Wind power equipment optimal decommissioning time window prediction method and device

Abstract

The invention provides a wind power equipment optimal decommissioning time window prediction method and device. According to the wind power equipment optimal decommissioning time window prediction method provided by the embodiment of the invention, the dynamic quantitative evaluation of the environmental influence in the full life cycle of the wind power equipment is realized, the optimal decommissioning time window is accurately identified through three-dimensional data fusion of technology, economy and environment and machine learning prediction, and the scientificity and environmental protection benefits of decommissioning decision are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind power equipment life cycle management, and in particular to a wind power equipment optimal retirement time window prediction method and device. BACKGROUND

[0002] Wind power equipment retirement decision-making is a core link of renewable energy asset management and is widely used in wind farm life cycle management. With the acceleration of global energy structure transformation, China's wind power industry is facing the first large-scale retirement wave, and the number of retired units is expected to increase significantly from 2025. In related technologies, through the collaborative work of condition monitoring systems (CMS), supervisory control and data acquisition systems (SCADA) and digital twin technology, an operation and maintenance decision-making system is constructed, mainly based on equipment health state evaluation and economic analysis. Specifically, this system covers the whole process from equipment performance monitoring, fault prediction to retirement disposal, including vibration analysis, residual useful life (RUL) prediction, spare parts cost accounting and other key links. Among them, the predictive maintenance technology based on machine learning has realized the dynamic optimization of equipment failure rate and maintenance cost, while the static life cycle assessment (LCA) method, as an industry standard for environmental impact assessment, is mainly used for carbon footprint accounting in the project planning stage.

[0003] However, in the existing retirement decision-making method, the LCA evaluation model directly uses fixed parameters and does not dynamically associate with the real-time running state of the equipment and the site environmental factors, which may cause the environmental cost accounting to lag behind or produce irreversible ecological impact during the equipment operation period. Specifically, traditional LCA evaluation only calculates the environmental load of material production, transportation, disassembly and other links in the retirement disposal stage, but cannot reflect the dynamic environmental problems such as soil heavy metal pollution and water microplastic accumulation caused by lubricating oil leakage and component wear during the 20-year operation cycle of the wind turbine. Based on this, the existing technology usually adopts a fragmented decision-making dimension, i.e. dealing with equipment reliability, economic benefits and environmental impact respectively, among which the technical dimension focuses on vibration, temperature and other physical parameter analysis, the economic dimension focuses on operation and maintenance cost and power generation income, and the environmental dimension stays at the static evaluation level. This fragmentation makes it difficult for decision-makers to quantitatively compare the dynamic trade-off relationship between "continuing operation" and "immediate retirement" in terms of environmental cost, economic benefit and technical performance. Due to the lack of real-time data fusion mechanism across dimensions, the existing system cannot predict the coupling trend of equipment performance degradation curve and environmental impact accumulation index (EII), which makes about 30% of wind power equipment retire at a non-optimal time, causing an annual economic loss of more than 10 billion yuan and an environmental burden of an increase of 40% in soil pollution risk. SUMMARY

[0004] The present application aims to at least partially solve one of the technical problems in the related art.

[0005] To this end, the first object of the present application is to provide a wind power equipment optimal retirement time window prediction method.

[0006] The second object of the present application is to provide a wind power equipment optimal retirement time window prediction device.

[0007] To achieve the above object, the first aspect of the present application provides a wind power equipment optimal retirement time window prediction method, comprising: S1, collecting soil pollution risk, water pollution risk, noise exceeding time and solid waste generation amount environmental monitoring data in real time through the environmental factor monitoring module deployed in the wind farm, and obtaining operation data and economic parameter data of the wind turbine generator; S2, calculating an environmental impact cumulative index EII based on the environmental monitoring data, wherein the EII is generated by a weighted integral formula EII=∫[Σ(Wn*Dn(t))]dt, wherein Wn is a preset weight of each environmental factor, and Dn(t) is a real-time deviation degree of each environmental factor at a preset time t; S3, inputting the operation data, economic parameter data and EII value into a multi-dimensional decision analysis engine, generating a comprehensive evaluation score through normalization processing and a multi-criteria decision analysis model, wherein the comprehensive evaluation score includes quantitative comparison of technical performance degradation, economic net income and environmental cost; S4, predicting the operation data, economic parameter data and EII change trend by using a machine learning model, constructing a comprehensive benefit function and optimizing to determine the optimal retirement time window that maximizes the comprehensive net present value of the whole life cycle.

[0008] In an embodiment of the present application, the S1 further comprises: S11, the soil pollution risk monitoring adopts a multi-spectral soil sensor array to detect the spatial and temporal distribution characteristics of three parameters of soil pH value, heavy metal ion concentration and organic matter content; S12, the solid waste generation amount monitoring includes a blade material aging prediction module to predict the remaining service life of the blade and the potential microplastic release amount through infrared spectrum analysis and mechanical property test data.

[0009] In an embodiment of the present application, the S2 further comprises: S21, when calculating the instantaneous deviation Dn(t) of each environmental factor, a dynamic reference value adjustment algorithm is adopted to dynamically update the reference value according to the historical environmental data of the wind farm and the real-time weather conditions; S22, the distribution of the weight Wn in the weighted integral formula adopts a multi-layer perception machine MLP model to automatically generate the dynamic weight coefficient of each environmental factor by training the correlation between the historical environmental impact events and the equipment operation data.

[0010] In one embodiment of the present application, the S3 further comprises: S31, using a positive index normalization formula to map the power generation efficiency positive index to the [0, 1] interval, and using a negative index normalization formula to map the operation and maintenance cost, EII negative index to the [0, 1] interval; S32, the weight distribution of the technology, economy and environment dimensions in the multi-criteria decision analysis model adopts the Monte Carlo simulation method, generates 1000 groups of weight combinations through random sampling, and selects the optimal weight configuration that minimizes the variance of the comprehensive evaluation score.

[0011] In one embodiment of the present application, the machine learning model includes a performance degradation prediction model and an economic benefit prediction model, and further comprises: S41, the performance degradation prediction model adopts the gradient boosting tree algorithm, and takes the device running time, component wear data and environmental corrosion index as input features, and outputs the future power generation efficiency degradation curve; S42, the economic benefit prediction model adopts the long short-term memory network algorithm, and takes the real-time electricity price fluctuation, spare parts price index and environmental cost discount coefficient as input features, and outputs the time series prediction of future net income.

[0012] In one embodiment of the present application, further comprising: S5, a visual decision report generation step, which displays the real-time comparison of device health, economic benefit curve and EII change trend through a three-dimensional dynamic dashboard, and outputs a PDF format evaluation report containing retirement time window, disposal scheme recommendation and risk warning.

[0013] To achieve the above purpose, the second embodiment of the present application proposes a wind power equipment optimal retirement time window prediction device, comprising: An environmental factor monitoring module is used to collect environmental monitoring data of soil pollution risk, water pollution risk, noise exceeding time and solid waste generation amount in real time through the environmental factor monitoring module deployed in the wind farm, and obtain operation data and economic parameter data of the wind turbine generator; An environmental impact cumulative index calculation module is used to calculate the environmental impact cumulative index EII based on the environmental monitoring data, and the EII is generated by a weighted integral formula EII=∫[Σ(Wn*Dn(t))]dt; A multi-dimensional decision analysis module is used to input the operation data, economic parameter data and EII value into a multi-dimensional decision analysis engine, generate a comprehensive evaluation score through normalization processing and a multi-criteria decision analysis model, and the comprehensive evaluation score contains quantitative comparison of technical performance degradation, economic net income and environmental cost; The machine learning prediction and optimization module is configured to utilize a machine learning model to predict the operation data, economic parameter data and EII change trend, construct a comprehensive benefit function and determine an optimal retirement time window that maximizes the whole life cycle comprehensive net present value. The visual decision report generation module is configured to generate a visual decision report, display a real-time comparison of the equipment health degree, economic benefit curve and EII change trend through a three-dimensional dynamic dashboard, and output a PDF format evaluation report containing the retirement time window, disposal scheme recommendation and risk warning.

