support self-organization and self-healing.

[0058] Through the deployment of the multimodal wireless sensor network, this system solves the problems of few data acquisition points and limited measurement range in traditional test systems, and significantly improves the comprehensiveness and accuracy of data.

[0059] When the quantum sensing integrated device is applied, nitrogen vacancy centers are used to achieve ultra-high sensitivity magnetic field detection, which specifically includes a quantum magnetometer based on nitrogen vacancy centers in single crystal diamond, a quantum state manipulation method using 532nm laser pulses and 2.87GHz microwaves, and a Bayesian magnetic field reconstruction algorithm based on compressed sensing; among them, the quantum magnetometer collects raw data, the quantum state manipulation method controls the sensing process, and the magnetic field reconstruction algorithm processes the collected data to obtain high-precision magnetic field information.

[0060] During the test of the variable pitch system, a magnetic field will be generated, which has a great impact on the test. Through analysis, the generation of magnetic fields specifically includes the following aspects: First, the magnetic field of the variable pitch motor. The motor in the wind turbine variable pitch system will generate an electromagnetic field when running. The strength of the magnetic field is related to parameters such as motor current and speed. The change of the motor magnetic field causes the control signal of the variable pitch system to be interfered, affecting the accuracy and stability of the variable pitch. Especially at the moment of motor start, stop or speed change, the change of the magnetic field is particularly drastic, and the impact on the test is more significant. Residual magnetic field of the variable pitch bearing: There may be weak magnetism in the bearing material, and it will be affected by the external magnetic field during long-term operation, resulting in changes in the magnetic field. Magnetic interference of cables and power electronic equipment: Power electronic equipment such as cables, inverters, and converters inside the wind turbine will generate additional magnetic field interference, which may affect the electromagnetic environment of the variable pitch system. Influence of the geomagnetic field: The installation environment of the wind turbine may be interfered by the earth's magnetic field. Especially in high-latitude areas, changes in the geomagnetic field may affect the magnetic measurement of the system.

[0061] In order to solve the magnetic field problem, the magnetic field data was collected by a quantum magnetometer based on the nitrogen vacancy center in single crystal diamond. During the collection process, since the nitrogen vacancy (NV) center is a point defect structure in single crystal diamond, it is formed by a nitrogen (N) replacing a carbon atom and accompanied by an adjacent carbon vacancy (V) to form a stable quantum defect. The NV center has the following characteristics: Controllable electron spin: The electron spin state of the NV center can be manipulated by external magnetic fields, light (laser) and microwaves, and can be used for precise magnetic field measurements.

[0062] High sensitivity: Due to their unique electronic structure, NV centers can achieve nanotesla (nT) or even picotesla (pT) level magnetic field detection at room temperature.

[0063] Long quantum coherence: NV centers have long spin coherence times (on the order of ms), making them suitable for precise magnetic field detection.

[0064] Strong anti-interference ability: NV center can maintain good magnetic detection performance in high noise environment, and is suitable for complex industrial environments such as wind power systems.

[0065] After the quantum magnetometer collects the magnetic field data, the operating state of the pitch motor can be monitored through the collected data. For example, by detecting the changes in the magnetic field inside the motor, the health state of the motor winding can be judged, and problems such as abnormal current, magnetic field offset, or winding short circuit can be identified; by analyzing the changes in the residual magnetic field inside the bearing, the wear degree and fatigue damage of the bearing can be identified; by detecting the current magnetic field distribution in the cable, whether there are risks of cable aging, breakage, or short circuit can be judged; by measuring the magnetic fields inside and outside the system, the normal working magnetic field can be distinguished from the noise magnetic field, improving the accuracy of magnetic field measurement.

[0066] When using the quantum state manipulation method with a 532 nm laser pulse and a 2.87 GHz microwave, since the selection of the 532 nm laser and the 2.87 GHz microwave is determined based on the physical properties and electron transition energy levels of the NV center, the 532 nm laser can initialize the electron spin state of the NV center to the |0> state and excite the photoluminescence (PL) signal of the NV center; the 2.87 GHz microwave can drive the electron spin transition of the NV center, causing it to resonate under the action of an external magnetic field, thereby realizing the precise measurement of the magnetic field.

[0067] When using the quantum state manipulation method, the sensing process is mainly controlled through the following steps: First, initialize the electron spin state of the NV center, and use a 532 nm laser pulse to excite the NV center to enter the initial spin state (usually the |0> state); Then, apply a microwave field. Through a 2.87 GHz microwave pulse, the electron spin of the NV center undergoes a transition, and energy level splitting (i.e., the Zeeman effect) occurs under the action of an external magnetic field; Secondly, optically read the magnetic field information. When the NV center is affected by the magnetic field, the fluorescence emission intensity thereof will change. By detecting the fluorescence intensity with a high-sensitivity photodetector, the change information of the magnetic field can be obtained; Finally, data processing and magnetic field calculation. By the frequency shift amount of the nuclear magnetic resonance signal, the magnitude of the external magnetic field is calculated, and the measurement accuracy is improved by combining the magnetic field reconstruction algorithm.

