Water-wind-solar complementary system unit state evaluation method based on cloud model

Abstract

The invention provides a water-wind-light complementary system unit state evaluation method based on a cloud model. Acquiring operation data; key features reflecting the running state of the unit are extracted from the high-quality data set, and a multi-dimensional unit state evaluation index system is constructed; carrying out multi-method weight fusion on the feature indexes by adopting a game theory combination weighting method, solving an optimal linear combination by optimizing a weight coefficient, and carrying out normalization processing to obtain a final combination weight of each feature index; constructing a cloud model, and generating a cloud picture by using a forward cloud generator; and according to the association degree between the cloud model and the unit operation state, determining the current state grade of the unit, and feeding back an evaluation result to a system operation end. The method can achieve the dynamic and accurate evaluation of the operation states of multiple types of units in the water-wind-light complementary system, effectively solves the evaluation problems caused by the complexity of a system structure, the fluctuation of energy and the variability of the environment, and improves the operation reliability and maintenance decision efficiency of the system.

Description

Technical Field

[0001] This invention relates to the field of condition assessment of hydro-wind-solar hybrid system units, and more particularly to a method for condition assessment of hydro-wind-solar hybrid system units based on a cloud model. Background Technology

[0002] Hydro-wind-solar hybrid systems, as an important technological form for the integrated development of renewable energy, are widely used in areas with high requirements for power supply stability. Related technologies utilize the coordinated operation of hydropower units, wind turbines, and photovoltaic arrays to construct a multi-energy complementary power generation system. Specifically, this system covers the entire process from energy harvesting and power regulation to grid connection control, including key aspects such as data monitoring, status analysis, and coordinated dispatch. With the development of clean energy technologies, hydro-wind-solar hybrid systems have shown significant advantages in improving energy utilization and grid adaptability; however, their operational status assessment still relies on traditional methods, making it difficult to meet the dynamic demands of complex operating environments.

[0003] However, existing unit condition assessment methods directly employ single-weight allocation or fixed-threshold judgments, failing to fully consider the coupling relationships between multi-source heterogeneous data and the dynamic changes in the operating environment. This may lead to lagging assessment results, high misjudgment rates, or an inability to accurately reflect the true health status of the unit under different operating conditions, thereby affecting the system's operational efficiency and the scientific nature of maintenance decisions. Specifically, hydropower units rely on head and flow rate, wind turbines on wind speed and pitch angle, and photovoltaic arrays on solar intensity and temperature, resulting in significant differences in data dimensions and making it difficult to establish a unified assessment model. Furthermore, traditional methods are significantly inadequate in handling fuzziness and randomness, making them ill-suited to the highly volatile and environmentally coupled operating characteristics of renewable energy. Summary of the Invention

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

[0005] Therefore, the first objective of this invention is to propose a method for assessing the status of hydro-wind-solar hybrid systems based on a cloud model.

[0006] The second objective of this invention is to propose a cloud-based hydro-wind-solar hybrid system unit status assessment device.

[0007] To achieve the above objectives, a first aspect of the present invention proposes a method for assessing the status of a hydro-wind-solar hybrid system based on a cloud model, comprising:

[0008] S1. Collect the operating data of hydropower units, wind power units and photovoltaic arrays, and clean, filter and normalize the operating data to obtain a high-quality dataset in a unified format.

[0009] S2, extract key features reflecting the unit's operating status from the high-quality dataset, and use feature selection algorithms to screen out feature indicators that have a significant impact on status assessment, and construct a multi-dimensional unit status assessment indicator system.

[0010] S3. A game theory-based combination weighting method is used to fuse the weights of the feature indicators using multiple methods. The optimal linear combination is solved by optimizing the weight coefficients, and the final combination weights of each feature indicator are obtained by normalization.

[0011] S4. Based on the final combined weights, construct a cloud model, calculate the expectation, entropy and hyperentropy of each state level, and use a positive cloud generator to generate a cloud map to achieve a fuzzy-random two-dimensional mapping of the unit's operating state.

[0012] S5. Based on the correlation between the cloud model and the unit's operating status, determine the current status level of the unit and feed the evaluation results back to the system operation terminal to trigger the online learning mechanism and dynamically update the cloud model parameters and feature weights.

[0013] In one embodiment of the present invention, the step of collecting operational data from hydropower units, wind power units, and photovoltaic arrays, and cleaning, filtering, and normalizing the operational data to obtain a high-quality dataset in a uniform format, further includes:

[0014] S11 uses the Z-score standardization method to normalize the head, flow rate, speed and power of the hydropower unit in order to eliminate the influence of different dimensions and magnitudes.

[0015] S12 performs wavelet filtering on the wind speed, wind direction, pitch angle, and power of the wind turbine to remove high-frequency noise and retain key operating characteristics; and processes the data of the photovoltaic array, including light intensity, temperature, voltage, and current.