[0014] In an embodiment of the present application, the environmental factor monitoring module is configured to: The soil pollution risk monitoring adopts a multi-spectral soil sensor array to detect the spatial and temporal distribution characteristics of the soil pH value, heavy metal ion concentration and organic matter content; The solid waste generation monitoring includes a blade material aging prediction module configured to predict the remaining service life of the blade and potential microplastic release amount through infrared spectrum analysis and mechanical property test data.

[0015] In an embodiment of the present application, the environmental impact cumulative index calculation module is further configured to: When the environmental monitoring data is standardized, the dynamic reference value adjustment algorithm is adopted to dynamically update the reference value according to the historical environmental data of the wind farm and the real-time weather conditions; The distribution of the weight Wn in the weighted integral formula adopts a multi-layer perception machine model to automatically generate the dynamic weight coefficient of each environmental factor by training the correlation between the historical environmental impact events and the equipment operation data.

[0016] In an embodiment of the present application, the multi-dimensional decision analysis module is further configured to: The positive index normalization formula is adopted to map the power generation efficiency positive index to the [0, 1] interval, and the negative index normalization formula is adopted to map the operation and maintenance cost and EII negative index to the [0, 1] interval; The weight distribution of the technology, economy and environment dimensions in the multi-standard decision analysis model adopts a Monte Carlo simulation method to generate 1000 groups of weight combinations through random sampling, and filter out the optimal weight configuration that minimizes the variance of the comprehensive evaluation score.

[0017] In an embodiment of the present application, the machine learning prediction and optimization module is further configured to: The performance degradation prediction model adopts a gradient boosting tree algorithm to take the equipment operation time, component wear data and environmental corrosive index as input features, and output the future power generation efficiency degradation curve; The economic benefit prediction model uses a long short-term memory network algorithm, taking real-time electricity price fluctuations, spare parts price index, and environmental cost discount factor as input features, and outputs a time series prediction of future net income.

[0018] In one embodiment of the present invention, it further includes: The 3D dynamic dashboard display module is used to display real-time comparisons of equipment health, economic benefit curves, and EII change trends through a 3D dynamic dashboard. The PDF assessment report generation module is used to output PDF assessment reports that include decommissioning time windows, recommended disposal plans, and risk warnings.

[0019] The method and apparatus of this invention enable dynamic quantitative assessment of the environmental impact of wind power equipment throughout its entire life cycle. By integrating data from three dimensions—technology, economy, and environment—and using machine learning prediction, the optimal decommissioning time window can be accurately identified, thereby improving the scientific rigor and foresight of decommissioning decisions.

[0020] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a method for predicting the optimal decommissioning time window for hydro-wind power equipment according to an embodiment of the present invention; Figure 2 This is an architecture diagram of an intelligent evaluation system for wind power equipment decommissioning decisions based on multidimensional environmental impact factors, according to an embodiment of the present invention. Figure 3 This is a structural diagram of an optimal decommissioning time window prediction device for hydro-wind power equipment according to an embodiment of the present invention. Detailed Implementation

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] The following description, with reference to the accompanying drawings, describes a method and apparatus for predicting the optimal decommissioning time window of wind power equipment according to an embodiment of the present invention.

[0025] Example 1 Figure 1 This is a flowchart of the optimal decommissioning time window prediction method for wind power equipment according to an embodiment of the present invention, as follows: Figure 1 As shown, it includes: S1 collects environmental monitoring data in real time, including soil pollution risk, water pollution risk, duration of noise exceeding standards, and amount of solid waste generated, through environmental factor monitoring modules deployed in wind farms, and also obtains wind turbine operation data and economic parameter data.

[0026] Specifically, this step involves real-time collection of environmental monitoring data on soil pollution risk, water pollution risk, duration of noise exceeding standards, and solid waste generation through environmental factor monitoring modules deployed at the wind farm, while simultaneously acquiring wind turbine operation data and economic parameter data. This step is one of the core functions of the data acquisition layer in the intelligent assessment system of this invention, providing basic data support for subsequent calculation of the Environmental Impact Accumulation Index (EII), multi-dimensional decision analysis, and prediction of optimal decommissioning time.

[0027] Furthermore, the environmental factor monitoring module adopts a distributed sensor network architecture, deployed in key ecologically sensitive areas of the wind farm, such as the soil around the wind turbine foundations, noise monitoring points at the site boundary, and sampling points in downstream water areas. Soil pollution risk is monitored in real time using soil pH sensors and heavy metal ion detectors, with a sampling frequency of once per hour and data accuracy reaching ±0.1 pH and ±5% heavy metal content. Water pollution risk is monitored by collecting indicators such as COD, ammonia nitrogen, and heavy metals using a multi-parameter water quality analyzer, with a sampling frequency of once every 2 hours, meeting the requirements of GB3838-2002 "Surface Water Environmental Quality Standard". The duration of noise exceeding the standard is recorded in real time by a sound level meter (sampling frequency of 1 second / time, with noise thresholds set at 55 dB(A) during the day and 45 dB(A) at night), meeting the requirements of GB12348-2008 "Emission Standard for Environmental Noise at the Boundary of Industrial Enterprises". Solid waste generation is automatically counted through an intelligent weighing system and waste classification and identification algorithm, with data updated daily.

[0028] While collecting data, the system also acquires real-time operating data of the wind turbines through the SCADA system interface, including power generation, equipment utilization, component vibration values, and lubricant consumption. The sampling frequency is once per minute, and the data format complies with the IEC61400-25 standard. Economic parameter data is obtained through the ERP system or manual entry, including real-time electricity prices, operation and maintenance costs, spare parts prices, and land lease fees. The data is updated daily or weekly to ensure the timeliness of the economic model.

[0029] This step, through the synchronous acquisition and standardized processing of multi-source heterogeneous data, provides high-quality and timely input data for subsequent calculation of the Environmental Impact Cumulative Index (EII) and Multi-Standard Decision Analysis (MCDA), and is a key prerequisite for realizing intelligent decision-making that integrates "environment, economy, and technology".

[0030] Furthermore, S1 includes: S11, soil pollution risk monitoring adopts a multispectral soil sensor array to detect the spatiotemporal distribution characteristics of three parameters: soil pH, heavy metal ion concentration, and organic matter content.

[0031] Specifically, the soil pollution risk monitoring step employs a multispectral soil sensor array. By detecting the spatiotemporal distribution characteristics of three key parameters—soil pH, heavy metal ion concentration, and organic matter content—it achieves dynamic assessment of the soil environmental quality in the wind power equipment operating area. In some implementations, the sensor array consists of multiple distributed nodes. Each node integrates a multispectral sensor, an electrochemical probe, and an infrared absorption module, used to measure soil pH, heavy metal ion concentration (such as lead, cadmium, mercury, and arsenic), and volatile organic compounds (VOCs) and total organic carbon (TOC) content, respectively. The sensor nodes upload data in real time to the system's central processing unit via wireless communication protocols (such as LoRaWAN or NB-IoT), forming a continuous spatiotemporal dataset.

