A method and apparatus for supervising an offshore wind turbine

Abstract

The specification provides a method and device for supervising offshore wind turbines. A preset risk prediction model is used to process operation data and operation and maintenance records of a plurality of target components in a first round of supervision cycles to obtain risk prediction information of the plurality of target components. The preset supervision cycle parameters are updated according to the risk prediction information of the plurality of target components to obtain updated preset supervision cycle parameters. The operation data and operation and maintenance records of the plurality of target components in offshore wind turbines in a target sea area in a second round of supervision cycles are collected according to the updated preset supervision cycle parameters. In the second round of supervision cycles, whether to generate on-site detection prompt information about offshore wind turbines is determined according to the collected operation data and operation and maintenance records of the plurality of target components in offshore wind turbines in the target sea area and the updated preset supervision cycle parameters. Thus, the supervision resource allocation is optimized, and the overall operation and maintenance efficiency is improved.

Description

Technical Field

[0001] This manual pertains to the field of offshore wind power generation technology, and in particular relates to a monitoring method and device for offshore wind turbines. Background Technology

[0002] Currently, most offshore wind power operation and maintenance still follow the regular inspection strategy of onshore wind power. The health status of large components such as gearboxes, blades, generators, and main bearings mainly relies on real-time monitoring data and fixed-period reminders. This approach cannot fully cover the dynamic operating conditions of floating units, nor can it achieve efficient and accurate preventive maintenance within the limited operation and maintenance window.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This specification provides a method and apparatus for monitoring offshore wind turbines. By performing risk prediction on the operational data and maintenance records of target components based on preset monitoring cycle parameters, and dynamically updating the monitoring cycle parameters accordingly, adaptive monitoring of the status of key components of offshore wind turbines can be achieved. Compared to fixed-cycle monitoring methods, this method can adjust the data collection frequency according to changes in component risk, optimizing the allocation of monitoring resources and improving overall maintenance efficiency while ensuring safe equipment operation. Furthermore, data analysis within the updated monitoring cycle can promptly identify potential anomalies, providing a basis for determining whether on-site inspections should be conducted, thereby enhancing risk warning and response capabilities.

[0005] This manual provides a method for supervising offshore wind turbines, including:

[0006] Based on preset regulatory cycle parameters, the operation data and maintenance records of multiple target components in offshore wind turbines within the target sea area are obtained during the first round of regulatory cycle.

[0007] By using a pre-set risk prediction model, the operational data and maintenance records of multiple target components during the first round of regulatory cycle are processed to obtain risk prediction information for multiple target components;

[0008] Based on the risk prediction information of the multiple target components, the preset regulatory cycle parameters are updated to obtain the updated preset regulatory cycle parameters;

[0009] Based on the updated preset regulatory cycle parameters, the operation data and maintenance records of multiple target components in the offshore wind turbines within the target sea area of ​​the second round of regulatory cycle are collected; and within the second round of regulatory cycle, based on the updated preset regulatory cycle parameters, the collected operation data and maintenance records of multiple target components in the offshore wind turbines within the target sea area are used to determine whether to generate on-site inspection prompt information for the offshore wind turbines.

[0010] In one embodiment, the operating data includes at least one of the following: blade operating data, gearbox operating data, main bearing operating data, tower operating data, and generator operating data;

[0011] The blade operating data includes the blade's vibration frequency, noise, blade root torque, and blade root strain; the gearbox operating data includes the gearbox's output shaft temperature, vibration, and lubricating oil impurity content; the main bearing operating data includes the main bearing's temperature and vibration; the tower operating data includes the tower's vibration and strain; and the generator operating data includes the generator's temperature, speed, and vibration.

[0012] In one embodiment, the maintenance record includes abnormal data, job type, and fault location information for multiple target components.

[0013] In one embodiment, the step of using a preset risk prediction model to process the operational data and maintenance records of multiple target components during the first round of regulatory cycles to obtain risk prediction information for multiple target components includes:

[0014] The operating data of multiple target components are aligned with the corresponding operation and maintenance records according to a preset timestamp. Key features are extracted and operation and maintenance events are encoded and fused to obtain a fused feature vector. The fused feature vector is then input into a preset risk prediction model to obtain risk prediction information for multiple target components. The risk prediction information includes the frequency of risk occurrence and the characteristics of risk occurrence.

[0015] In one embodiment, updating the preset regulatory period parameters based on the risk prediction information of the plurality of target components to obtain the updated preset regulatory period parameters includes:

[0016] The weighted risk levels of multiple target components are determined based on the frequency and characteristics of risk occurrence in the risk prediction information and the preset periodic adjustment rules.

[0017] The corresponding adjustment coefficient is determined based on the weighted risk level;

[0018] The preset regulatory cycle parameters of the multiple target components are multiplied by the corresponding adjustment coefficients to obtain the updated preset regulatory cycle parameters.

[0019] In one embodiment, based on updated preset regulatory cycle parameters, and utilizing the collected operational data and maintenance records of multiple target components within the offshore wind turbine in the target sea area, it is determined whether to generate on-site inspection alerts for the offshore wind turbine, including:

[0020] Based on the collected operational data and maintenance records of multiple target components in the offshore wind turbine within the target sea area, as well as the last on-site inspection notification time and sea condition information of the offshore wind turbine's location, it is determined whether to generate on-site inspection notification information for the offshore wind turbine.

[0021] In one embodiment, the method further includes:

[0022] Based on the on-site detection prompts, a corresponding manual inspection work order is generated on the operation and maintenance management platform;

[0023] The inspection work order is issued to the on-site maintenance personnel, and the time of issuance of the work order is recorded;

[0024] After the maintenance personnel complete the inspection and upload the on-site inspection results, the inspection results are stored in the risk and incident database to update the maintenance records.

[0025] This specification provides a monitoring device for offshore wind turbines, including:

[0026] The data acquisition module is used to acquire the operating data and maintenance records of multiple target components in the offshore wind turbine in the target sea area during the first round of the regulatory cycle, based on preset regulatory cycle parameters.

[0027] The risk prediction module is used to process the operational data and maintenance records of multiple target components during the first round of supervision using a preset risk prediction model, and obtain risk prediction information for multiple target components.

[0028] The parameter update module is used to update the preset regulatory cycle parameters based on the risk prediction information of the multiple target components, so as to obtain the updated preset regulatory cycle parameters.

[0029] The information prompt module is used to collect operational data and maintenance records of multiple target components in offshore wind turbines within the target sea area for the second round of supervision, based on updated preset supervision cycle parameters. Within the second round of supervision, based on the updated preset supervision cycle parameters and using the collected operational data and maintenance records of multiple target components in the offshore wind turbines within the target sea area, it determines whether to generate on-site inspection prompt information for the offshore wind turbines. This specification also provides an electronic device, including a processor and a memory for storing processor-executable instructions, wherein the processor executes the instructions to implement a method for supervising offshore wind turbines.

[0030] This specification also provides a computer-readable storage medium storing computer instructions that, when executed, implement a method for monitoring offshore wind turbines.

[0031] Based on the offshore wind turbine monitoring method provided in this specification, the following steps are taken: First, based on preset monitoring cycle parameters, operational data and maintenance records of multiple target components within the offshore wind turbine in the target sea area are acquired during the first monitoring cycle. Then, using a preset risk prediction model, the operational data and maintenance records of the multiple target components during the first monitoring cycle are processed to obtain risk prediction information for the multiple target components. Based on the risk prediction information of the multiple target components, the preset monitoring cycle parameters are updated to obtain updated preset monitoring cycle parameters. Based on the updated preset monitoring cycle parameters, operational data and maintenance records of the multiple target components within the offshore wind turbine in the target sea area are collected during the second monitoring cycle. Finally, during the second monitoring cycle, based on the updated preset monitoring cycle parameters and using the collected operational data and maintenance records of the multiple target components within the offshore wind turbine in the target sea area, it is determined whether to generate on-site inspection prompts for the offshore wind turbine. In this way, by performing risk prediction on the operational data and maintenance records of the target components based on preset monitoring cycle parameters, and dynamically updating the monitoring cycle parameters accordingly, adaptive monitoring of the status of key components of the offshore wind turbine can be achieved. Compared to fixed-cycle monitoring, this method can adjust the data collection frequency based on changes in component risk, optimizing the allocation of regulatory resources and improving overall operation and maintenance efficiency while ensuring safe equipment operation. Furthermore, data analysis within the updated regulatory cycle can promptly identify potential anomalies, providing a basis for determining whether on-site inspections should be conducted, thereby enhancing risk warning and response capabilities. Attached Figure Description

[0032] To more clearly illustrate the embodiments of this specification, the accompanying drawings used in the embodiments will be briefly introduced below. The drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart illustrating a method for monitoring offshore wind turbines, provided in one embodiment of this specification.