[0068] The magnetic field reconstruction algorithm is based on Compressed Sensing (CS) and Bayesian inference, and the main process is as follows: Data acquisition: The quantum magnetometer obtains the original magnetic field data of multiple sampling points to form a time series signal; Signal denoising: Using wavelet transform or adaptive filtering method to remove the high-frequency noise and environmental interference signals in the measurement data; Sparse sampling and data compression: Using the compressed sensing method to perform dimensionality reduction processing on the data, extracting key information points, and reducing the computational complexity; Bayesian Inference Magnetic Field Estimation: Using the Bayesian optimization method, calculate the optimal estimated value of the magnetic field distribution based on historical data and current measurement values; Magnetic Field Reconstruction: Based on sparsely sampled data, use the L1-norm optimization method to reconstruct the magnetic field distribution and generate a high-precision magnetic field image; Error Correction: Through a feedback control mechanism, adjust system parameters (such as laser power, microwave frequency, etc.) to improve the magnetic field measurement accuracy and ensure long-term stable operation; Finally, this algorithm can achieve high-precision magnetic field measurements at the nano-tesla (nT) or even pico-tesla (pT) level, providing key data support for the health monitoring of wind turbine pitch systems.

[0069] In this embodiment, the acoustic imaging analysis device includes a spherical array composed of 64 microphones for collecting omnidirectional acoustic signals; the acoustic signal processing flow based on deep learning includes data preprocessing, feature extraction, sound source localization, and 3D reconstruction. These steps are executed sequentially to convert the original acoustic signals into three-dimensional acoustic images available for analysis. Among them, the sources of acoustic signals mainly include the following aspects: Mechanical noise during the operation of the pitch system During the operation of the pitch drive mechanism (motor, hydraulic system), motor noise, hydraulic noise, gear meshing noise, etc. will be generated. These signals can be used to monitor the operation status and health of the equipment.

[0070] For example, abnormal motors may cause high-frequency whistling sounds, while hydraulic system leaks may cause changes in low-frequency noise.

[0071] Aerodynamic noise of the wind on the blades The aerodynamic characteristics of the blades in the wind field determine the interaction between the wind and the blades, thus generating turbulent noise.

[0072] Through acoustic signal analysis, the wind flow characteristics, the distribution of aerodynamic noise on the blade surface, and possible blade surface damage (such as cracks or wear) can be monitored.

[0073] Noise caused by structural vibration During the operation of the blades, bearings, and support structures, vibrations will occur, and these vibrations are transmitted through the air to form acoustic wave signals.

[0074] For example, bearing wear or looseness may cause periodic impact noises, while blade fatigue cracks may change the acoustic resonance characteristics.

[0075] External environmental noise Environmental noises around the wind turbine, such as wind speed changes, thunderstorms, bird impacts, etc., may also be captured by the microphone array.

[0076] The noise filtering algorithm based on deep learning can effectively distinguish the internal noise of the pitch system from the external environmental noise, improving the analysis accuracy of the system.

[0077] In summary, the acoustic signals mainly come from the mechanical operation of the pitch system, the aerodynamic noise under the action of wind, the noise caused by structural vibration, and the external environmental noise. Using a spherical microphone array, these signals can be collected and analyzed, and a three-dimensional acoustic image can be generated through 3D reconstruction technology for equipment health monitoring and fault diagnosis.

[0078] When the data preprocessing device is applied, a customized real-time Linux system and containerized deployment method are adopted. Its data preprocessing algorithms include: using the Isolation Forest algorithm for real-time anomaly detection, adopting the method of wavelet transform combined with adaptive quantization for data compression and extracting time domain, frequency domain, and time-frequency domain features; among them, these algorithms are executed in sequence to process the original data from the multi-modal wireless sensor network.

[0079] Isolation Forest algorithm is used for anomaly detection. The time complexity of this algorithm is O(nlogn), which is suitable for real-time processing at the edge. The anomaly score calculation formula is as follows:

[0080] where is the anomaly score, is the sample, is the number of samples, is the sample 's average path length, is the average path length of the binary tree. Wavelet transform combined with adaptive quantization is used for data compression. Daubechies 4th-order wavelet is used in wavelet transform to balance the compression ratio and signal fidelity. The compression ratio can reach 10:1, while keeping the root mean square error (RMSE) of the reconstructed signal less than 1%.

[0081] The introduction of the data preprocessing device solves the problems of data transmission delay and excessive network load caused by the traditional centralized processing method, improving the real-time performance and reliability of the system.

[0082] When performing real-time anomaly detection, specifically, the following aspects of the collected data are detected: Mechanical anomalies: Detect the data of the pitch sensor and acceleration sensor to judge whether there are phenomena such as overload stress, abnormal vibration, or mechanical shock.

[0083] Electrical anomalies: Detect the current, voltage, and magnetic field data of the pitch motor to identify fault signals such as overcurrent, undervoltage, and magnetic field drift.

[0084] Temperature anomaly: Detect the data of the temperature sensor to determine whether there is abnormal temperature rise, so as to warn of overheating problems in the motor, bearing or hydraulic system.

[0085] Pressure anomaly: Detect the data of the pressure sensor in the hydraulic system to determine whether there are faults such as hydraulic oil leakage, blockage or actuator jamming.

[0086] Acoustic anomaly: Analyze the data of the acoustic sensor array to detect whether there are abnormal noises, such as bearing wear, abnormal gear meshing or sudden changes in motor noise.

[0087] Magnetic field anomaly: Detect the data of the quantum magnetometer, monitor the change of motor magnetic flux, and identify problems such as magnetic saturation, magnetic offset or winding faults.

