A wind turbine health state assessment method and device based on hint engineering

Abstract

The application provides a wind turbine health state evaluation method and device based on prompt engineering, the method comprising: collecting multi-source heterogeneous data of a wind turbine and preprocessing the data; constructing a four-layer knowledge system based on the preprocessed data; according to the real-time running situation of the wind turbine, calling a prompt template corresponding to the four-layer knowledge system to generate a situational prompt chain; inputting the situational prompt chain into a large language model, performing cross-level knowledge reasoning through the large language model, and outputting sub-item evaluation results of technical state, economic influence and risk level; based on the sub-item evaluation results, constructing three-dimensional health degree comprehensive evaluation indexes of technical health degree, economic health degree and risk health degree; generating an evaluation report according to the three-dimensional health degree comprehensive evaluation indexes, and providing operation and maintenance suggestions. The application realizes the real-time, comprehensiveness, self-adaptability and explainability of the wind turbine health evaluation, and provides scientific and comprehensive quantitative basis for operation and maintenance decision-making.

Description

Technical Field

[0001] This application relates to the field of wind turbine condition monitoring and health management technology, and in particular to a method and device for assessing the health status of wind turbines based on prompting engineering. Background Technology

[0002] With the continuous growth of wind power installed capacity, the reliable operation and efficient maintenance of wind turbine units have become crucial to ensuring the economic benefits of wind farms. Currently, the health status assessment of wind turbine units mainly relies on multi-source data generated by their Supervisory Control and Data Acquisition (SCADA) systems and vibration monitoring systems. However, existing assessment methods face many challenges in practical applications and are insufficient to meet the needs for real-time, comprehensive, and accurate understanding of the unit's status.

[0003] The existing technical solutions have the following limitations: Firstly, expert systems based on preset rules are a relatively traditional assessment method. These systems set static thresholds for alarms and diagnoses based on industry standards (such as IEC 61400-25) or operational experience. Their drawbacks are: the rules are mostly statically set, making it difficult to adapt to dynamic changes in wind turbines caused by equipment aging, environmental fluctuations, and complex operating conditions (such as wind speed and load variations), leading to false alarms or missed alarms; furthermore, these systems typically only process structured data such as SCADA and sensor data, neglecting unstructured text containing crucial information, such as fault logs, maintenance records, and maintenance personnel experience, resulting in insufficient information utilization and a one-sided assessment perspective.

[0004] Secondly, traditional machine learning methods, such as Support Vector Machines (SVM) and Random Forests, are applied to fault classification and prediction. While these methods can learn patterns from data to some extent, their effectiveness heavily relies on complex and specialized feature engineering, resulting in high implementation barriers and cumbersome operations. More importantly, these models are often considered "black boxes," with their internal decision-making logic difficult to interpret, leading to insufficient trust from operations and maintenance personnel in their outputs. Furthermore, these models struggle to effectively integrate multimodal heterogeneous information such as vibration signals, temperature data, and textual descriptions.

[0005] Furthermore, advanced technologies such as Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) models and knowledge graphs are being explored. Deep learning models perform excellently in time series prediction, but their training typically relies on massive amounts of high-quality, labeled historical fault data, resulting in extremely high labeling costs and limiting their application in wind power scenarios where fault samples are relatively scarce. While knowledge graphs can structurally represent domain knowledge, their construction, maintenance, and updates require a continuous investment of significant expert manpower, and the complex reasoning process based on graphs incurs high computational costs, making it difficult to meet the urgent need of wind farms for online, real-time assessment of unit health status.

[0006] In summary, the current field of wind turbine health assessment faces core challenges, including difficulties in integrating heterogeneous data, translating expert experience into practical applications, and ensuring model generalization. Existing technologies either suffer from poor adaptability and interpretability, or are costly to implement and lack real-time capabilities. There is a lack of effective means to integrate multi-source information efficiently and cost-effectively, and to transform implicit operational knowledge into dynamic and interpretable assessment logic. Summary of the Invention

[0007] In view of this, this application proposes a method, device, equipment, and storage medium for wind turbine health status assessment based on prompting engineering, which realizes the real-time, comprehensive, adaptive, and interpretable nature of wind turbine health assessment, and provides a scientific and comprehensive quantitative basis for operation and maintenance decisions.