[0034] Figure 2 This is a schematic diagram of the electronic device structure provided in one embodiment of this specification;

[0035] Figure 3 This is a schematic diagram of the structural composition of a monitoring device for an offshore wind turbine provided in one embodiment of this specification;

[0036] Figure 4 This is a schematic diagram of a self-learning floating offshore wind turbine major component damage early warning process provided in one embodiment of this specification;

[0037] Figure 5 This is a schematic diagram of another self-learning floating offshore wind turbine major component damage early warning process provided in one embodiment of this specification. Detailed Implementation

[0038] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0039] Existing technologies mostly follow the inspection mode of onshore wind turbines, mainly setting early warnings for major components that are regularly inspected, or relying on real-time monitoring data for status assessment. However, they are difficult to provide early warnings of failures in critical major components, and cannot provide effective preventive maintenance suggestions for operation and maintenance personnel. Although some solutions monitor components inside the nacelle through means such as infrared thermometry, they lack a global perception of the overall operating status of the machine, making it difficult to make comprehensive judgments and accurate diagnoses of fault trends.

[0040] To address the root causes of the aforementioned problems, this manual employs a risk prediction method based on preset regulatory cycle parameters, analyzing operational data and maintenance records of target components. This method dynamically updates the regulatory cycle parameters accordingly, enabling adaptive monitoring of the status of critical components in offshore wind turbines. Compared to fixed-cycle monitoring, this approach adjusts data collection frequency based on changes in component risk, optimizing regulatory resource allocation and improving overall maintenance efficiency while ensuring safe equipment operation. Furthermore, data analysis within the updated regulatory cycle allows for timely identification of potential anomalies, providing a basis for on-site inspections and enhancing risk warning and response capabilities.

[0041] See Figure 1 As shown in the embodiments of this specification, a method for monitoring offshore wind turbines is provided, wherein the method is specifically applied to the server side. In specific implementation, the method may include the following:

[0042] S101: Based on preset regulatory cycle parameters, obtain the operation data and maintenance records of multiple target components in offshore wind turbines within the target sea area during the first round of regulatory cycle;

[0043] S102: Using a pre-set risk prediction model, process the operational data and maintenance records of multiple target components during the first round of regulatory cycle to obtain risk prediction information for multiple target components;

[0044] S103: Based on the risk prediction information of the multiple target components, update the preset regulatory cycle parameters to obtain the updated preset regulatory cycle parameters;

[0045] S104: Based on the updated preset regulatory cycle parameters, collect the operating data and maintenance records of multiple target components in the offshore wind turbines within the target sea area for the second round of regulatory cycle; and within the second round of regulatory cycle, based on the updated preset regulatory cycle parameters, use the collected operating data and maintenance records of multiple target components in the offshore wind turbines within the target sea area to determine whether to generate on-site inspection prompt information for the offshore wind turbines.

[0046] The aforementioned regulatory cycle parameter can be used to set the time interval for risk monitoring of each target component, usually in "days" or "months," representing the time span from the completion of one regulatory operation (such as inspection or maintenance) to the start of the next regulatory task for a particular component. This parameter can be preset according to the importance of different components, historical risk levels, and operational resource status, and can be dynamically adjusted during operation.

[0047] The aforementioned risk prediction information can be a comprehensive status assessment result obtained by processing the operational data and maintenance records of the target component collected within a specified regulatory period through a risk prediction model. The results may include, but are not limited to: risk level (e.g., low, medium, high), risk trend (e.g., stable, fluctuating, continuously rising), and details of abnormal indicators (e.g., exceeding limits for temperature, vibration, strain, etc.), which are used to support updates and adjustments for subsequent regulatory periods.

[0048] The aforementioned on-site inspection prompts can be automatically generated by the system based on the component's operating status and historical trend changes within the current monitoring period, when it is determined that one or more components have potential faults or signs of degradation. The information may include: the name of the component to be inspected, the triggering reason, the suggested inspection time limit or priority, and the optional inspection methods (such as manual inspection, drone observation, or installation of temporary sensors), which are used to remind maintenance personnel to reasonably arrange on-site inspection tasks.

[0049] In some embodiments, the step of using a preset risk prediction model to process the operational data and maintenance records of multiple target components during the first round of regulatory cycle to obtain risk prediction information for multiple target components may, in specific implementation, include:

[0050] The operational data and maintenance records are grouped according to the target components, and the operational feature data in each group are normalized and processed by moving average. The processed data is input into a risk prediction model based on a convolutional neural network (CNN) to extract the risk feature parameters of the target components. The feature parameters output by the model are compared with the preset risk threshold to determine the risk level of the target component in the current regulatory period and generate corresponding risk prediction information.

[0051] For example, in the first round of supervision, operational data such as temperature, vibration and lubricating oil particle content of a gearbox component of a wind turbine are collected, and it is also recorded whether the component has been maintained during this period. After normalizing and filtering these data, they are input into a CNN model. The model output shows that the vibration intensity of the gearbox continues to rise and the temperature fluctuates at a high level. The overall assessment is "medium to high risk". Based on this, the system generates risk prediction information for the component.

[0052] By using operational data and maintenance records as joint inputs, and employing a pre-defined deep learning model for feature extraction and risk assessment, the accuracy and sensitivity of fault risk prediction can be improved. This enables early identification of potential faults in key components of wind turbines, helps to dynamically adjust the regulatory cycle, assists maintenance personnel in formulating reasonable maintenance strategies, and improves overall maintenance efficiency and the stability of wind turbines.

[0053] In some embodiments, updating the preset regulatory period parameters based on the risk prediction information of the plurality of target components to obtain the updated preset regulatory period parameters may specifically include:

[0054] Risk prediction information for each target component is extracted to obtain its corresponding risk level, risk occurrence frequency, and abnormal trend characteristic value. Based on preset cycle adjustment rules, a cycle adjustment coefficient for each target component is calculated. These rules are used to determine different cycle shortening ratios from low to high risk levels, and dynamic adjustment weights are set in conjunction with risk frequency. The original regulatory cycle parameters are multiplied by the cycle adjustment coefficients to obtain updated regulatory cycle parameters, which are then recorded in the system scheduling module for scheduling and early warning alerts in the next regulatory cycle.

[0055] For example, for a wind turbine blade component, the risk prediction information shows that three abnormal events exceeding the vibration threshold occurred within the current monitoring cycle, and the prediction result is determined to be of medium risk level. The system determines its cycle adjustment coefficient to be 0.8 based on the preset cycle adjustment rules, that is, updating the original 30-day monitoring cycle to a new 24-day cycle, and simultaneously updating it to the reminder and scheduling system to ensure that subsequent inspection tasks are arranged before the end of the new cycle.

[0056] By introducing a periodic adjustment mechanism based on risk level and frequency, differentiated dynamic management of components in different operating states can be achieved, effectively increasing the frequency of supervision of high-risk components while avoiding excessive testing of low-risk components. This improves the efficiency of operation and maintenance resource utilization, reduces unnecessary operation and maintenance costs, and enhances the overall safety and reliability of wind turbine operation.

