Multi-time scale hierarchical operation and maintenance scheduling method and system for new energy station coping with extreme weather

Abstract

The application provides a new energy station multi-time scale hierarchical cascade operation and maintenance scheduling method and system for coping with extreme weather, first, multi-source data is acquired and the current scene is identified according to the multi-source data; the multi-source data is input into the mechanism model corresponding to the current scene to obtain extreme weather quantization results; then the data is input into a multi-time scale hierarchical cascade decision model to generate a decision scheme; wherein the multi-time scale hierarchical cascade decision model has long-term, medium-term and short-term execution layers and a cross-layer coordination hub; the cross-layer coordination hub is used for cooperatively adjusting the parameters of the long-term planning layer, the medium-term scheduling layer and the short-term execution layer according to the extreme weather quantization results and the mechanism model state. The application precisely quantizes the influence of extreme weather by using the mechanism model, and then dynamically cooperatively adjusts the parameters of the multi-time scale hierarchical cascade decision model by using the quantization results and the model state, thereby effectively improving the adaptability of the operation and maintenance of the new energy station under extreme weather.

Description

Technical Field

[0001] This invention relates to the field of power operation and maintenance technology, and in particular to a multi-timescale hierarchical operation and maintenance scheduling method and system for new energy power plants to cope with extreme weather. Background Technology

[0002] Against the backdrop of escalating global climate change, extreme weather events (such as blizzards, severe dust storms, low-temperature icing, freezing rain, and frost) are occurring more frequently, affecting wider areas, and causing greater damage, posing severe non-traditional safety challenges to renewable energy power plants, particularly solar and wind power plants. These power plants are mostly located in open, outdoor areas, with equipment exposed to the natural environment for extended periods. Extreme weather not only directly causes a sharp drop in power generation, resulting in huge economic losses, but also inflicts irreversible physical damage on the plant's infrastructure. Examples include snow collapses on solar modules due to blizzards, hot spot effects on modules caused by dust storms, and excessive vibration and mechanical failures in wind turbine blades due to icing, seriously threatening the safety of renewable energy assets throughout their entire lifecycle and the operational safety of maintenance personnel.

[0003] Although there are currently many advanced operation and maintenance technologies such as drone de-icing, photovoltaic panel heating, and superhydrophobic coating protection, these technologies are mostly used in isolation. Their activation timing, applicable scope, and combination methods are not embedded in an automated decision-making process based on cost-benefit analysis. At the same time, the decision-making process does not fully consider the dynamics and uncertainties of extreme weather evolution, making it difficult to achieve a balance between safety and economy. Summary of the Invention

[0004] This invention provides a multi-timescale hierarchical operation and maintenance scheduling method and system for new energy power stations to cope with extreme weather, which solves the problem of poor adaptability of existing operation and maintenance for new energy power stations under extreme scenarios.

[0005] The first aspect of this invention provides a multi-timescale hierarchical operation and maintenance scheduling method for new energy power plants to cope with extreme weather, including: Acquire multi-source data and identify the current scene based on the multi-source data; By inputting multi-source data into the mechanism model corresponding to the current scenario, the quantitative results of extreme weather are obtained; Multi-source data and extreme weather quantification results are input into a multi-timescale hierarchical decision-making model to generate decision-making schemes; The multi-timescale hierarchical decision-making model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer coordination hub. The cross-layer coordination hub is used to coordinate the adjustment of parameters of the long-term planning layer, the medium-term scheduling layer, and the short-term execution layer based on the quantitative results of extreme weather and the state of the mechanism model.

[0006] In one possible implementation, multi-source data is input into the mechanistic model corresponding to the current scenario to obtain the quantification results of extreme weather, including: Multi-source data is input into the mechanism model, and the extreme weather quantitative results are output based on the multi-field coupling algorithm built into the model. Among them, the mechanism models include, but are not limited to, snow shedding cycle model, dust accumulation rate model, and ice growth and shedding prediction model; Multi-field coupling algorithms are used to analyze the interaction mechanism of multi-physics parameters in multi-source data, in order to quantify the nonlinear correlation between various physics parameters under extreme weather conditions.

[0007] In one possible implementation, the mechanistic model includes a coupling correlation module and a solution quantization module; the coupling correlation module has coupling parameters; the coupling parameters of the mechanistic model are different for each scenario; multi-source data are input into the mechanistic model, and the extreme weather quantification results are output according to the multi-field coupling algorithm built into the model, including: The multi-source data is processed to obtain the input matrix; The coupling parameter adjustment values ​​are determined based on the input matrix and the multi-field coupling algorithm. Based on multi-source data and coupling parameter adjustment values, the time series curve of the dynamic correlation between extreme weather factors and equipment operating status parameters is determined; Based on the dynamic correlation time series curve and the solution quantification module, the quantification results of extreme weather are determined.

[0008] In one possible implementation, multi-source data and extreme weather quantification results are input into a multi-timescale hierarchical decision model to generate decision schemes, including: Extract the state of the mechanistic model after adjusting the coupling parameters; By fusing multi-source data and extreme weather quantification results, short-term decision feature sets, medium-term decision feature sets, and long-term decision feature sets are obtained. The quantitative results of extreme weather and the status of the mechanism model are input into the cross-layer collaborative center to determine short-term, medium-term and long-term decision constraints. Based on the long-term planning feature set, long-term decision constraints, and long-term planning layer, a preventive maintenance strategy and resource allocation plan are generated. Based on the mid-term scheduling feature set, mid-term decision constraints, and mid-term scheduling layer, an optimal resource scheduling scheme is generated. Based on the short-term execution feature set, short-term decision constraints, optimal resource scheduling scheme, acquired real-time feedback data, and short-term execution layer, a short-term dynamic adjustment scheme is determined. The decision-making scheme is determined based on the preventive maintenance strategy, resource allocation plan, optimal resource scheduling scheme, and short-term dynamic adjustment scheme.

[0009] In one possible implementation, the mechanistic model state after coupling parameter adjustment is extracted, including: Collect the dynamic adjustment trajectory of the coupling parameters and the corresponding adjustment trigger source data; wherein, the dynamic adjustment trajectory includes the numerical change information and timestamp information of the coupling parameters before and after adjustment, and the adjustment trigger source data includes the key feature threshold and deviation threshold that trigger the adjustment of the coupling parameters; Obtain the operational status data of the coupling and correlation module based on the adjusted parameters; wherein, the operational status data includes at least one of the following: the modeling fit of the multiphysics coupling effect, the fitting degree of the dynamic correlation time series curve, the nonlinear correlation calculation error, and the resource consumption parameters; The trajectory, trigger source data, and running status data will be dynamically adjusted to determine the state of the mechanism model.

[0010] In one possible implementation, the quantification results of extreme weather and the state of the mechanistic model are input into a cross-layer collaborative hub to determine short-term, medium-term, and long-term decision constraints, including: Scenario-based analysis of extreme weather quantitative results is performed to extract core quantitative indicators and critical conditions for equipment failure. Based on the dynamically adjusted trajectory and operational status data, the quantitative reliability level of the mechanism model for extreme weather scenarios is determined. Based on the core quantitative indicators and quantitative reliability levels of the scenario, decision constraints at each time scale level are determined.

[0011] In one possible implementation, the quantitative results of extreme weather are analyzed in a scenario-based manner to extract core quantitative indicators and critical conditions for equipment failure, including: Screening basic quantitative data directly related to the physical characteristics of the scene from the quantitative results of extreme weather events; Based on the basic quantitative data, calculate the core quantitative indicators for the scenario; Based on the technical parameters of new energy power station equipment, historical failure data, and dynamic correlation time series curves, the critical conditions for equipment failure in each scenario are determined.

[0012] In one possible implementation, based on dynamically adjusted trajectory and operational status data, the quantitative reliability level of the mechanism model for extreme weather scenarios is determined, including: Based on dynamically adjusted trajectory and operational status data, a multi-dimensional quantitative accuracy evaluation index system is constructed. We set the weights and grading thresholds for each evaluation indicator in the multi-dimensional quantitative accuracy evaluation index system, and use the weighted summation method to calculate the comprehensive score of quantitative reliability in order to determine the level of quantitative reliability.

[0013] In one possible implementation, the method also includes: The preventive maintenance strategy, optimal resource scheduling scheme, and short-term dynamic adjustment scheme in the decision-making scheme are broken down into instruction sets; Execute the instruction set, evaluate the execution process of the instruction set, and obtain the evaluation results; Based on the evaluation results, the multi-timescale hierarchical decision-making model is iteratively optimized.