[0088] When the data processing device is applied, a distributed computing framework and a time series database are used for data processing and storage. Its data analysis pipeline includes a data access layer, a data processing layer, a model service layer and a storage layer, and the data access layer, the data processing layer, the model service layer and the storage layer are sequentially connected by signals; among them, the data access layer is used to receive the parameter data transmitted by the data preprocessing device; the data processing layer is used to further process the parameter data; the model service layer is used to deploy and manage the analysis model; the storage layer is used to store the original data of the parameter data and the processing results of the parameter data.

[0089] In the data processing layer, the received data is further processed, specifically including the following steps: Data cleaning: Remove outliers, fill in missing data, and align data at different time scales.

[0090] Noise reduction processing: Apply filtering algorithms (such as Kalman filtering, wavelet noise reduction, etc.) to remove noise in the sensor signal and improve data quality.

[0091] Feature extraction: Extract time-domain, frequency-domain and time-frequency-domain features, such as mean, variance, spectral energy, short-time Fourier transform (STFT) features, etc.

[0092] Data standardization: Normalize or standardize the data to meet the input requirements of different machine learning and deep learning models.

[0093] Data correlation analysis: Establish the relationship between data from different sensors, such as the relationship between blade angle change and stress distribution.

[0094] Data annotation: Combine the existing fault database or manual intervention to label some data for model training or optimization.

[0095] Real-time calculation: Use streaming processing technologies (such as Flink, Kafka Streams) to perform real-time calculation and update of key data.

[0096] The construction process of the model service layer is as follows: Based on the microservices architecture, use Docker and Kubernetes for container management to ensure the flexible deployment and expansion of the analysis model.

[0097] Provide model inference capabilities based on deep learning frameworks (such as TensorFlow Serving, PyTorch Serve).

[0098] Adopt an API gateway (such as Kong, Traefik) to manage the request routing of the model and ensure load balancing under high-concurrency access.

[0099] Integrate MLOps (Machine Learning Operations), and use MLflow or Kubeflow to implement the lifecycle management of the model.

[0100] The construction process of the analysis model is as follows: Data collection: Extract training data from the storage layer, including sensor historical data, environmental data, and historical fault case data.

[0101] Data preprocessing: Use feature engineering methods (such as PCA dimensionality reduction, time series decomposition, signal processing) to optimize data quality.

[0102] Model selection: Select appropriate algorithms according to different tasks, such as: Pitch system performance prediction: LSTM + attention mechanism Fault diagnosis and classification: XGBoost + random forest ensemble learning Remaining useful life prediction: Physics-informed neural networks (PINNs) Hyperparameter optimization: Use grid search, Bayesian optimization, or genetic algorithms to adjust model parameters and improve the generalization ability of the model Model training: Adopt GPU-accelerated training and use cross-validation methods to evaluate model performance Model validation: Use an independent test set for validation and calculate key metrics (such as RMSE, F1-score, AUC, etc.) Model storage: Store the trained model in a time series database or a model repository for deployment and use.

[0103] In deployment and management, the model service layer is carried out according to the following conditions: System operating status: When the wind turbine pitch system detects an anomaly or a drastic environmental change, a new model deployment request is triggered.

[0104] Model performance evaluation: Regularly compare the accuracy of the currently running model. If the accuracy of the new model is improved, replace the old model.

[0105] Computing resource usage: Based on the load conditions of cloud computing or edge computing, decide whether to perform local inference or cloud inference.

[0106] Data change trend: When the distribution of the input data drifts (concept drift), automatically trigger model update or migration.

[0107] The locations for analyzing model deployment are as follows: Edge computing side: Analysis tasks with low latency requirements (such as anomaly detection, short-term prediction) are preferentially deployed on the edge computing unit of the wind turbine to reduce data transmission latency.

[0108] Cloud server: Large-scale training tasks (such as life prediction, complex fault analysis) are deployed in the cloud computing center to optimize the model using high-performance computing resources.

[0109] Hybrid architecture: Some models (such as pitch angle optimization) can run on both the edge side and the cloud simultaneously. The edge side performs real-time inference, and the cloud performs long-term analysis and model update.

[0110] During the management process, the service layer manages the analysis model as follows: Version control: Use MLflow or DVC (Data Version Control) to record the training parameters, performance metrics, and applicable scenarios of different versions of the model.

[0111] Automatic update: Set a threshold. When the prediction error of the existing model exceeds the set range, automatically trigger the deployment process of the new model.

[0112] Load balancing: Adopt container orchestration (such as Kubernetes) to dynamically allocate computing resources to ensure that multiple model inference tasks can be executed in parallel efficiently.

[0113] Log monitoring: Integrate Prometheus + Grafana to monitor the inference time, call frequency, prediction accuracy, etc. of the model to prevent model failure.

[0114] Rollback mechanism: When an abnormal situation (such as an increase in the false alarm rate) occurs after the new model is deployed, it can be quickly rolled back to the previous stable version.

[0115] In this embodiment, the analysis device includes a pitch system performance prediction model, a fault diagnosis and classification model, and a life prediction model. The pitch system performance prediction model, the fault diagnosis and classification model, and the life prediction model are signal-connected to the data processing device. Among them, the pitch system performance prediction model is established based on the long short-term memory network and the attention mechanism. The pitch system performance prediction model is constructed by combining the long short-term memory network (LSTM) with the attention mechanism. The LSTM can effectively process the time series data of the pitch system and capture long-term dependence relationships, while the attention mechanism (Attention) can enhance the attention to key data points and improve the prediction accuracy.