[0057] In some embodiments, during the second regulatory cycle, based on updated preset regulatory cycle parameters, and utilizing the collected operational data and maintenance records of multiple target components in the offshore wind turbine within the target sea area, it is determined whether to generate on-site inspection prompts for the offshore wind turbine. Specifically, this may include:

[0058] The operational data collected for each target component within the current regulatory cycle is normalized and processed using a moving average, and key feature indicators are extracted. These feature indicators are then compared with the historical operational threshold range and abnormal state characteristics corresponding to the target component. If the current feature indicator meets the set prompt generation conditions (including: abnormal fluctuation amplitude exceeding the limit, trend change rate exceeding the threshold, or consecutive abnormal number exceeding the limit), then on-site detection prompt information for the target component is generated and associated with the operation and maintenance scheduling platform.

[0059] For example, within the updated 20-day monitoring period, the Supervisory Control and Data Acquisition (SCADA) system recorded vibration values ​​exceeding the warning threshold five consecutive times for the main bearing of a floating wind turbine. Based on the characteristic change trend and abnormal duration of this component, a potential structural wear problem was determined, triggering the alert generation conditions. Based on this, on-site inspection alert information for the main bearing was generated and synchronized to the operation and maintenance platform to recommend scheduling on-site inspection tasks.

[0060] By combining updated cycle parameters with real-time operational data trends, a field detection and alert generation mechanism based on dynamic monitoring was implemented. This effectively improved the responsiveness to the operational status of target components and enabled timely early warning alerts to be generated before fault risks manifested as serious consequences. This enhanced the feasibility of preventative maintenance and improved the initiative and accuracy of overall operation and maintenance.

[0061] In some embodiments, the method may further include the following:

[0062] After obtaining the updated preset regulatory cycle parameters, if the duration corresponding to the updated regulatory cycle parameters is significantly reduced compared to the duration corresponding to the previous regulatory cycle parameters, and is lower than the preset standard duration threshold, then non-periodic external influence characteristics are further introduced to dynamically calibrate the execution timing of on-site regulatory tasks. Specifically, this may include: collecting current sea state information of the sea area where the target wind turbine is located, including parameters such as wind speed, wave height, tidal changes, visibility, and ocean current speed; comparing the sea state information with a preset seaworthiness threshold to determine whether the conditions for offshore operations are met; simultaneously acquiring the operating status information of the target wind turbine, including the current load level, alarm status, operating mode, and communication online status; further acquiring scheduling resource information, including the status of available operational vessels, the on-duty status of personnel, and whether the work permit is valid; if the current time is within the updated regulatory cycle range, and the sea state is seaworthy, the turbine is in a detectable state, and the scheduling resources meet the requirements, then an on-site detection prompt message is generated; otherwise, it is recorded as a delayed state and enters the next cycle evaluation window.

[0063] Among them, the aforementioned non-periodic external influence characteristics refer to external environmental or resource factors that are independent of the internal operating cycle of wind turbine units. They usually do not have a fixed regularity but have a direct impact on operation and maintenance. These include current sea state information (such as wind speed, wave height, tidal changes, ocean current speed and visibility), wind turbine unit operating status (such as load level, alarm status, operating mode and communication status), and operation and maintenance scheduling resource status (such as the ability to dispatch operating vessels, the availability of on-site personnel and the status of operation permits).

[0064] By introducing non-periodic external influence characteristics for dynamic decision-making when the updated regulatory cycle parameters suddenly drop in duration and fall below the predetermined standard cycle, it is possible to avoid the problem of erroneously triggering on-site inspection tasks when the operating conditions are not met.

[0065] In some embodiments, the method may further include the following:

[0066] After combining risk prediction information from multiple target components, it is determined whether any of the target components meet the preset attention rules. If a component of concern exists, during the second round of supervision, in addition to collecting the operating data and maintenance records of the multiple target components according to the updated preset supervision cycle parameters, before the end of the second round of supervision and before the arrival of the next collection cycle, one or more supplementary data collection operations are performed on the component of concern based on its risk level and characteristic evolution trend. The supplementary data collection operation includes collecting multi-source operating data and maintenance image records such as gearbox oil impurity concentration, transmission system current fluctuation, generator power output deviation, pitch system response delay, and structural corrosion degree images of the component of concern of the target offshore wind turbine at a specified time point or when specific triggering conditions are met (e.g., abnormal fluctuation rate of key operating parameters, deterioration of lubricating oil quality, abnormal change of current load, etc.) to enhance the comprehensive perception capability of changes in the status of the component of concern.

[0067] The aforementioned preset attention rules are used to identify the attention components that need to be strengthened from multiple target components. The judgment criteria may include: the risk level of the target component reaches the high risk threshold, the frequency of risk occurrence is significantly higher than the historical average, key operating indicators (such as power deviation, oil contaminant content, etc.) show a continuous abnormal change trend, or the component has abnormalities in multiple consecutive cycles but has not received sufficient intervention.

[0068] By introducing a mechanism for identifying and supplementing data collection of components of concern, additional data collection support can be provided for high-risk or unstable components outside of the regular regulatory cycle. This will improve the frequency and accuracy of monitoring key risk components, help achieve earlier fault prediction and operation and maintenance response, reduce the risk of major failures, and enhance the initiative and flexibility of offshore wind turbine operation and maintenance.

[0069] Based on the above embodiments, by performing risk prediction on the operational data and maintenance records of target components using preset regulatory cycle parameters, and dynamically updating the regulatory cycle parameters accordingly, adaptive monitoring of the status of key components of offshore wind turbines can be achieved. Compared to fixed-cycle monitoring methods, this method can adjust the data collection frequency according to changes in component risk, optimizing the allocation of regulatory resources and improving overall operation and maintenance efficiency while ensuring safe equipment operation. Furthermore, data analysis within the updated regulatory cycle can promptly identify potential anomalies, providing a basis for determining whether on-site inspections should be conducted, thereby enhancing risk warning and response capabilities.

[0070] In some embodiments, the operating data includes at least one of the following: blade operating data, gearbox operating data, main bearing operating data, tower operating data, and generator operating data;

[0071] The blade operating data includes the blade's vibration frequency, noise, blade root torque, and blade root strain; the gearbox operating data includes the gearbox's output shaft temperature, vibration, and lubricating oil impurity content; the main bearing operating data includes the main bearing's temperature and vibration; the tower operating data includes the tower's vibration and strain; and the generator operating data includes the generator's temperature, speed, and vibration.

[0072] By collecting and classifying operational data from multiple key components of offshore wind turbines, comprehensive monitoring of the operational status of major components such as blades, gearboxes, main bearings, towers, and generators can be achieved. This improves the accuracy and response speed of fault identification, supports efficient training of subsequent risk assessment and prediction models, and ultimately enables early warning of potential faults and scientific operation and maintenance decisions, significantly improving the operational reliability and power generation efficiency of wind turbines.

[0073] In some embodiments, the maintenance records include abnormal data, job types, and fault location information for multiple target components.

[0074] In some embodiments, the maintenance record may include abnormal blade root strain data found during a certain inspection of a wind turbine blade, the operation type being "high-altitude manual replacement of blade tip components," and the fault location information being marked as "loose bolts at the blade root." Similarly, for gearbox components, the maintenance record may include maintenance tasks triggered by excessive gearbox vibration, the operation type being "replacement of lubrication system," and the fault location information being "oil circuit blockage."

[0075] By structuring and recording maintenance information such as abnormal data, job types, and fault locations and linking it to each target component, it not only facilitates subsequent model learning of risk characteristics and trend prediction, but also improves the ability to trace the historical status of components and the accuracy of classification analysis, thereby enhancing the scientific nature of regulatory cycle adjustments and the pertinence of maintenance strategies.

[0076] In some embodiments, the method of using a preset risk prediction model to process the operational data and maintenance records of multiple target components during the first round of regulatory cycle to obtain risk prediction information for multiple target components may further include the following:

[0077] S1: Align the running data of multiple target components with the corresponding operation and maintenance records according to the preset timestamp, extract key features, and encode and fuse operation and maintenance events to obtain the fused feature vector;

[0078] S2: Input the fused feature vector into a preset risk prediction model to obtain risk prediction information for multiple target components; wherein, the risk prediction information includes the frequency of risk occurrence and the characteristics of risk occurrence.