[0016] In one embodiment of the present invention, the step of extracting key features reflecting the unit's operating status from the high-quality dataset, and using a feature selection algorithm to screen out feature indicators that have a significant impact on status assessment, thereby constructing a multi-dimensional unit status assessment indicator system, further includes:

[0017] S21, extract the vibration frequency, temperature change rate and power fluctuation coefficient of the hydropower unit as key features reflecting its operating status;

[0018] S22 uses an information entropy-based feature selection algorithm to filter the features of wind turbines and photovoltaic arrays, retaining feature indicators that are highly sensitive to state changes.

[0019] In one embodiment of the present invention, the step of employing a game theory-based combinatorial weighting method to perform multi-method weight fusion on the feature indicators, solving for the optimal linear combination by optimizing the weight coefficients, and obtaining the final combined weights of each feature indicator through normalization processing further includes:

[0020] S31, using principal component analysis, analytic hierarchy process and entropy weight method as three basic weight calculation methods, construct a basic weight vector set;

[0021] S32 constructs the optimal linear combination based on the basic weight vector set and solves for the final combination weights.

[0022] In one embodiment of the present invention, the step of constructing a cloud model based on the final combined weights, calculating the expectation, entropy, and hyperentropy of each state level, and generating a cloud map using a forward cloud generator to achieve a fuzzy-random two-dimensional mapping of the unit's operating state further includes:

[0023] S41, Calculate the expectation of the cloud model based on the weighted scores of each feature indicator;

[0024] S42 uses a normal distribution cloud generator to generate cloud maps, and the distribution density and dispersion of cloud droplets intuitively reflect the fuzziness and randomness of the unit's status.

[0025] In one embodiment of the present invention, it further includes:

[0026] S6 generates maintenance recommendations based on the unit's current status level and historical operating data, and automatically pushes them to the operation and maintenance management system to achieve intelligent maintenance scheduling based on status assessment results.

[0027] To achieve the above objectives, a second aspect of the present invention provides a cloud-model-based hydro-wind-solar hybrid system unit condition assessment device, comprising:

[0028] The data acquisition and preprocessing module is used to collect the operating data of hydropower units, wind power units and photovoltaic arrays, and to clean, filter and normalize the operating data to obtain a high-quality dataset in a unified format.

[0029] The feature extraction and filtering module is used to extract key features reflecting the unit's operating status from the high-quality dataset, and to use a feature selection algorithm to filter out feature indicators that have a significant impact on the status assessment, thereby constructing a multi-dimensional unit status assessment indicator system.

[0030] The game theory weight fusion module is used to perform multi-method weight fusion on the feature indicators using a game theory combinatorial weighting method. It solves the optimal linear combination by optimizing the weight coefficients and normalizes the result to obtain the final combined weight of each feature indicator.

[0031] The cloud model construction and mapping module is used to construct a cloud model based on the final combined weights, calculate the expectation, entropy and hyperentropy of each state level, and generate a cloud map using a positive cloud generator to achieve a fuzzy-random two-dimensional mapping of the unit's operating state.

[0032] The status assessment and model update module is used to determine the current status level of the unit based on the correlation between the cloud model and the unit's operating status, and to feed the assessment results back to the system operation terminal to trigger the online learning mechanism, dynamically update the cloud model parameters and feature weights, and improve the adaptability and accuracy of the assessment model.

[0033] The methods and apparatus of this invention can achieve dynamic and accurate assessment of the operating status of multiple types of units in a hydro-wind-solar hybrid system, effectively address the assessment challenges caused by data heterogeneity and environmental uncertainty, and improve the reliability of system operation and the scientific nature of maintenance decisions.

[0034] 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

[0035] 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:

[0036] Figure 1 This is a flowchart of a cloud-based hydro-wind-solar hybrid system unit status assessment method according to an embodiment of the present invention;

[0037] Figure 2 This is a system architecture diagram of a cloud-based hydro-wind-solar hybrid system unit status assessment system according to an embodiment of the present invention;

[0038] Figure 3 This is an evaluation result diagram of a water-wind-solar hybrid system according to an embodiment of the present invention;

[0039] Figure 4 This is a diagram showing the subsystem evaluation results according to an embodiment of the present invention;

[0040] Figure 5 This is a structural diagram of a cloud-based hydro-wind-solar hybrid system unit condition assessment device according to an embodiment of the present invention. Detailed Implementation

[0041] 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.

[0042] 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.

[0043] The following description, with reference to the accompanying drawings, describes a method and apparatus for assessing the status of a hydro-wind-solar hybrid system based on a cloud model, according to an embodiment of the present invention.