[0014] A second aspect of the present invention provides a multi-timescale hierarchical operation and maintenance scheduling system for new energy power stations to cope with extreme weather, comprising: The acquisition module is used to acquire multi-source data and identify the current scene based on the multi-source data; The quantization module is used to input multi-source data into the mechanism model corresponding to the current scenario to obtain the quantification results of extreme weather. The decision module is used to input multi-source data and extreme weather quantification results into a multi-timescale hierarchical decision model to generate decision schemes; The multi-timescale hierarchical decision-making model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer coordination hub. The cross-layer coordination hub is used to coordinate the adjustment of parameters of the long-term planning layer, the medium-term scheduling layer, and the short-term execution layer based on the quantitative results of extreme weather and the state of the mechanism model.

[0015] Compared to traditional technologies, this invention provides a multi-timescale hierarchical operation and maintenance scheduling method and system for new energy power plants to cope with extreme weather. First, multi-source data is acquired and the current scenario is identified based on this data. The multi-source data is then input into a mechanism model corresponding to the current scenario to obtain quantified results of extreme weather. Finally, the multi-source data and the quantified results are input into a multi-timescale hierarchical decision model to generate a decision scheme. This model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer coordination hub. The cross-layer coordination hub is used to coordinately adjust the parameters of the long-term planning layer, medium-term scheduling layer, and short-term execution layer based on the quantified results of extreme weather and the state of the mechanism model. This invention utilizes a mechanism model to accurately quantify the impact of extreme weather, thereby constructing a multi-timescale hierarchical decision model containing a cross-layer coordination hub. By using the quantified results and model state to dynamically and coordinately adjust the parameters at the long, medium, and short-term levels, it effectively improves the adaptability of new energy power plants to operation and maintenance under extreme weather conditions. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the implementation of the multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather provided in this embodiment of the invention. Figure 2 This is a schematic diagram of the structure of a multi-timescale hierarchical operation and maintenance scheduling system for new energy power stations to cope with extreme weather, provided in an embodiment of the present invention. Detailed Implementation

[0017] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart illustrating the implementation of a multi-timescale hierarchical operation and maintenance scheduling method for new energy power plants in response to extreme weather, as provided in this embodiment of the invention. Figure 1 As shown, the method includes: S110, acquire multi-source data and identify the current scene based on the multi-source data; S120: Input multi-source data into the mechanism model corresponding to the current scenario to obtain the quantification results of extreme weather; S130 inputs multi-source data and extreme weather quantification results into a multi-timescale hierarchical decision model to generate decision schemes; The multi-timescale hierarchical decision-making model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer coordination hub. The cross-layer coordination hub is used to coordinate the adjustment of parameters of the long-term planning layer, the medium-term scheduling layer, and the short-term execution layer based on the quantitative results of extreme weather and the state of the mechanism model.

[0019] In this embodiment of the invention, multi-source data covering meteorology, equipment, environment, and operation and maintenance are comprehensively collected through multiple channels such as distributed sensor networks, meteorological satellite receiving terminals, equipment operation and maintenance management platforms, and geographic information systems. Specifically, this includes real-time meteorological parameters (such as snowfall, dust concentration, temperature, and wind speed), equipment operating status data (such as photovoltaic module power, wind turbine blade stress, and electrical equipment insulation performance), geographical environment data (such as station altitude, terrain slope, and windward area distribution), and historical operation and maintenance data (such as records of past extreme weather events and equipment fault repair data). After collection, the system performs preprocessing on the multi-source heterogeneous data, including cleaning, noise reduction, and spatiotemporal alignment, to eliminate data redundancy and errors. Then, based on a pre-set extreme weather scenario feature library, the system analyzes data features using machine learning classification algorithms to accurately identify the types of extreme weather scenarios currently occurring or about to occur, such as blizzards, strong dust storms, low-temperature icing, and freezing rain / frost, thus clarifying the core meteorological disaster types that operation and maintenance scheduling needs to address. Specifically, machine learning classification algorithms can include random forest, support vector machine, gradient boosting decision tree, long short-term memory network, or convolutional neural network algorithms. The choice of algorithm can be flexibly configured based on the characteristics of the extreme weather scenario data and the recognition requirements.

[0020] For identified extreme weather scenarios, the system automatically matches suitable dedicated mechanism models from its built-in model library. Different scenarios correspond to different mechanism models; for example, a snow accumulation impact quantification mechanism model is matched for blizzard scenarios, and a dust-irradiation attenuation linkage accumulation model is matched for severe dust storm scenarios. Subsequently, preprocessed multi-source data is input into the matched mechanism model. The model uses a built-in multi-field coupling algorithm to perform refined modeling and analysis of the coupling mechanism between multiple physical fields such as fluid field, structural field, thermal field, and particulate matter field under extreme weather scenarios, quantifying the comprehensive impact of extreme weather on the operating status, power generation efficiency, and safety risks of new energy power generation stations. Finally, the mechanism model outputs quantitative results of extreme weather, including the intensity level of extreme weather impact, core impact parameters (such as icing thickness, snow load, etc.), key equipment status parameter thresholds, quantified values ​​of power generation efficiency attenuation, safety risk probability, and critical triggering conditions for operation and maintenance intervention.

[0021] The collected multi-source data and the quantitative results of extreme weather are synchronously input into a multi-timescale hierarchical decision-making model. This model has a complete architecture with a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer collaborative hub. Each layer has a clear division of labor and works in concert.

[0022] Among them, the long-term planning layer focuses on preventive operation and maintenance on a monthly / quarterly scale. Based on the long-term impact trends and security risk distribution in the quantitative results, it formulates equipment protection upgrade plans for high-risk areas, long-term reservation plans for operation and maintenance resources, and operation and maintenance priority planning to achieve risk prevention and control.

[0023] The mid-term scheduling layer focuses on hourly / daily-scale operation and maintenance optimization, with the core objective of minimizing "operation and maintenance costs + power generation loss costs." It combines parameters such as power generation efficiency degradation and impact duration from the quantified results to conduct economic and feasibility assessments of feasible operation and maintenance solutions, including manual cleaning, drone operations, and heating de-icing, generating optimal resource scheduling schemes and resource allocation plans. The short-term execution layer focuses on minute-scale real-time response. Based on the critical trigger conditions for operation and maintenance intervention, real-time changes in impact parameters, and the mid-term scheduling scheme from the quantified results, it breaks down the data into directly executable equipment control commands, work path planning commands, and personnel deployment commands to ensure rapid implementation of operation and maintenance actions.

[0024] As the "collaborative core" for decision-making at all levels, the cross-layer collaborative hub plays a crucial role in linkage and regulation. It receives real-time data on extreme weather quantification results and the operational status of mechanistic models (such as coupling parameter adjustment trajectories and quantification errors), and distributes this data and decision constraints synchronously to each level. Simultaneously, it continuously collects decision-making process data and execution feedback data from each level, dynamically adjusting decision parameters based on the mechanistic model status. For example, when the short-term execution layer reports unsatisfactory operational results or when extreme weather quantification results suddenly change, the cross-layer collaborative hub immediately coordinates with the medium-term scheduling layer to adjust resource allocation schemes and the long-term planning layer to update risk prevention and control strategies, achieving coordinated adaptation of decisions at all levels. Ultimately, the system integrates long-term preventative maintenance strategies, medium-term optimal resource scheduling schemes, and short-term dynamic adjustment schemes to form a closed-loop decision-making scheme covering the entire process of prevention, scheduling, and execution, ensuring the accuracy, timeliness, and economy of new energy power plant operation and maintenance scheduling under extreme weather conditions.

[0025] In some embodiments, inputting multi-source data into a mechanism model corresponding to the current scenario to obtain extreme weather quantification results includes: inputting multi-source data into the mechanism model, and outputting extreme weather quantification results according to the multi-field coupling algorithm built into the model; wherein, the mechanism model includes, but is not limited to, a snow shedding cycle model, a dust accumulation rate model, and an ice growth and shedding prediction model; the multi-field coupling algorithm is used to analyze the interaction mechanism of multi-physics parameters in multi-source data, so as to quantify the nonlinear correlation between various physical field parameters under extreme weather.