[0079] The aforementioned key characteristics can include temperature characteristics, vibration characteristics, and strain characteristics. Specifically, temperature characteristics can be the operating temperatures of various target components (such as gearboxes, main bearings, generators, etc.), such as gearbox oil temperature, main bearing temperature, and generator winding temperature. These temperature indicators reflect the thermal state of the equipment and help identify potential risks such as overheating and abnormal lubrication. Vibration characteristics can be the overall vibration value of the target component or the vibration acceleration value in a specific direction, such as blade vibration frequency, gearbox axial / radial vibration, and main bearing vibration amplitude, which can effectively indicate component loosening, wear, or imbalance. Strain characteristics can be the strain generated in structural parts under stress, often including blade root strain, tower strain, and submarine cable tensile strain, used to assess the degree of structural fatigue or the risk of abnormal load impact.

[0080] By extracting and fusing the key operational features such as temperature, vibration, and strain, a more accurate and engineering-interpretable feature vector can be constructed.

[0081] Specifically, the operational data of each target component is resampled at a time granularity of minutes or hours to generate a continuous and unified timestamp sequence, which serves as the standard time reference. The operation time field in each operation and maintenance record is matched with the unified timestamp. When the difference between the time of the operation and maintenance record and the timestamp of the operational data is within a preset tolerance range (e.g., ±30 minutes), it is considered a successful alignment, and the operation and maintenance event is appended to the operational data at the corresponding time point. If an operation and maintenance event cannot be matched with any timestamp, it is marked as an isolated event and does not participate in the fusion analysis of the current regulatory cycle.

[0082] After time alignment is completed, key feature dimensions are extracted from the operational data corresponding to the successfully aligned time points. These include temperature indicators (such as generator winding temperature, gearbox oil temperature, main bearing temperature, etc.), vibration indicators (such as blade vibration frequency, bearing acceleration, tower vibration amplitude, etc.), and strain indicators (such as tower strain, blade root strain, submarine cable tensile strain, etc.). Simultaneously, discrete event codes with fixed dimensions are generated based on the operation type (such as replacement, repair, inspection) and fault location (such as gearbox, high-speed bearing, generator end cover, etc.) marked in the maintenance records. Finally, the above operational features and event codes are concatenated to form a unified feature vector, which serves as the input to the risk prediction model. The model outputs risk prediction information for each target component within the current regulatory period, including the frequency and characteristics of risk occurrence.

[0083] For example, during a certain regulatory cycle, temperature and vibration signals collected from generator components are time-aligned with a previous "abnormal bearing replacement" maintenance event. Monitoring features are extracted and fused to generate a feature vector, which is then input into a risk prediction model. The model output shows that the generator exhibits a "high-frequency vibration anomaly" trend, and similar historical events have occurred frequently, indicating a high probability of recurrence in the short term. Therefore, it is inferred that the subsequent regulatory cycle needs to be shortened.

[0084] By aligning the operational data with the maintenance records using time and fusing features, the constructed input vector can comprehensively reflect the operational status and historical maintenance background of each target component in the current cycle. This helps improve the risk prediction model's ability to identify potential fault characteristics, enhances the accuracy and practicality of prediction results, and thus provides a data foundation and decision support for the dynamic adjustment of the regulatory cycle.

[0085] In some embodiments, the method of updating the preset regulatory cycle parameters based on the risk prediction information of the plurality of target components to obtain the updated preset regulatory cycle parameters may further include the following:

[0086] S1: Determine the weighted risk level of multiple target components based on the risk occurrence frequency and risk occurrence characteristics in the risk prediction information and the preset periodic adjustment rules;

[0087] S2: Determine the corresponding adjustment coefficient based on the weighted risk level;

[0088] S3: Multiply the preset monitoring cycle parameters of the multiple target components by the corresponding adjustment coefficients to obtain the updated preset monitoring cycle parameters.

[0089] The aforementioned risk occurrence characteristics are used to characterize the abnormal performance of the target component during the current regulatory cycle. These characteristics include, but are not limited to, the abnormal trend, duration, and magnitude of key monitoring indicators within a certain time interval. The abnormal trend includes upward, fluctuating, or abrupt trends; the duration reflects the length of time the abnormal state persists; and the magnitude measures the degree of indicator change. These risk occurrence characteristics reflect the development trend of potential failures in the target component, supporting the determination of the target component's risk level and the adjustment of parameters in subsequent regulatory cycles.

[0090] Specifically, firstly, risk prediction information for multiple target components within the current regulatory cycle is extracted, and the risk occurrence frequency and risk characteristics of each component are obtained. The risk occurrence frequency refers to the number of times a monitored indicator (such as temperature or vibration) of the component exceeds a warning threshold per unit time. Risk characteristics include abnormal trends in the indicator (such as sudden increases or sustained rises), abnormal duration (such as exceeding a set number of hours), and abnormal magnitude (such as a percentage deviation from a benchmark value). Next, based on the aforementioned frequency and characteristic information, and in conjunction with preset cycle adjustment rules, the weighted risk level of multiple target components is calculated, and the corresponding cycle adjustment coefficient is obtained based on the weighted risk level. Finally, the currently set regulatory cycle parameter for each target component is multiplied by its corresponding cycle adjustment coefficient to obtain the updated regulatory cycle parameter. When the adjustment coefficient is less than 1, it is used to shorten the cycle; when it is greater than 1, it is used to extend the cycle; and when it is equal to 1, it remains unchanged, thereby achieving dynamic optimization and refined adjustment of the regulatory plan.

[0091] The aforementioned preset periodic adjustment rules refer to a set of strategies that convert the frequency and characteristics of risk occurrence of target components into risk levels according to a certain logical mapping relationship, and determine the periodic adjustment coefficient accordingly. Specifically, the rules can set multiple risk level classification thresholds. For example, when the risk frequency exceeds the upper limit of the number of warnings, or when a sudden change trend occurs and continues for more than a set duration with a significant deviation, it is rated as "high risk," with a corresponding adjustment coefficient of 0.8; if there is only a short-term anomaly but the magnitude is small, it is rated as "medium risk," with an adjustment coefficient of 0.95; when the operating state is stable and there are no anomalies, it is rated as "low risk," with an adjustment coefficient of 1.2. These rules can be preset as tables, functions, or logical judgment structures for automatic system judgment and invocation.

[0092] For example, during a certain regulatory cycle, temperature monitoring data for a generator component showed a rapid temperature rise, sustained high temperatures exceeding 48 hours, and temperatures exceeding the set upper limit multiple times within that cycle. The calculated risk frequency was 5 times, with an abnormality exceeding 10%. According to the cycle adjustment rules, this component was classified as "medium-high risk," with a corresponding adjustment coefficient of 0.75. The original regulatory cycle was 16 weeks, so the updated cycle is adjusted to 12 weeks. In contrast, the main bearing operated smoothly, with only one minor exceedance of the limit—a small magnitude and short duration—and was assessed as "low risk," with an adjustment coefficient of 1.0, and the regulatory cycle remained unchanged.

[0093] By using a risk-level-based adjustment mechanism to differentiate the monitoring cycles of different components, their current health status and risk level can be effectively reflected, enabling precise operation and maintenance strategies. Compared to a fixed-cycle mechanism, this method helps improve the timeliness and coverage of early warnings, reduces unnecessary investment in inspection resources, and lowers the risk of failures due to monitoring lag, thereby improving the overall operational reliability and economy of wind turbines.

[0094] In some embodiments, based on updated preset regulatory cycle parameters, and utilizing the collected operational data and maintenance records of multiple target components in the offshore wind turbine within the target sea area, it is determined whether to generate on-site inspection prompts for the offshore wind turbine. In specific implementations, the method may further include the following:

[0095] Based on the collected operational data and maintenance records of multiple target components in the offshore wind turbine within the target sea area, as well as the last on-site inspection notification time and sea condition information of the offshore wind turbine's location, it is determined whether to generate on-site inspection notification information for the offshore wind turbine.