[0008] Firstly, this application provides a method for assessing the health status of wind turbine generators based on prompting engineering, comprising the following steps: Collect multi-source heterogeneous data from wind turbine generators and preprocess the data. The multi-source heterogeneous data includes structured data and unstructured text data. A four-layer knowledge system is constructed based on the preprocessed data. The four-layer knowledge system includes a physical layer, a state layer, a history layer, and a strategy layer. Based on the real-time operating scenario of the wind turbine, the prompt templates corresponding to the four-layer knowledge system are invoked to generate a contextualized prompt chain. The contextualized cue chain is input into the large language model, and cross-level knowledge reasoning is performed through the large language model to output the sub-evaluation results of technology status, economic impact and risk level; Based on the results of the sub-items assessment, a three-dimensional comprehensive health assessment index is constructed, comprising technological health, economic health, and risk health. An assessment report is generated based on the three-dimensional health comprehensive assessment indicators, and operation and maintenance suggestions are provided.

[0009] Therefore, the wind turbine health status assessment method based on prompting engineering provided in this application constructs a four-layer knowledge system through multi-source heterogeneous data fusion (structured and unstructured text), realizing cross-level knowledge association of physical, state, history, and strategy. Combined with contextualized prompting chains driving large language model reasoning, it outputs three-dimensional sub-assessments of technology, economics, and risk, as well as a comprehensive health index, ultimately generating an assessment report that includes operation and maintenance decisions. This application breaks through the limitations of traditional single-dimensional assessments, forming a full-link intelligent assessment system encompassing data collection, knowledge modeling, contextual reasoning, comprehensive assessment, and decision-making closed loops. This enhances the comprehensiveness, dynamism, and decision support capabilities of wind turbine health status assessment, providing a scientific and comprehensive quantitative basis for operation and maintenance decisions.

[0010] Optionally, the data preprocessing includes: An improved 3σ criterion, combined with wind speed threshold conditions, was used to remove outliers from structured data. Structured data is subjected to adaptive standardization by introducing a working condition correction factor. Keyword extraction was performed on unstructured text data based on a wind power professional dictionary, and the data was then converted into standardized entries.

[0011] Based on the above, outliers in structured data are removed using the improved 3σ criterion and wind speed threshold conditions. Adaptive standardization of operating conditions is achieved by combining this with a working condition correction coefficient. Then, keywords from unstructured text are extracted using a professional dictionary and transformed into standardized entries. This improves the quality and consistency of multi-source data, solves the problems of high sensitivity of structured data to operating conditions and semantic fragmentation of unstructured text, and provides a high-quality data foundation for subsequent knowledge system construction.

[0012] Optionally, the construction of the four-layer knowledge system includes: Physical layer knowledge: Organize the standardized sensor data into structured entries, including parameter names, standardized values, acquisition time, associated components, and operating condition information; State layer knowledge: Based on physical layer data, the operating state indicators of wind turbines are derived through calculation formulas. The operating state indicators include at least load rate and component degradation trend. Historical layer knowledge: Associate historical fault text entries with historical fault impact coefficients, which are calculated based on the historical frequency of occurrence and average downtime of the fault. Strategy layer knowledge: Transform operational experience into configurable trigger rules that include threshold conditions.

[0013] Based on the above, a hierarchical knowledge framework consisting of a physical layer, a state layer, a historical layer, and a strategy layer enables the step-by-step accumulation and structured transformation from raw data to decision-making knowledge. The physical layer clarifies the core attributes and relationships of the data, providing fundamental data support for assessment. The state layer accurately depicts the unit's operating status and performance changes through quantitative indicators, achieving a leap from data collection to state awareness. The historical layer quantifies the severity of faults and their recurrence reference value through historical fault impact coefficients, making experiential knowledge reusable. The strategy layer transforms operation and maintenance rules into configurable trigger conditions, realizing the implementation of experience into decision-making. The construction of this four-layer knowledge system breaks down the barriers between data and knowledge, transforming multi-source information into structured knowledge that can be reasoned about and reused by the model. This provides core support for the generation of contextualized prompt chains and precise LLM reasoning, improving the logic and coherence of the assessment.

[0014] Optionally, the formula for calculating the load rate is: ; in, For load rate, For real-time output power, The rated power of the unit, This is the wind speed correction factor; The formula for calculating the degradation trend of the component is: ; in, For the trend of component deterioration, This represents the number of samples taken within a preset time window in the near future. The number of sampling times is preset for a future time window; , Standardize the data for the corresponding window; Weighted by years of service; The inherent attenuation coefficient is configured based on the component type.

[0015] As described above, the load rate formula incorporates a wind speed correction coefficient to address the bias issue of traditional load calculations neglecting the influence of wind speed, accurately reflecting the actual operating load of the unit under different wind speed conditions. The component degradation trend formula, through comparison of recent and long-term data, combined with the weight of service life and the inherent attenuation coefficient of the component, quantifies the degree of component performance degradation, extending from real-time status monitoring to trend prediction. These two formulas provide a scientific and reproducible quantitative basis for state-level knowledge, avoiding subjective judgments on operating status and degradation trends in traditional assessments, making technical condition assessments more objective and accurate, and providing reliable quantitative support for subsequent fault hazard identification and risk prediction.