[0026] In this embodiment of the invention, based on the identified extreme weather scenarios (such as blizzards, severe sandstorms, and low-temperature icing), corresponding specialized mechanism models are matched from the built-in model library. Different scenarios correspond to different core mechanism models. For example, a snow shedding cycle model is adapted for blizzard scenarios, a dust accumulation rate model is matched for severe sandstorm scenarios, and an ice growth and shedding prediction model is matched for low-temperature icing scenarios, ensuring that the model can accurately fit the physical characteristics of the current extreme weather. At the same time, the multi-source data collected in the early stage is standardized and preprocessed to screen out effective data that matches the input parameter dimensions of the selected mechanism model, including meteorological field parameters, equipment physical field parameters, and geographical environmental field parameters, eliminating data redundancy and format differences, and laying a high-quality data foundation for subsequent quantitative analysis.

[0027] The core quantification process is driven by a multi-field coupling algorithm built into the mechanistic model, which is the key to achieving accurate quantification of the impact of extreme weather. The impact of extreme weather on new energy power plants is not the result of a single physical field, but a comprehensive effect of the interaction of multiple physical fields, such as fluid field, structural field, thermal field, and particulate field. The multi-field coupling algorithm is designed specifically for this complex characteristic and is used to analyze the interaction mechanism of multi-physical field parameters in multi-source data. For example, in the icing scenario, the algorithm can analyze the coupling relationship between the meteorological field (temperature, wind speed) and the structural field (equipment surface morphology, stress distribution) and the thermal field (equipment heat dissipation efficiency); in the dust storm scenario, it can interpret the interaction law between the particulate field (dust concentration, particle size) and the light field (irradiance intensity) and the thermal field (component temperature), thereby breaking through the limitations of single physical field analysis.

[0028] In extreme weather environments, the parameters of various physical fields do not change independently, but rather exhibit complex nonlinear coupling effects. For example, a small change in wind speed can lead to significant fluctuations in the rate of icing growth, and an increase in dust concentration can nonlinearly affect the degree of irradiance degradation of photovoltaic modules. Multi-field coupling algorithms, by establishing refined coupling models, quantitatively calculate these nonlinear relationships, transforming the originally abstract multi-field interactions into quantifiable numerical correlations. This avoids the simplistic handling of the complex characteristics of extreme weather by traditional linear analysis, significantly improving the accuracy of quantitative analysis.

[0029] Through in-depth analysis and calculation using multi-field coupling algorithms, the mechanistic model ultimately outputs comprehensive quantitative results for extreme weather. These results are not a single indicator, but rather a multi-dimensional dataset reflecting the comprehensive impact of extreme weather on renewable energy power plants. They encompass key information such as the intensity level of extreme weather impact, the evolution rate of core physical field parameters, the threshold values ​​for changes in critical equipment state parameters, the quantitative value of power generation efficiency degradation, and the probability of safety risks.

[0030] In some embodiments, the mechanistic model includes a coupling correlation module and a solution quantization module; the coupling correlation module is configured with coupling parameters; the coupling parameters of the mechanistic model are different for each scenario; multi-source data is input into the mechanistic model, and extreme weather quantification results are output according to the multi-field coupling algorithm built into the model, including: processing the multi-source data to obtain an input matrix; determining the adjustment value of the coupling parameters according to the input matrix and the multi-field coupling algorithm; determining the time series curve of the dynamic correlation degree between extreme weather factors and equipment operating status parameters according to the multi-source data and the adjustment value of the coupling parameters; and determining the extreme weather quantification results according to the time series curve of the dynamic correlation degree and the solution quantization module.

[0031] In this embodiment of the invention, the core configuration of the coupling and correlation module is the coupling parameters. These parameters are key variables characterizing the interaction strength and mode of action between different physical fields (such as fluid fields, structural fields, thermal fields, and particulate fields), and directly determine the accuracy of multi-physics coupling relationship modeling. Since the physical mechanisms of different extreme weather scenarios (such as blizzards, severe sandstorms, and low-temperature icing) are fundamentally different, the corresponding multi-physics coupling modes are also drastically different. Therefore, the mechanism model corresponding to each scenario needs to be configured with exclusive coupling parameters. For example, the coupling parameters for a blizzard scenario focus on the coupling coefficient between snow load and wind load, and the adhesion parameters between snow and equipment surfaces; while for a severe sandstorm scenario, the focus is on the correlation coefficient between sandstorm concentration and irradiance attenuation, and the adsorption coefficient of sandstorm particles. Through scenario-specific parameter configuration, accurate matching between coupling modeling and the actual scenario is achieved.

[0032] After inputting multi-source data into the mechanistic model, the first step in quantitative analysis is to standardize the multi-source data to generate an input matrix. The system first cleans, denoises, and removes outliers from the collected heterogeneous multi-source data, such as meteorological data, equipment operation data, and geographical environment data, eliminating data redundancy and errors. Then, it extracts core parameters that match the current scenario and model input dimensions through feature filtering, completing the spatiotemporal alignment and dimensional unification of the data. Finally, the processed structured data is organized into an input matrix that meets the model's computational requirements.

[0033] The multi-field coupling algorithm first analyzes the multi-physics parameter characteristics contained in the input matrix, and combines this with the core physical mechanisms of the current extreme weather scenario to determine the degree of fit between the initial coupling parameters and the actual scenario. Then, through algorithms such as gradient descent and Bayesian optimization, it calculates parameter adjustment values ​​that can optimize the accuracy of coupling modeling. This process allows the coupling parameters to be dynamically corrected based on real-time multi-source data, avoiding the shortcomings of fixed parameters that are difficult to adapt to dynamic changes in the scenario.

[0034] After obtaining the adjusted coupling parameters, the system constructs a time-series curve of the dynamic correlation between extreme weather factors and equipment operating status parameters by combining multi-source data. Then, the final extreme weather quantification result is output through the solution quantification module. First, based on the adjusted coupling parameters, a multi-field coupling algorithm quantifies the dynamic correlation between extreme weather factors (such as snowfall and dust concentration) and equipment operating status parameters (such as component power and structural stress), intuitively presenting the correlation strength between the two over time in the form of a time-series curve. Subsequently, the solution quantification module calls numerical solving, simulation, and other algorithms to conduct in-depth analysis of the dynamic correlation time-series curve, transforming it into specific quantification indicators. Finally, it outputs the extreme weather quantification result, which includes the intensity level of extreme weather impact, the threshold for equipment status changes, the quantification value of power generation efficiency degradation, and the probability of safety risks.

[0035] In some embodiments, multi-source data and extreme weather quantification results are input into a multi-timescale hierarchical decision model to generate decision schemes, including: extracting the mechanistic model state after coupling parameter adjustment; fusing multi-source data and extreme weather quantification results to obtain short-term decision feature sets, medium-term decision feature sets, and long-term decision feature sets; inputting extreme weather quantification results and mechanistic model state into a cross-layer collaborative hub to determine short-term, medium-term, and long-term decision constraints; generating preventive maintenance strategies and resource allocation plans based on the long-term planning feature set, long-term decision constraints, and the long-term planning layer; generating optimal resource scheduling schemes based on the medium-term scheduling feature set, medium-term decision constraints, and the medium-term scheduling layer; determining short-term dynamic adjustment schemes based on the short-term execution feature set, short-term decision constraints, optimal resource scheduling schemes, acquired real-time feedback data, and the short-term execution layer; and determining decision schemes based on preventive maintenance strategies, resource allocation plans, optimal resource scheduling schemes, and short-term dynamic adjustment schemes.

[0036] In this embodiment of the invention, multi-source data and extreme weather quantification results are input into a multi-timescale hierarchical decision model to generate a decision scheme. The first step is to deeply extract the state of the mechanism model after the coupling parameters are adjusted. First, it is necessary to accurately collect the full-cycle data of the coupling parameter adjustment in the mechanism model, including the initial baseline value before adjustment, the intermediate transition values ​​of each iteration during the adjustment process, and the final convergence value that tends to stabilize. At the same time, each set of parameter data is marked with a precise timestamp to form a complete dynamic change trajectory of the parameters. Simultaneously, the trigger source data that drives the parameter adjustment is collected to clarify the key feature thresholds of the multi-source data that trigger the parameter adjustment (such as the snowfall mutation threshold in the blizzard scenario and the temperature critical threshold in the icing scenario) and the deviation threshold between the extreme weather quantification results and the historical measured baseline values, so as to clarify the causal logic of the parameter adjustment. Based on this, further acquisition of operational status data of the coupling and correlation module based on adjusted parameters is required. The focus is on collecting modeling fit of multi-physics coupling effects (evaluating the accuracy of parameters in characterizing inter-field interactions), fitting degree of dynamic correlation time series curves (such as R² goodness of fit, measuring the consistency between the curve and actual data), and nonlinear correlation calculation errors (such as mean absolute error and root mean square error, quantifying the accuracy of the model in handling complex nonlinear relationships). At the same time, resource consumption parameters of module operation (such as computation time and memory usage) are recorded. In addition, the core operational indicators of the solution quantization module need to be extracted, including computation convergence speed (number of iterations, convergence time) and output stability (fluctuation amplitude of core quantization indicators).