[0044] Figure 1 This is a flowchart of a cloud-based hydro-wind-solar hybrid system unit condition assessment method according to an embodiment of the present invention, such as... Figure 1 and Figure 2 As shown, it includes:

[0045] S1. Collect operating data from hydropower units, wind power units, and photovoltaic arrays, and clean, filter, and normalize the operating data to obtain a high-quality dataset in a unified format.

[0046] Specifically, this step involves the acquisition and preprocessing of operational data from hydropower units, wind turbines, and photovoltaic arrays in a hydro-wind-solar hybrid system, and is a fundamental step in the entire condition assessment method. In some implementations, data acquisition utilizes SCADA systems, PLC controllers, smart sensors, and other devices to obtain key operating parameters of each unit in real time. For example, hydropower units collect data on head (in meters) and flow rate (in cubic meters per second). 3 Data collected includes: wind speed (r / min), rotational speed (r / min), and active power (MW); wind turbine data includes: hub wind speed (m / s), wind direction (°), pitch angle (°), generator speed (r / min), and output power (kW); and photovoltaic array data includes: irradiance (W / m²). 2 The data includes component temperature (°C), DC voltage (V), and current (A). The acquisition frequency is typically set from 1 second to 5 minutes to balance real-time performance with data integrity.

[0047] At the parameter level, the data cleaning process includes missing value imputation (e.g., using linear interpolation or the KNN algorithm), outlier detection (e.g., identifying data points deviating from the mean by three times the standard deviation based on the 3σ principle or the isolated forest algorithm), and data consistency verification (e.g., checking timestamp synchronization and sensor sampling frequency consistency). Filtering can use low-pass filters (cutoff frequency set to 0.1Hz to 1Hz) or wavelet transforms (e.g., db4 wavelet basis functions, with a decomposition level of 5) to remove high-frequency noise. Normalization typically uses Min-Max normalization or Z-score standardization to map the data of each unit to the [0,1] interval or a distribution with a mean of 0 and a standard deviation of 1, to eliminate dimensional differences and improve the convergence speed and stability of subsequent evaluation models.

[0048] At the application level, this step is widely used in centralized monitoring platforms, intelligent operation and maintenance systems, and status prediction modules of hydro-wind-solar hybrid systems, and is particularly suitable for complex scenarios involving the fusion of multi-source heterogeneous data. By using high-quality datasets in a unified format, reliable input is provided for subsequent feature extraction, combined weighting, and cloud model evaluation, thereby enabling dynamic and qualitative assessment of the unit's status.

[0049] From a technical perspective, this step effectively improves data quality, reduces the interference of noise and outliers on the evaluation results, enhances the robustness and generalization ability of the model, and lays a solid foundation for achieving accurate unit status identification and fault early warning.

[0050] Furthermore, S1 includes:

[0051] S11 uses the Z-score standardization method to normalize the head, flow rate, speed and power of the hydropower unit in order to eliminate the influence of different dimensions and magnitudes.

[0052] Specifically, in this invention, the Z-score standardization method is used to normalize key operating parameters of hydropower units such as head, flow rate, speed and power. Its core purpose is to eliminate the data comparability problem caused by the difference in dimensions and the inconsistency in numerical magnitude between different physical quantities, thereby providing a unified and standardized data input basis for subsequent state assessment based on cloud models.

[0053] At the application level, this step is typically deployed in the data preprocessing module of a hydro-wind-solar hybrid system as a preliminary step in the state assessment process. In actual operation, the system collects real-time operating data of the hydropower units via SCADA or PLC. After filtering and outlier removal, the data is input into the Z-score standardization algorithm for processing, providing standardized feature vectors for the cloud model.

[0054] In terms of technical effectiveness, Z-score standardization effectively improves the comparability of data and the robustness of models, providing high-quality input data for subsequent game theory combination weighting and cloud model evaluation, thereby enhancing the accuracy and reliability of unit status assessment and serving as a key technical support for realizing intelligent operation and maintenance of the system.

[0055] S12 performs wavelet filtering on the wind speed, wind direction, pitch angle, and power of the wind turbine to remove high-frequency noise and retain key operating characteristics; and processes the data of the photovoltaic array, including light intensity, temperature, voltage, and current.

[0056] Specifically, in some implementations, wavelet filtering of wind turbine speed, wind direction, pitch angle, and power is one of the key technical steps in the data preprocessing stage of this invention. This step aims to effectively remove high-frequency noise from the data by leveraging the multi-resolution analysis characteristics of wavelet transform, while retaining key feature information reflecting the unit's operating status, thus providing a high-quality data foundation for subsequent feature extraction and cloud model status assessment.