[0016] Optionally, the generation of the contextualized cue chain includes: The current operating status of the wind turbine is identified as normal monitoring, abnormal alarm, or regular inspection scenario. The four types of prompt templates corresponding to the four-layer knowledge system are invoked, and the knowledge and calculation results of each layer are embedded in the templates. The prompts are integrated according to the logical order of the physical layer, state layer, history layer, and strategy layer to form a complete set of reasoning instructions.

[0017] Based on the above, a prompt chain generation mechanism that integrates scenario recognition, template adaptation, and logical integration addresses the challenges of translating experience into concrete results and lacking focus in traditional assessments. Specifically, it adaptively invokes prompt templates based on real-time unit operation scenarios, avoiding redundancy and improving the targeting of reasoning; it embeds four layers of knowledge and quantitative calculation results into corresponding templates, automating the transformation of implicit experience into structured instructions and reducing reliance on large amounts of labeled data; and it logically integrates prompt content according to physics, state, history, and strategy, guiding LLM to focus on the core dimensions of wind power scenarios and significantly improving the accuracy and efficiency of cross-level knowledge integration.

[0018] Optionally, the formula for calculating the technological health status is: ; in, For the health of technology, For the first The dynamic weight of each component is related to the historical failure impact coefficient and the component's degradation trend. Score the health of the components. To assess the number of parts; The formula for calculating the economic health level is: ; in, For economic health, To estimate the total loss, For daily revenue per machine, Preset recycling cycle; The formula for calculating the risk health level is as follows: ; in, For risk health, The probability of failure occurring. For risk coupling coefficient, To estimate downtime, To preset a safe operating cycle.

[0019] Based on the above, a three-dimensional health assessment system encompassing technology, economics, and risk is constructed using three quantitative formulas. This breaks through the limitations of traditional single-technical-indicator assessments, achieving synergistic evaluation across these three dimensions and supporting the generation of multi-dimensional decision-making basis. The technology health formula introduces dynamic weights, strongly correlating component weights with historical failure impacts and degradation trends, accurately reflecting the health status of core components and severely degraded components. The economic health formula quantifies the impact of failures on economic benefits by comparing potential total losses with expected gains, providing an economic feasibility basis for operation and maintenance decisions. The risk health formula introduces a risk coupling coefficient, considering the inter-component correlation effects, and accurately assessing the probability of failure occurrence and the risk of its spread.

[0020] Optionally, after generating the assessment report based on the three-dimensional health comprehensive assessment indicators, the report may also include: Based on the assessment report, different levels of early warning notifications will be automatically triggered; Based on operation and maintenance feedback data, the attenuation coefficient, risk coupling coefficient, and prompt template used in constructing the three-dimensional health status are dynamically optimized.

[0021] Based on the above, different levels of early warning notifications are automatically triggered based on the assessment report, and the attenuation coefficient, risk coupling coefficient and prompt template are dynamically optimized through operation and maintenance feedback data, forming an adaptive closed-loop mechanism of assessment, early warning, feedback and optimization. This improves the system's response speed to changes in unit status and its continuous optimization capability, and enhances the robustness and long-term applicability of the assessment system.

[0022] Secondly, this application provides a wind turbine health status assessment device based on prompting engineering, comprising: The data collection and preprocessing module is used to collect multi-source heterogeneous data from wind turbine generators and preprocess the data. The multi-source heterogeneous data includes structured data and unstructured text data. The four-layer knowledge system construction module is used to construct a four-layer knowledge system based on preprocessed data. The four-layer knowledge system includes a physical layer, a state layer, a history layer, and a strategy layer. The contextualized prompt fusion engine module is used to call prompt templates corresponding to the four-layer knowledge system based on the real-time operating scenario of the wind turbine and generate a contextualized prompt chain. The large language model reasoning module is used to input the contextualized cue chain into the large language model, perform cross-level knowledge reasoning through the large language model, and output the sub-evaluation results of technical status, economic impact and risk level; The three-dimensional health assessment module is used to construct a comprehensive three-dimensional health assessment index of technological health, economic health, and risk health based on the results of the sub-assessments. The results output and decision-making module is used to generate an assessment report based on the three-dimensional health comprehensive assessment indicators and provide operation and maintenance suggestions.

[0023] Thirdly, this application provides a computing device, the computing device comprising: processor; Memory, used to store one or more programs; When the processor executes one or more programs, it enables the processor to implement the above-described method for assessing the health status of wind turbine generators based on prompting engineering.