[0037] Next, the system will perform deep fusion processing on the input multi-source data and extreme weather quantification results. Through multi-dimensional data integration and targeted feature filtering, it will generate three types of decision feature sets: short-term, medium-term, and long-term. First, a cross-dimensional data fusion algorithm will be used to organically integrate real-time meteorological dynamic data (such as real-time changes in wind speed, temperature, and dust concentration), basic equipment asset data (such as equipment model, service life, and structural parameters), geographical environment data (such as station altitude, terrain slope, and regional distribution), and historical operation and maintenance time-series data (such as past extreme weather handling records and equipment failure repair costs) from the multi-source data with core information from the extreme weather quantification results, such as the impact intensity level, power generation efficiency decay quantification curve, safety risk probability distribution, and critical threshold for operation and maintenance intervention. This will eliminate data redundancy, format differences, and spatiotemporal misalignment issues, forming a unified fusion dataset. Subsequently, feature filtering and extraction will be performed based on the core objectives and needs of decisions at different time scales. The long-term decision-making feature set focuses on the core dimension of "risk trends and asset suitability," extracting long-term information such as the annual frequency of extreme weather events, the spatial distribution of high-risk areas, the matching between equipment aging and extreme weather tolerance, and the historical trend of risk disposal costs. The medium-term decision-making feature set emphasizes the key element of "cost-effectiveness and resource constraints," selecting core medium-term data such as the duration of extreme weather impacts, the estimated total power generation loss, the real-time inventory of operation and maintenance resources (personnel, equipment, and materials), and the fluctuation curve of resource leasing costs. The short-term decision-making feature set focuses on the core requirement of "real-time response and precise control," extracting short-term dynamic indicators such as the real-time change rate of core extreme weather parameters (e.g., the rate of increase in icing thickness and the magnitude of snow load change), real-time monitoring data of equipment operating status, and real-time feedback values ​​of operation and maintenance progress.

[0038] Subsequently, the cross-layer collaborative hub conducts in-depth scenario-based analysis of the extreme weather quantification results. Combining the identified extreme weather types (such as blizzards, severe sandstorms, and low-temperature icing), it extracts scenario-specific core quantification indicators (such as snow thickness and snow load in blizzard scenarios, and dust accumulation density and irradiance attenuation coefficient in sandstorm scenarios). This clarifies the critical conditions for equipment structural failure and performance degradation under different scenarios, identifying key risk targets that need to be addressed at each time scale. Simultaneously, based on the extracted mechanistic model state data, a model quantification reliability assessment is conducted. Through indicators such as the rationality of coupling parameter adjustments, model fit, and quantification error level, the current quantification reliability level (excellent, good, average, poor) for the extreme weather scenario is determined. If the assessment result is "average" or below, a constraint redundancy mechanism is immediately triggered to reserve a safety buffer space for subsequent decision constraints. Based on this, and combining the scenario-based analysis results with the model reliability level, differentiated decision constraints at each level are formulated: Long-term decision constraints focus on the core objective of "risk prevention and control in advance," specifying constraints such as the coverage rate of equipment protection upgrades in high-risk areas being ≥ a preset threshold (e.g., 90%), the long-term reserve ratio of operation and maintenance resources not being lower than the corresponding impact level benchmark value, and the matching degree between the implementation window of preventive maintenance plans and periods with low incidence of extreme weather being ≥ 80%. When the model's quantification accuracy is "poor," an additional 30%-50% safety redundancy coefficient is added; medium-term decision constraints revolve around the objective of "balancing cost-effectiveness and response timeliness," setting a comprehensive optimal threshold for "operation and maintenance costs + power generation loss costs," The resource allocation response delay upper limit (≤1 hour for fast evolution rate scenarios, ≤3 hours for medium evolution rate scenarios) and the minimum requirement for safety risk reduction rate (≥80% for high impact level) are set. When the model quantification accuracy is insufficient, the cost optimization weight is reduced and the safety constraint weight is increased. Short-term decision constraints focus on "response timeliness and execution safety", specifying that the instruction execution response delay is ≤5 minutes, the equipment operation threshold does not exceed the safety operation boundary in extreme weather (e.g., heating device power ≤70% of equipment load), and the fault tolerance rate for work progress deviation is ≤10%. When extreme weather evolves rapidly or the model reliability is low, the instruction adjustment cycle is shortened to 50% of the original cycle, and the frequency of safety operation threshold verification is increased. At the same time, the cross-layer collaborative hub conducts cross-calibration of the three types of constraints to eliminate conflicts between constraints at different levels (e.g., when there is a conflict between long-term resource reservation and medium-term resource demand, dynamic adjustment is made based on the risk level priority), ensuring the coordination and adaptation of the constraint system in risk prevention and control, cost optimization, and response timeliness.

[0039] Based on long-term planning feature sets and long-term decision constraints, the long-term planning layer focuses on preventative maintenance decisions at the monthly / quarterly scale. Through risk classification matching and resource coordination planning, it generates targeted preventative maintenance strategies and resource allocation plans to reduce the long-term risks brought by extreme weather from the source. First, based on the long-term impact trends of extreme weather and the spatial distribution data of high-risk areas from the long-term planning feature set, combined with the matching analysis of equipment aging and extreme weather tolerance, the long-term planning layer uses a risk classification algorithm to divide the site into high, medium, and low-risk areas. Subsequently, according to the requirements of protection coverage and resource reservation ratio in the long-term decision constraints, differentiated preventative maintenance strategies are formulated for different risk areas: For high-risk areas, specific technical solutions for equipment anti-extreme weather modifications are defined (such as adding anti-icing coatings to photovoltaic modules, strengthening the structural support of wind turbine blades, and upgrading the insulation protection of electrical equipment), and the specific window for implementation (such as completing it 1-2 months before the autumn snowstorm); for medium-risk areas, routine maintenance and upgrade plans are planned (such as regularly cleaning modules and replacing aging seals); for low-risk areas, basic inspection plans are formulated (such as monthly inspections and quarterly performance tests). At the same time, based on historical operation and maintenance cost trends and the total annual operation and maintenance budget, a resource allocation plan is formulated in a coordinated manner. The plan clarifies the long-term reserve ratio of various operation and maintenance resources (professional operation and maintenance personnel, dedicated de-icing / cleaning equipment, cold-proof / dust-proof materials) (e.g., the reserve ratio of de-icing equipment in high-risk seasons is ≥40%) and the budget allocation plan (the budget allocation ratio for renovation in high-risk areas is ≥60%). The implementation progress of maintenance strategies and resource allocation plans is precisely matched with periods of low incidence of extreme weather to ensure that long-term risks are prevented and controlled in advance without affecting the normal operation of the site.

[0040] The mid-term scheduling layer, based on the mid-term scheduling feature set and mid-term decision constraints, generates optimal resource scheduling schemes at the hourly / daily scale through multi-objective optimization modeling and scheme evaluation and screening, achieving efficient allocation of operation and maintenance resources and reasonable control of operation and maintenance costs. First, based on data such as the duration range of extreme weather impacts, power generation efficiency decay curves, and estimated total power generation losses from the mid-term scheduling feature set, the mid-term scheduling layer clarifies the core objectives of operation and maintenance operations. Under the premise of ensuring safety, it minimizes the "operation and maintenance cost + power generation loss cost." Subsequently, combining requirements such as resource call response delay and safety risk reduction rate in the mid-term decision constraints, a multi-objective optimization decision model is constructed. Multiple feasible operation and maintenance technologies, such as manual cleaning, drone operations, heating de-icing, and remote control, are used as candidate schemes and input into the model to conduct full-cycle economic simulation and efficiency evaluation. During the evaluation process, the focus is on analyzing key indicators such as total resource consumption (personnel salaries, equipment energy consumption, material losses, and rental costs), operation completion time, power generation recovery efficiency, and reduction in safety risks for different candidate schemes. Using a weighted assignment method (with safety indicator weights dynamically adjusted based on the impact level of extreme weather) and a scheme ranking algorithm, the optimal resource scheduling scheme is selected from the candidate schemes. The final output of the optimal resource scheduling scheme clearly includes: the resource combination method (e.g., "2 drones + 3 professionals" collaboratively conducting de-icing operations), the specific timing of scheme activation (e.g., immediate activation when snow depth reaches the critical threshold for operation and maintenance intervention), the specific scope of the operation (e.g., all components / wind turbines in high-risk areas), the expected economic benefits (e.g., estimated reduction in power generation losses and percentage savings in operation and maintenance costs), and the triggering conditions for alternative schemes (e.g., alternative schemes when the main scheme's resources cannot arrive in time), providing clear resource scheduling guidance for precise operations at the short-term execution level.