[0096] Specifically, firstly, the system acquires the operational data and corresponding maintenance records of multiple target components and extracts key parameter information, including characteristic indicators such as temperature, vibration, and strain. Secondly, it acquires the timestamp corresponding to the last on-site detection prompt information generated by the wind turbine and compares it with the current time to calculate the interval since the last on-site prompt. Thirdly, it acquires real-time sea condition information of the sea area where the current target wind turbine is located, including data such as wind speed, wave height, and current velocity, and compares it with a preset maintenance adaptation threshold. Finally, a new on-site detection prompt information is generated when any of the following conditions are met: the current component's operational data shows a significant abnormal trend and the time since the last prompt exceeds a preset interval threshold, or the sea conditions are suitable during the current maintenance window, and at least one component shows potential risk signs.

[0097] For example, during a certain regulatory cycle, if the vibration data of the wind turbine main bearing exceeds the set risk threshold for three consecutive days, and more than 14 days have passed since the last detection alert, and the predicted wind speed for the next 48 hours is less than 5 m / s and the wave height is less than 1.2 m, meeting the preset operational thresholds, then it is determined that the conditions for offshore detection are met, a new on-site detection alert is generated, and pushed to the operation and maintenance platform.

[0098] By dynamically determining whether to generate detection alerts by integrating operational data, maintenance history, and sea condition information, unnecessary on-site operations can be avoided, the timeliness and scientific validity of alerts can be improved, risk prediction and maintenance resources can be effectively coordinated, and floating offshore wind turbines can achieve a refined, window-driven intelligent maintenance strategy.

[0099] Furthermore, the generation of the aforementioned on-site inspection alerts is based on multi-source data fusion analysis: operational data provides current status indicators for each target component, such as temperature, vibration, and strain, to identify any abnormal trends; maintenance records reflect historical maintenance activities and fault information, assisting in assessing the evolution of potential risks; the last on-site inspection alert time is used to determine whether the recommended minimum inspection interval has been exceeded; and sea area information (such as wind speed, wave height, and current velocity) is used to determine whether environmental conditions are suitable for conducting on-site operations. Based on this information, the on-site inspection alerts can be populated with fields including: component number, anomaly type and intensity, recommended inspection timing, current environmental accessibility assessment, historical maintenance background, and timestamp, thereby achieving comprehensive risk perception and refined support for maintenance response.

[0100] In some embodiments, the method may further include the following:

[0101] S1: Based on the on-site detection prompts, generate a corresponding manual inspection work order on the operation and maintenance management platform;

[0102] S2: Issue the inspection work order to the on-site maintenance personnel and record the time the work order is issued;

[0103] S3: After the maintenance personnel complete the inspection and upload the on-site inspection results, the inspection results will be stored in the risk and incident database to update the maintenance records.

[0104] In specific implementation, this can include: upon receiving automatically generated on-site inspection prompts from the system, the operation and maintenance management platform automatically generates a manual inspection work order. This work order includes information such as the target offshore wind turbine number, target component name, warning type, risk level, inspection requirements, and time constraints. The work order is then sent to the corresponding on-site operation and maintenance personnel's terminals via a preset platform interface, while simultaneously recording the sending time and reception status. On-site operation and maintenance personnel conduct inspections according to the work order and upload inspection process records via mobile terminals as inspection results. These results include fault point confirmation, handling suggestions, photo records, and text descriptions. Simultaneously, during the inspection process, operation and maintenance personnel use portable measuring devices (such as handheld vibration analyzers, infrared thermometers, and bearing diagnostic instruments) to acquire the operating status parameters of the target components, forming corresponding on-site measured data. After receiving the uploaded data, the platform matches and binds the inspection results with the on-site test data recorded during the inspection, and archives them in the risk and incident database. The inspection time, wind turbine number, component identification and operator information are marked accordingly, which is used to update the operation and maintenance records of the target component and subsequent risk analysis.

[0105] Through the above implementation methods, the synchronous archiving and unified storage of on-site inspection results and objective measured data were achieved, a structured risk and accident data chain was constructed, and the credibility and completeness of operation and maintenance records were improved.

[0106] As can be seen from the above, the offshore wind turbine monitoring method provided in this specification, based on preset monitoring cycle parameters, acquires the operating data and maintenance records of multiple target components in the offshore wind turbine within the target sea area during the first monitoring cycle; utilizes a preset risk prediction model to process the operating data and maintenance records of the multiple target components during the first monitoring cycle to obtain risk prediction information for the multiple target components; updates the preset monitoring cycle parameters based on the risk prediction information of the multiple target components to obtain updated preset monitoring cycle parameters; collects the operating data and maintenance records of the multiple target components in the offshore wind turbine within the target sea area during the second monitoring cycle based on the updated preset monitoring cycle parameters; and during the second monitoring cycle, uses the collected operating data and maintenance records of the multiple target components in the offshore wind turbine within the target sea area based on the updated preset monitoring cycle parameters to determine whether to generate on-site inspection prompt information for the offshore wind turbine. Thus, by performing risk prediction on the operating data and maintenance records of the target components based on preset monitoring cycle parameters and dynamically updating the monitoring cycle parameters accordingly, adaptive monitoring of the status of key components of the offshore wind turbine can be achieved. Compared to fixed-cycle monitoring, this method can adjust the data collection frequency based on changes in component risk, optimizing the allocation of regulatory resources and improving overall operation and maintenance efficiency while ensuring safe equipment operation. Furthermore, data analysis within the updated regulatory cycle can promptly identify potential anomalies, providing a basis for determining whether on-site inspections should be conducted, thereby enhancing risk warning and response capabilities.

[0107] See Figure 2 As shown in the embodiments of this specification, a specific electronic device is also provided, wherein the electronic device includes a network communication port 201, a processor 202 and a memory 203, and the above structures are connected by internal cables so that the various structures can perform specific data interaction.

[0108] Specifically, the network communication port 201 can be used to acquire the operating data and maintenance records of multiple target components in the offshore wind turbine in the target sea area during the first round of the regulatory cycle, based on preset regulatory cycle parameters.

[0109] The processor 202 can specifically be used to process the operating data and maintenance records of multiple target components in the first round of supervision using a preset risk prediction model to obtain risk prediction information for multiple target components; update the preset supervision period parameters based on the risk prediction information of the multiple target components to obtain updated preset supervision period parameters; collect the operating data and maintenance records of multiple target components in the offshore wind turbine in the target sea area in the second round of supervision based on the updated preset supervision period parameters; and determine whether to generate on-site inspection prompt information for the offshore wind turbine based on the collected operating data and maintenance records of multiple target components in the offshore wind turbine in the target sea area during the second round of supervision, according to the updated preset supervision period parameters.

[0110] The memory 203 can be used to store the corresponding instruction program.

[0111] Based on the above method, the relevant structural performance of electronic equipment can be effectively utilized to improve the data processing speed of electronic equipment and efficiently realize a method for monitoring offshore wind turbines.

[0112] In this embodiment, the network communication port 201 can be a virtual port bound to different communication protocols, thereby enabling the sending or receiving of different data. For example, the network communication port can be a port responsible for web data communication, a port responsible for FTP data communication, or a port responsible for email data communication. Furthermore, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM or CDMA; it can also be a Wi-Fi chip; or it can be a Bluetooth chip.

[0113] In this embodiment, the processor 202 can be implemented in any suitable manner. For example, the processor can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers, etc. This specification is not limiting.

[0114] In this embodiment, the memory 203 may include a hierarchy. In a digital system, anything that can store binary data can be a memory. In an integrated circuit, a circuit with storage function but no physical form is also called a memory, such as RAM, FIFO, etc. In a system, a storage device with a physical form is also called a memory, such as a memory stick, TF card, etc.