[0024] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a computer, implements the aforementioned method for assessing the health status of wind turbine generators based on prompting engineering.

[0025] These and other aspects of this application will become more apparent in the description of the following embodiments(s). Attached Figure Description

[0026] Figure 1A schematic diagram illustrating a wind turbine health status assessment method based on prompting engineering, provided for an embodiment of this application; Figure 2 A flowchart of a wind turbine health status assessment method based on prompting engineering provided for embodiments of this application; Figure 3 A structural diagram of a wind turbine health status assessment device based on prompting engineering provided in this application embodiment; Figure 4 This is a structural diagram of a computing device provided in an embodiment of this application.

[0027] It should be understood that the dimensions and shapes of the block diagrams in the above structural diagrams are for reference only and should not constitute an exclusive interpretation of the embodiments of this application. The relative positions and inclusion relationships between the block diagrams presented in the structural diagrams are only schematic representations of the structural relationships between the block diagrams, and are not intended to limit the physical connection methods of the embodiments of this application. Detailed Implementation

[0028] The technical solutions provided in this application will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the system architecture and business scenarios provided in the embodiments of this application are mainly for illustrating possible implementations of the technical solutions of this application and should not be construed as the sole limitation on the technical solutions of this application. Those skilled in the art will recognize that the technical solutions provided in this application are equally applicable to similar technical problems as system architectures evolve and new business scenarios emerge.

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. In case of any inconsistency, the meaning set forth in this specification or derived from the content described herein shall prevail. Furthermore, the terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit the scope of this application.

[0030] The solutions provided in this application will now be described in detail with reference to the accompanying drawings and embodiments.

[0031] This application proposes a method and device for assessing the health status of wind turbine units based on prompting engineering. Through three major innovations—hierarchical knowledge system, contextualized prompt templates, and interpretable reasoning process—it achieves real-time, comprehensive, adaptive, and interpretable health assessment of wind turbine units, providing a scientific and comprehensive quantitative basis for operation and maintenance decisions.

[0032] like Figure 1 The diagram shown is a schematic representation of a wind turbine health status assessment method based on prompting engineering, provided in an embodiment of this application. Figure 2The diagram shown is a flowchart of a wind turbine health status assessment method based on prompting engineering provided in this embodiment. (Refer to...) Figure 1 and Figure 2 As shown, the method includes the following steps: S110: Collect multi-source heterogeneous data from wind turbine generators and preprocess the data.

[0033] Reference Figure 2 As shown, the multi-source heterogeneous data acquisition and preprocessing in this embodiment specifically includes: First, operational data of the target wind turbines is collected from the wind farm data system. Data sources include structured and unstructured text data. Structured data includes operational parameters such as wind speed, engine speed, power, and pitch angle obtained from the SCADA system, as well as physical quantity monitoring data from sensors such as vibration, temperature, and fluid levels. Unstructured text data includes fault alarm logs, maintenance work orders, and component replacement records obtained from the asset management system, and maintenance manuals and descriptions of historical typical fault cases obtained from the knowledge base.

[0034] Subsequently, the collected raw data undergoes preprocessing to lay the foundation for knowledge construction. This includes outlier removal and adaptive standardization for structured data, and keyword extraction and structuring for unstructured text data. Specifically: Outlier Removal: For structured data, an improved 3σ criterion combined with a wind speed threshold is used to remove outliers. To avoid misjudgment under special operating conditions, a wind speed threshold is set. Conditions. For example, the cut-in wind speed can be configured. Rated wind speed When the real-time wind speed or At that moment, power, vibration, and other data are not included in the anomaly detection, but are only marked to avoid misjudgment under low or high wind speed conditions. For the data involved in the detection, the mean and standard deviation within the sliding window are calculated, and data points that deviate from the mean by more than 3 times the standard deviation are removed.

[0035] Adaptive Standardization: For structured data, an operating condition correction factor adapted to wind speed is introduced. , load-adaptive This addresses the issue of inconsistent data units under different operating conditions. For any monitoring parameter... (e.g., bearing temperature), its standardized value The calculation formula is as follows:

[0036] in, The original data, , Based on the current wind speed Load rate Dynamically adjusted reasonable range of extreme values, Correction factor for component type (e.g., rotating components) Static components ).final, ,Right now It is normalized to the interval [0, 1].

[0037] Keyword extraction and structuring: For unstructured text data, a pre-built wind power professional dictionary (containing component and fault terms such as gearbox, generator, bearing, pitting, and overheating) can be loaded. Dictionary matching and named entity recognition technologies can be used to extract key information from fault logs or maintenance records and fill it into a standard template, transforming unstructured text into structured entries of "fault location - fault phenomenon - triggering condition - handling measures".