[0032] Furthermore, the pH measurement accuracy is typically ±0.1 pH, with a sampling frequency of once per hour; heavy metal ion concentration is detected using electrochemical stripping or X-ray fluorescence spectrometry (XRF), with a detection limit of up to 0.1 mg / kg and a resolution better than 0.05 mg / kg; organic matter content is quantitatively analyzed using Fourier transform infrared spectroscopy (FTIR) or gas chromatography-mass spectrometry (GC-MS), with a detection accuracy of ±5%. The collection of these parameters complies with the relevant requirements of the "Technical Specification for Soil Environmental Monitoring" (HJ967-2018) and the "Soil Environmental Quality Standard" (GB15618-2018).

[0033] Furthermore, this step is deployed around the wind farm equipment bases, underground cable routes, and key areas such as lubricating oil / coolant discharge points to monitor potential soil pollution risks during equipment operation over the long term. By constructing spatial interpolation models (such as Kriging interpolation), the system can generate pollution heat maps to help identify pollution diffusion paths and high-risk areas.

[0034] The technical advantage of this step lies in providing key input data for the calculation of the "Cumulative Environmental Impact Index" (EII), thereby enabling dynamic quantification of the environmental load throughout the entire life cycle of wind power equipment. Through high-precision, high-frequency monitoring using a multispectral sensor array, the system can capture changes in the soil environment in real time, providing a scientific basis for subsequent multi-dimensional decision analysis and significantly improving the environmental sensitivity and data support capabilities of decommissioning decisions.

[0035] S12, solid waste generation monitoring includes a blade material aging prediction module, which predicts the remaining usable years of the blades and the potential amount of microplastics released through infrared spectroscopy analysis and mechanical performance test data.

[0036] Specifically, the step of "monitoring solid waste generation, including a blade material aging prediction module, which predicts the remaining usable years of the blades and potential microplastic release through infrared spectroscopy analysis and mechanical performance test data" described in this invention is one of the key technical links in realizing dynamic assessment of the environmental impact of wind power equipment throughout its entire life cycle. This module provides the system with a scientific basis for blade retirement risk through quantitative analysis of material aging status, thereby supporting intelligent decision-making regarding the optimal retirement time window.

[0037] Furthermore, this module first employs Fourier transform infrared spectroscopy (FTIR) to detect changes in the chemical structure of blade materials (such as epoxy resin, glass fiber, and other composite materials). FTIR identifies aging characteristics such as functional group cracking, oxidation, or hydrolysis in materials by analyzing their absorption spectra in the 2.5–15 μm wavelength range. For example, epoxy resin undergoes ester bond breakage under long-term ultraviolet radiation and humid heat conditions, and its characteristic peak (e.g., 1730 cm⁻¹) will show these changes. -1 The intensity change (in the vicinity) can serve as a quantitative indicator of the degree of aging. In addition, the module also integrates mechanical performance tests, such as the three-point bending test (ASTM D790 standard) and tensile strength test (ASTM D638 standard), to assess the decay trend of key mechanical parameters such as the elastic modulus, flexural strength and elongation at break of the material.

[0038] Furthermore, the system sets aging thresholds; for example, when the infrared characteristic peak intensity of the epoxy resin decreases by more than 15% or the flexural strength is lower than 70% of the initial value, the material is determined to have entered a high-risk aging stage. Simultaneously, combined with a material aging rate model (such as the Arrhenius model), the remaining service life (RUL) is predicted. Microplastic release is estimated using an empirical model established based on material surface wear rate and environmental exposure parameters (such as wind speed and rain erosion frequency), outputting the microplastic release per unit time (mg / m²·year).

[0039] Furthermore, this module can be deployed in the digital twin system of a wind farm to periodically collect blade samples or remotely monitor material status, making it suitable for blade health management in both offshore and onshore wind farms. Its technological advantage lies in the fact that it is the first to incorporate blade material aging and microplastic release risks into the decommissioning decision-making system, enabling forward-looking prediction of solid waste generation and providing key input for the Environmental Impact Accumulation Index (EII), thereby improving the accuracy and scientific rigor of the system's environmental assessment.

[0040] S2, calculate the cumulative environmental impact index EII based on the environmental monitoring data. The EII is generated by the weighted integral formula EII=∫[Σ(Wn*Dn(t))]dt.

[0041] Specifically, in this step, the Environmental Impact Cumulative Index (EII) is calculated based on environmental monitoring data. The technical principle behind this is to introduce a weighted integral model to dynamically quantify and accumulate the real-time deviations of multiple environmental factors during operation, thereby achieving continuous assessment of the environmental load throughout the entire lifecycle of wind power equipment. Specifically, firstly, the system collects environmental factor data in real time through sensors deployed around the wind farm, such as noise levels, pollutant concentrations, and soil acidification levels. Each environmental factor has a compliance benchmark value. At the data processing layer, the system calculates the difference between the real-time monitoring value and the benchmark value of each environmental factor to obtain the instantaneous deviation Dn(t). Subsequently, the weighted deviation Σ(Wn×Dn(t)) of each factor is integrated over time, with the integration interval from the equipment commissioning time t0 to the current time t. The integration step size can be set according to the data acquisition frequency. In a discrete system, the integration can be transformed into a cumulative calculation of the time series, i.e., EII=∫[Σ(Wn*Dn(t))]dt. This step transforms the originally static environmental impact assessment into a continuous and traceable quantitative indicator through dynamic integration, providing crucial input for subsequent multi-dimensional decision analysis. The EII calculation results can serve as an important basis for whether equipment should be decommissioned early. When it exceeds a preset threshold, the system will trigger an environmental risk early warning mechanism. This step acts as a bridge in the system, transforming environmental monitoring data into actionable decision variables, realizing real-time visibility and controllability of environmental costs. It is one of the core supporting technologies for this invention to achieve integrated intelligent decision-making encompassing the "environment-economy-technology" framework.

[0042] Furthermore, S2 includes: S21, the environmental monitoring data is standardized and the instantaneous deviation Dn(t) of each environmental factor is calculated. A dynamic benchmark adjustment algorithm is used to dynamically update the benchmark value based on the historical environmental data of the wind farm and real-time meteorological conditions.

[0043] Specifically, when standardizing the environmental monitoring data and calculating the instantaneous deviation Dn(t) of each environmental factor, a dynamic benchmark adjustment algorithm is employed. This algorithm is based on the fusion analysis of real-time and historical data to adapt to the dynamic changes of wind farm environmental factors with time, season, and meteorological conditions. In some implementations, the algorithm first obtains the current monitoring data from the environmental factor monitoring module, including but not limited to noise levels, soil pH, heavy metal concentration, and water pollutant concentration. Subsequently, the system calls the historical environmental database to extract historical environmental factor data of the wind farm under the same or similar meteorological conditions (such as wind speed, temperature, humidity, and rainfall), and constructs a benchmark dynamic adjustment model.

[0044] The dynamic baseline adjustment algorithm can optionally employ a sliding window average method or a weighted moving average (WMA) method, combined with multivariate regression analysis using meteorological parameters, to determine a reasonable baseline value for the current environmental factors. For example, for noise factors, the system can calculate its baseline value based on noise data from the past 30 days under the same wind speed and operating load conditions, and then adjust it with weights according to real-time wind speed and operating status. In specific parameter settings, the sliding window length can be set to 7–30 days, and the weighting coefficients are dynamically allocated based on the correlation between meteorological parameters and environmental factors. The correlation coefficient R² must be greater than 0.75 to ensure model reliability.

[0045] This step plays a crucial role in the entire system, providing accurate and dynamic deviation input for the subsequent calculation of the Environmental Impact Cumulative Index (EII). By updating the baseline value in real time, the system can more accurately identify abnormal changes in environmental factors, thereby achieving dynamic quantification and early warning of the environmental impact of wind power equipment. Furthermore, this method effectively solves the problem that traditional static LCA assessments cannot reflect real-time environmental loads, improving the environmental sensitivity and scientific rigor of decommissioning decisions.

