Preventive maintenance and fault diagnosis system and method for offshore wind turbine generator

Abstract

The invention discloses a preventive maintenance and fault diagnosis system and method for an offshore wind turbine generator, and relates to the technical field of offshore wind turbine generators, a sound collector hardware deployment module adopts a high-strength corrosion-resistant bracket and a waterproof sealing design for a severe offshore environment, and collects voiceprint signals and environmental data of each component; the fault real-time prediction module is used for calculating a fault evaluation value based on multi-source data fusion so as to pre-judge a fault in advance; the fault type analysis module accurately identifies a fault type and a danger level through comparison with a database; and the preventive maintenance strategy analysis module respectively executes dynamic monitoring and hidden danger pre-control, planned shutdown and accurate repair, and emergency shutdown and emergency rescue strategies according to the fault danger level. The system can improve the fault diagnosis accuracy, optimize the operation and maintenance strategy, guarantee the stable operation of the offshore wind turbine generator, and reduce the operation and maintenance cost and safety risk.

Description

Technical Field

[0001] This invention relates to the field of offshore wind turbine technology, specifically to a preventive maintenance and fault diagnosis system and method for offshore wind turbines. Background Technology

[0002] With the continued growth of global demand for clean energy, offshore wind power has become a key development area in the energy sector due to its abundant resources and high power generation efficiency. However, offshore wind turbines operate in harsh environments such as high salt spray, strong winds and waves, and high humidity. Key components, such as blades, gearboxes, and generators, are prone to wear and failure, leading to unplanned downtime. This not only reduces power generation efficiency but also significantly increases operation and maintenance costs. Therefore, a preventive maintenance and fault diagnosis system and method for offshore wind turbines is needed.

[0003] Existing technologies, such as the invention application patent with publication number CN118898474A, disclose a wind farm maintenance and operation system and method. The system includes an output power prediction module, an anomaly comparison and analysis module, an impact level calculation module, a priority ranking module, a maintenance route calculation module, and a component inspection and maintenance module. In this invention, by analyzing the relationship between various environmental data and wind power generation, energy output is predicted and compared with actual power output data. Abnormal power generation sites and faulty equipment are identified, enabling early warning and intervention of faults, reducing the fault incidence rate. By calculating the impact scores of various fault types, the severity and repair time of various faults are assessed. Fault handling is prioritized based on maintenance time and resources, improving the utilization efficiency of maintenance resources, reducing equipment downtime, optimizing maintenance routes according to wind farm layout and maintenance resources, reducing maintenance costs and improving operation and maintenance efficiency, and ensuring the long-term stable operation of equipment.

[0004] Regarding the above-mentioned solutions, the inventors of this application have discovered at least the following technical problems: 1. Traditional methods rely on single sensor data or simple signal processing techniques, making it difficult to extract effective fault features from complex mechanical noise and environmental interference. In scenarios with small samples and unlabeled data, existing machine learning models have low prediction accuracy and cannot accurately calculate fault assessment values, easily leading to missed or misjudged faults, resulting in delayed fault detection, missing the best repair opportunity, and causing more serious equipment damage.

[0005] 2. Existing maintenance strategies are mostly reactive or periodic inspections, lacking a dynamic, tiered response mechanism based on real-time fault status. Reactive maintenance leads to frequent unplanned outages, severely impacting power generation; periodic inspections, on the other hand, suffer from over-maintenance or under-maintenance, increasing operating costs and failing to guarantee safe equipment operation. Furthermore, existing strategies cannot accurately match maintenance plans according to fault hazard levels, making it difficult to balance maintenance efficiency, power generation benefits, and equipment safety.

[0006] 3. The existing system does not fully consider the impact of complex environmental factors such as high salt spray, strong winds and waves, and electromagnetic interference on fault diagnosis. Environmental interference can easily lead to the masking or misjudgment of fault characteristics, reducing the reliability of the diagnostic system and making it difficult to meet the requirements of long-term stable operation of offshore wind turbines. Summary of the Invention

[0007] To address the aforementioned technical shortcomings, the purpose of this invention is to provide a preventive maintenance and fault diagnosis system and method for offshore wind turbines.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: In the first aspect, the present invention provides a preventive maintenance and fault diagnosis system for offshore wind turbines, including: a sound acquisition hardware deployment module: used to deploy sound acquisition hardware on each component of the target offshore wind turbine, and after the deployment is completed, several acquisition time points are set, and then the acoustic signature signal data and environmental impact data corresponding to each component are collected at each acquisition time point.