[0038] S120: Construct a four-layer knowledge system based on preprocessed data.

[0039] In this step, a four-layer knowledge system, including a physical layer, a state layer, a history layer, and a strategy layer, can be constructed based on the preprocessed data to achieve a step-by-step accumulation and structured transformation from raw data to decision knowledge.

[0040] Physical layer knowledge: Organizing standardized sensor data into structured entries, including parameter names and standardized values. Data collection time, associated components, and operating conditions.

[0041] State-level knowledge: Based on physical layer data, the operating state indicators of wind turbines are derived through calculation formulas. These operating state indicators include at least load rate and component degradation trend. The formula for calculating the load rate is:

[0042] in, For load rate, For real-time output power, The rated power of the unit, This is the wind speed correction factor, used to correct the theoretical impact of wind speed fluctuations on power output; The component degradation trend is used to quantify the rate of performance degradation of the component, and the calculation formula is as follows:

[0043] in, For the trend of component deterioration, This represents the number of samples taken within a preset time window in the near future. The number of sampling times is preset for a future time window; , Standardize the data for the corresponding window; Weighted by years of service; The inherent attenuation coefficient is configured based on the component type.

[0044] Historical layer knowledge: Associate historical fault text entries with historical fault impact coefficients, which are calculated based on the historical frequency of occurrence and average downtime of the fault.

[0045] Should The coefficient reflects the historical severity of a specific fault, forming entries such as: {fault location, fault phenomenon,} Handling measures, historical maintenance costs.

[0046] Strategy layer knowledge: Transforming operational experience into configurable trigger rules that include threshold conditions, such as a maintenance trigger rule: "When..." and "At that time, a maintenance suggestion will be triggered". For parameter threshold, The degradation trend threshold can be configured.

[0047] S130: Based on the real-time operating scenario of the wind turbine, call the prompt template corresponding to the four-layer knowledge system to generate a contextualized prompt chain.

[0048] In this step, the contextualized prompt fusion engine automatically determines the scenario (such as normal monitoring, abnormal alarm, or periodic inspection) based on the unit's current status and calls the corresponding prompt template. It embeds hierarchical knowledge and the calculation results of the aforementioned formulas, and integrates the prompt content according to the logical order of the physical layer, state layer, history layer, and policy layer to form a complete inference instruction set. An example of a prompt template is as follows: The physical layer indicates: "The normalized value of the [parameter name] of the [associated component] is known to be..." (Working condition correction factor) , ( ), determine if there are any abnormal signals? Status layer prompt: "Current [component] load rate" Deterioration trend _, combined with the service life of this component In 2018, will there be any potential faults and their severity? Historical layer prompt: "Impact coefficient of similar historical failures" What are the corresponding handling measures and maintenance costs? Considering the current situation, what is the probability of this fault recurring? The strategy layer prompts: "Based on the above status data and historical impact, does the preset maintenance trigger rule meet? Please deduce the potential economic loss and risk level." The contextualized prompt fusion engine fills the hierarchical knowledge and calculation results generated in step S120 into the corresponding template and forms a complete inference instruction set according to the logic of "physical, state, history, and strategy" to guide the Large Language Model (LLM) to perform inference.

[0049] S140: Input the contextualized cue chain into the large language model, perform cross-level knowledge reasoning through the large language model, and output the sub-evaluation results of technology status, economic impact and risk level.

[0050] This step uses the LLM inference module to input the generated contextualized cue chain into a large language model fine-tuned with wind power-related text (e.g., the Llama 2-7B model fine-tuned using numerous maintenance manuals and fault reports). This model can understand the domain terminology and computational logic in the cue, perform cross-level knowledge integration and reasoning, and output structured reasoning results, mainly including: Technical status reasoning: Output the health level (Excellent / Good / Medium / Poor) of each component and potential hazards, with correlations in the reasoning process. and Quantitative relationships; Economic impact reasoning: Based on historical maintenance costs, load factor formulas, and combined with the daily revenue per wind turbine unit. Derive the potential total loss:

[0051] in, For maintenance costs, To estimate downtime, For grid connection electricity price, Current load rate; Risk level: Based on the probability of failure. In relation to the scope of impact, a configurable risk coupling coefficient is introduced. It indicates the interrelationship between components, outputs the risk level and the spread trend.

[0052] S150: Based on the results of the sub-items assessment, construct a three-dimensional comprehensive health assessment index encompassing technological health, economic health, and risk health.