[0046] S22, the weight Wn in the weighted integral formula is allocated using a multilayer perceptron (MLP) model. By training the correlation between historical environmental impact events and equipment operation data, dynamic weight coefficients of each environmental factor are automatically generated.

[0047] Specifically, in some implementations, the weights Wn in the weighted integral formula are allocated using a multilayer perceptron (MLP) model. This model is based on a supervised learning framework, which trains the correlation between historical environmental impact events and equipment operation data to automatically generate dynamic weight coefficients for each environmental factor. This step is the core component of the "Environmental Impact Cumulative Index (EII)" algorithm in this invention, used to quantify the relative impact of different environmental factors at different time points and operating conditions, thereby improving the accuracy and dynamic adaptability of environmental cost assessment.

[0048] The specific operation method is as follows: First, the system extracts environmental monitoring data (such as noise level, soil pH value, and water pollutant concentration) and equipment operating status data (such as lubricating oil replacement frequency, component wear degree, and power generation efficiency) from the historical database during the operation of the wind turbine, and constructs a training sample set. Each sample contains real-time monitoring values ​​of multiple environmental factors and their corresponding equipment operating status and environmental events (such as pollution exceeding standards, ecological disturbances, etc.). Subsequently, these data are input into the MLP model for training. The MLP model usually consists of an input layer, several hidden layers, and an output layer, using activation functions such as ReLU or Sigmoid. The loss function can be the mean squared error (MSE) or cross-entropy loss. The optimizer can be Adam or SGD. The learning rate is generally set to 0.001~0.01, and the number of training epochs is controlled between 50 and 200 to ensure that the model converges and has good generalization ability.

[0049] Furthermore, the generation of weights Wn depends on the model's accuracy in fitting the nonlinear relationship between various environmental factors and equipment operating status. For example, the weight of noise factors may increase significantly during nighttime operation or at sites near ecological protection zones, while the weight of soil pollution factors may dynamically adjust with the equipment's operating age and the frequency of lubricating oil leaks. The weight coefficients output by the model must satisfy the normalization constraint, i.e., ΣWn=1, to ensure that the relative contribution of each factor in the cumulative index is reasonable.

[0050] In practical applications, this step is deployed in the intelligent decision-making system of wind farms, working in conjunction with real-time environmental monitoring modules and equipment status databases to achieve dynamic quantification of environmental impact. Its technical advantage lies in automatically learning the changes in the importance of environmental factors through machine learning, replacing the traditional subjective method of manually setting weights. This significantly improves the scientific rigor and adaptability of environmental impact assessments, providing crucial inputs for subsequent multi-criteria decision analysis (MCDA) and optimal decommissioning time window prediction.

[0051] S3. Input the operating data, economic parameter data and EII value into the multi-dimensional decision analysis engine, and generate a comprehensive evaluation score through normalization processing and multi-standard decision analysis model. The comprehensive evaluation score includes a quantitative comparison of technical performance degradation, net economic benefits and environmental costs.

[0052] Specifically, inputting the operational data, economic parameter data, and cumulative environmental impact index (EII) into the multi-dimensional decision analysis engine is the core step in achieving a comprehensive assessment of the "environment-economy-technology" triad in this invention. In some implementations, the engine uses a combination of normalization processing and a multi-standard decision analysis model (MCDA) to standardize and weight the three types of heterogeneous data, thereby generating a comparable comprehensive assessment score.

[0053] Furthermore, a central control system or cloud-based decision-making platform deployed in the wind farm can generate a dynamic comprehensive evaluation score for each wind turbine by combining real-time data streams with historical data. For example, in the 18th year of equipment operation, the system can output a score comparison between the two options: continued operation and immediate decommissioning, to assist decision-makers in determining whether to proceed with the decommissioning process.

[0054] Furthermore, this step enables the fusion analysis of cross-dimensional data, allowing for the quantitative comparison of environmental costs, economic benefits, and technical performance within a unified framework, thereby significantly improving the scientific rigor and foresight of decommissioning decisions. Through this model, the system can identify the optimal decommissioning time window for comprehensive benefits, avoiding economic losses or environmental burdens caused by single-dimensional decision-making, and providing a standardized and replicable technical path for the intelligent management of wind power assets.

[0055] Furthermore, S3 includes: S31 uses a positive index normalization formula to map the positive power generation efficiency index to the [0,1] interval, and uses a negative index normalization formula to map the operation and maintenance cost and EII negative index to the [0,1] interval.

[0056] Specifically, this step involves normalizing the positive and negative indicators involved in the decision-making process for wind power equipment decommissioning, so as to achieve the fusion analysis of multi-dimensional data under a unified dimension.

[0057] This normalization step plays a crucial role in data preprocessing within the system, providing standardized input for the subsequent multi-dimensional decision analysis engine. By unifying indicators of different dimensions to the [0,1] range, the system can calculate a comprehensive score based on a weighted sum model, thereby achieving intelligent decision-making integrating the "environment-economy-technology" framework. The implementation of this step significantly improves the accuracy and comparability of data fusion, serving as a fundamental step in predicting the optimal decommissioning time window for wind power equipment and conducting dynamic cost-benefit analysis.

[0058] S32, In the Multi-Standard Decision Analysis (MCDA) model, the weight allocation of the technology, economic and environmental dimensions adopts the Monte Carlo simulation method. 1000 weight combinations are generated by random sampling, and the optimal weight configuration that minimizes the variance of the comprehensive evaluation score is selected.

[0059] Specifically, in the Multi-Criterion Decision Analysis (MCDA) model of this invention, the weight allocation of the three dimensions of technology, economy, and environment is optimized using Monte Carlo simulation to achieve the stability and scientific nature of the comprehensive evaluation score. The technical implementation principle of this step is based on probability statistics and optimization algorithms. By randomly sampling within the weight space, the impact of different weight combinations on the comprehensive evaluation result is simulated, ultimately selecting the optimal weight configuration that minimizes the variance of the evaluation score.

[0060] The specific operation is as follows: First, set the value range of each dimension's weight, usually within the range [0,1], and satisfy the constraint that the sum of the weights is 1. In some implementations, Latin hypercube sampling (LHS) or uniform random sampling methods can be used to generate 1000 weight combinations from this constrained space. Each weight combination is used to calculate the comprehensive evaluation score of the current equipment under the two decision options of "continue operation" and "immediate retirement". Normalization is performed using min-max normalization or Z-score standardization to ensure the comparability of indicators with different dimensions.

[0061] Furthermore, the system performs statistical analysis on the comprehensive scores under 1000 weights, calculating their standard deviation or variance. The weight combination with the smallest variance is the optimal configuration, which minimizes the differences in evaluation results between different decision-making schemes, thereby improving the robustness and consistency of the decision. In practical applications, this step can be combined with external factors such as the environmental sensitivity of the wind farm, local environmental policies, and electricity price fluctuation cycles to dynamically adjust the weight value range and sampling strategy to adapt to the needs of different regions and operational stages.

[0062] This step plays a crucial role in decision optimization within the overall technical solution, ensuring the system's scientific rigor and interpretability during the fusion of multi-dimensional indicators. Through Monte Carlo simulation, the system can effectively avoid biases caused by subjective weight settings, improving the intelligence and environmental friendliness of decommissioning decisions.

[0063] S4. Using machine learning models, predict the operational data, economic parameter data, and EII change trends, construct a comprehensive benefit function, and optimize to determine the optimal retirement time window that maximizes the comprehensive net present value over the entire life cycle.