[0009] Real-time fault prediction module: It is used to analyze the acoustic signal data and environmental impact data of each component at each acquisition time point to obtain the fault assessment value of each component at each acquisition time point, and then predict whether a fault will be found in each component in the future cycle at each acquisition time point.

[0010] Fault Type Analysis Module: When it is predicted that a fault will be found in a certain component within a future period at a certain data acquisition time, the module analyzes the fault type corresponding to that component at that data acquisition time and the fault hazard level corresponding to that component at that data acquisition time.

[0011] Preventive maintenance strategy analysis module: This module is used to analyze the preventive maintenance strategy corresponding to the component at the time of data collection based on the fault hazard level of the component at that time.

[0012] In a second aspect, the present invention provides a method for preventive maintenance and fault diagnosis of offshore wind turbines, comprising: Step 1, deployment of sound acquisition hardware: deploying sound acquisition hardware for each component in the target offshore wind turbine; after deployment, setting several acquisition time points, and then acquiring acoustic signature data and environmental impact data corresponding to each component at each acquisition time point.

[0013] Step 2: Real-time Fault Prediction: Based on the acoustic signal data and environmental impact data corresponding to each component at each acquisition time point, the fault assessment value corresponding to each component at each acquisition time point is obtained, and then it is predicted whether a fault will be found in each component in the future cycle at each acquisition time point.

[0014] Step 3: Fault type analysis: When it is predicted that a fault will be found in a certain component within a future period at a certain data acquisition time point, the fault type corresponding to that component at that data acquisition time point is analyzed, and the fault hazard level corresponding to that component at that data acquisition time point is also analyzed.

[0015] Step 4: Analysis of preventive maintenance strategies: Based on the fault hazard level of the component at the time of data collection, the preventive maintenance strategies corresponding to the component at the time of data collection are analyzed.

[0016] The beneficial effects of this invention are as follows: 1. In the embodiments of this invention, the sound acquisition device hardware deployment module adopts a high-strength corrosion-resistant alloy bracket, combined with a sealed, waterproof, and salt spray-proof design, ensuring stable acquisition of acoustic signature signal data and environmental impact data in harsh marine environments. By deploying acquisition devices on various key components of the wind turbine, acoustic signature signal data such as wavelet packet energy values ​​and linear prediction cepstral coefficients, as well as environmental data such as wind speed and ocean current speed, can be obtained, providing a comprehensive and reliable data foundation for subsequent analysis. Compared with traditional monitoring methods, the accuracy and stability of data acquisition are significantly improved.

[0017] 2. In this embodiment of the invention, efficient fault prediction and diagnosis are achieved: The real-time fault prediction module calculates fault assessment values ​​based on multi-source data fusion. Through scientific normalization processing and calculation formulas, and by comprehensively considering acoustic signal assessment values ​​and environmental impact factors, it can accurately predict whether each component will fail in the future cycle. The fault type analysis module, in conjunction with database comparison, quickly determines the fault type and hazard level, changing the previous situation of relying on manual experience for judgment, which was inefficient and prone to errors. It realizes early warning and accurate diagnosis of faults, and improves the accuracy of fault identification.

[0018] 3. In this embodiment of the invention, intelligent maintenance strategy formulation: The preventive maintenance strategy analysis module executes different strategies based on the fault hazard level. Dynamic monitoring and hazard prevention strategies at the early warning level can promptly curb fault development; planned shutdown and precise repair strategies at the fault level balance maintenance and power generation needs; and emergency shutdown and emergency rescue strategies at the severe fault level ensure unit safety. This tiered strategy makes maintenance work more targeted and planned, reduces unplanned downtime, lowers operation and maintenance costs, and improves the operational reliability and power generation efficiency of offshore wind turbines.

[0019] 4. The embodiments of the present invention have strong environmental adaptability: The system fully considers the complex environment of high salt spray and strong winds and waves at sea. From hardware protection to data analysis, it is designed to be adaptable, effectively reducing the interference of environmental factors on monitoring and diagnostic results. Compared with similar systems, it has better stability and reliability in harsh environments, providing strong technical support for the continuous and stable development of offshore wind power. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the system module connections of the present invention.