[0116] The pitch system performance prediction uses the long short-term memory network (LSTM) combined with the attention mechanism. The network structure includes: an input layer (24 features), two LSTM layers (128 and 64 units), an attention layer, a fully connected layer (32 neurons), and an output layer (1 neuron). The Adam optimizer is used in the training process, the learning rate is set to 0.001, the batch size is 64, the number of training epochs is 100, and an early stopping strategy is adopted to prevent overfitting.

[0117] Fault diagnosis and classification use an ensemble learning method, combining three models: Random Forest, XGBoost, and LightGBM. The ensemble strategy uses soft voting, which is a weighted average based on the probability distributions output by each model.

[0118] Life prediction uses a deep learning method based on a physical model, namely Physics-Informed Neural Networks (PINNs). The loss function is as follows:

[0119] Among them, is the data-driven loss, is the physical model constraint loss, is the L2 regularization term, and are weight hyperparameters. The introduction of the analysis device solves the limitation of traditional analysis methods in dealing with complex nonlinear problems and greatly improves the prediction accuracy and fault diagnosis ability of the system.

[0120] During specific operation, the pitch system performance prediction model mainly predicts the following performances of the pitch system: Blade pitch angle response characteristics: Predict the change trend of the pitch angle under different wind speeds and loads to optimize the control strategy.

[0121] Operating state of the actuator: Predict the dynamic response of the pitch drive motor or hydraulic actuator to judge whether there is delay or abnormality.

[0122] Vibration characteristics: Analyze the vibration signals of the blades during the pitch-changing process to predict potential structural fatigue or resonance risks.

[0123] Temperature change trend: Predict the temperature rise of the pitch motor, hydraulic system, or bearings to prevent performance degradation or failures caused by overheating.

[0124] Energy consumption characteristics: Predict the energy consumption of the pitch system under different working loads to optimize the energy management strategy.

[0125] Fault diagnosis and classification model, established based on ensemble learning, specifically including the following steps: Random Forest: Used to extract features of sensor data and perform preliminary classification.

[0126] Gradient Boosting Decision Trees (XGBoost, LightGBM): Combine multiple features for optimized classification to improve the accuracy of fault identification.

[0127] Support Vector Machine (SVM): Used to handle non-linear fault patterns and improve the fault identification ability in boundary cases.

[0128] Deep Neural Network (DNN): Trained on a large-scale dataset to improve the classification ability for complex faults.

[0129] The fault diagnosis and classification model established based on the above steps is mainly used to identify whether there are abnormalities during the operation of the pitch system, such as sensor signal drift, motor overload, etc.; then distinguish different types of faults, such as blade jamming, pitch actuator malfunction, sensor failure, etc.; in addition, combine sensor data analysis to find the cause of the fault, such as motor overheating may be caused by poor lubrication or overloaded operation; combine long-term monitoring data to predict the health status of each component of the pitch system and provide preventive maintenance suggestions.

[0130] Deep learning life prediction model, the specific construction process is as follows: Constructed based on the method combining physical models (Physics-Informed Neural Networks, PINNs) and deep learning: Physical model part: Use finite element analysis (FEA) to simulate the stress conditions of the pitch system and provide an initial life prediction curve.

[0131] Deep learning part: Use an LSTM+CNN hybrid architecture and train with historical operation data to optimize the life prediction accuracy.

[0132] Adopt the Residual Learning method to correct the error between the physical model and the real data to improve the prediction accuracy.

[0133] The deep learning life prediction model established through the above steps is mainly used for blade fatigue life prediction, bearing wear trend assessment, motor life estimation, and hydraulic system aging analysis. Among them, for blade fatigue life prediction: analyze the stress changes of the blade during long-term operation and predict possible fatigue damage.

[0134] Bearing wear trend assessment: Predict the remaining service life of the bearing through data such as magnetic field, vibration, and temperature.

[0135] Motor life estimation: Based on operating current, temperature, and magnetic field data, evaluate the health status of the pitch motor and predict when replacement or maintenance is required.

[0136] Hydraulic system aging analysis: Predict the aging degree of the hydraulic system through hydraulic oil pressure, temperature, and viscosity data, and optimize the maintenance plan.

[0137] In this embodiment, the policy generation device dynamically generates test policies based on the reinforcement learning framework, which is implemented using the SoftActor-Critic algorithm, and performs knowledge-learning fusion through the expert knowledge system described by the ontology. Among them, the reinforcement learning framework generates test policies according to the system state and historical data, and the expert knowledge system optimizes and adjusts the generated policies.

[0138] The specific implementation of the SAC algorithm includes two neural networks: a policy network and a Q-value network. The policy network uses a three-layer fully connected network with hidden layer sizes of 256 and 128 neurons respectively, and uses the ReLU activation function. The Q-value network also uses a similar structure, but the output layer uses a linear activation function. The learning rate is set to 3e-4, and the discount factor γ is 0.99.

[0139] The reward function is designed as follows:

[0140] Among them, is the test coverage rate, is the abnormal detection effect, is the resource consumption, 、 and are weight coefficients.