[0041] The short-term execution layer focuses on real-time response and dynamic adjustment on a minute-by-minute scale. Combining short-term execution feature sets, short-term decision constraints, optimal medium-term resource scheduling schemes, and real-time feedback data during operation and maintenance, it determines short-term dynamic adjustment schemes through task decomposition and dynamic optimization, ensuring the accuracy and flexibility of operation and maintenance actions. The short-term execution layer first decomposes the optimal resource scheduling scheme output by the medium-term scheduling layer into a series of directly executable basic execution instructions, including basic equipment control instructions (such as initial start-up and shutdown power of photovoltaic panel heating devices, operating parameters of wind turbine de-icing systems), basic work path planning (such as the initial path of drone swarm cleaning, and the division of personnel work areas), and basic personnel allocation instructions (such as the number of personnel in each area and task assignments). Subsequently, combining the real-time change rate of core extreme weather parameters in the short-term execution feature set (such as the rate of increase in ice thickness and the magnitude of snow load changes), real-time monitoring data of equipment operating status, and the requirements of short-term decision constraints regarding instruction response timeliness and equipment safe operation boundaries, the basic execution instructions are initially optimized. During the operation and maintenance process, the short-term execution layer collects feedback data in real time, including real-time operation progress data (such as the number of components that have been cleaned / de-iced, and the deviation value of operation progress), equipment status recovery data (such as the recovery value of component power generation, and the vibration parameters of wind turbine operation), and real-time change data of extreme weather parameters (such as a sudden increase in wind speed and a sharp drop in temperature). If the feedback data shows that the operation progress deviation exceeds the fault tolerance rate, the equipment status recovery does not meet expectations, or there are sudden changes in extreme weather (such as an accelerated evolution rate), the short-term execution layer immediately activates the dynamic adjustment mechanism to optimize the operation path (such as adjusting the order of drone operations and prioritizing areas with increased risk levels), adjust resource input (such as increasing the number of operators and increasing the operating power of heating devices), and correct equipment control parameters (such as adjusting the wind turbine de-icing rhythm according to real-time wind speed). This ensures that the operation and maintenance actions can accurately match the dynamic evolution of extreme weather, guaranteeing operational safety and effectiveness.

[0042] Finally, through the overall coordination of the cross-layer collaborative hub, the system comprehensively integrates long-term preventive maintenance strategies, resource allocation plans, medium-term optimal resource scheduling schemes, and short-term dynamic adjustment schemes, forming a closed-loop decision-making scheme covering the entire process of "long-term prevention - medium-term scheduling - short-term execution," providing comprehensive and systematic support for the operation and maintenance scheduling of new energy power plants under extreme weather conditions. During the integration process, the cross-layer collaborative hub first conducts a correlation analysis of the core content of each scheme to ensure the synergy and adaptability of each scheme in terms of time dimension, resource allocation, and risk control: the long-term preventive maintenance strategy clarifies the overall direction of risk control and long-term resource reserve requirements for medium-term scheduling and short-term execution; the medium-term optimal resource scheduling scheme provides specific resource combination methods and operational objectives for short-term execution; and the short-term dynamic adjustment scheme ensures the accurate implementation of medium-term scheduling objectives by flexibly adapting to real-time scenario changes. Addressing potential conflicts between schemes (such as mismatches between long-term resource reservation ratios and medium-term resource needs, or conflicts between short-term emergency response and medium-term cost constraints), the cross-layer collaborative hub dynamically adjusts the scheme content based on the principle of "safety first, risk-oriented." In the event of resource conflicts, the long-term resource reservation ratio is adjusted based on the priority of the impact level of extreme weather. In the event of cost and safety conflicts, short-term safety constraints are prioritized, and feedback is simultaneously sent to the medium-term level to adjust the resource combination plan. At the same time, the system eliminates redundant content between the various plans (such as avoiding duplicate resource allocation and overlapping tasks) and clarifies the implementing entity, responsibility boundaries, connection nodes, and feedback mechanisms for each plan.

[0043] In some embodiments, extracting the mechanistic model state after coupling parameter adjustment includes: collecting the dynamic adjustment trajectory of the coupling parameters and the corresponding adjustment trigger source data; wherein, the dynamic adjustment trajectory includes the numerical change information and timestamp information of the coupling parameters before and after adjustment, and the adjustment trigger source data includes the key feature threshold and deviation threshold that trigger the adjustment of the coupling parameters; obtaining the operating status data of the coupling correlation module based on the adjusted parameters; wherein, the operating status data includes at least one of the following: the modeling fit of the multiphysics coupling effect, the fitting degree of the dynamic correlation time series curve, the nonlinear correlation calculation error and resource consumption parameters; and determining the mechanistic model state by combining the dynamic adjustment trajectory, the adjustment trigger source data, and the operating status data.

[0044] In this embodiment of the invention, the core change information of the coupling parameters is recorded with millisecond-level time precision. Specifically, this includes the initial baseline value before adjustment, the intermediate transition values ​​of each iteration during the adjustment process, and the final converged stable value. Simultaneously, each set of parameter data is assigned a unique timestamp, forming a dynamic trajectory curve containing two-dimensional information of "time-parameter value". For example, in a blizzard scenario, each fluctuation of the snow accumulation-wind load coupling parameter is recorded. Through the trajectory curve, the response pattern of the parameters to changes in multi-source data such as snowfall and wind speed can be clearly traced, clarifying the step size, frequency, and convergence time of parameter adjustment.

[0045] While collecting data on the dynamic adjustment trajectory, it is essential to simultaneously collect corresponding adjustment trigger source data to accurately pinpoint the core causes driving the coupled parameter adjustments and avoid "causeless" or "erroneous" parameter adjustments. Trigger source data is mainly divided into two categories: one is key feature thresholds in multi-source data, such as the sudden change threshold of snowfall ≥5mm / h and the excessive threshold of dust concentration ≥150μg / m³ in meteorological data, and the warning value of structural stress ≥ the safety threshold in equipment data. These thresholds are the direct external conditions triggering parameter adjustments. The other category is the deviation threshold of extreme weather quantification results, i.e., the deviation between the model's current output quantification index (such as the predicted value of icing thickness) and historical measured data and real-time monitoring data. When the deviation exceeds a preset threshold (such as 10%), parameter adjustments are triggered to correct the deviation. By correlating the trigger source data with the dynamic adjustment trajectory, the causal link of "data change - parameter response" can be clarified, verifying whether the parameter adjustments match the actual evolution trend of extreme weather.

[0046] After adjusting the coupling parameters, the system needs to collect operational status data of the coupled modules to evaluate the actual effect of the parameter adjustments from three dimensions: modeling accuracy, operational efficiency, and stability. Specifically, the modeling fit of multiphysics coupling is determined by calculating the fit between the coupled model constructed with the adjusted parameters and the actual physical scenario. For example, in an icing scenario, the model's calculated ice adhesion force is compared with the measured data from the equipment to determine whether the parameters accurately characterize the coupling mechanism of the thermal field and the structural field. The fit of the dynamic correlation time-series curve is quantified using indicators such as R² goodness of fit and mean absolute error (MAE) to evaluate the accuracy of the curve's representation of the correlation between extreme weather factors and equipment operating parameters. The nonlinear correlation calculation error focuses on the model's ability to handle strong coupling and nonlinear relationships, measured by the deviation rate between theoretical and measured values. Resource consumption parameters include the module's computation time, peak memory usage, and CPU utilization to evaluate whether the module's operational efficiency after parameter adjustments meets real-time quantization requirements.