[0057] At the technical implementation level, this step employs Discrete Wavelet Transform (DWT) to perform multi-scale decomposition of the original data. Specifically, appropriate wavelet basis functions (such as db4, sym8, etc.) are selected to perform three- or four-level wavelet decomposition on the wind speed, wind direction, pitch angle, and power signals, decomposing the signals into low-frequency approximation coefficients and high-frequency detail coefficients. The high-frequency detail coefficients mainly reflect instantaneous disturbances and noise components in the signal; therefore, they are thresholded or discarded during the reconstruction process, retaining only the low-frequency approximation coefficients to reconstruct the denoised signal. In some implementations, soft or hard thresholding methods can be combined to perform nonlinear processing on the detail coefficients to further improve the filtering effect.

[0058] At the parameter level, the number of wavelet decomposition layers is usually determined based on the sampling frequency and signal characteristics, generally set to 3 to 5 layers to balance noise removal and key feature preservation. The selection of wavelet basis functions must satisfy orthogonality, symmetry, and good time-frequency localization characteristics. The db4 wavelet has good adaptability in wind turbine signal processing. Threshold calculation can use a universal threshold or an adaptive thresholding method based on Stein unbiased estimation.

[0059] In terms of application scenarios, this step is suitable for long-term operation data processing of wind turbines in complex natural environments. Especially under conditions such as sudden wind changes, sensor drift, or electromagnetic interference, it can effectively improve the signal-to-noise ratio and stability of the data, providing reliable input for subsequent health status assessment.

[0060] From a technical perspective, wavelet filtering can significantly reduce high-frequency noise interference in data, improve signal smoothness and continuity, thereby enhancing the accuracy of feature extraction and the robustness of cloud model evaluation. Similarly, wavelet filtering is also applicable to photovoltaic array data, including light intensity, temperature, voltage, and current. By reducing noise interference in this data, it can improve the accuracy of photovoltaic array performance evaluation and play an important supporting role in identifying the health status of photovoltaic systems.

[0061] Therefore, this embodiment of the invention collects operational data from each unit in a hydro-wind-solar hybrid system, including head, flow rate, rotational speed, and power of the hydropower units; wind speed, wind direction, pitch angle, and power of the wind turbines; irradiance, temperature, voltage, and current of the photovoltaic array; and grid operation data such as system frequency, voltage, and power. The collected data undergoes preprocessing operations such as cleaning, filtering, and normalization to remove outliers and noise interference, ensuring data quality and consistency and providing a reliable data foundation for subsequent condition assessment.

[0062] S2 extracts key features reflecting the unit's operating status from high-quality datasets and uses feature selection algorithms to screen out feature indicators that have a significant impact on status assessment, thus constructing a multi-dimensional unit status assessment indicator system.

[0063] Specifically, this step aims to extract key features reflecting the operational status of various types of units (hydropower units, wind turbines, and photovoltaic arrays) in a hydro-wind-solar hybrid system from the preprocessed high-quality dataset. A feature selection algorithm is then used to identify feature indicators that significantly impact the status assessment, ultimately constructing a multi-dimensional unit status assessment indicator system. In some implementations, the feature extraction process is based on time-domain, frequency-domain, and statistical feature analysis. Examples include the vibration frequency, power fluctuation rate, and efficiency change rate of hydropower units; the wind speed-power curve deviation, yaw error, and pitch angle response time of wind turbines; and the current-voltage curve slope, temperature coefficient, and maximum power point tracking efficiency of photovoltaic arrays. These features can effectively characterize the performance degradation trend and operational anomalies of the units.

[0064] Furthermore, feature selection algorithms can employ methods such as feature importance assessment based on information entropy, random forest feature scoring, or LASSO regression to reduce the dimensionality and optimize the extracted features. By calculating the correlation coefficients (e.g., Pearson correlation coefficient, Spearman rank correlation coefficient) between each feature and the unit's health status, as well as the feature contribution (e.g., game theory analysis based on Shapley values), feature indicators that have a significant impact on the status assessment are selected. Optionally, a feature selection threshold can be set, such as retaining features with a correlation coefficient greater than 0.7 or a contribution higher than 0.15, to ensure the accuracy and generalization ability of the assessment model.

[0065] This step constructs a multi-dimensional unit condition assessment index system, covering multiple dimensions such as performance, operational stability, and environmental adaptability. This provides a structured and quantifiable input foundation for subsequent game theory combinatorial weighting and cloud model evaluation. Its technical value lies in effectively improving the robustness and accuracy of the condition assessment model through scientific feature extraction and selection methods, providing data-driven decision support for system operation and maintenance.

[0066] Furthermore, S2 includes:

[0067] S21, extract the vibration frequency, temperature change rate and power fluctuation coefficient of the hydropower unit as key features reflecting its operating status.