[0141] The introduction of the policy generation device solves the problem that traditional fixed test policies are difficult to adapt to complex changing environments, and improves the pertinence and efficiency of testing.

[0142] Specifically, the analysis results all include pitch system performance analysis results, fault detection results, health status assessment results, and environmental impact factor analysis. Pitch system performance analysis results: include the response time of pitch angle adjustment, power consumption of the actuator, vibration characteristics, etc.

[0143] Fault detection result: Identify whether there are abnormalities in the system, such as motor overload, abnormal blade vibration, etc.

[0144] Health status assessment result: Provide the current health status score of the system, including the wear degree of the actuator, sensor error, etc.

[0145] Analysis of environmental impact factors: Include the impact of wind speed, temperature, humidity, etc. on the operation of the pitch system.

[0146] The test strategy includes the following: Adjustment of sensor sampling frequency: Adjust the data acquisition frequency according to the current system state to improve data acquisition efficiency.

[0147] Dynamic adjustment strategy of pitch angle: Change the pitch angle during the test and observe the system response.

[0148] Load test strategy: Apply different wind loads to the pitch system and analyze the system performance under extreme working conditions.

[0149] Environmental simulation test: Verify the system adaptability by simulating the operation under different temperature, humidity, and wind speed conditions through data.

[0150] Intelligent fault injection: Actively simulate certain faults (such as sensor data offset, abnormal motor current) to test the system's fault detection ability.

[0151] Among them, different analysis results will trigger corresponding test strategies to optimize the system performance and ensure its reliability, specifically as follows: When the analysis result shows that there is a lag in pitch angle adjustment, the test strategy will adjust the sampling frequency of the sensor to improve the time accuracy of data acquisition, so as to analyze the response delay of the pitch system and detect whether there is mechanical resistance of the actuator or delay of control instructions.

[0152] If abnormal blade vibration is detected, the test strategy will conduct high-frequency vibration tests, and analyze the vibration mode of the system through high-resolution data of the acceleration sensor to determine whether the vibration comes from blade resonance, bearing wear, or mechanical instability of the actuator.

[0153] When it is found that the pitch motor temperature is too high, the test strategy will trigger a long-term load test, simulate working conditions with different wind speeds and pitch angle adjustment frequencies, to evaluate the heat dissipation ability of the motor and determine whether there is overload operation or cooling system failure.

[0154] If the analysis results indicate a fatigue trend in the blade structure, the test strategy will apply periodic loads, monitor the long-term stress conditions of the blade through strain sensors and acceleration sensors, evaluate the material fatigue degree, and predict the possible future structural failure time.

[0155] When it is detected that the response time of the actuator becomes longer, the test strategy will increase the rapid switching operation of the pitch actuator to observe its reaction speed, and analyze whether there is mechanical jamming, poor lubrication, or response delay in the control system through an acoustic sensor array and an angle encoder.

[0156] If sensor signal drift is found, the test strategy will initiate the sensor self-calibration process, including multi-sensor data fusion and environmental compensation algorithms, to eliminate the influence of external interference on sensor data and ensure the accuracy of measurement data.

[0157] Through the dynamic adjustment of the above targeted test strategy, the system can timely detect potential faults and improve the operating efficiency and safety of the pitch system.

[0158] In this embodiment, the calibration device realizes self-calibration by using multi-sensor data fusion technology, including a multi-sensor data fusion algorithm based on unscented Kalman filter; environmental factor compensation methods for temperature, humidity, and vibration; an anomaly detection method based on a finite element analysis and data-driven hybrid model; these methods work together to calibrate and detect anomalies in sensor data and ensure the long-term accuracy of measurement.

[0159] Through the multi-sensor data fusion algorithm of unscented Kalman filter, self-calibration is performed on the temperature sensor (monitoring the environment and equipment temperature), acceleration sensor (monitoring the vibration impact), angle encoder (temperature change may cause drift), and acoustic sensor array (humidity may affect the acoustic characteristics); the specific calibration process is as follows: Environmental data acquisition: Real-time monitor the temperature, humidity, and vibration level of the wind power pitch system, and establish a time-series database to record the change trend.

[0160] Model construction: Adopt a regression model based on historical data to calculate the relationship between sensor readings and environmental parameters, such as the drift impact of temperature on the angle encoder, the impact of humidity on the attenuation of acoustic signals, etc.

[0161] Temperature compensation: Use linear interpolation or polynomial fitting methods to correct the sensor drift caused by temperature changes.

[0162] Humidity compensation: Perform adaptive gain adjustment on the sensitivity decline of the acoustic sensor array to improve signal stability.

[0163] Vibration Compensation: Based on the data from the acceleration sensor, analyze the impact of vibration on the angle encoder, and perform signal filtering or offset correction.

[0164] Real-time Update of Calibration Parameters: Apply the compensation model to real-time measurement data, dynamically adjust the sensor output, and make it maintain accurate measurement under different environmental conditions.

[0165] Self-calibration of Strain Sensors, Pressure Sensors, and Acceleration Sensors Using an Anomaly Detection Method Based on a Hybrid Model of Finite Element Analysis and Data-Driven. The calibration process is as follows: Establish a finite element analysis model: Based on the structural parameters of the wind turbine pitch system, use the finite element method (FEA) to simulate the stress distribution, vibration response, and pressure changes of the blade under different wind speeds and pitch angles.