[0115] This specification also provides a computer-readable storage medium based on the above-described method for monitoring offshore wind turbines. Based on preset monitoring cycle parameters, it acquires operational data and maintenance records of multiple target components within an offshore wind turbine in a target sea area during the first monitoring cycle. Using a preset risk prediction model, it processes the operational data and maintenance records of the multiple target components during the first monitoring cycle to obtain risk prediction information for the multiple target components. Based on the risk prediction information of the multiple target components, it updates the preset monitoring cycle parameters to obtain updated preset monitoring cycle parameters. Based on the updated preset monitoring cycle parameters, it collects operational data and maintenance records of the multiple target components within the offshore wind turbine in the target sea area during the second monitoring cycle. During the second monitoring cycle, based on the updated preset monitoring cycle parameters and using the collected operational data and maintenance records of the multiple target components within the offshore wind turbine in the target sea area, it determines whether to generate on-site inspection prompt information for the offshore wind turbine.

[0116] In this embodiment, the storage medium includes, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), cache, hard disk drive (HDD), or memory card. The memory can be used to store computer program instructions. The network communication unit can be an interface configured according to standards specified in the communication protocol for network connection communication.

[0117] In this embodiment, the specific functions and effects implemented by the program instructions stored in the computer-readable storage medium can be explained in comparison with other embodiments, and will not be repeated here.

[0118] See Figure 3 At the software level, this specification also provides a monitoring device for offshore wind turbines, which may specifically include the following structural modules:

[0119] The data acquisition module 301 is used to acquire the operating data and maintenance records of multiple target components in the offshore wind turbine in the target sea area during the first round of the regulatory cycle, based on preset regulatory cycle parameters.

[0120] The risk prediction module 302 is used to process the operating data and maintenance records of multiple target components during the first round of supervision using a preset risk prediction model, and obtain risk prediction information for multiple target components.

[0121] The parameter update module 303 is used to update the preset regulatory cycle parameters based on the risk prediction information of the multiple target components, so as to obtain the updated preset regulatory cycle parameters.

[0122] The information prompt module 304 is used to collect the operation data and maintenance records of multiple target components in the offshore wind turbine in the target sea area of ​​the second round of supervision based on the updated preset supervision cycle parameters; and within the second round of supervision, based on the updated preset supervision cycle parameters, it uses the collected operation data and maintenance records of multiple target components in the offshore wind turbine in the target sea area to determine whether to generate on-site inspection prompt information about the offshore wind turbine.

[0123] In some embodiments, the device further includes, when implemented, the operating data including at least one of the following: blade operating data, gearbox operating data, main bearing operating data, tower operating data, and generator operating data;

[0124] The blade operating data includes the blade's vibration frequency, noise, blade root torque, and blade root strain; the gearbox operating data includes the gearbox's output shaft temperature, vibration, and lubricating oil impurity content; the main bearing operating data includes the main bearing's temperature and vibration; the tower operating data includes the tower's vibration and strain; and the generator operating data includes the generator's temperature, speed, and vibration. In some embodiments, the device further includes: the maintenance record includes abnormal data, operation type, and fault location information for multiple target components.

[0125] In some embodiments, the risk prediction module 302, when specifically implemented, aligns the operating data of multiple target components with the corresponding operation and maintenance records according to a preset timestamp, extracts key features, and encodes and fuses operation and maintenance events to obtain a fused feature vector; the fused feature vector is input into a preset risk prediction model to obtain risk prediction information for multiple target components; wherein, the risk prediction information includes the frequency of risk occurrence and the characteristics of risk occurrence.

[0126] In some embodiments, the parameter update module 303, in its specific implementation, determines the weighted risk level of multiple target components based on the risk occurrence frequency and risk occurrence characteristics in the risk prediction information and the preset periodic adjustment rules, and determines the corresponding adjustment coefficient based on the weighted risk level; and multiplies the preset regulatory period parameters of the multiple target components by the corresponding adjustment coefficient to obtain the updated preset regulatory period parameters.

[0127] In some embodiments, the information prompting module 304, in specific implementation, determines whether to generate on-site inspection prompt information about the offshore wind turbine based on the collected operating data and maintenance records of multiple target components in the offshore wind turbine within the target sea area, as well as the last on-site inspection prompt time and sea condition information of the sea area where the offshore wind turbine is located.

[0128] In some embodiments, the device further includes: generating a corresponding manual inspection work order on the operation and maintenance management platform based on the on-site detection prompt information; issuing the inspection work order to the on-site operation and maintenance personnel and recording the work order issuance time; and storing the inspection results in the risk and incident database after the operation and maintenance personnel complete the inspection and upload the on-site inspection results to update the operation and maintenance records.

[0129] It should be noted that the units, devices, or modules described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. For ease of description, the above devices are described by dividing them into various modules according to their functions. Of course, in implementing this specification, the functions of each module can be implemented in the same software and / or hardware, or modules that implement the same function can be implemented by a combination of sub-modules or sub-units, etc. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and there may be other division methods in actual implementation. For example, units or components can be combined or integrated into another system, or some features can be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection between the devices or units shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0130] As can be seen from the above, the monitoring device for offshore wind turbines provided in the embodiments of this specification can adaptively monitor the status of key components of offshore wind turbines by predicting risks based on the operating data and maintenance records of target components according to preset monitoring cycle parameters, and dynamically updating the monitoring cycle parameters accordingly. Compared with fixed-cycle monitoring methods, this method can adjust the data collection frequency according to changes in component risks, optimizing the allocation of monitoring resources and improving overall operation and maintenance efficiency while ensuring the safe operation of equipment. Furthermore, through data analysis within the updated monitoring cycle, potential anomalies can be identified in a timely manner, providing a basis for determining whether on-site inspections should be conducted, thereby enhancing risk warning and response capabilities.

[0131] In a specific scenario example, the monitoring method and device for offshore wind turbines provided in this specification can be applied. By performing risk prediction on the operational data and maintenance records of target components based on preset monitoring cycle parameters, and dynamically updating the monitoring cycle parameters accordingly, adaptive monitoring of the status of key components of offshore wind turbines can be achieved. Compared to fixed-cycle monitoring methods, this method can adjust the data collection frequency according to changes in component risk, optimizing the allocation of monitoring resources and improving overall operation and maintenance efficiency while ensuring the safe operation of equipment. Simultaneously, data analysis within the updated monitoring cycle can promptly identify potential anomalies, providing a basis for determining whether on-site inspections should be conducted, thereby enhancing risk warning and response capabilities. The specific implementation process may include the following:

[0132] This manual also provides a monitoring system for offshore wind turbines; see [link / reference]. Figure 4 As shown, the system conducts early warning work through big data self-learning. The system includes: a risk supervision definition module, a risk supervision reminder module, a SCADA database, a risk record module, a risk supervision learning module, and a risk incident database.

[0133] The risk monitoring definition module is used to customize risk monitoring items for floating wind turbines, submarine cables, foundation floats, and mooring systems, and can be dynamically updated and maintained. For each risk monitoring item, if there are direct or indirect monitoring points in the integrated monitoring subsystem, video security, fire monitoring, etc., they can be selected for configuration; if there are no monitoring methods or it is difficult to supplement them, and regular on-site inspections and investigations are required, a special configuration for risk inspection and maintenance can be made in the smart inspection module.

[0134] S1: Let R be the set of all risk regulatory items, R = {r1, r2, ..., r...} n ,}where r i This represents the i-th risk regulatory item, such as r1 representing the wind turbine blade regulatory item and r2 representing the dynamic submarine cable regulatory item.

[0135] S2: Let M be the set of all detection methods, M = {m1, m2, ..., m} k}, where m j This indicates the j-th monitoring method, such as m1 being an integrated detection subsystem and m2 being video security.

[0136] S3: For each risk regulatory item r i There exists a subset Indicates the relationship with r i A set of related monitoring methods. If Then a special configuration for risk inspection needs to be made in the intelligent inspection module.

[0137] The aforementioned risk regulatory items are divided into periodic risk regulatory items and non-periodic risk regulatory items. The initial periodic risk regulatory items are defined by the wind turbine manufacturer in terms of the regulatory cycle, while the non-periodic risk regulatory items are determined based on the periodic risk regulatory items.