[0064] Specifically, in this step, a machine learning model is used to predict the changing trends of operational data, economic parameter data, and the Environmental Impact Accumulation Index (EII), and a comprehensive benefit function is constructed to determine the optimal decommissioning time window that maximizes the comprehensive net present value over the entire life cycle. This step is the core decision-making mechanism of the intelligent assessment system of this invention, realizing a shift from a "passive response" to a "proactive prediction" management model.

[0065] Furthermore, through multi-dimensional data fusion and machine learning prediction, the system can accurately identify the peak range of future comprehensive benefits, thereby determining the optimal retirement time window and maximizing both economic and environmental benefits throughout the entire life cycle.

[0066] In summary, the wind power equipment optimal decommissioning time window prediction method of this invention realizes dynamic quantification and multi-dimensional comprehensive evaluation of the environmental impact of wind power equipment decommissioning decisions, improves the scientificity and foresight of decision-making, accurately predicts the optimal decommissioning time window, and optimizes economic and environmental benefits.

[0067] Furthermore, S4 includes: S41, the performance degradation prediction model adopts the gradient boosting tree (GBDT) algorithm, with equipment operating time, component wear data and environmental corrosion index as input features, and outputs the degradation curve of future power generation efficiency.

[0068] Specifically, in some implementations, the performance degradation prediction model employs the Gradient Boosted Decision Tree (GBDT) algorithm, using equipment operating time, component wear data, and environmental corrosion index as input features to output a degradation curve of future power generation efficiency. This model is based on a supervised learning framework, trained using historical operating data and the performance degradation trajectories of similar units, thereby achieving accurate prediction of the future power generation efficiency trend of the target equipment.

[0069] Furthermore, the model first preprocesses the input features, including missing value imputation, outlier detection and correction, and feature normalization. Equipment operating time is measured in cumulative operating hours. Component wear data includes gearbox tooth wear, bearing vibration amplitude, and blade surface roughness. The environmental corrosion index is provided by the environmental factor monitoring module, typically based on corrosion levels (C1-C5) defined in ISO 9223, and dynamically calculated using on-site meteorological data (such as humidity, salt spray concentration, and acid rain frequency). The model input feature dimensions are typically 10-15, covering equipment status, operating conditions, and key environmental impact variables.

[0070] Furthermore, the GBDT model employs the efficient frameworks XGBoost or LightGBM, setting key hyperparameters such as the learning rate to 0.05-0.2, the maximum tree depth to 4-8, and the subsample ratio to 0.6-0.9 to balance the model's fitting ability and generalization performance. The model output is the power generation efficiency degradation curve for the next 3-5 years, with a monthly time granularity, and the prediction error is controlled within ±3%, meeting the accuracy requirements for wind farm operation and management.

[0071] Furthermore, this model is deployed in the wind farm's intelligent decision-making system, interfacing in real time with the SCADA system, Condition Monitoring System (CMS), and environmental monitoring sensors to assist maintenance personnel in formulating equipment decommissioning or upgrade strategies. By predicting performance degradation trends, the system can identify when equipment enters a phase of "significantly declining power generation efficiency," thus providing crucial input for subsequent comprehensive economic and environmental assessments.

[0072] Furthermore, this step enables dynamic modeling of the equipment performance degradation process, providing a data foundation for subsequent multi-criteria decision analysis (MCDA) and optimal decommissioning time window prediction. By introducing an environmental corrosion index, the model not only considers the aging process of the equipment itself but also incorporates the long-term impact of the external environment on equipment performance, thereby improving the comprehensiveness and accuracy of the prediction and providing a scientific basis for the full life cycle management of wind power assets.

[0073] S42, the economic benefit prediction model adopts the Long Short-Term Memory (LSTM) network algorithm, with real-time electricity price fluctuations, spare parts price index and environmental cost discount factor as input features, and outputs time series prediction of future net income.

[0074] Specifically, the economic benefit prediction model employs a Long Short-Term Memory (LSTM) network algorithm, using real-time electricity price fluctuations, spare parts price indices, and environmental cost discount factors as input features to output a time series prediction of future net income. This step is one of the core functions of the decision analysis and output layer in the "Intelligent Evaluation System for Wind Power Equipment Decommissioning Decisions Based on Multi-Dimensional Environmental Impact Factors" of this invention, aiming to achieve dynamic modeling and prediction of the future economic benefits of wind power equipment, thereby providing a key basis for intelligent decision-making on decommissioning timing.

[0075] Furthermore, the LSTM model, as a special type of recurrent neural network (RNN), possesses the ability to capture long-term dependencies, making it particularly suitable for time series forecasting tasks. The model's input layer receives three key features: real-time electricity price fluctuations (unit: yuan / kWh), spare parts price indices (normalized based on industry standards such as the CPI-energy index), and environmental cost discount factors (set according to national environmental policies and NPV calculation standards, typically ranging from 0.05 to 0.10). After standardization, the input data is fed into the LSTM network for training and prediction. The model structure typically contains 2-4 layers of LSTM units, with each layer containing 64-256 nodes, adjusted according to data dimensionality and prediction accuracy. The output layer is a fully connected layer, outputting the net income time series (unit: ten thousand yuan / year) for the next 1-5 years.

[0076] Furthermore, the model training employs the Adam optimizer with a learning rate of 0.001–0.005 and a loss function of mean squared error (MSE) to ensure the stability and accuracy of the prediction results. The time window length for the input features is typically set to 36–72 months to fully reflect the cyclical and trending nature of electricity price fluctuations, spare parts cost changes, and environmental cost accumulation. The prediction step size can be set from 1 to 12 months according to actual needs to support short-term and medium-term economic decision-making.

[0077] Furthermore, this model is deployed in the intelligent decision-making system of wind farms, working in conjunction with equipment performance degradation models and environmental impact cumulative index models. By accessing real-time grid electricity price data, supply chain spare parts price changes, and environmental cost parameters, the system can dynamically update future net income forecasts, providing a quantitative economic basis for "continued operation" or "immediate retirement."

[0078] Furthermore, this step enables high-precision prediction of the future economic benefits of wind power equipment, solving the problems of lagging and static economic assessment in traditional methods. By modeling multi-dimensional time-series data using the LSTM algorithm, the system can identify the inflection points of revenue at different operating stages of the equipment, thereby helping to determine the optimal retirement time window and improving the comprehensive economic and environmental benefits throughout the asset's entire life cycle.

[0079] The wind power equipment optimal decommissioning time window prediction method of this invention realizes dynamic quantification and multi-dimensional comprehensive evaluation of the environmental impact of wind power equipment decommissioning decisions, improves the scientificity and foresight of decision-making, accurately predicts the optimal decommissioning time window, and optimizes economic and environmental benefits.

[0080] Furthermore, it also includes: S5 generates a visual decision-making report. It displays a real-time comparison of equipment health, economic benefit curves, and EII change trends through a 3D dynamic dashboard, and outputs a PDF assessment report that includes decommissioning time windows, recommended disposal plans, and risk warnings.

[0081] This step, "Generating a Visualized Decision Report," focuses on the real-time comparison and display of the health status, economic benefit curve, and Environmental Impact Accumulation Index (EII) of wind power equipment through a 3D dynamic dashboard. It outputs a PDF assessment report containing decommissioning time windows, recommended disposal plans, and risk warnings. This step is a crucial output component in the intelligent assessment system of this invention, enabling multi-dimensional decision support.