[0022] Figure 2 This is a flowchart illustrating the implementation steps of the method of the present invention. Detailed Implementation

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0024] Examples of embodiments of the present invention Figure 1 As shown, a preventive maintenance and fault diagnosis system for offshore wind turbines includes: a sound acquisition hardware deployment module, a real-time fault prediction module, a fault type analysis module, a preventive maintenance strategy analysis module, and a database.

[0025] The real-time fault prediction module is connected to the sound acquisition device hardware deployment module and the fault type analysis module, respectively. The fault type analysis module is connected to the preventive maintenance strategy analysis module, and the database is connected to the fault type analysis module.

[0026] It should be noted that the database is used to store the fault assessment value range corresponding to each fault type.

[0027] Sound acquisition hardware deployment module: used to deploy sound acquisition hardware on each component of the target offshore wind turbine. After deployment, several acquisition time points are set, and then the acoustic signal data and environmental impact data of each component are collected at each acquisition time point.

[0028] In a specific embodiment, the hardware deployment of sound acquisition devices for each component of the target offshore wind turbine is carried out as follows: When deploying the sound acquisition devices for each component of the target offshore wind turbine, a high-strength corrosion-resistant alloy bracket is used. The bracket is firmly fixed to the surface of each component or adjacent structure by means of expansion bolts, welding or buckles. The sound acquisition device is installed on the bracket. The connection between the sound acquisition device and the bracket is sealed with a silicone sealing ring, and the interface is wrapped with waterproof tape to ensure the waterproof and salt spray resistance of the equipment.

[0029] It should be noted that the main components of an offshore wind turbine include, but are not limited to, wind turbine blades, hub, main shaft system, gearbox, generator, nacelle, tower, converter, and yaw system. Among them, the wind turbine blades are responsible for capturing wind energy, the hub connects the blades to the main shaft, the main shaft system transmits rotational motion, the gearbox increases the speed, the generator realizes the conversion of mechanical energy into electrical energy, the nacelle houses the core components, the tower provides support, the converter regulates electrical energy, and the yaw system controls the steering of the nacelle.

[0030] The sound acquisition devices can be installed on the outer wall of the nacelle at the root of the wind turbine blades or on the top of the nacelle near the pitch system to collect blade noise; near the main shaft bearing behind the hub inside the nacelle or on the outside of the hub cover to monitor the main shaft system sound; close to the top, side, or bottom of the gearbox housing, or the top or side of the generator housing to acquire their operating noise; near the flange at the connection between the nacelle and the tower, or on the nacelle support structure of the tower top platform to monitor the yaw system and nacelle vibration sound; and on the side or top of the converter cabinet inside the tower, as well as on the middle and bottom platforms of the tower to collect converter electromagnetic noise and tower structural vibration noise.

[0031] In one specific embodiment, the acoustic signature signal data includes wavelet packet energy, linear prediction cepstral coefficients, and generalized frequency cepstral coefficients, while the environmental impact data includes wind speed, ocean current speed, and electromagnetic field intensity.

[0032] It should be noted that when acquiring voiceprint signal data, the sound acquisition devices installed in various components of the wind turbine are used in the sound acquisition hardware deployment module to acquire sound signals in real time at preset acquisition time points. After the acquisition device converts the sound wave vibration into an electrical signal, it uses digital signal processing technology to preprocess the original electrical signal, including pre-emphasis, framing, and windowing operations, to enhance high-frequency signals and reduce spectral leakage. Subsequently, the signal is decomposed into multiple scales through wavelet packet transform to calculate the energy of each frequency band and obtain the wavelet packet energy value; linear predictive coding technology is used to model the signal spectrum and then extract the linear predictive cepstral coefficients; the signal is processed by generalized frequency transform, and the generalized frequency cepstral coefficients are obtained through cepstral calculation, finally completing the extraction of voiceprint signal data.

[0033] When acquiring environmental impact data, wind speed data is collected using an ultrasonic anemometer or mechanical wind speed sensor installed on the top of the nacelle, recording real-time wind speed and direction at set time intervals. Ocean current speed data is measured using a Doppler current meter installed in the underwater part of the tower, utilizing the acoustic Doppler effect. Electromagnetic field strength is monitored by electromagnetic field sensors placed near the unit's electrical equipment. The sensors convert electromagnetic signals into electrical signals, which are then conditioned and converted from analog to digital to output the electromagnetic field strength data in digital form. The data collected by these sensors is transmitted to the system database in real time according to the set collection time points, serving as the basis for environmental impact analysis.