[0138] The risk monitoring and reminder module allows for customized monitoring cycles for each risk monitoring item, distinguishing between periodic and non-periodic monitoring items. Reminders for periodic and non-periodic risk monitoring items do not conflict with each other. After determining that a risk monitoring item is due for a reminder, the risk monitoring module will comprehensively assess the weather by combining information such as the fault status of major components, the last time the item was at sea, and sea conditions, and generate a suggested maintenance work order to remind maintenance personnel to check the relevant monitoring data or arrange on-site inspections and troubleshooting.

[0139] S1: Major component failure status is divided into three levels: 1, 2, and 3, corresponding to good, faulty, and damaged, respectively.

[0140] S2: Weather information is divided into four levels: 1, 2, 3, and 4, which correspond to suitable for going to sea, can go to sea, not recommended for going to sea, and not suitable for going to sea, respectively.

[0141] S3: Let Tr i For risk supervision items r i The regulatory cycle is divided into periodic regulatory items (Tr). i = Fixed periodic value (unit: month) and non-periodic regulatory items (special setting value).

[0142] S4: Let t now t represents the current time. last-checkri For the last check r i The time.

[0143] S5: When t now -t last-checkri ≥Tr i When a fault alarm is triggered, the system automatically retrieves the fault information of major components. When a fault alarm is triggered, the system retrieves the Weather level. When Weather ≤ 3, the system automatically issues a reminder and generates a sea departure work order. When Weather = 4, if the fault = 1, the system issues a reminder but does not generate a sea departure work order. If the fault = 2 or 3, the system continuously issues reminders until the conditions for sea departure are met.

[0144] The risk recording module, for non-periodic risk items requiring monitoring, uses data provided by the wind turbine SCADA system. The risk recording module inputs data to the risk monitoring module including blade data, gearbox data, main bearing data, tower data, and generator data. Blade data includes vibration frequency, blade noise, blade root torque, and blade root strain. Gearbox data includes gearbox output shaft temperature, gearbox vibration, and gearbox lubricating oil impurity content. Main bearing data includes main bearing temperature and main bearing vibration. Tower data includes tower vibration and tower strain. Generator data includes generator temperature, generator speed, and generator vibration. The risk recording module synchronously inputs the above data, along with the corresponding maintenance time for different data types, into the risk database.

[0145] Let S be the data set provided by the SCADA system of the wind turbine, S = {s} 叶片 s 齿轮箱 s 主轴承 s 塔筒 s 发电机 s 动态海缆}

[0146] ·s 叶片 ={v 叶片振动 n 噪音 , t 叶根扭矩 , ε 叶片应变}

[0147] ·s 齿轮箱 ={T 齿轮箱温度 v 齿轮箱振动 c 杂质含量}

[0148] ·s 主轴承 ={T 主轴承温度 v 主轴承振动}

[0149] ·s 塔筒 ={v 塔筒振动 , ε 塔筒应变}

[0150] ·s 发电机 ={T 发电机温度 ω 发电机转速 v 发电机振动}

[0151] ·s 动态海缆 ={T 海缆温度 , ε 海缆应变}

[0152] The data that the risk recording module synchronizes with the risk database can be represented as follows:

[0153] D={s 叶片 s齿轮箱 s 主轴承 s 塔筒 s 发电机 s 动态海缆}

[0154] The risk and incident database includes blade data, gearbox data, main bearing data, tower data, generator data, and maintenance time.

[0155] The risk supervision learning module acquires SCADA data of risk information recorded by the risk recording module, learns the frequency and characteristics of risk occurrence through algorithms, and redefines the next alert cycle for non-periodic regulatory items by analyzing the SCADA data. The risk supervision learning module uses a CNN model to learn and predict the risk characteristics of each component. Initially, periodic and non-periodic risk regulatory items are manually defined; subsequently, based on the learning results of the risk supervision learning module, periodic risk regulatory items can be changed to non-periodic items, and the learning results are fed back to the risk supervision alert module.

[0156] S1: Data preprocessing.

[0157] Normalization: Normalize each feature x in the data D obtained from the risk record module, using the formula:

[0158]

[0159] In this way, different feature values ​​are mapped to the [0,1] interval, which makes it easier for the model to process.

[0160] Smoothing: The moving average method is used to smooth the data. For a one-dimensional data sequence y = [y1, y2, ..., y...], ... n The result after moving average is:

[0161]

[0162] Where ω is the window size, used to remove noise from the data.

[0163] S2: Construct the CNN model structure.

[0164] Input layer: Converts the processed data D into a tensor form suitable for CNN input. Assuming the processed data D forms an n*m matrix (n is the number of samples, m is the number of features), it is reshaped as input = reshape(D, [n,m,1]), with the number of channels set to 1.

[0165] Convolutional layer: The first convolutional layer C onv1 We use C1 convolution kernels of size F1×1. The convolution operation formula is:

[0166]

[0167] The feature map output1 = C is obtained. onv1 (input), with size [n, m-F1+1, C1], and the second convolutional layer C onv2 Using C2 convolution kernels of size F2×1, convolution is performed on output1 to obtain output2 = C onv2 (output1), with a size of [n, (m-F1+1)-F2+1, C2].

[0168] Pooling layer: Taking average pooling as an example, a pooling window of size P×1 is used. For the input feature map y, the average pooling formula is:

[0169]

[0170] Pooling output2 yields pooled = AveragePool(output2, P), with a size of...

[0171] Fully connected layer: Flattens the pooled feature map into a one-dimensional vector and connects it to the fully connected layer. Let the length of the flattened vector be L. The first fully connected layer FC1 has N1 neurons, and its output is:

[0172] output3=FC1(flatten(pooled))=W1×flatten(pooled)+b1

[0173] Output layer: Output layer FC2 has 1 neuron, output:

[0174] O=FC2(output3)=W2×output3+b2

[0175] The output value O is the latest monitoring cycle of the major component (unit: month) O = ReLU(W2×output3+b2), and the output value is non-negative.

[0176] The risk supervision learning module constructs a simple convolutional neural network (CNN) model, including an input layer, convolutional layers, ReLU activation layers, pooling layers, and fully connected layers. The model is trained using the CNN, and the alert period for non-periodic regulatory items is dynamically adjusted based on the model's prediction accuracy.

[0177] Based on the above embodiments, a self-learning algorithm for the health status of large components in non-periodic monitoring of floating wind turbines is proposed. By utilizing machine learning, the algorithm can discover the relationship between operating characteristics with inconspicuous features and the status of large components, redefine the next reminder cycle for non-periodic monitoring information, and help maintenance personnel perform preventive maintenance on large components.

[0178] In some embodiments, see Figure 5 As shown, firstly, the risk recording module obtains the wind turbine's operating data from the SCADA module and transmits this data to the risk supervision learning module. Subsequently, the risk supervision learning module initiates a self-learning program to train and analyze the collected data, and transmits updated supervision cycle information to the risk supervision reminder module based on the learning results. At the same time, the risk incident database synchronously receives data from the risk recording module for the accumulation of historical risk data and the improvement of model accuracy. Finally, based on the reminder information and the actual sea conditions in the area, the operation and maintenance personnel decide whether to carry out on-site operation and maintenance work.

[0179] Based on the above embodiments, by introducing a self-learning mechanism, the risk cycle of major components of floating offshore wind turbines is intelligently updated. The monitoring cycle can be dynamically adjusted according to actual operating data, improving the accuracy and timeliness of early warning response. At the same time, combined with sea state information, operation and maintenance resources are rationally scheduled to reduce unnecessary offshore operations, thereby improving operation and maintenance efficiency, reducing costs, and enhancing the risk management capabilities of offshore wind farms.

[0180] While this specification provides the steps of operation for the methods described in the embodiments or flowcharts, more or fewer steps may be included based on conventional or non-inventive means. The order of steps listed in the embodiments is merely one possible order of execution among many steps and does not represent the only possible order. In actual device or client product execution, the methods shown in the embodiments or drawings may be executed sequentially or in parallel (e.g., in a parallel processor or multi-threaded processing environment, or even a distributed data processing environment). The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, product, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, product, or apparatus. Without further limitations, the presence of other identical or equivalent elements in a process, method, product, or apparatus that includes said elements is not excluded. The terms "first," "second," etc., are used to denote names and do not indicate any particular order.