[0047] After collecting the three types of core data, the system initiates a multi-source data integration and feature enhancement process, performing cross-dimensional correlation and standardization on the dynamic trajectory adjustment data, adjustment trigger source data, and coupled module operation status data. First, the data format is standardized, converting data of different dimensions and units into standardized indicators. Then, through feature extraction algorithms, core features such as parameter adjustment step size, convergence speed, and fluctuation amplitude are extracted from the dynamic trajectory; features such as trigger frequency and trigger intensity of key thresholds are extracted from the trigger source data; and features such as modeling fit, goodness of fit, and error rate are extracted from the operation status data. Finally, a three-dimensional state feature matrix containing "parameter change features - trigger cause features - operation performance features" is constructed through data fusion algorithms, eliminating data redundancy and noise to form a high-quality mechanistic model state dataset.

[0048] Finally, based on the integrated three-dimensional state feature matrix, the system uses a multi-criteria decision evaluation algorithm to accurately define the state of the mechanistic model after the coupling parameters are adjusted. Specifically, it combines the convergence of the dynamic trajectory to judge the effectiveness of the parameter adjustment, the rationality of the trigger source data to judge the necessity of the parameter adjustment, and the accuracy and efficiency of the operating state data to judge the reliability of the model. The final output is a comprehensive state evaluation result that includes "parameter adjustment effectiveness level, modeling fit level, operating stability level, and quantization accuracy level".

[0049] In some embodiments, the extreme weather quantification results and the mechanistic model status are input into the cross-layer collaborative hub to determine short-term, medium-term, and long-term decision constraints, including: performing scenario-based analysis on the extreme weather quantification results to extract the core quantitative indicators of the scenario and the critical conditions for equipment failure; determining the current quantification reliability level of the mechanistic model for the extreme weather scenario based on the dynamically adjusted trajectory and operating status data; and determining the decision constraints at each time scale level based on the core quantitative indicators of the scenario and the quantification reliability level.

[0050] In this embodiment of the invention, the cross-layer collaborative hub first matches the identified extreme weather types (such as blizzards, severe sandstorms, and low-temperature icing), extracts key data directly related to the physical mechanisms of the scene from the quantification results, and then extracts the core quantification indicators of the scene in a targeted manner. For example, the blizzard scene focuses on the real-time value and growth rate of snow thickness and the snow load distribution coefficient; the severe sandstorm scene focuses on the temporal change rate of sandstorm concentration, the cumulative density of dust attached to photovoltaic modules and the irradiance attenuation coefficient; and the low-temperature icing scene focuses on the icing thickness and the ice adhesion coefficient. At the same time, combined with the technical parameters of new energy power station equipment (such as the load-bearing limit of modules and the vibration threshold of wind turbines) and historical failure data, the critical conditions for equipment failure are defined from the quantification results, and the specific quantification thresholds for structural failure (such as snow load exceeding the load-bearing limit), performance failure (such as irradiance attenuation exceeding the efficiency interruption threshold), and functional failure (such as the energy consumption of de-icing equipment exceeding the limit) are clarified.

[0051] Next, the cross-layer collaborative hub, based on the dynamic adjustment trajectory and operational status data of the mechanistic model, systematically evaluates the current quantitative reliability level of the model for this extreme weather scenario. The process begins by analyzing the dynamic adjustment trajectory to determine whether the step size, frequency, and convergence of the coupling parameter adjustments align with the changing patterns of the scenario. For example, does the parameter adjustment accurately respond to triggering conditions such as sudden changes in snowfall or excessive dust concentration? Then, combining core indicators from the operational status data, such as the multi-physics coupling modeling fit, the time-series curve fitting degree of dynamic correlation, and the nonlinear correlation calculation error, a multi-dimensional evaluation system is constructed. By setting weights and grading thresholds for each indicator, a weighted summation method is used to calculate the comprehensive score, classifying the quantitative reliability into four levels: "Excellent," "Good," "Average," and "Poor." For example, a fit ≥90% and a calculation error ≤5% is considered "Excellent," while a fit <65% and an error >15% is considered "Poor," thus clarifying the credibility of the model's quantitative results.

[0052] Based on the core quantitative indicators and critical conditions for equipment failure derived from scenario-based analysis, as well as the quantitative reliability level determined through assessment, the cross-layer collaborative hub begins to coordinate and formulate decision constraints at various time scales, achieving precise adaptation of constraints to scenarios and model states. For long-term decision constraints, constraints related to proactive risk prevention are set in conjunction with the long-term risk trends reflected by the core quantitative indicators of the scenario (such as annual high-risk periods and regional distribution), such as the coverage rate of equipment protection upgrades in high-risk areas and the proportion of long-term reserved operation and maintenance resources. If the reliability level is "average" or below, the safety redundancy coefficient is increased. Medium-term decision constraints revolve around the impact duration and power generation loss prediction in the core quantitative indicators of the scenario, setting constraints that balance cost-effectiveness and response time, such as the upper limit of resource call delay and the minimum requirement for safety risk reduction rate. When the model reliability is insufficient, the cost optimization weight is reduced.

[0053] The formulation of short-term decision constraints focuses more on real-time response and safety assurance. It combines the real-time change rate of core quantitative indicators of the scenario (such as the rate of increase in icing thickness) and the critical threshold for equipment failure to set constraints such as command response delay, equipment operation safety boundaries, and operational progress tolerance. If the extreme weather evolves rapidly or the model reliability level is low, the command adjustment cycle will be further shortened and the frequency of safety threshold verification will be increased. Simultaneously, the cross-layer collaborative hub will conduct cross-calibration of the three types of constraints to eliminate constraint conflicts between levels. For example, if long-term resource reservations do not match medium-term needs, dynamic adjustments will be made based on the priority of the scenario's impact level, ensuring that constraints at each time scale form a collaborative closed loop in risk prevention, cost optimization, and response timeliness, providing clear and feasible criterion boundaries for subsequent layered decision-making.

[0054] In some embodiments, the extreme weather quantification results are analyzed in a scenario-based manner to extract the core quantification indicators and critical conditions for equipment failure in the scenario. This includes: screening basic quantification data directly related to the physical characteristics of the scenario from the extreme weather quantification results; calculating the core quantification indicators of the scenario based on the basic quantification data; and determining the critical conditions for equipment failure in each scenario based on the technical parameters of the new energy power station equipment, historical failure data, and dynamic correlation time series curves.

[0055] In this embodiment of the invention, the extreme weather quantification results are analyzed in a scenario-based manner. The first step is to accurately screen basic quantitative data directly related to the physical characteristics of the scenario, laying a solid data foundation for subsequent indicator extraction and critical condition definition. The cross-layer collaborative hub first identifies the currently identified extreme weather scenario types, such as blizzards, severe sandstorms, and low-temperature icing. Based on the core physical mechanisms of different scenarios, irrelevant data is extracted from the massive extreme weather quantification results, and key basic data is selected. Taking the blizzard scenario as an example, the focus is on screening data directly related to snow formation and accumulation, such as snow thickness, snowfall intensity, ambient temperature, and wind speed; for the severe sandstorm scenario, the focus is on core basic data such as sandstorm particle concentration, particle size distribution, wind direction and speed, and radiation intensity, ensuring that the selected data accurately matches the essential characteristics of the scenario.

[0056] After selecting the basic quantitative data, core quantitative indicators for each scenario need to be generated through specialized calculations. This transforms the scattered basic data into core parameters that intuitively reflect the degree of impact and evolution trend of the scenario. Different calculation logics are adopted for different scenarios: in blizzard scenarios, the snow load distribution coefficient is calculated based on snow thickness and wind speed data, and the snow growth rate is derived by combining snowfall intensity and time data; in severe dust storm scenarios, the cumulative dust density is calculated using dust concentration and time data, and the irradiance attenuation coefficient is calculated using irradiance intensity change data; in low-temperature icing scenarios, the ice adhesion coefficient is calculated based on icing thickness and temperature data, and the icing unevenness is analyzed by combining wind speed data. These core quantitative indicators accurately characterize the core impact dimensions of extreme weather on new energy power plants.

[0057] Determining the critical failure conditions for equipment is one of the core objectives of scenario-based analysis. This process requires close integration of the technical parameters of new energy power plant equipment, historical failure data, and dynamic correlation time-series curves to ensure the scientific validity and practicality of the critical conditions. First, the technical specifications of various equipment at the power plant are reviewed to clarify the basic parameters such as the structural load-bearing limits, performance thresholds, and safe operating boundaries of photovoltaic modules, wind turbines, and electrical equipment. Next, historical equipment failure records under similar extreme weather scenarios are retrieved, and the extreme weather parameters and equipment status parameters at the time of failure are analyzed. Simultaneously, by combining the dynamic correlation time-series curves, nodes where the correlation strength between extreme weather factors and equipment operating status parameters abruptly changes are identified, and these are used as the basis for defining the critical failure conditions for equipment.