[0082] Furthermore, the 3D dynamic dashboard employs WebGL or Three.js graphics rendering technology, combined with real-time data streaming interfaces (such as WebSocket or MQTT protocols), to achieve synchronous visualization of equipment health (such as Remaining Useful Life (RUL) and critical component failure rates), economic benefits (such as power generation revenue, operation and maintenance costs, and Net Present Value (NPV)), and EII trends. In the dashboard, health is represented by a color gradient (such as green to red), the economic benefit curve uses a time series graph to show the dynamic changes in future revenue and costs, and EII reflects the cumulative process of environmental load through area maps or heat maps. All data is updated every 15 minutes to ensure the timeliness of decision-making information.

[0083] Furthermore, when generating PDF reports, the system uses a template engine to structure and output key data from the dashboard, predictive model outputs, and risk warning information. The report content includes: the decommissioning time window (e.g., Q2 2028 to Q1 2029), recommended disposal plans (e.g., "large-to-small" renovation, overall sale, dismantling and recycling, etc.) and their corresponding economic and environmental benefit assessments, as well as warning prompts when EII exceeds the threshold (e.g., EII>80).

[0084] Furthermore, this step can be deployed in the wind farm's central control room or remote operation and maintenance platform for use by asset managers, environmental compliance departments, and decision-makers. Through this report, users can quickly obtain comprehensive assessment results throughout the equipment's entire lifecycle, assisting in the development of scientific, economical, and environmentally friendly decommissioning strategies.

[0085] Furthermore, this step enables the unified presentation and output of multi-source heterogeneous data, enhances the visualization and traceability of decision-making, and provides a standardized and reusable assessment tool for the intelligent management of wind power assets, which has significant engineering practical value and environmental benefits.

[0086] Example 2 This invention proposes a device for predicting the optimal decommissioning time window for wind power equipment. The device, as described in this invention, enables dynamic quantitative assessment of the environmental impact throughout the entire lifecycle of wind power equipment. By fusing data from three dimensions—technical, economic, and environmental—and incorporating machine learning predictions, it accurately identifies the optimal decommissioning time window, thereby improving the scientific rigor and environmental benefits of decommissioning decisions.

[0087] In one embodiment of the present invention, dynamic quantification and early warning of environmental impact are achieved: An innovative "environmental impact accumulation index" algorithm quantifies and accumulates the continuous environmental impact during wind turbine operation. When the index exceeds a preset threshold, an early warning can be issued, making environmental costs visible and manageable, and empowering comprehensive and forward-looking scientific decision-making. The system integrates three dimensions: equipment technical status, economic benefits, and dynamic environmental impact, providing decision-makers with a clear, data-driven cost-benefit comparison platform. It can intuitively compare the comprehensive advantages and disadvantages of different retirement time points, thereby making more scientific decisions and accurately predicting the optimal retirement window: Using machine learning models, the system predicts the changing trends of equipment performance degradation curves, economics, and the environmental impact accumulation index, thereby accurately locking in an "optimal retirement time window." Retirement within this window maximizes economic and environmental benefits, improving the level of intelligent asset management and environmental benefits. This elevates wind farm asset management from a passive "repair when it breaks, retire when it expires" model to a proactive, predictive, and environmentally responsible intelligent management system, ultimately maximizing the full life-cycle value of wind power assets and reducing their potential negative environmental impacts.

[0088] This invention systematically solves the shortcomings and problems existing in the prior art through system architecture and core algorithms: In one embodiment of the present invention, addressing the static nature of environmental impact assessment, a multi-dimensional environmental impact factor real-time monitoring module is designed. This module can connect to existing or added sensors in the wind farm and, combined with the equipment's own operational data, uses an environmental impact accumulation index (EII) algorithm to weight, quantify, and accumulate this discrete, multi-dimensional data. This transforms the originally static, one-off LCA assessment into a dynamic, real-time environmental load assessment specific to the equipment and site, solving the problems of assessment lag and non-dynamic nature. The detailed calculation process of the Environmental Impact Accumulation Index (EII) involves the following steps to achieve dynamic quantification of environmental impact: Data standardization and deviation calculation: For each environmental factor n (e.g., noise, pollutant concentration), actual... The time monitoring value is recorded as Set its compliance benchmark value. Calculate the instantaneous deviation: .

[0089] Instantaneous environmental impact index Calculation: Assign weights to each factor that reflect the degree of harm. ( ), calculate the weighted deviation: .

[0090] Cumulative Index (EII) calculation: Integrating the instantaneous impact over time yields the result from equipment commissioning ( ) to the present ( Cumulative environmental load:

[0091] In discrete systems, the calculation is as follows: .

[0092] In one embodiment of the present invention, addressing the problem of fragmented decision-making dimensions, the present invention constructs a multi-source data fusion decision algorithm architecture. This architecture, as the core of the system, is capable of simultaneously receiving and processing data streams from three different dimensions: Furthermore, in terms of technical dimensions: equipment performance degradation curves, real-time operating conditions, and failure rate data; in terms of economic dimensions: real-time electricity prices, operation and maintenance costs, spare parts prices, decommissioning costs, and forecasts of upgrade and renovation benefits; and in terms of environmental dimensions: the real-time calculated value of the "cumulative environmental impact index".

[0093] The algorithm architecture, through normalization and a multi-criteria decision analysis (MCDA) model, integrates data from these three dimensions into a unified evaluation framework, outputting a comprehensive evaluation score. This allows decision-makers to clearly see the advantages and disadvantages of different decisions across each dimension, resolving the problem of fragmented decision-making dimensions. The normalization and MCDA process further involves data normalization: mapping the indicators from the three dimensions of technology, economy, and environment to a unified standard. Range. For positive indicators (such as power generation efficiency), use... Negative indicators (such as cost, EII) are adopted The Comprehensive Performance Index (MCDA) is calculated using a weighted sum model. The weights for technology, economy, and environment are as follows: ( ), then a certain decision scheme The overall score is: .

[0094] In one embodiment of the present invention, to address the limitation of predictive capability, the present invention integrates a machine learning-based decommissioning time prediction model. This model uses historical operating data, attenuation data of similar units, and economic and environmental parameters as training sets to learn and construct the following three core sub-models: This invention uses a machine learning regression algorithm to train the following model: Performance degradation prediction model: Predicts the future decline in equipment power generation efficiency. Processing: Based on operating time. As input, predict future power generation efficiency .expression: .

[0095] Economic benefit prediction model: Predicts the future upward trend of operation and maintenance costs and the downward trend of power generation revenue. Processing: Input , and time Forecast net income Operation and maintenance costs With equipment aging and Increase and rise. Expression:

[0096] Environmental Impact Cumulative Prediction Model: Predicts the future growth rate of the "Cumulative Environmental Impact Index". Processing: Predicts future instantaneous impacts. And by integrating, we obtain .expression: .

[0097] Furthermore, by coupling the results of these three prediction models, the system can deduce the curve of comprehensive benefits (power generation revenue - operation and maintenance costs - environmental costs) over a future period and automatically identify the interval where the curve shifts from its peak to a decline, i.e., the "optimal decommissioning time window," thus achieving a leap from "failure prediction" to "optimal decommissioning timing prediction." The coupling analysis establishes a full life-cycle comprehensive net present value (NPV) model. For any future decommissioning time point... Calculate its comprehensive benefit function :

[0098] in, For environmental costs (by Incremental conversion), To cover the costs of decommissioning and disposal, Let be the discount rate. The system uses an optimization algorithm to find the discount rate. The largest This is known as the "optimal retirement time window".

[0099] This invention proposes an intelligent evaluation system for wind power equipment decommissioning decisions based on multidimensional environmental impact factors. Figure 2 The system architecture diagram according to an embodiment of the present invention is as follows: Figure 2 As shown, it mainly includes a data acquisition layer 100, a data processing and model layer 200, and a decision analysis and output layer 300.