[0034] Real-time fault prediction module: It is used to analyze the acoustic signal data and environmental impact data of each component at each acquisition time point to obtain the fault assessment value of each component at each acquisition time point, and then predict whether a fault will be found in each component in the future cycle at each acquisition time point.

[0035] In a specific embodiment, the analysis obtains the fault assessment values ​​corresponding to each component at each acquisition time point. The specific analysis process is as follows: Analyze the acoustic signature signal assessment values ​​and environmental impact factors corresponding to each component at each acquisition time point, and denote them as Z. hg and V hg Where h represents the number corresponding to each acquisition time point, h = 1, 2, ..., q, where q is a positive integer and also the set of all acquisition time points; g represents the number corresponding to each component, g = 1, 2, ..., w, where w is a positive integer and also the set of all components. Simultaneously, normalization is performed, and the values ​​are substituted into the calculation formula: In the process, the fault assessment value φ corresponding to each component at each data collection time point is obtained. hg , where ω1 and ω2 are the weighting factors corresponding to the component acoustic signal evaluation value and the environmental impact factor, respectively, and e represents the natural constant.

[0036] It should be noted that, firstly, a large amount of historical data was collected, covering the acoustic signature signal evaluation values ​​and environmental impact factor data under normal operation of various components of the wind turbine and under different fault types and degrees. Principal component analysis and grey relational analysis were used to quantify the contribution of acoustic signature signals and environmental factors to fault assessment, and the weight ratios were initially determined. Then, experts in the offshore wind power field were invited to adjust the data calculation results based on equipment structural characteristics, environmental sensitivity, and operation and maintenance experience, focusing on the impact of environmental interference on acoustic signature signals to ensure that the weight settings conform to actual scenarios. Finally, through cross-validation and real-world case testing, the weight parameters were continuously optimized so that the fault assessment values ​​accurately reflect the true state of the equipment and balance the roles of acoustic signature signals and environmental factors in fault diagnosis.

[0037] In a specific embodiment, the analysis of the voiceprint signal evaluation value and environmental impact factor corresponding to each component at each collection time point is carried out as follows: The voiceprint signal data and environmental impact data corresponding to each component at each collection time point are normalized and then input into the voiceprint signal evaluation value analysis model and the environmental impact factor analysis model, respectively, and the voiceprint signal evaluation value and environmental impact factor corresponding to each component at each collection time point are output.

[0038] It should be noted that the analysis process for the voiceprint signal evaluation values ​​corresponding to each component at each acquisition time point is as follows: The wavelet packet energy value, linear prediction cepstral coefficient, and generalized frequency cepstral coefficient corresponding to each component at each acquisition time point are respectively denoted as D. hg 、F hg and R hg Substitute into the calculation formula: In this process, the voiceprint signal evaluation values ​​corresponding to each component at each acquisition time point are obtained, and the environmental impact factors corresponding to each component at each acquisition time point are analyzed according to the analysis process of the voiceprint signal evaluation values.

[0039] In a specific embodiment, the prediction process for whether a fault will be found in each component at each acquisition time point in the future cycle is as follows: the fault assessment value corresponding to each component at each acquisition time point is compared with the fault assessment value range corresponding to the set standard component. If the fault assessment value corresponding to a component at a certain acquisition time point is within the set fault assessment value range corresponding to the standard component, it is predicted that no fault will be found in the component in the future cycle at that acquisition time point. If the fault assessment value corresponding to a component at a certain acquisition time point is not within the set fault assessment value range corresponding to the standard component, it is predicted that a fault will be found in the component in the future cycle at that acquisition time point.

[0040] Fault Type Analysis Module: When it is predicted that a fault will be found in a certain component within a future period at a certain data acquisition time, the module analyzes the fault type corresponding to that component at that data acquisition time and the fault hazard level corresponding to that component at that data acquisition time.

[0041] In a specific embodiment, the analysis of the fault type corresponding to the component at the acquisition time point is carried out as follows: the fault assessment value corresponding to the component at the acquisition time point is compared with the fault assessment value range corresponding to each fault type in the database. If the fault assessment value corresponding to the component at the acquisition time point is within the fault assessment value range corresponding to a certain fault type in the database, then the fault type in the database is recorded as the fault type corresponding to the component at the acquisition time point.