[0053] This step breaks through the limitations of traditional single-technical indicator assessment by quantifying the LLM output into three-dimensional health indicators, achieving a synergistic assessment of technical, economic, and risk dimensions, and supporting the generation of multi-dimensional decision-making basis. Specifically: Technology Health The formula reflects the overall technical condition and is as follows:

[0054]

[0055] in, For the health of technology, For the first Dynamic weights of each component As the initial weights, For the first Historical impact coefficient of each component; For the first The deterioration trend of individual components Score the health of the components. To assess the number of parts; Economic health The formula for measuring the potential economic loss is as follows:

[0056] in, For economic health, To estimate the total loss, For daily revenue per machine, Preset recycling cycle; Risk Health The formula for assessing operational safety risks is as follows:

[0057] in, For risk health, The higher the score, the lower the risk. The probability of failure occurring. For risk coupling coefficient, To estimate downtime, Preset safe operating cycle (configurable).

[0058] S160: Generate an assessment report based on the three-dimensional health comprehensive assessment indicators and provide operation and maintenance suggestions.

[0059] This step combines the three-dimensional health assessment index calculated in step S150 with the hidden danger points and reasoning basis obtained from LLM reasoning to generate a three-dimensional health index ( , , ), a list of key hidden dangers, and a visual assessment report on the reasoning basis, and according to , , The system prioritizes key performance indicators (KPIs) and provides specific maintenance recommendations (e.g., "Prioritize the maintenance of the spindle bearing; estimated repair cost: XX yuan; potential loss of XX yuan can be avoided"). This visual assessment report and specific maintenance recommendations can be pushed to maintenance personnel via a web interface or mobile device.

[0060] In some embodiments, different levels of early warning notifications can be automatically triggered based on the visual assessment report, for example, if If so, a level-two warning notification will be sent to the relevant responsible person; and If this occurs, a Level 1 emergency warning will be triggered.

[0061] In some embodiments, by recording each evaluation result and subsequent actual operation and maintenance feedback (such as whether a failure actually occurred, actual repair costs, and downtime), and using this feedback data, key parameters in the model are periodically and automatically optimized, such as adjusting the attenuation coefficients of different components. Risk coupling coefficient Furthermore, it can optimize the wording of prompt templates to achieve adaptive evolution of the evaluation model and enhance the robustness and long-term applicability of the evaluation system.

[0062] In summary, the wind turbine health status assessment method based on prompting engineering provided in this application constructs a four-layer knowledge system through multi-source heterogeneous data fusion (structured and unstructured text), achieving cross-level knowledge association of physical, state, historical, and strategic aspects. Combined with contextualized prompting chains driving large language model reasoning, it outputs three-dimensional sub-assessments of technology, economics, and risk, as well as a comprehensive health index, ultimately generating an assessment report that includes operation and maintenance decisions. This application breaks through the limitations of traditional single-dimensional assessments, forming a full-link intelligent assessment system encompassing data collection, knowledge modeling, contextual reasoning, comprehensive assessment, and decision-making closed loops. This enhances the comprehensiveness, dynamism, and decision support capabilities of wind turbine health status assessment, providing a scientific and comprehensive quantitative basis for operation and maintenance decisions.

[0063] like Figure 3 As shown, this application also provides a wind turbine health status assessment device based on prompting engineering. This device can be used to implement any step of the above-described wind turbine health status assessment method based on prompting engineering and its optional embodiments, as shown below. Figure 3 As shown, the device includes a data collection and preprocessing module 210, a four-layer knowledge system construction module 220, a contextualized prompting fusion engine module 230, a large language model reasoning module 240, a three-dimensional health assessment module 250, and a result output and decision-making module 260.

[0064] The system includes the following modules: a data collection and preprocessing module 210, which collects multi-source heterogeneous data from wind turbines and preprocesses the data, including structured and unstructured text data; a four-layer knowledge system construction module 220, which constructs a four-layer knowledge system based on the preprocessed data, comprising a physical layer, a state layer, a historical layer, and a strategy layer; a contextualized prompting fusion engine module 230, which generates a contextualized prompting chain by calling prompt templates corresponding to the four-layer knowledge system based on the real-time operating scenario of the wind turbines; a large language model reasoning module 240, which inputs the contextualized prompting chain into a large language model, performs cross-level knowledge reasoning through the large language model, and outputs sub-item evaluation results for technical status, economic impact, and risk level; a three-dimensional health assessment module 250, which constructs a three-dimensional comprehensive health assessment index for technical health, economic health, and risk health based on the sub-item evaluation results; and a result output and decision-making module 260, which generates an assessment report based on the three-dimensional comprehensive health assessment index and provides operation and maintenance suggestions.