[0166] Data Acquisition and Comparison: Obtain the real-time data of strain, pressure, and vibration sensors, and compare them with the results of finite element simulation calculations to analyze data deviations.

[0167] Anomaly Detection: Adopt a data-driven hybrid model, such as PCA (Principal Component Analysis) + Random Forest, to detect abnormal patterns in sensor data. For example, if the error between the measured value of the strain sensor and the calculated value of the finite element analysis exceeds the threshold, it may indicate sensor offset or failure. Abnormal measured values of the pressure sensor may indicate leakage or blockage in the hydraulic system. Deviation of the vibration data of the acceleration sensor from the normal spectrum may indicate looseness or fatigue cracks in the system.

[0168] Correct Sensor Data: Correct abnormal sensor data, such as adjusting the strain measurement value using an error compensation curve, or performing adaptive offset compensation on the output of the pressure sensor.

[0169] Sensor Health Assessment: If the deviation exceeds the set threshold and cannot be corrected through calibration, trigger an alarm for sensor replacement or maintenance to prevent incorrect measurements from affecting system control.

[0170] In this embodiment, the energy management device includes a piezoelectric energy harvesting component, a wireless charging component, and a sleep component. The piezoelectric energy harvesting component, the wireless charging component, and the sleep component are electrically connected to the control device. Among them, the piezoelectric energy harvesting component is used to convert the mechanical vibration of the pitch system into electrical energy; the wireless charging component is used to provide electrical energy; the sleep component is used to adjust the working mode of the multimodal wireless sensing device. The piezoelectric energy harvesting component converts mechanical vibration into electrical energy with a power density of 10 - 100 μW / cm³; the wireless charging component provides additional energy when needed, and the transmission efficiency at a distance of 3 meters is 40%; the sleep component dynamically adjusts the sensor working mode according to the system state, and can reduce the average power consumption of the sensor by more than 50%.

[0171] In a second aspect, a wind turbine pitch testing method based on a wireless sensor network is provided. The method is carried out by using the wind turbine pitch testing system based on a wireless sensor network as described above, and includes: S101. Using a multimodal wireless sensing device to collect parameter data of key parts in the wind turbine pitch system; Exemplarily, strain sensors are arranged on the support structure of the pitch system bearing and the blade root to detect the mechanical stress borne by the blade during the pitching process; acceleration sensors are arranged on the pitch system motor housing and the blade root to detect the vibration characteristics during the operation of the pitch system; pressure sensors are arranged on the hydraulic pipelines, cylinder inner cavities and blade surfaces of the hydraulic actuator and electric actuator of the pitch system to monitor pressure changes; temperature sensors are arranged on the drive motor, cylinder and main bearing seat of the pitch system to monitor temperature changes; an angle encoder is arranged on the rotating shaft and the blade root of the pitch system to measure the blade pitch angle; an acoustic sensor array is arranged on the pitch system housing, main bearing and motor to collect the acoustic characteristics of the pitch system operation; a quantum magnetometer is arranged in the bearing cavity and motor stator of the pitch system to measure magnetic field changes.

[0172] S102. Based on a data preprocessing device, performing real-time processing and anomaly detection on the parameter data; Exemplarily, anomaly detection includes detecting the data of the strain sensor and the acceleration sensor to judge whether there are phenomena such as overload stress, abnormal vibration or mechanical shock; detecting the current, voltage and magnetic field data of the pitch motor to identify fault signals such as overcurrent, undervoltage and magnetic field drift; detecting the data of the temperature sensor to judge whether there is abnormal temperature rise to warn of overheating problems of the motor, bearing or hydraulic system; detecting the pressure sensor data of the hydraulic system to judge whether there are faults such as hydraulic oil leakage, blockage or actuator jamming; analyzing the data of the acoustic sensor array to detect whether there are abnormal noises, such as bearing wear, abnormal gear meshing or sudden change of motor noise; detecting the data of the quantum magnetometer to monitor the change of motor magnetic flux and identify problems such as magnetic saturation, magnetic offset or winding faults. By arranging a variety of sensors at key parts of the wind turbine pitch system and performing real-time processing and anomaly detection on the collected data based on the data preprocessing device, not only can the quality and accuracy of the data be improved, but also the self-diagnosis ability of the system can be enhanced, and the system performance and maintenance strategy can be optimized. S103. Deeply analyze the parameter data after real-time processing through a data processing device. Exemplarily, through the data processing device, it is possible to comprehensively, meticulously, and deeply analyze the parameter data that has been real-time collected and preliminarily processed. It not only has efficient data processing capabilities but also applies advanced algorithms and models to ensure the accuracy and reliability of the analysis. During the deep analysis process, the data processing device will conduct multi-dimensional and multi-level mining on the parameter data after real-time processing, thereby revealing the hidden laws and characteristics behind the data. For example, through the data analysis of the strain sensor, it is possible to deeply understand the changes in the mechanical stress borne by the blade during the pitch change process and the impact of these changes on the system performance; through the data analysis of the acceleration sensor, it is possible to master the dynamic evolution of the system vibration characteristics and timely identify potential vibration faults; through the data analysis of the pressure, temperature, and angle encoder, it is possible to comprehensively monitor the operating state of the system and ensure that each component operates within the normal working range. In addition, the data processing device also conducts intelligent analysis and prediction on the parameter data after real-time processing. It can predict the future operating state of the system and possible problems based on the change trends of historical data and current data, providing forward-looking decision-making support for the operation and maintenance personnel. The results of the deep analysis can not only provide strong data support for the operation and maintenance management of the system but also provide a scientific basis for the performance optimization, fault diagnosis, and intelligent upgrade of the system. The operation and maintenance personnel can adjust the operation and maintenance strategy in a timely manner according to the analysis results, optimize the system performance, and improve the reliability and security of the system. At the same time, the deep analysis can also provide valuable feedback information for the research and development and improvement of the system, promoting the continuous innovation and development of the system.