[0068] In some implementations, vibration frequencies are typically extracted in real time by using vibration sensors (such as piezoelectric accelerometers) installed on critical components like turbine bearings, generator stators, and rotors to acquire vibration signals. By performing a Fast Fourier Transform (FFT) or Short-Time Fourier Transform (STFT) on the raw vibration signal, the main frequency components can be extracted, with particular attention paid to 1x, 2x, and other rotational frequencies and their harmonic components, as well as abnormal frequencies that may be caused by mechanical faults (such as gear meshing frequencies and shaft resonance frequencies). The vibration frequency acquisition frequency is generally set to above 1000Hz to ensure signal integrity and spectral resolution.

[0069] The extraction of the temperature change rate is based on temperature sensors installed on key components of the hydropower unit (such as bearings, stator windings, and cooling systems). The time interval for collecting temperature data is typically 1 to 5 minutes. The temperature change rate, expressed in °C / min, is obtained by performing first-order difference calculations on the continuous time-series temperature data. This parameter reflects whether the unit exhibits an overheating trend or abnormal local temperature rise during operation, and is an important basis for judging potential faults such as mechanical wear and insulation aging.

[0070] The power fluctuation coefficient is calculated based on time-series data of the hydropower unit's output power. A sliding window method (such as a 5-minute or 10-minute window) is typically used to calculate the ratio of the power's standard deviation to its mean, i.e., power fluctuation coefficient = σ_P / μ_P, where σ_P is the power standard deviation and μ_P is the power mean. This coefficient is used to quantify the stability of the unit's output power, and its value generally ranges from 0.01 to 0.1. A larger value indicates more severe power fluctuations, which may indicate abnormalities in the speed control system or unstable hydraulic conditions.

[0071] In practical applications, this step is typically deployed in the SCADA system or edge computing nodes of hydropower stations, working in conjunction with feature extraction modules for wind turbines and photovoltaic arrays to form a unified state assessment data input. By extracting the three key features mentioned above and combining them with game theory-based weighting methods, the assessment model's ability to identify the operating status of hydropower units can be effectively improved. This provides high-quality input data for subsequent cloud model evaluation, thereby achieving dynamic and qualitative assessment of the unit's health status and improving the safety and economy of system operation.

[0072] S22 uses an information entropy-based feature selection algorithm to filter the features of wind turbines and photovoltaic arrays, retaining feature indicators that are highly sensitive to state changes.

[0073] Specifically, in the feature extraction and selection stage, this invention employs an information entropy-based feature selection algorithm to screen the operational features of wind turbines and photovoltaic arrays. The aim is to retain key feature indicators that are highly sensitive to changes in the unit's state, thereby improving the accuracy and efficiency of the subsequent state assessment model. In some implementations, this algorithm first performs feature discretization processing on the preprocessed multidimensional operational data of the wind turbines and photovoltaic arrays to accommodate the computational requirements of information entropy. Specifically, the features of the wind turbines include wind speed, wind direction, pitch angle, generator speed, active power, and reactive power, while the features of the photovoltaic arrays include irradiance, module temperature, output voltage, and output current.

[0074] Information entropy, as a measure of data uncertainty, indicates that the smaller the value, the greater the information content of the feature and the higher its sensitivity to state changes. Therefore, the algorithm calculates the information entropy of each feature and filters out features with low entropy values ​​and strong discriminative ability. Furthermore, conditional entropy and joint entropy can be optionally introduced to evaluate the correlation and redundancy between features, thereby eliminating duplicate or inefficient features and achieving dimensionality reduction and optimization of the feature set. Regarding parameter settings, feature discretization typically employs the equal-width method or the equal-frequency method, dividing continuous variables into several intervals. The number of intervals is generally set to 5–10 based on the data distribution characteristics to balance computational efficiency and information retention.

[0075] This step plays a crucial role in the status assessment method for hydro-wind-solar hybrid power systems. Through a scientific feature selection mechanism, it effectively reduces the dimensionality of the model input and improves the computational efficiency and discriminative ability of the cloud model assessment. In practical applications, this algorithm can be deployed on a system operation monitoring platform. By combining real-time collected data from wind turbines and photovoltaic arrays, it dynamically updates the feature set, providing a high-quality input foundation for subsequent game theory-based weighting and cloud model evaluation. This enhances the system's ability to identify and warn of the unit's health status.

[0076] Therefore, key features reflecting the unit's operating status, such as performance indicators and the rate of change of operating parameters, are extracted from the preprocessed data. Feature selection algorithms are then used to filter and optimize the extracted features, removing redundant features and retaining only the most influential features for assessing the unit's status, thereby improving the efficiency and accuracy of the assessment model.

[0077] S3. A game theory-based combination weighting method is used to fuse the weights of the feature indicators using multiple methods. The optimal linear combination is solved by optimizing the weight coefficients, and the final combination weights of each feature indicator are obtained by normalization.