[0058] The final equipment failure critical conditions cover multiple dimensions and are precisely adapted to specific scenarios: structural failure critical conditions include snow load on photovoltaic modules ≥ preset bearing limit and uneven icing of wind turbine blades ≥ vibration trigger threshold; performance failure critical conditions include irradiance attenuation coefficient of photovoltaic modules ≥ efficiency interruption threshold and flashover voltage reduction of electrical equipment ≥ safe operation threshold; functional failure critical conditions include energy consumption of de-icing equipment ≥ upper limit threshold and visibility of drone operations ≤ minimum operation threshold.

[0059] In some embodiments, the quantitative reliability level of the mechanism model for extreme weather scenarios is determined based on dynamically adjusted trajectory and operational status data, including: constructing a multi-dimensional quantitative accuracy evaluation index system based on dynamically adjusted trajectory and operational status data; setting the weight and grading threshold of each evaluation index in the multi-dimensional quantitative accuracy evaluation index system; and calculating the comprehensive score of quantitative reliability using a weighted summation method to determine the quantitative reliability level.

[0060] In this embodiment of the invention, dynamic trajectory adjustment and operational status data are used as the core data sources, and an indicator framework is built around three core dimensions: parameter adaptability, modeling accuracy, and operational stability. Specifically, based on the dynamic trajectory adjustment, three types of indicators are extracted: "coupled parameter adaptability," "parameter adjustment convergence," and "trigger response rationality." Coupled parameter adaptability measures the degree of matching between the adjusted parameters and the physical mechanisms of the current extreme weather scenario; parameter adjustment convergence assesses the convergence speed and fluctuation amplitude of the parameters from their initial values ​​to stable values; and trigger response rationality judges whether the parameter adjustment accurately matches the triggering conditions of the scenario change. Based on the operational status data, four types of indicators are extracted: "multi-physics coupling modeling adaptability," "dynamic correlation time-series curve fitting degree," "nonlinear correlation calculation error," and "output stability." These comprehensively cover the accuracy and stability of the model throughout the entire process from coupled modeling to quantized output, forming a multi-dimensional, full-chain evaluation indicator matrix.

[0061] After constructing the multi-dimensional quantitative accuracy evaluation index system, it is necessary to scientifically set the weights and grading thresholds of each evaluation index, taking into account the characteristics of extreme weather scenarios and the needs of model operation. The weight setting adopts the analytic hierarchy process (AHP), allocating weights according to the degree of influence of each index on the model's quantitative reliability. For example, in strongly coupled scenarios such as blizzards and icing, the weight of "multiphysics coupling modeling fit" is set at 25%, and the weight of "dynamic correlation time-series curve fitting degree" is set at 20%, both higher than other indicators. Meanwhile, indicators ensuring real-time performance, such as "parameter adjustment convergence" and "output stability," have their weights appropriately reduced in slower-evolving scenarios such as sandstorms. The grading thresholds are determined by combining historical model operation data from similar scenarios, actual equipment measurement data, and industry standards. For example, the threshold for "dynamic correlation time-series curve fitting degree" (R²) is divided into four levels: ≥90%, 80%-90%, 65%-80%, and <65%, and the threshold for "nonlinear correlation calculation error" is divided into four levels: ≤5%, 5%-10%, 10%-15%, and >15%, providing clear standards for subsequent score calculation and grade determination.

[0062] After setting the weights and grading thresholds, a weighted summation method is used to calculate the comprehensive quantitative reliability score, thereby achieving a quantitative assessment of the model's quantitative reliability. During the calculation, firstly, based on the actual monitoring values ​​of each evaluation indicator, the score for each indicator is determined against the preset grading thresholds (e.g., a goodness-of-fit R² = 92% corresponds to a full score of 100 points for a single indicator, and an error rate of 12% corresponds to 60 points). Then, each single indicator score is multiplied by its corresponding weight to obtain the weighted score for each indicator. Finally, the weighted scores of all indicators are summed to obtain the comprehensive quantitative reliability score. For example, in an extreme weather scenario, the coupling parameter fit score is 90 points (weight 15%), the modeling fit score is 85 points (weight 25%), and the goodness-of-fit score is 92 points (weight 20%). The weighted summation yields a comprehensive score of 88.75 points, directly reflecting the model's current level of quantitative reliability.

[0063] Based on the comprehensive score results, the quantification reliability of the current mechanistic model for extreme weather scenarios is divided into four levels: "Excellent," "Good," "Average," and "Poor." The specific grading criteria are as follows: A comprehensive score ≥ 85 is "Excellent," indicating high model quantification accuracy and good scenario adaptability; the quantification results can be directly used as the core basis for decision-making. A comprehensive score ≤ 70 < 85 is "Good," indicating that the model's quantification accuracy meets basic decision-making requirements and only requires a small amount of safety redundancy. A comprehensive score ≤ 75 < 70 is "Average," indicating that the model has some quantification deviation and the safety redundancy coefficient of decision constraints needs to be increased. A comprehensive score < 55 is "Poor," indicating insufficient model quantification reliability; a strengthened verification mechanism needs to be triggered, and the quantification results should be corrected based on real-time monitoring data before being used for decision-making.

[0064] In some embodiments, the method further includes: decomposing the preventive maintenance strategy, the optimal resource scheduling scheme, and the short-term dynamic adjustment scheme in the decision scheme into an instruction set; executing the instruction set and evaluating the execution process of the instruction set to obtain an evaluation result; and iteratively optimizing the multi-timescale hierarchical decision model based on the evaluation result.

[0065] In this embodiment of the invention, long-term preventive maintenance strategies are broken down to generate long-term basic instructions, including equipment modification lists, material procurement plans, and annual inspection calendars. Subsequently, for the optimal resource scheduling scheme in the medium term, these are refined into specific resource allocation instructions, specifying personnel shift schedules, equipment scheduling routes, material delivery points, and operation start-up time windows. Finally, for the short-term dynamic adjustment scheme, real-time control instructions are generated, covering flight parameters for UAV operations, power adjustment thresholds for heating devices, and yaw and pitch instructions for wind turbines. These instruction sets are arranged according to priority and execution logic to form a standardized and process-oriented operation and maintenance manual, ensuring that operation and maintenance personnel and automated equipment can execute them accurately.

[0066] During instruction set execution, the system constructs a comprehensive, multi-dimensional performance evaluation mechanism to track and quantify the actual effectiveness of maintenance operations in real time. First, there is the operation progress evaluation, which uses GPS positioning and equipment sensor feedback to monitor indicators such as drone cleaning area, personnel inspection mileage, and de-icing equipment runtime in real time, calculating the deviation rate between actual and planned progress. Second, there is the performance recovery evaluation, which compares key indicators such as power generation data, equipment vibration levels, and component cleanliness before and after the operation to quantify the actual contribution of maintenance measures to restoring equipment performance. Finally, there is the cost and safety evaluation, which calculates actual labor costs, equipment energy consumption, and rental fees, and monitors safety hazards and violations during the operation process to ensure that maintenance activities are carried out efficiently within safety boundaries.

[0067] Based on the collected execution data, the system utilizes big data analytics and machine learning algorithms to generate a comprehensive evaluation report covering four dimensions: progress, efficiency, cost, and safety. This report not only compares the actual and target values ​​for each indicator but also delves into the underlying causes of deviations. For example, if power generation recovery after de-icing operations falls short of expectations, the system analyzes whether it's due to prediction errors in ice thickness, insufficient de-icing equipment power, or delayed operation timing. If operating costs exceed the budget, the system identifies whether it's due to unreasonable resource allocation or fluctuations in external leasing prices. This evaluation report serves as a "litmus test" for the effectiveness of the decision-making model, providing precise targets for subsequent iterative optimizations.

[0068] Based on the comprehensive evaluation results, the system initiates full-link iterative optimization of the multi-timescale hierarchical decision-making model to achieve self-evolution and continuous improvement of the operation and maintenance scheduling system. In the long term, based on the identified shortcomings in high-risk area protection, the system adjusts the preventive maintenance strategy of the long-term planning layer, optimizes the priority of equipment upgrades and the proportion of long-term resource reservations. In the medium term, based on the evaluation data of operation costs and efficiency, the system corrects the cost optimization algorithm and resource allocation model of the medium-term scheduling layer, and adjusts the weight coefficients of different operation and maintenance schemes. In the short term, based on the response delay and deviation data of command execution, the system optimizes the dynamic adjustment logic and threshold settings of the short-term execution layer, improving the system's sensitivity to extreme weather changes.