[0042] In a specific embodiment, the analysis of the fault hazard level corresponding to the component at the data acquisition time point is carried out as follows: the fault assessment value corresponding to the component at the data acquisition time point is compared with the fault assessment value range corresponding to each set fault hazard level. If the fault assessment value corresponding to the component at the data acquisition time point is within the set fault assessment value range corresponding to a certain fault hazard level, then the set fault hazard level is recorded as the fault hazard level corresponding to the component at the data acquisition time point. The fault hazard level includes warning level, fault level and severe fault level.

[0043] Preventive maintenance strategy analysis module: This module is used to analyze the preventive maintenance strategy corresponding to the component at the time of data collection based on the fault hazard level of the component at that time.

[0044] In a specific embodiment, the preventive maintenance strategy corresponding to the component at the time of data collection is analyzed. The specific analysis process is as follows: A1. If the fault hazard level corresponding to the component at the time of data collection is the warning level, then the dynamic monitoring and hidden danger prevention and control strategy is executed.

[0045] It should be noted that the dynamic monitoring and hazard prevention strategy is as follows: First, intensified monitoring is implemented, increasing the regular data collection frequency of this component to once every 15-30 minutes, focusing on collecting acoustic signals, vibration data, temperature parameters, etc. Data trends are tracked in real time through a visualization platform to analyze whether abnormal characteristics continue to develop or fluctuate due to environmental interference. Second, targeted investigation is conducted. Maintenance personnel use mobile terminals to access historical health data of this component and similar early warning cases of similar units, combined with AR remote assistance functions, to conduct rapid inspections of the component's appearance and connection structure; portable diagnostic equipment is used to conduct precise detection of mechanical parts corresponding to abnormal acoustic signal frequency bands to identify potential fault sources. Finally, dynamic early warning is implemented, recording the current early warning event and investigation results into the equipment health management database, triggering the platform's "early warning tracking" task flow. Subsequent data collection automatically compares with historical early warning characteristics; if the abnormal trend intensifies, the maintenance strategy is automatically upgraded; if the abnormality is due to environmental interference and subsequent data returns to normal, intensified monitoring can be deactivated, and regular inspections can resume.

[0046] A1. If the fault hazard level corresponding to the component at the time of data collection is a fault level, then execute the planned shutdown and precise repair strategy.

[0047] It should be noted that the planned shutdown and precise repair strategy is as follows: First, shutdown preparation: Prioritize assessing the current sea conditions and whether the unit's power generation is during a critical period of grid peak shaving. Once the "safe shutdown window" is met (e.g., wave height ≤ 3 meters, wind speed ≤ 12 m / s), and coordination with grid dispatch is established, a "planned shutdown" command is issued, putting the unit into maintenance mode. Simultaneously, the related pitch, yaw, and power transmission systems are locked to prevent fault propagation. Second, in-depth diagnosis: After shutdown, the "fault reproduction and location" process is initiated. Using offline data acquisition equipment, full-dimensional characteristic data is collected under static and simulated operating conditions. Real-time data is imported using a digital twin model, combined with fault tree analysis, to precisely locate the fault point and assess the secondary impact of the fault on adjacent components. The third step is maintenance execution. Based on the diagnostic results, if it is a readily replaceable, easily damaged part, the spare parts warehouse of the offshore wind farm is used. The maintenance vessel carries the replacement team and completes the replacement work with the help of tower climbing equipment and nacelle cranes. The component life parameters of the digital twin model are updated simultaneously. If it is a complex fault, the collaborative process of "in-plant maintenance + maritime transportation" is initiated. The faulty part is marked, disassembled, and sent to the manufacturer for repair via a dedicated sea freight line. During this period, spare parts are used or the unit operation strategy is adjusted to ensure continuous power generation. After the maintenance is completed, a 72-hour load test is performed to collect data such as acoustic signatures and vibration to verify the repair effect. Once the test is passed, the maintenance mode is lifted.

[0048] A1. If the fault hazard level corresponding to the component at the time of data collection is a severe fault level, then an emergency shutdown and emergency rescue strategy shall be implemented.