[0100] For example, data acquisition layer 100: This layer is responsible for collecting the raw data needed for decision-making from multiple sources. It includes: Equipment Operation Database 110: Real-time data is collected or obtained from the wind farm's SCADA (Supervisory and Data Acquisition) system and CMS (Condition Monitoring System), including but not limited to: wind turbine power curves, power generation, equipment utilization rate, component vibration, temperature, speed, lubricating oil / coolant consumption and replacement records, and fault records. This data reflects the equipment's technical performance and health status.

[0101] Economic Parameters Database 120: Stores and updates various types of data related to economic analysis, such as: real-time grid electricity price, operation and maintenance labor cost, spare parts price, scrap material recycling price, new equipment investment cost, and land lease fee.

[0102] Environmental Factor Monitoring Module 130: This is one of the key data sources of this invention. It monitors environmental indicators related to wind turbine operation in real time through sensors deployed at specific locations within the wind farm (such as around the turbine base and downstream waterways). These include: soil pH and heavy metal content (for monitoring oil spills), water pollutant concentrations, site boundary noise levels, and monitoring of surrounding ecological activities (such as bird activity). It also records the environmental behavior of the equipment itself, such as the amount of waste generated.

[0103] For example, the data processing and modeling layer 200: This layer is the "brain" of the system, responsible for processing, analyzing, and modeling the collected data. It includes: Data preprocessing and fusion module 210: Cleans, denoises, unifies the format and aligns the time of heterogeneous data from different sources, providing high-quality input data for subsequent algorithm models.

[0104] Equipment performance degradation model 220: Using historical operating data 110, a unique performance degradation curve model for each wind turbine is established through time series analysis and regression analysis. This model can quantitatively describe the downward trend of power generation efficiency over time.

[0105] Environmental Impact Cumulative Index Algorithm 230: This algorithm first sets baseline values ​​and weights (Wn) for different environmental impact factors (such as soil pollution risk, duration of noise exceeding standards, water pollution risk, and solid waste generation). The weights can be adjusted according to national environmental protection standards and the environmental sensitivity of the site.

[0106] The algorithm receives environmental factor monitoring data 130 in real time and calculates the deviation (Dn) of each factor relative to the baseline value. The Environmental Impact Cumulative Index (EII) is calculated as follows: EII = ∫[Σ(Wn*Dn(t))]dt, which is the integral of the deviation of all weighted environmental factors over time. This index dynamically and quantitatively reflects the cumulative environmental load caused by the equipment from the start of operation to the present moment. This algorithm solves the problem that environmental impact cannot be dynamically quantified in existing technologies.

[0107] Machine learning-based decommissioning time prediction model 240: This model, trained using historical data, can predict the future performance degradation trend of equipment (based on model 220), the upward trend of operation and maintenance costs (based on data 110 and 120), and the growth trend of the Environmental Impact Accumulation Index (EII) (based on model 230) over a future period. By comprehensively analyzing these three prediction curves, the model can predict the future comprehensive benefit curve, thereby determining an optimal decommissioning time window. This solves the problem that existing technologies cannot proactively predict the optimal decommissioning timing.

[0108] For example, the decision analysis and output layer 300 is responsible for integrating the analysis results of the model layer and presenting them to the user in a user-friendly manner to support the final decision.

[0109] Multi-dimensional Decision Analysis Engine 310: Another core component of this invention. This engine constructs a dynamic cost-benefit comparison model. For the "continue operation" option, it calculates in real time: expected power generation revenue - expected operation and maintenance costs - expected additional environmental impact costs (converted from EII increments); for the "immediate decommissioning" option, it calculates: equipment residual value recovery revenue - dismantling and disposal costs - land restoration costs. This engine unifies the three separate dimensions of technical status, economic benefits, and environmental impact into a comparable comprehensive cost-benefit framework, solving the problem of single-dimensional decision-making in existing technologies.

[0110] The visualization and report generation module 320 displays the real-time health status, performance degradation curve, cumulative environmental impact index (EII) and its changing trend of the equipment through dashboards and charts. It clearly provides the optimal decommissioning time window recommended by the system, and then provides recommendations for different disposal schemes and their corresponding economic and environmental benefit assessment reports.

[0111] Example 3 To achieve the above embodiments, such as Figure 3 As shown, this embodiment also provides a wind power equipment optimal decommissioning time window prediction device 10, including: The Environmental Factor Monitoring Module 1000 is used to collect environmental monitoring data in real time, including soil pollution risk, water pollution risk, duration of noise exceeding standards, and amount of solid waste generated, by deploying environmental factor monitoring modules in wind farms, and to obtain wind turbine operation data and economic parameter data. The Environmental Impact Cumulative Index Calculation Module 2000 is used to calculate the Environmental Impact Cumulative Index EII based on the environmental monitoring data. The EII is generated by the weighted integral formula EII=∫[Σ(Wn*Dn(t))] dt. The multi-dimensional decision analysis module 3000 is used to input the operating data, economic parameter data and EII value into the multi-dimensional decision analysis engine, and generate a comprehensive evaluation score through normalization processing and multi-standard decision analysis model. The comprehensive evaluation score includes a quantitative comparison of technical performance degradation, net economic benefits and environmental costs. The machine learning prediction and optimization module 4000 is used to use machine learning models to predict the operating data, economic parameter data and EII change trends, construct a comprehensive benefit function and optimize to determine the optimal retirement time window that maximizes the comprehensive net present value over the entire life cycle. The Visual Decision Report Generation Module 5000 is used to generate visual decision reports. It displays real-time comparisons of equipment health, economic benefit curves, and EII change trends through a 3D dynamic dashboard, and outputs a PDF evaluation report that includes decommissioning time windows, recommended disposal plans, and risk warnings.

[0112] Furthermore, the environmental factor monitoring module 1000 is also used for: Soil pollution risk monitoring employs a multispectral soil sensor array to detect the spatiotemporal distribution characteristics of three parameters: soil pH, heavy metal ion concentration, and organic matter content. The solid waste generation monitoring includes a blade material aging prediction module, which uses infrared spectroscopy analysis and mechanical performance test data to predict the remaining usable years of the blades and the potential amount of microplastics released.

[0113] Furthermore, the environmental impact cumulative index calculation module 2000 is also used to: standardize the environmental monitoring data, and when calculating the instantaneous deviation degree Dn(t) of each environmental factor, adopt a dynamic benchmark value adjustment algorithm to dynamically update the benchmark value according to the historical environmental data of the wind farm and real-time meteorological conditions; The weight Wn in the weighted integral formula is assigned using a multilayer perceptron model. By training the correlation between historical environmental impact events and equipment operation data, dynamic weight coefficients of each environmental factor are automatically generated.

[0114] Furthermore, the multi-dimensional decision analysis module 3000 is also used to: map the positive power generation efficiency index to the [0,1] interval using the positive index normalization formula, and map the operation and maintenance cost and EII negative index to the [0,1] interval using the negative index normalization formula; In the multi-criteria decision analysis model, the weight allocation of the technology, economy, and environment dimensions adopts the Monte Carlo simulation method. 1000 weight combinations are generated by random sampling, and the optimal weight configuration that minimizes the variance of the comprehensive evaluation score is selected.

[0115] Furthermore, the machine learning prediction and optimization module 4000 is also used for: the performance degradation prediction model adopts the gradient boosting tree algorithm, takes equipment running time, component wear data and environmental corrosion index as input features, and outputs the degradation curve of future power generation efficiency; The economic benefit prediction model uses a long short-term memory network algorithm, taking real-time electricity price fluctuations, spare parts price index, and environmental cost discount factor as input features, and outputs a time series prediction of future net income.