[0069] The cross-layer collaborative hub plays a crucial "coordinator" and "decision-maker" role in the iterative optimization process. It is responsible for mapping the evaluation results from the lower-level execution feedback back to the mechanistic model and decision constraint system. For the quantification errors identified in the evaluation model, the system automatically fine-tunes the coupling parameters in the mechanistic model to improve the accuracy of extreme weather predictions. For unreasonable decision constraints, such as excessively high safety redundancy leading to cost waste or excessively low redundancy resulting in risk exposure, the system dynamically adjusts the constraint thresholds set by the cross-layer collaborative hub.

[0070] Figure 2 This is a schematic diagram of the multi-timescale hierarchical operation and maintenance scheduling system for new energy power plants in response to extreme weather, provided in an embodiment of the present invention. Figure 2 As shown, the multi-timescale hierarchical operation and maintenance scheduling system for new energy power stations to cope with extreme weather includes: The acquisition module 210 is used to acquire multi-source data and identify the current scene based on the multi-source data; The quantization module 220 is used to input multi-source data into the mechanism model corresponding to the current scenario to obtain the quantification results of extreme weather. The decision module 230 is used to input multi-source data and extreme weather quantification results into a multi-timescale hierarchical decision model to generate decision schemes; The multi-timescale hierarchical decision-making model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer coordination hub. The cross-layer coordination hub is used to coordinate the adjustment of parameters of the long-term planning layer, the medium-term scheduling layer, and the short-term execution layer based on the quantitative results of extreme weather and the state of the mechanism model.

[0071] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0072] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations to cope with extreme weather, characterized in that, include: Acquire multi-source data and identify the current scene based on the multi-source data; The multi-source data is input into the mechanism model corresponding to the current scenario to obtain the quantification results of extreme weather. The multi-source data and the extreme weather quantification results are input into a multi-timescale hierarchical decision model to generate a decision scheme; The multi-timescale hierarchical decision-making model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer collaborative hub. The cross-layer collaborative hub is used to coordinate the adjustment of parameters of the long-term planning layer, medium-term scheduling layer, and short-term execution layer based on the quantification results of extreme weather and the status of the mechanism model.

2. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 1, characterized in that, The multi-source data is input into the mechanism model corresponding to the current scenario to obtain the extreme weather quantification results, including: The multi-source data is input into the mechanism model, and the extreme weather quantification results are output according to the multi-field coupling algorithm built into the model. The mechanism models include, but are not limited to, snow shedding cycle models, dust accumulation rate models, and ice growth and shedding prediction models. The multi-field coupling algorithm is used to analyze the interaction mechanism of multi-physics parameters in multi-source data, so as to quantify the nonlinear correlation between various physical field parameters under extreme weather conditions.

3. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 2, is characterized in that... The mechanism model includes a coupling correlation module and a solution quantization module; the coupling correlation module is configured with coupling parameters; the coupling parameters of the mechanism model are different for each scenario. The multi-source data is input into the mechanism model, and the extreme weather quantification results are output based on the multi-field coupling algorithm built into the model, including: The multi-source data is processed to obtain the input matrix; Based on the input matrix and the multi-field coupling algorithm, determine the coupling parameter adjustment value; Based on the multi-source data and the coupling parameter adjustment value, determine the time series curve of the dynamic correlation between extreme weather factors and equipment operating status parameters; The extreme weather quantification results are determined based on the dynamic correlation time series curve and the solution quantification module.

4. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 3, is characterized in that... The multi-source data and the quantification results of extreme weather are input into a multi-timescale hierarchical decision model to generate a decision scheme, including: Extract the state of the mechanistic model after adjusting the coupling parameters; The multi-source data and the extreme weather quantification results are fused to obtain a short-term decision feature set, a medium-term decision feature set, and a long-term decision feature set. The extreme weather quantification results and mechanism model status are input into the cross-layer collaborative hub to determine short-term decision constraints, medium-term decision constraints and long-term decision constraints. Based on the long-term planning feature set, long-term decision constraints, and the long-term planning layer, a preventative maintenance strategy and resource allocation plan are generated. Based on the intermediate scheduling feature set, intermediate decision constraints, and the intermediate scheduling layer, an optimal resource scheduling scheme is generated. Based on the short-term execution feature set, short-term decision constraints, optimal resource scheduling scheme, acquired real-time feedback data, and the short-term execution layer, a short-term dynamic adjustment scheme is determined; Based on the aforementioned preventive maintenance strategy, resource allocation plan, optimal resource scheduling scheme, and short-term dynamic adjustment scheme, a decision-making scheme is determined.

5. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 4, characterized in that, Extract the mechanistic model state after coupling parameter adjustment, including: Collect the dynamic adjustment trajectory of the coupling parameter and the corresponding adjustment trigger source data; wherein, the dynamic adjustment trajectory includes the numerical change information and timestamp information of the coupling parameter before and after adjustment, and the adjustment trigger source data includes the key feature threshold and deviation threshold that trigger the adjustment of the coupling parameter; Obtain the operational status data of the coupling and correlation module based on the adjusted parameters; wherein, the operational status data includes at least one of the following: the modeling fit of the multiphysics coupling effect, the fitting degree of the dynamic correlation time series curve, the nonlinear correlation calculation error, and the resource consumption parameters; The dynamic adjustment trajectory, adjustment trigger source data, and running status data are used to determine the state of the mechanism model.

6. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 5, is characterized in that... The quantification results of extreme weather and the state of the mechanism model are input into the cross-layer collaborative hub to determine short-term, medium-term, and long-term decision constraints, including: The extreme weather quantification results are analyzed in a scenario-based manner to extract the core quantitative indicators of the scenario and the critical conditions for equipment failure. Based on the dynamically adjusted trajectory and operational status data, the quantitative reliability level of the mechanism model for the extreme weather scenario is determined. Based on the core quantitative indicators of the scenario and the quantitative reliability level, decision constraints at each time scale level are determined.

7. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 5, is characterized in that, The extreme weather quantification results are analyzed in a scenario-based manner to extract core quantitative indicators and critical conditions for equipment failure, including: Screening basic quantitative data directly related to the physical characteristics of the scene from the quantitative results of extreme weather events; Based on the aforementioned basic quantitative data, calculate the core quantitative indicators for the scenario; Based on the technical parameters of the new energy power station equipment, historical failure data, and the dynamic correlation time series curve, the critical conditions for equipment failure in each scenario are determined.

8. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 5, is characterized in that, Based on the dynamically adjusted trajectory and operational status data, the quantitative reliability level of the mechanism model for the extreme weather scenario is determined, including: Based on dynamically adjusted trajectory and operational status data, a multi-dimensional quantitative accuracy evaluation index system is constructed. We set the weights and grading thresholds for each evaluation indicator in the multi-dimensional quantitative accuracy evaluation index system, and use the weighted summation method to calculate the comprehensive score of quantitative reliability in order to determine the level of quantitative reliability.

9. The multi-timescale hierarchical operation and maintenance scheduling method for new energy power stations in response to extreme weather as described in claim 1, characterized in that, The method further includes: The preventive maintenance strategy, optimal resource scheduling scheme, and short-term dynamic adjustment scheme in the decision-making scheme are broken down into instruction sets; Execute the instruction set, evaluate the execution process of the instruction set, and obtain the evaluation result; Based on the evaluation results, the multi-timescale hierarchical decision-making model is iteratively optimized.

10. A multi-timescale hierarchical operation and maintenance scheduling system for new energy power stations to cope with extreme weather, characterized in that, include: The acquisition module is used to acquire multi-source data and identify the current scene based on the multi-source data; The quantization module is used to input the multi-source data into the mechanism model corresponding to the current scenario to obtain the quantification results of extreme weather. The decision module is used to input the multi-source data and the extreme weather quantification results into a multi-timescale hierarchical decision model to generate a decision scheme; The multi-timescale hierarchical decision-making model includes a long-term planning layer, a medium-term scheduling layer, a short-term execution layer, and a cross-layer collaborative hub. The cross-layer collaborative hub is used to coordinate the adjustment of parameters of the long-term planning layer, medium-term scheduling layer, and short-term execution layer based on the quantification results of extreme weather and the status of the mechanism model.

Images