[0078] Specifically, the game-theoretic combined weighting method employed in this invention to fuse the weights of feature indicators using multiple methods is a crucial step in achieving the scientific rigor and objectivity of the hydro-wind-solar hybrid system unit condition assessment model. In some implementations, this step first assigns independent weights to n feature indicators based on Q different weight calculation methods (such as entropy weighting, principal component analysis, and analytic hierarchy process), thereby constructing a basic weight vector set. Each method reflects the importance of the indicators from different perspectives; for example, entropy weighting reflects objective weights based on the degree of data variation, while analytic hierarchy process reflects subjective weights through expert judgment.

[0079] In application scenarios, this step is suitable for the multi-source heterogeneous data fusion evaluation of hydropower units, wind power units, and photovoltaic arrays in hydro-wind-solar hybrid systems. By combining weighting methods, the advantages and disadvantages of different weighting methods can be effectively balanced, improving the sensitivity and robustness of the evaluation model to changes in unit status.

[0080] In terms of technical effectiveness, this method overcomes the subjectivity or dependence on data distribution of a single weighting method, achieves synergistic optimization of weights from multiple methods, improves the accuracy and reliability of unit status assessment, and provides a scientific and reasonable weighting basis for subsequent cloud model evaluation.

[0081] Furthermore, S3 includes:

[0082] S31 uses principal component analysis, analytic hierarchy process (AHP), and entropy weighting as three basic weight calculation methods to construct a basic weight vector set.

[0083] S32 constructs the optimal linear combination based on the basic weight vector set and solves for the final combination weights.

[0084] Specifically, in some implementations, constructing the basic weight vector set k is one of the key steps of the game theory-based combined weighting method of this invention. The technical implementation principle is to introduce three classic weight calculation methods, namely principal component analysis (PCA), analytic hierarchy process (AHP), and entropy weighting (EW), to independently assign weights to n evaluation indicators in the water-wind-solar hybrid system from the perspectives of data dimensionality reduction, expert experience judgment, and information entropy, thereby constructing a weight vector set with a multi-dimensional perspective, providing basic support for subsequent combined weighting.

[0085] The specific operation method is as follows: First, for the preprocessed unit operation data, the PCA method is used to perform linear dimensionality reduction on the indicators, extract principal components, and calculate the contribution rate of each indicator in the principal components, thereby obtaining a weight vector based on the data variance interpretation capability. Second, the AHP method is used to construct a judgment matrix and perform a consistency test (CR < 0.1), and calculate the hierarchical weight vector AHP for each indicator. This method fully integrates expert experience and subjective judgment and is suitable for scenarios where there are obvious logical relationships between indicators. Finally, the entropy weight method is used to calculate the information entropy based on the dispersion of indicator values, thereby obtaining the objective weight vector. Its core lies in reflecting the importance of indicators through information uncertainty.

[0086] Game theory combinatorial weighting:

[0087] a. Construct the basic weight vector set ω k Suppose we use Q weighting methods to assign weights to an indicator system containing n evaluation indicators based on game theory, and we can obtain the corresponding weight vector ω. k ={ω k1 ,ω k2 ,......ω kn (k = 1, 2, ..., Q). Furthermore, by performing arbitrary linear combinations of these n weight vectors, a complete set of weight vectors can be constructed:

[0088]

[0089] In the formula, α k These are the weighting coefficients.

[0090] b. Construct the optimal linear combination by optimizing the weight coefficient α. k To minimize the relationship between ω and each ω kn The deviation between them is expressed by the following formula:

[0091]

[0092] Using the differential properties of matrices, the system of linear differential equations for the optimal condition of the above equation is derived as follows:

[0093]

[0094] From the above equation, the optimal linear combination can be obtained as (α1, α2, ..., α Q ), and then normalize it:

[0095]

[0096] c. Solve for the final combined weight ω:

[0097]

[0098] Understandably, this step is widely used in the multi-index fusion process of unit condition assessment in hydro-wind-solar hybrid systems. For example, when assessing indicators such as vibration, temperature, and power of hydropower units, different weighting methods may assign different weights to the same indicator. By combining weighting methods using game theory, the advantages of each method can be integrated, improving the adaptability and stability of the assessment model. Figure 3 The image shows the evaluation results of the hydro-wind-solar hybrid system, as follows: Figure 4 The results of the subsystem evaluation are shown below.

[0099] In terms of technical effectiveness, this step effectively solves the problems of strong subjectivity and poor adaptability of a single weighting method. Through multi-method collaborative optimization, it enhances the scientificity and rationality of weight allocation, provides a solid data foundation for subsequent cloud model evaluation, and thus significantly improves the accuracy and reliability of unit status assessment.

[0100] Therefore, considering the varying importance of different indicators in unit status assessment, a game theory approach is used to combine and weight the indicators. By calculating the contribution of each indicator under different combinations, its role in the overall assessment is determined.