[0116] Furthermore, the visualization decision report generation module 5000 is also used for: The 3D dynamic dashboard display module is used to display real-time comparisons of equipment health, economic benefit curves, and EII change trends through a 3D dynamic dashboard. The PDF assessment report generation module is used to output PDF assessment reports that include decommissioning time windows, recommended disposal plans, and risk warnings.

[0117] The wind power equipment optimal decommissioning time window prediction method of this invention realizes dynamic quantitative assessment of the environmental impact of wind power equipment throughout its entire life cycle, and accurately identifies the optimal decommissioning time window through the fusion of technical, economic and environmental data and machine learning prediction, thereby improving the scientific nature and environmental benefits of decommissioning decisions.

[0118] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0119] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for predicting an optimal retirement time window of a wind power device, characterized in that, Comprise: S1, real-time collection of soil pollution risk, water pollution risk, noise exceeding time and solid waste generation amount environmental monitoring data by environmental factor monitoring module deployed in wind farm, and acquisition of operation data and economic parameter data of wind turbine; S2, calculation of environmental impact cumulative index EII based on the environmental monitoring data, the EII being generated by a weighted integral formula EII=∫[Σ(Wn*Dn(t))]dt; S3, input of operation data, economic parameter data and EII value into a multi-dimensional decision analysis engine, generation of a comprehensive evaluation score by normalization processing and a multi-criteria decision analysis model, the comprehensive evaluation score including quantitative comparison of technical performance attenuation, economic net benefit and environmental cost; S4, prediction of operation data, economic parameter data and EII trend by a machine learning model, construction of a comprehensive benefit function and optimization to determine an optimal retirement time window that maximizes the full life cycle comprehensive net present value.

2. The method of claim 1, wherein, The S1 further comprises: S11, soil pollution risk monitoring using a multi-spectral soil sensor array to detect the spatio-temporal distribution characteristics of three parameters of soil pH value, heavy metal ion concentration and organic matter content; S12, solid waste generation amount monitoring including a blade material aging prediction module to predict blade remaining service life and potential microplastic release amount through infrared spectrum analysis and mechanical property test data.

3. The method of claim 1, wherein, The S2 further comprises: S21, when calculating the instantaneous deviation Dn(t) of each environmental factor, a dynamic reference value adjustment algorithm is used to dynamically update the reference value according to historical environmental data of the wind farm and real-time weather conditions; S22, the distribution of weights Wn in the weighted integral formula uses a multi-layer perception machine MLP model to automatically generate dynamic weight coefficients of each environmental factor by training the correlation between historical environmental impact events and equipment operation data.

4. The method of claim 1, wherein, The S3 further comprises: S31, using a positive index normalization formula to map the positive index of power generation efficiency to the [0, 1] interval, and using a negative index normalization formula to map the operation and maintenance cost and EII negative index to the [0, 1] interval; S32, the weight distribution of technology, economy and environment dimensions in the multi-criteria decision analysis model uses a Monte Carlo simulation method to generate 1000 groups of weight combinations by random sampling, and selects the optimal weight configuration that minimizes the variance of the comprehensive evaluation score.

5. The method of claim 1, wherein, The machine learning model includes a performance attenuation prediction model and an economic benefit prediction model, and further comprises: S41, the performance attenuation prediction model uses a gradient boosting tree algorithm, with device operation time, component wear data and environmental corrosivity index as input features, and outputs a future power generation efficiency attenuation curve; S42, the economic benefit prediction model uses a long short-term memory network algorithm, with real-time electricity price fluctuation, spare parts price index and environmental cost discount factor as input features, and outputs a time series prediction of future net benefit.

6. The method of claim 1, wherein, Further comprise: S5, generating a visual decision report step, through a three-dimensional dynamic dashboard to show the real-time comparison of equipment health, economic benefit curve, EII change trend, and output a PDF format evaluation report containing the retirement time window, disposal scheme recommendation and risk warning.

7. An optimal decommissioning time window prediction device for a wind power plant, characterized in that, Comprise: An environmental factor monitoring module for collecting environmental monitoring data of soil pollution risk, water pollution risk, noise exceeding time and solid waste generation amount in real time through the environmental factor monitoring module deployed in the wind farm, and obtaining operation data and economic parameter data of the wind turbine generator; An environmental impact cumulative index calculation module for calculating an environmental impact cumulative index EII based on the environmental monitoring data, the EII being generated by a weighted integral formula EII = ∫[Σ(Wn*Dn(t))]dt; A multi-dimensional decision analysis module for inputting the operation data, economic parameter data and EII value into a multi-dimensional decision analysis engine, generating a comprehensive evaluation score through normalization processing and a multi-criteria decision analysis model, the comprehensive evaluation score including quantitative comparison of technical performance degradation, economic net benefit and environmental cost; A machine learning prediction and optimization module for predicting the operation data, economic parameter data and EII change trend using a machine learning model, constructing a comprehensive benefit function and determining the optimal retirement time window that maximizes the overall life cycle comprehensive net present value; A visual decision report generation module for generating a visual decision report, showing the real-time comparison of equipment health, economic benefit curve, EII change trend through a three-dimensional dynamic dashboard, and outputting a PDF format evaluation report containing the retirement time window, disposal scheme recommendation and risk warning.

8. The apparatus of claim 7, wherein, The environmental factor monitoring module is also used for: Soil pollution risk monitoring uses a multi-spectral soil sensor array to detect the spatio-temporal distribution characteristics of three parameters of soil pH value, heavy metal ion concentration and organic matter content; Solid waste generation monitoring includes a blade material aging prediction module that predicts the remaining usable life of the blade and the potential microplastic release amount through infrared spectrum analysis and mechanical property test data.

9. The apparatus of claim 7, wherein, The environmental impact cumulative index calculation module is also used for: When calculating the instantaneous deviation Dn(t) of each environmental factor, a dynamic reference value adjustment algorithm is used to dynamically update the reference value according to historical environmental data of the wind farm and real-time weather conditions; The allocation of weights Wn in the weighted integral formula uses a multi-layer perceptron model to automatically generate dynamic weight coefficients for each environmental factor by training the correlation between historical environmental impact events and device operation data.

10. The apparatus of claim 7, wherein, The multi-dimensional decision analysis module is also used for: A positive index normalization formula is used to map the positive index of power generation efficiency to the [0, 1] interval, and a negative index normalization formula is used to map the operation and maintenance cost and EII negative index to the [0, 1] interval; The weight distribution of the technology, economy and environment dimensions in the multi-criteria decision analysis model uses a Monte Carlo simulation method to generate 1000 groups of weight combinations through random sampling, and selects the optimal weight configuration that minimizes the variance of the comprehensive evaluation score.

11. The apparatus of claim 7, wherein, The machine learning prediction and optimization module is also used for: The performance degradation prediction model adopts a gradient boosting tree algorithm, takes device running time, component wear data, and environmental corrosion index as input features, and outputs a future power generation efficiency degradation curve; The economic benefit prediction model adopts a long short-term memory network algorithm, takes real-time electricity price fluctuations, spare parts price index, and environmental cost discount coefficient as input features, and outputs a future net income time series prediction.

12. The apparatus of claim 7, wherein, Also includes: A three-dimensional dynamic dashboard display module for displaying the health of the device, the economic benefit curve, and the real-time comparison of the EII change trend through a three-dimensional dynamic dashboard; A PDF format evaluation report generation module for outputting a PDF format evaluation report containing the retirement time window, disposal scheme recommendation, and risk warning.

Images