[0173] S104. Predict the lifespan of the wind power pitch system by relying on the analysis device and combining the analysis data. Exemplarily, by relying on the analysis device, in-depth and detailed analysis can be carried out on the massive parameter data generated during the operation of the wind power pitch system. During the analysis process, the real-time data collected by various sensors such as strain sensors, acceleration sensors, pressure sensors, temperature sensors, and angle encoders, as well as the results after preprocessing of these data, are fully utilized. Through the comprehensive analysis and comparison of these data, the state changes of the wind power pitch system at each operation stage can be comprehensively understood, as well as the impact of these changes on the system lifespan. Specifically, by analyzing the data of the strain sensor, the impact of the mechanical stress borne by the blade during the pitching process on the material fatigue lifespan can be evaluated; by the data of the acceleration sensor, the potential threat of the system vibration characteristics to the structural integrity can be judged; by the data of the pressure and temperature sensors, the operation states of the hydraulic system and the drive motor can be monitored to timely detect possible abnormalities such as leakage and overheating; by the data of the angle encoder, the accuracy and stability of the blade pitching angle can be mastered to ensure the reliability of the system operation. Combining these analysis data, a prediction model can be used to quantitatively evaluate the lifespan of the wind power pitch system, and the lifespan situation of the system in a future period can be predicted relatively accurately. Through the prediction results, the lifespan status of the system can be timely understood, providing targeted maintenance suggestions for the operation and maintenance personnel, thereby extending the service life of the system and reducing the operation and maintenance costs. In addition, the lifespan prediction results can be continuously compared and verified with the actual operation data of the system to continuously optimize the accuracy and reliability of the prediction model. Through continuous iteration and optimization, the lifespan prediction can be made more in line with the actual operation situation of the system, providing more scientific and effective decision-making support for the operation and maintenance management of the wind power pitch system.

[0174] S105. Dynamically adjust the test strategy according to the life prediction results. Exemplarily, when the analysis results indicate that there is a lag in the pitch angle adjustment, the test strategy will adjust the sampling frequency of the sensor to improve the time accuracy of data acquisition, so as to analyze the response delay of the pitch system and detect whether there is mechanical resistance in the actuator or delay in the control instruction. If abnormal blade vibration is detected, the test strategy will conduct high-frequency vibration tests, analyze the vibration mode of the system through the high-resolution data of the acceleration sensor, to determine whether the vibration comes from blade resonance, bearing wear, or mechanical instability of the actuator. When it is found that the temperature of the pitch motor is too high, the test strategy will trigger a long-term load test, simulate the working conditions of different wind speeds and pitch angle adjustment frequencies, to evaluate the heat dissipation capacity of the motor and determine whether there is overload operation or cooling system failure. If the analysis results indicate that there is a fatigue trend in the blade structure, the test strategy will apply periodic loads, monitor the long-term stress conditions of the blade through strain sensors and acceleration sensors, evaluate the material fatigue degree, and predict the possible future structural failure time. When it is detected that the response time of the actuator becomes longer, the test strategy will increase the rapid switching operations of the pitch actuator to observe its reaction speed, and analyze whether there is mechanical jamming, poor lubrication, or response delay of the control system through the acoustic sensor array and angle encoder. If sensor signal drift is found, the test strategy will start the sensor self-calibration process, including multi-sensor data fusion and environmental compensation algorithms, to eliminate the influence of external interference on sensor data and ensure the accuracy of measurement data. Through the dynamic adjustment of the above targeted test strategies, the system can timely detect potential faults and improve the operation efficiency and safety of the pitch system.

[0175] In a third aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned wind power pitch test method based on a wireless sensor network are implemented.

[0176] In a fourth aspect, a computer program product is provided, including computer instructions, and the computer instructions direct a computing device to perform the operations corresponding to the above-mentioned wind power pitch test method based on a wireless sensor network.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than limit the scope of its protection. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that after reading the present invention, various changes, modifications, or equivalent replacements can still be made to the specific implementation manners of the invention. However, these changes, modifications, or equivalent replacements are all within the scope of the protection of the pending claims of the invention.

Claims

1. A wind turbine pitch test system based on a wireless sensor network, characterized in that Including: A multimodal wireless sensing device, a quantum sensing integration device, an acoustic imaging analysis device, a data preprocessing device, a data processing device, an analysis device, a strategy generation device, a calibration device, an energy management device, and a control device. The multimodal wireless sensing device, the quantum sensing integration device, the acoustic imaging analysis device, the data preprocessing device, the data processing device, the analysis device, the strategy generation device, the calibration device, and the energy management device are electrically connected to the control device; the multimodal wireless sensing device, the quantum sensing integration device, and the acoustic imaging analysis device are signal-connected to the data preprocessing device, and the data preprocessing device, the analysis device, the strategy generation device, and the calibration device are all signal-connected to the data processing device; Among them, the multimodal wireless sensing device is used to collect parameter data of the wind power pitch system; The data preprocessing device is used to preprocess the parameter data; The data processing device is used to analyze the preprocessed parameter data and obtain analysis data; The analysis device is used to receive the analysis data and predict the life of the wind power pitch system using the analysis data; The strategy generation device is used to generate a test strategy according to the analysis data; The calibration device is used to calibrate the multimodal wireless sensing device; The quantum sensing integration device is used to detect the magnetic field of the wind power pitch system; The acoustic imaging analysis device is used to capture and analyze the acoustic characteristics of the wind power pitch system; The energy management device is used to provide energy for the multimodal wireless sensing device.