[0065] It should be understood that the apparatus or module in the embodiments of this application can be implemented by software, for example, by a computer program or instruction having the above-described functions. The corresponding computer program or instruction can be stored in the internal memory of the terminal, and the above functions can be implemented by the processor reading the corresponding computer program or instruction in the memory. Alternatively, the apparatus or module in the embodiments of this application can also be implemented by hardware. Or, the apparatus or module in the embodiments of this application can also be implemented by a combination of a processor and a software module.

[0066] It should be understood that the processing details of the apparatus or module in the embodiments of this application can be found by referring to... Figures 1-2 The descriptions of the embodiments and related extended embodiments shown will not be repeated in this application.

[0067] Figure 4 This is a structural diagram of a computing device 1000 provided in an embodiment of this application. The computing device 1000 includes: a processor 1010, a memory 1020, a communication interface 1030, and a bus 1040.

[0068] It should be understood that Figure 4 The communication interface 1030 in the computing device 1000 shown can be used to communicate with other devices.

[0069] The processor 1010 can be connected to the memory 1020. The memory 1020 can be used to store the program code and data. Therefore, the memory 1020 can be a storage unit inside the processor 1010, an external storage unit independent of the processor 1010, or a component that includes both the storage unit inside the processor 1010 and the external storage unit independent of the processor 1010.

[0070] Optionally, the computing device 1000 may also include a bus 1040. The memory 1020 and communication interface 1030 can be connected to the processor 1010 via the bus 1040. The bus 1040 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus 1040 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The symbol is represented by only one line, but this does not mean that there is only one bus or one type of bus.

[0071] It should be understood that in the embodiments of this application, the processor 1010 may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. Alternatively, the processor 1010 may employ one or more integrated circuits to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0072] The memory 1020 may include read-only memory and random access memory, and provides instructions and data to the processor 1010. A portion of the processor 1010 may also include non-volatile random access memory. For example, the processor 1010 may also store device type information.

[0073] When the computing device 1000 is running, the processor 1010 executes the computer execution instructions in the memory 1020 to perform the operation steps of the above method.

[0074] It should be understood that the computing device 1000 according to the embodiments of this application can correspond to the corresponding subject in executing the methods according to the various embodiments of this application, and the other operations and / or functions of each module in the computing device 1000 are respectively for implementing the corresponding processes of the methods of this embodiment. For the sake of brevity, they will not be described in detail here.

[0075] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0076] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0080] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is used to perform the above-described method, which includes at least one of the schemes described in the above embodiments.

[0082] The computer storage medium in this application embodiment can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0083] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0084] The program code contained on a computer-readable medium may be transmitted using any suitable medium, including, but not limited to, wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0085] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0086] It should be noted that the embodiments described in this application are merely some embodiments, not all embodiments. The components of the embodiments of this application typically described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the above detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0087] The terms "first, second, third, etc." or similar terms such as module A, module B, module C, etc., used in the specification and claims are only used to distinguish similar objects and do not represent a specific ordering of objects. It is understood that a specific order or sequence may be interchanged where permitted so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0088] In the above description, the labels indicating the steps do not necessarily mean that the steps will be executed. They may include intermediate steps or be replaced by other steps. Where permissible, the order of the steps may be interchanged or executed simultaneously.

[0089] The term "comprising" as used in the specification and claims should not be construed as limiting itself to what follows; it does not exclude other elements or steps. Therefore, it should be interpreted as specifying the presence of the mentioned feature, integral, step, or component, but does not exclude the presence or addition of one or more other features, integrals, steps, or components, or groups thereof. Thus, the statement "device comprising means A and B" should not be limited to a device consisting solely of components A and B.

[0090] The terms "an embodiment" or "an embodiment" as used in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in at least one embodiment of this application. Therefore, the terms "in one embodiment" or "in an embodiment" appearing throughout this specification do not necessarily refer to the same embodiment, but may refer to the same embodiment. Furthermore, in the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions between different embodiments are consistent and can be mutually referenced. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0091] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present application has been described in detail through the above embodiments, the present invention is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of the present invention, all of which fall within the scope of protection of the present invention.

Claims

1. A method for assessing the health status of wind turbine generators based on prompting engineering, characterized in that, Includes the following steps: Collect multi-source heterogeneous data from wind turbine generators and preprocess the data. The multi-source heterogeneous data includes structured data and unstructured text data. A four-layer knowledge system is constructed based on the preprocessed data. The four-layer knowledge system includes a physical layer, a state layer, a history layer, and a strategy layer. Based on the real-time operating scenario of the wind turbine, the prompt templates corresponding to the four-layer knowledge system are invoked to generate a contextualized prompt chain. The contextualized cue chain is input into the large language model, and cross-level knowledge reasoning is performed through the large language model to output the sub-evaluation results of technology status, economic impact and risk level; Based on the results of the sub-items assessment, a three-dimensional comprehensive health assessment index is constructed, comprising technological health, economic health, and risk health. An assessment report is generated based on the three-dimensional health comprehensive assessment indicators, and operation and maintenance suggestions are provided.