[0101] S4 constructs a cloud model based on the final combined weights, calculates the expectation, entropy n, and hyperentropy of each state level, and uses a positive cloud generator to generate a cloud map to achieve a fuzzy-random two-dimensional mapping of the unit's operating state.

[0102] Specifically, in some implementations, constructing a cloud model based on the final combined weights and calculating the expectation, entropy, and hyperentropy of each state level are the core steps in this invention for realizing the fuzzy-stochastic two-dimensional mapping of the unit's operating state. This step combines the indicator weights obtained by game theory combination weighting with the digital features of the cloud model to construct a state assessment system with both qualitative and quantitative characteristics, thereby more accurately reflecting the health status of the unit under complex operating environments.

[0103] At the technical implementation level, firstly, the health status of each unit is divided into several levels, such as "normal," "attention," and "abnormal," with each level corresponding to a set of digital features of the cloud model. The cloud model is an uncertainty modeling tool that unifies fuzziness and randomness. Its core lies in describing the quantitative characteristics of qualitative concepts through three parameters: expectation, entropy, and hyperentropy.

[0104] Furthermore, S4 includes:

[0105] S41, Calculate the expectation of the cloud model based on the weighted scores of each feature indicator.

[0106] Specifically, in this invention, this step is based on a weighted score of feature indicators after combination and weighting. By mapping the fuzziness and randomness of the unit's operating status through the digital feature map of the cloud model, a multi-dimensional and dynamic assessment of the health status of hydropower units, wind power units, and photovoltaic arrays can be achieved.

[0107] At the technical implementation level, firstly, each characteristic indicator x i After data preprocessing and feature selection, the final weight ω has been obtained using a game theory combinatorial weighting method. i These weights reflect the relative importance of each indicator in the unit status assessment. The weighted scoring results are then substituted into the mathematical expression of the cloud model to calculate the expectation Ex, which is the central value of all weighted characteristic indicators, used to characterize the overall trend of the unit status. Entropy En measures the dispersion of characteristic values ​​around the expectation, reflecting the fuzziness of the assessment results; hyperentropy He further describes the uncertainty of entropy values, reflecting the randomness in the assessment process.

[0108] At the parameter level, the expected value of Ex is usually between [0,1], representing the comprehensive score of the unit status; the larger the value of entropy En is, the more ambiguous the status is and the lower the credibility of the evaluation result; the reasonable setting of super-entropy He needs to be combined with the distribution characteristics of actual operating data, and its threshold range is generally determined through statistical analysis of historical data to ensure the stability and sensitivity of the cloud model.

[0109] In application scenarios, this step is embedded in the real-time status assessment process of hydro-wind-solar hybrid systems to transform multi-source heterogeneous data into semantically meaningful health status levels (such as normal, alert, and abnormal). Through cloud model visualization (such as cloud maps) and membership calculation, system operators can intuitively determine whether the unit is in a critical state, thereby providing early warnings and formulating maintenance strategies.

[0110] The technical advantage of this step lies in its effective integration of fuzziness and randomness in the assessment process through the introduction of the cloud model's three-parameter system, thereby improving the accuracy and robustness of the condition assessment. Compared to traditional single-indicator or deterministic models, the cloud model can more realistically reflect the state changes of the unit under complex operating environments, providing solid technical support for the intelligent operation and maintenance of the system.

[0111] S42 uses a normal distribution cloud generator to generate cloud maps, and the distribution density and dispersion of cloud droplets intuitively reflect the fuzziness and randomness of the unit's status.

[0112] Specifically, in this invention, using a normal distribution cloud generator to generate cloud maps is a key step in realizing a visual assessment of the fuzziness and stochasticity of the unit's status. This step is based on cloud model theory, transforming a quantitative health status score into an uncertain cloud droplet distribution, thereby intuitively reflecting the uncertain characteristics of the unit's operating status.

[0113] At the technical implementation level, a normal distribution cloud generator generates a set of cloud droplets following a normal distribution based on given digital features of the cloud model (expectation Ex, entropy En, and hyperentropy He). Specifically, expectation Ex represents the central tendency of the unit's state, entropy En reflects the fuzziness of the state, i.e., the distribution range of cloud droplets near Ex, while hyperentropy He describes the randomness of entropy, i.e., the discreteness of the cloud droplet distribution. In some implementations, the cloud droplet generation process can be represented as follows: first, a cloud droplet's expected value is randomly generated based on Ex and En; then, the entropy value of the cloud droplet is generated based on En and He; finally, the specific cloud droplet value is generated through a normal distribution function. This process can be repeated to generate multiple cloud droplets, forming a cloud map.