[0181] Those skilled in the art will also know that, besides implementing the controller using purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller function as logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices within it used to implement various functions can also be considered structures within that hardware component. Alternatively, the devices used to implement various functions can be considered as both software modules implementing the method and structures within a hardware component.

[0182] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this specification can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions of this specification can essentially be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of this specification.

[0183] Although this specification has been described by way of examples, those skilled in the art will recognize that many variations and modifications are possible without departing from the spirit of this specification, and it is intended that the appended claims cover such variations and modifications without departing from the spirit of this specification.

Claims

1. A method for monitoring offshore wind turbine generators, characterized in that, include: Based on preset regulatory cycle parameters, the operation data and maintenance records of multiple target components in offshore wind turbines within the target sea area are obtained during the first round of regulatory cycle. By using a pre-set risk prediction model, the operational data and maintenance records of multiple target components during the first round of regulatory cycle are processed to obtain risk prediction information for multiple target components; Based on the risk prediction information of the multiple target components, it is determined whether there are any target components that meet the preset attention rules; wherein, the preset attention rules include at least one of the following: the risk level of the target component reaches a high risk threshold, the key operating indicators show a continuous abnormal change trend, and the target component has anomalies in multiple consecutive regulatory cycles but has not received sufficient intervention; Based on the risk prediction information of the multiple target components, the preset regulatory cycle parameters are updated to obtain the updated preset regulatory cycle parameters; Based on the updated preset regulatory cycle parameters, operational data and maintenance records of multiple target components in offshore wind turbines within the target sea area will be collected during the second round of regulatory cycles. If the monitored component exists, before the second round of monitoring cycle ends and the next data collection cycle begins, one or more supplementary data collection operations will be performed on the monitored component based on its risk level and characteristic evolution trend. The supplementary data collection operations include collecting multi-source operational data and maintenance image records for the monitored component at a specified time point or when specific triggering conditions are met. The specific triggering conditions include at least one of the following: abnormal fluctuation of key operating parameters, deterioration of lubricating oil quality, or abnormal change in current load. During the second regulatory cycle, based on the updated preset regulatory cycle parameters, the system will use the collected operating data and maintenance records of multiple target components in the offshore wind turbines within the target sea area to determine whether to generate on-site inspection prompts for the offshore wind turbines. The step of determining whether to generate on-site inspection prompt information for the offshore wind turbine includes: The operational data collected for each target component during the second round of monitoring are normalized and averaged, and key feature indicators are extracted. The key feature indicators are compared with the historical operating threshold range and abnormal state characteristics corresponding to the target component; If the key feature indicators meet the set prompt generation conditions, then on-site detection prompt information about the target component is generated; wherein, the prompt generation conditions include at least one of abnormal fluctuation amplitude exceeding the limit, trend change rate exceeding the threshold, or consecutive abnormal number exceeding the limit.

2. The method according to claim 1, characterized in that, The operating data includes at least one of the following: blade operating data, gearbox operating data, main bearing operating data, tower operating data, and generator operating data; The blade operating data includes the blade's vibration frequency, noise, blade root torque, and blade root strain; the gearbox operating data includes the gearbox's output shaft temperature, vibration, and lubricating oil impurity content; the main bearing operating data includes the main bearing's temperature and vibration; the tower operating data includes the tower's vibration and strain; and the generator operating data includes the generator's temperature, speed, and vibration.

3. The method according to claim 2, characterized in that, The maintenance records include abnormal data, operation types, and fault location information for multiple target components.

4. The method according to claim 3, characterized in that, The process utilizes a pre-defined risk prediction model to process the operational data and maintenance records of multiple target components during the first round of regulatory cycles, obtaining risk prediction information for these components, including: Align the operational data of multiple target components with the corresponding maintenance records according to the preset timestamps, extract key features, and encode and fuse maintenance events to obtain the fused feature vector. The fused feature vector is input into a preset risk prediction model to obtain risk prediction information for multiple target components; wherein, the risk prediction information includes the frequency of risk occurrence and the characteristics of risk occurrence.

5. The method according to claim 4, characterized in that, The step of updating the preset regulatory cycle parameters based on the risk prediction information of the multiple target components to obtain the updated preset regulatory cycle parameters includes: The weighted risk levels of multiple target components are determined based on the frequency and characteristics of risk occurrence in the risk prediction information and the preset periodic adjustment rules. The corresponding adjustment coefficient is determined based on the weighted risk level; The preset regulatory cycle parameters of the multiple target components are multiplied by the corresponding adjustment coefficients to obtain the updated preset regulatory cycle parameters.

6. The method according to claim 5, characterized in that, Based on the updated preset regulatory cycle parameters, and utilizing the collected operational data and maintenance records of multiple target components within the offshore wind turbines in the target sea area, it is determined whether to generate on-site inspection alerts for the offshore wind turbines, including: Based on the collected operational data and maintenance records of multiple target components in the offshore wind turbine within the target sea area, as well as the last on-site inspection notification time and sea condition information of the offshore wind turbine's location, it is determined whether to generate on-site inspection notification information for the offshore wind turbine.

7. The method according to claim 6, characterized in that, The method further includes: Based on the on-site detection prompts, a corresponding manual inspection work order is generated on the operation and maintenance management platform; The inspection work order is issued to the on-site maintenance personnel, and the time of issuance of the work order is recorded; After the maintenance personnel complete the inspection and upload the on-site inspection results, the inspection results are stored in the risk and incident database to update the maintenance records.

8. A monitoring device for offshore wind turbines, characterized in that, include: The data acquisition module is used to acquire the operating data and maintenance records of multiple target components in the offshore wind turbine in the target sea area during the first round of the regulatory cycle, based on preset regulatory cycle parameters. The risk prediction module is used to process the operational data and maintenance records of multiple target components during the first round of supervision using a preset risk prediction model, and obtain risk prediction information for multiple target components. Based on the risk prediction information of the multiple target components, it is determined whether there are any target components that meet the preset attention rules; wherein, the preset attention rules include at least one of the following: the risk level of the target component reaches a high risk threshold, the key operating indicators show a continuous abnormal change trend, and the target component has anomalies in multiple consecutive regulatory cycles but has not received sufficient intervention; The parameter update module is used to update the preset regulatory cycle parameters based on the risk prediction information of the multiple target components, so as to obtain the updated preset regulatory cycle parameters. The information prompt module is used to collect the operation data and maintenance records of multiple target components in the offshore wind turbine in the target sea area during the second round of supervision, based on the updated preset supervision cycle parameters. If the monitored component exists, before the second round of monitoring cycle ends and the next data collection cycle begins, one or more supplementary data collection operations will be performed on the monitored component based on its risk level and characteristic evolution trend. The supplementary data collection operations include collecting multi-source operational data and maintenance image records for the monitored component at a specified time point or when specific triggering conditions are met. The specific triggering conditions include at least one of the following: abnormal fluctuation of key operating parameters, deterioration of lubricating oil quality, or abnormal change in current load. During the second regulatory cycle, based on the updated preset regulatory cycle parameters, the system will use the collected operating data and maintenance records of multiple target components in the offshore wind turbines within the target sea area to determine whether to generate on-site inspection prompts for the offshore wind turbines. The step of determining whether to generate on-site inspection prompt information for the offshore wind turbine includes: The operational data collected for each target component during the second round of monitoring are normalized and averaged, and key feature indicators are extracted. The key feature indicators are compared with the historical operating threshold range and abnormal state characteristics corresponding to the target component; If the key feature indicators meet the set prompt generation conditions, then on-site detection prompt information about the target component is generated; wherein, the prompt generation conditions include at least one of abnormal fluctuation amplitude exceeding the limit, trend change rate exceeding the threshold, or consecutive abnormal number exceeding the limit.

9. An electronic device, characterized in that, It includes a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements the steps of the monitoring method for an offshore wind turbine as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores computer instructions that, when executed by a processor, implement the steps of a monitoring method for an offshore wind turbine as described in any one of claims 1 to 7.

Images