2. The wind power pitch test system based on a wireless sensor network according to claim 1, wherein The multimodal wireless sensing device includes a strain sensor, an acceleration sensor, a pressure sensor, a temperature sensor, an angle encoder, an acoustic sensor array, and a quantum magnetometer; Among them, the strain sensor is arranged on the support structure of the pitch system bearing and the blade root to detect the mechanical stress borne by the blade during the pitching process; The acceleration sensor is arranged on the pitch system motor housing and the blade root to detect the vibration characteristics during the operation of the pitch system; The pressure sensor is arranged on the hydraulic pipelines, the inner cavity of the oil cylinder, and the blade surface of the hydraulic actuator and the electric actuator of the pitch system to monitor the pressure change; The temperature sensor is arranged on the drive motor, the oil cylinder, and the main bearing seat of the pitch system to monitor the temperature change; The angle encoder is arranged on the rotating shaft and the blade root of the pitch system to measure the blade pitch angle; The acoustic sensor array is arranged on the pitch system housing, the main bearing, and the motor to collect the acoustic characteristics during the operation of the pitch system; The quantum magnetometer is arranged in the bearing cavity and the motor stator of the pitch system to measure the magnetic field change.

3. The wind power pitch test system based on a wireless sensor network according to claim 1, characterized in that, The data processing device includes a data access layer, a data processing layer, a model service layer, and a storage layer. The data access layer, the data processing layer, the model service layer, and the storage layer are signal-connected in sequence; Among them, the data access layer is used to receive the parameter data transmitted by the data preprocessing device; The data processing layer is used to further process the parameter data; The model service layer is used to deploy and manage the analysis model; The storage layer is used to store the original data of parameter data and the processing results of parameter data.

4. The wind power pitch testing system based on a wireless sensor network according to claim 1, wherein The analysis device includes a pitch system performance prediction model, a fault diagnosis and classification model, and a life prediction model. The pitch system performance prediction model, the fault diagnosis and classification model, and the life prediction model are signal-connected to the data processing device.

5. The wind power pitch testing system based on the wireless sensor network according to claim 1, wherein The energy management device includes a piezoelectric energy harvesting component, a wireless charging component, and a sleep component. The piezoelectric energy harvesting component, the wireless charging component, and the sleep component are electrically connected to the control device; Among them, the piezoelectric energy harvesting component is used to convert the mechanical vibration of the pitch system into electrical energy; The wireless charging component is used to provide electrical energy; The sleep component is used to adjust the working mode of the multi-modal wireless sensing device.

6. A wind power pitch testing method based on a wireless sensor network, characterized in that, The method is carried out by using the wind power pitch test system based on a wireless sensor network according to any one of claims 1-5, and includes: Using a multi-modal wireless sensing device to collect parameter data of key parts in the wind power pitch system; Based on the data preprocessing device, performing real-time processing and anomaly detection on the parameter data; Through the data processing device, performing in-depth analysis on the parameter data after real-time processing; Relying on the analysis device and combining with the analysis data, predicting the life of the wind power pitch system; According to the life prediction result, dynamically adjusting the test strategy.

7. The wind power pitch testing method based on a wireless sensor network according to claim 6, characterized in that Based on the data preprocessing device, performing real-time processing and anomaly detection on the parameter data, specifically including: Using the isolation forest algorithm to perform real-time anomaly detection on the parameter data; Adopting wavelet transform combined with an adaptive quantization method to perform data compression on the parameter data after anomaly detection; Extracting time-domain, frequency-domain, and time-frequency domain features from the parameter data after data compression.

8. The method for testing the pitch control of a wind turbine based on a wireless sensor network according to claim 7, wherein During the process of performing real-time anomaly detection, specifically: Detecting mechanical anomalies of the pitch system to determine whether there is overloading stress, abnormal vibration, or mechanical shock in the pitch system; Detecting the current, voltage, and magnetic field data of the pitch motor to identify overcurrent, undervoltage, and magnetic field drift fault signals; Detecting the temperature of the pitch system to determine whether there is abnormal temperature rise in the pitch system; Detecting the pressure data of the pitch system to determine whether there are faults such as hydraulic oil leakage, blockage, or actuator jamming; Analyzing the acoustic data of the pitch system to detect whether there is abnormal noise; Detecting the magnetic field data of the pitch system to identify magnetic saturation, magnetic offset, or winding faults.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the wind power pitch test method based on a wireless sensor network according to any one of claims 1-6.

10. A computer program product comprising computer instructions, characterized in that, The computer instructions instruct the computing device to perform the operations corresponding to the wind power pitch test method based on a wireless sensor network according to any one of claims 1-6.

Images