2. The method according to claim 1, characterized in that, The data preprocessing includes: An improved 3σ criterion, combined with wind speed threshold conditions, was used to remove outliers from structured data. Structured data is subjected to adaptive standardization by introducing a working condition correction factor. Keyword extraction was performed on unstructured text data based on a wind power professional dictionary, and the data was then converted into standardized entries.

3. The method according to claim 1, characterized in that, The construction of the four-layer knowledge system includes: Physical layer knowledge: Organize the standardized sensor data into structured entries, including parameter names, standardized values, acquisition time, associated components, and operating condition information; State layer knowledge: Based on physical layer data, the operating state indicators of wind turbines are derived through calculation formulas. The operating state indicators include at least load rate and component degradation trend. Historical layer knowledge: Associate historical fault text entries with historical fault impact coefficients, which are calculated based on the historical frequency of occurrence and average downtime of the fault. Strategy layer knowledge: Transform operational experience into configurable trigger rules that include threshold conditions.

4. The method according to claim 3, characterized in that, The formula for calculating the load rate is: ； in, For load rate, For real-time output power, The rated power of the unit, This is the wind speed correction factor; The formula for calculating the degradation trend of the component is: ； in, For the trend of component deterioration, This represents the number of samples taken within a preset time window in the near future. The number of sampling times is preset for a future time window; , Standardize the data for the corresponding window; Weighted by years of service; This is the inherent attenuation coefficient configured based on component type.

5. The method according to claim 1, characterized in that, The generation of the contextualized cue chain includes: The current operating status of the wind turbine is identified as normal monitoring, abnormal alarm, or regular inspection scenario. The four types of prompt templates corresponding to the four-layer knowledge system are invoked, and the knowledge and calculation results of each layer are embedded in the templates. The prompts are integrated according to the logical order of the physical layer, state layer, history layer, and strategy layer to form a complete set of reasoning instructions.

6. The method according to claim 1, characterized in that, The formula for calculating the health of the technology is as follows: ； in, For the health of technology, For the first The dynamic weight of each component is related to the historical failure impact coefficient and the component's degradation trend. Score the health of the components. To assess the number of parts; The formula for calculating the economic health level is: ； in, For economic health, To estimate the total loss, For daily revenue per machine, Preset recycling cycle; The formula for calculating the risk health level is as follows: ； in, For risk health, The probability of failure occurring. For risk coupling coefficient, To estimate downtime, To preset a safe operating cycle.

7. The method according to claim 1, characterized in that, After generating the assessment report based on the three-dimensional health comprehensive assessment indicators, it also includes: Based on the assessment report, different levels of early warning notifications will be automatically triggered; Based on operation and maintenance feedback data, the attenuation coefficient, risk coupling coefficient, and prompt template used in constructing the three-dimensional health status are dynamically optimized.

8. A wind turbine health status assessment device based on prompting engineering, characterized in that, include: The data collection and preprocessing module is used to collect multi-source heterogeneous data from wind turbine generators and preprocess the data. The multi-source heterogeneous data includes structured data and unstructured text data. The four-layer knowledge system construction module is used to construct a four-layer knowledge system based on preprocessed data. The four-layer knowledge system includes a physical layer, a state layer, a history layer, and a strategy layer. The contextualized prompt fusion engine module is used to call prompt templates corresponding to the four-layer knowledge system based on the real-time operating scenario of the wind turbine and generate a contextualized prompt chain. The large language model reasoning module is used to input the contextualized cue chain into the large language model, perform cross-level knowledge reasoning through the large language model, and output the sub-evaluation results of technical status, economic impact and risk level; The three-dimensional health assessment module is used to construct a comprehensive three-dimensional health assessment index of technological health, economic health, and risk health based on the results of the sub-assessments. The results output and decision-making module is used to generate an assessment report based on the three-dimensional health comprehensive assessment indicators and provide operation and maintenance suggestions.

9. A computing device, characterized in that, include: processor; Memory, used to store one or more programs; When the processor executes the one or more programs, the processor implements a wind turbine health status assessment method based on prompting engineering as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a computer, the computer program implements a wind turbine health status assessment method based on prompting engineering as described in any one of claims 1 to 7.

Images