[0049] It should be noted that the emergency shutdown and emergency rescue strategy is as follows: First, emergency shutdown and safety isolation trigger the unit's "emergency shutdown" hard logic, cutting off the main circuit, locking the pitch system to feather position, and braking the yaw system to quickly disconnect the unit from the power grid; simultaneously, the "safety isolation" procedure is initiated, shutting down the hydraulic, lubrication, and cooling circuits related to the affected component, and using the fire protection system for pre-installed protection to prevent secondary accidents such as fires or component detachment caused by the fault; the operation and maintenance center simultaneously reports the fault and shutdown plan to the maritime authorities and power grid dispatch. Second, emergency repair dispatch activates the "maritime emergency repair plan": dispatching an operation and maintenance vessel capable of operating in adverse sea conditions, carrying a repair team, emergency spare parts, and rescue equipment, to prioritize overcoming sea condition restrictions and rush to the scene; at the same time, the manufacturer's technical experts access the unit's data in real time through a remote diagnostic platform to guide on-site personnel in predicting the risk of the fault escalating and developing a "minimum damage repair plan." Finally, in the fault handling and debriefing process, on-site emergency repairs focused on "minimizing losses and ensuring safety." Through temporary spare parts replacement and physical isolation of faulty components, the unit's basic safety status was quickly restored, allowing temporary grid connection under reduced power and limited operating conditions. After the fault was resolved, a full lifecycle tracing of the faulty components was conducted, and an expert team carried out "serious fault root cause analysis." The fault level judgment threshold of the voiceprint diagnostic model and the triggering conditions for preventative maintenance strategies were optimized, and the spare parts reserve plan was updated to prevent the recurrence of similar faults.

[0050] Examples of embodiments of the present invention Figure 2 As shown, a preventive maintenance and fault diagnosis method for offshore wind turbines includes: Step 1, deployment of sound acquisition hardware: deploying sound acquisition hardware for each component in the target offshore wind turbine. After deployment, setting several acquisition time points, and then collecting the corresponding acoustic signature data and environmental impact data of each component at each acquisition time point.

[0051] Step 2: Real-time Fault Prediction: Based on the acoustic signal data and environmental impact data corresponding to each component at each acquisition time point, the fault assessment value corresponding to each component at each acquisition time point is obtained, and then it is predicted whether a fault will be found in each component in the future cycle at each acquisition time point.

[0052] Step 3: Fault type analysis: When it is predicted that a fault will be found in a certain component within a future period at a certain data acquisition time point, the fault type corresponding to that component at that data acquisition time point is analyzed, and the fault hazard level corresponding to that component at that data acquisition time point is also analyzed.

[0053] Step 4: Analysis of preventive maintenance strategies: Based on the fault hazard level of the component at the time of data collection, the preventive maintenance strategies corresponding to the component at the time of data collection are analyzed.

[0054] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. A preventive maintenance and fault diagnosis system for offshore wind turbines, characterized in that, include: Sound acquisition hardware deployment module: used to deploy sound acquisition hardware on each component of the target offshore wind turbine. After deployment, several acquisition time points are set, and then the acoustic signature signal data and environmental impact data of each component are collected at each acquisition time point. Real-time fault prediction module: Based on the acoustic signal data and environmental impact data of each component at each acquisition time point, it analyzes and obtains the fault assessment value of each component at each acquisition time point, and then predicts whether a fault will be found in each component in the future period at each acquisition time point. Fault type analysis module: When it is predicted that a fault will be found in a certain component within a future period at a certain data acquisition time, the module analyzes the fault type corresponding to that component at that data acquisition time and the fault hazard level corresponding to that component at that data acquisition time. Preventive maintenance strategy analysis module: This module is used to analyze the preventive maintenance strategy corresponding to the component at the time of data collection based on the fault hazard level of the component at that time.

2. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 1, characterized in that, The specific deployment process for sound acquisition devices in various components of the target offshore wind turbine is as follows: When deploying the sound acquisition device hardware on each component of the target offshore wind turbine, a high-strength corrosion-resistant alloy bracket is used. The bracket is firmly fixed to the surface of each component or adjacent structure by means of expansion bolts, welding or buckles. The sound acquisition device is installed on the bracket. The connection between the sound acquisition device and the bracket is sealed with a silicone sealing ring, and the interface is wrapped with waterproof tape to ensure the waterproof and salt spray resistance of the equipment.

3. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 2, characterized in that, The acoustic signature data includes wavelet packet energy, linear prediction cepstral coefficients, and generalized frequency cepstral coefficients, while the environmental impact data includes wind speed, ocean current speed, and electromagnetic field strength.

4. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 3, characterized in that, The analysis yielded fault assessment values ​​for each component at each data collection time point. The specific analysis process is as follows: Analyze the acoustic signature signal evaluation values ​​and environmental impact factors corresponding to each component at each acquisition time point, and denote them as Z. hg and V hg Where h represents the number corresponding to each acquisition time point, h = 1, 2, ..., q, where q is a positive integer and also the set of all acquisition time points; g represents the number corresponding to each component, g = 1, 2, ..., w, where w is a positive integer and also the set of all components. Simultaneously, normalization is performed, and the values ​​are substituted into the calculation formula: In the process, the fault assessment value φ corresponding to each component at each data collection time point is obtained. hg , where ω1 and ω2 are the weighting factors corresponding to the component acoustic signal evaluation value and the environmental impact factor, respectively, and e represents the natural constant.

5. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 4, characterized in that, The analysis of the acoustic signature signal evaluation values ​​and environmental impact factors corresponding to each component at each acquisition time point is as follows: The voiceprint signal data and environmental impact data corresponding to each component at each collection time point are normalized and then used as inputs into the voiceprint signal evaluation value analysis model and the environmental impact factor analysis model, respectively, to output the voiceprint signal evaluation value and environmental impact factor corresponding to each component at each collection time point.

6. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 5, characterized in that, The specific prediction process for determining whether a fault will be detected in each component at each data collection point within a future period is as follows: The fault assessment values ​​of each component at each data acquisition time point are compared with the set fault assessment value range of the standard component. If the fault assessment value of a component at a certain data acquisition time point is within the set fault assessment value range of the standard component, it is predicted that no fault will be found in the component in the future period at that data acquisition time point. If the fault assessment value of a component at a certain data acquisition time point is not within the set fault assessment value range of the standard component, it is predicted that a fault will be found in the component in the future period at that data acquisition time point.

7. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 6, characterized in that, The analysis of the fault type corresponding to the component at the time point of data acquisition is as follows: The fault assessment value corresponding to the component at the time of data collection is compared with the fault assessment value range corresponding to each fault type in the database. If the fault assessment value corresponding to the component at the time of data collection is within the fault assessment value range corresponding to a certain fault type in the database, then the fault type in the database is recorded as the fault type corresponding to the component at the time of data collection.

8. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 7, characterized in that, The analysis of the fault hazard level corresponding to the component at the data acquisition point is as follows: The fault assessment value corresponding to the component at the time of data collection is compared with the fault assessment value range corresponding to each set fault hazard level. If the fault assessment value corresponding to the component at the time of data collection is within the fault assessment value range corresponding to a certain set fault hazard level, then the set fault hazard level is recorded as the fault hazard level corresponding to the component at the time of data collection. The fault hazard level includes warning level, fault level and severe fault level.

9. The preventive maintenance and fault diagnosis system for offshore wind turbines as described in claim 8, characterized in that, The analysis of the preventive maintenance strategy corresponding to this component at the time of data collection is as follows: A1. If the fault hazard level corresponding to the component at the time of data collection is the warning level, then the dynamic monitoring and hidden danger prevention and control strategy shall be implemented. A1. If the fault hazard level corresponding to the component at the time of data collection is a fault level, then execute the planned shutdown and precise repair strategy. A1. If the fault hazard level corresponding to the component at the time of data collection is a severe fault level, then an emergency shutdown and emergency rescue strategy shall be implemented.

10. A method for preventive maintenance and fault diagnosis of offshore wind turbines implementing the preventive maintenance and fault diagnosis system of any one of claims 1-9, characterized in that, include: Step 1: Deployment of sound acquisition hardware: Deploy sound acquisition hardware for each component in the target offshore wind turbine. After deployment, set several acquisition time points and then collect the acoustic signature data and environmental impact data corresponding to each component at each acquisition time point. Step 2: Real-time Fault Prediction: Based on the acoustic signal data and environmental impact data corresponding to each component at each acquisition time point, the fault assessment value corresponding to each component at each acquisition time point is obtained, and then it is predicted whether a fault will be found in each component in the future cycle at each acquisition time point. Step 3: Fault type analysis: When it is predicted that a fault will be found in a certain component within a future period at a certain data collection time point, the fault type corresponding to that component at that data collection time point is analyzed, and the fault hazard level corresponding to that component at that data collection time point is also analyzed. Step 4: Analysis of preventive maintenance strategies: Based on the fault hazard level of the component at the time of data collection, the preventive maintenance strategies corresponding to the component at the time of data collection are analyzed.

Images