[0114] At the parameter level, the numerical characteristics of the cloud model need to be calculated based on the actual unit status score. For example, Ex can be taken as the mean of the unit health status score, En as the standard deviation of the score, and He as the standard deviation of En. In this invention, the health status score is usually calculated by weighting multiple operating parameters (such as power, temperature, vibration, etc.), and the weights are determined by a game theory combination weighting method. The distribution density and dispersion of cloud droplets in the cloud map can serve as a visual representation of the unit status. For example, dense cloud droplets with a small distribution range indicate a stable unit status; conversely, there may be anomalies or aging trends.

[0115] In application scenarios, this step is mainly used for real-time status assessment of hydropower units, wind turbines, and photovoltaic arrays in hydro-wind-solar hybrid systems. By displaying the assessment results in the form of a cloud map, operation and maintenance personnel can quickly identify the health level of the units and assist in formulating maintenance strategies.

[0116] The technical effect of this step is that by introducing the uncertainty expression mechanism of the cloud model, it effectively addresses the fuzziness and randomness issues in unit status assessment, improves the intuitiveness and credibility of the assessment results, and provides a scientific basis for system operation and maintenance.

[0117] Specifically, the digital features of the cloud model include expectation Ex, entropy En, and hyperentropy He. The formulas for calculating the digital parameters of the cloud model are as follows:

[0118]

[0119]

[0120] In the formula, ν i It is the health status score of the i-th group.

[0121] By combining the numerical parameters obtained from the solution with a positive cloud generator, a cloud map of the indicator is generated, thereby analyzing the health level mapped by the parameters.

[0122] Therefore, this invention establishes a cloud model that categorizes the unit's operating status into multiple levels, such as normal, alert, and abnormal. Each level is represented by a cloud droplet, and the droplet's expected value, entropy, and hyperentropy characterize the central location, fuzziness, and randomness of that level, respectively. Based on the weighted values ​​after combination, the correlation between the unit's status and the cloud droplets at each level is calculated. The current operating status level of the unit is determined according to the membership calculation principle, thus achieving a qualitative assessment of the unit's status.

[0123] In terms of application scenarios, this step is widely used in the real-time status assessment of hydropower units, wind power units, and photovoltaic arrays in hydro-wind-solar hybrid systems. By generating cloud maps through a forward cloud generator, the distribution of units at different status levels can be intuitively displayed, helping operation and maintenance personnel to identify potential fault trends and optimize maintenance strategies.

[0124] S5. Based on the correlation between the cloud model and the unit's operating status, determine the current status level of the unit and feed the evaluation results back to the system operation terminal to trigger the online learning mechanism, dynamically update the cloud model parameters and feature weights, and improve the adaptability and accuracy of the evaluation model.

[0125] Specifically, the core of this step lies in dynamically determining the current state level of the unit based on the correlation between the cloud model and the unit's operating status, and feeding the evaluation results back to the system operation terminal to trigger an online learning mechanism. This enables dynamic updates of the cloud model parameters and feature weights, thereby improving the adaptability and accuracy of the evaluation model. In some implementations, this process classifies the state level based on the degree of matching between the digital features of the cloud model (expectation Ex, entropy En, hyperentropy He) and the actual operating data. Specifically, the system first inputs the preprocessed unit operating data into the constructed cloud model, generates a cloud map through a forward cloud generator, and calculates the correlation between cloud droplets of each state level and actual data points. This is typically done using methods such as Euclidean distance, membership functions, or fuzzy similarity for quantitative evaluation.

[0126] Furthermore, the classification of status levels is typically based on preset health status thresholds. For example, the unit status is divided into three levels: "normal," "attention," and "abnormal." The cloud model parameters corresponding to each level need to be initialized based on historical operating data and expert experience. For instance, the expected value Ex for a normal state can be set to 90-100, the entropy En to 1-3, and the super-entropy He to 0.5-1.5; while the Ex for an abnormal state may be lower than 70, with En and He increasing accordingly to reflect the ambiguity and uncertainty of the status. In practical applications, this step can be deployed on the central monitoring platform or edge computing nodes of the hydro-wind-solar hybrid system, receiving real-time operating data streams from hydropower units, wind turbines, and photovoltaic arrays, and performing status assessment by combining game theory-based weighted feature weights.

[0127] The evaluation results are fed back to the system runtime via a communication interface, triggering the online learning mechanism. The online learning module, based on incremental machine learning algorithms (such as online random forests and incremental support vector machines), dynamically adjusts the Ex, En, and He parameters of the cloud model, as well as the weight coefficients of each feature, by combining newly collected operational data and evaluation feedback. For example, when the system detects that a wind turbine frequently enters an "attention" state under high wind speeds, it can automatically adjust the weights of wind speed-related features and update the entropy value of its corresponding cloud model to enhance the model's sensitivity to abnormal states. This step significantly improves the evaluation model's adaptability in complex operating environments, reduces the false positive and false negative rates, and provides scientific and real-time decision support for system operation and maintenance.

[0128] Furthermore, it also includes:

[0129] S6 generates maintenance recommendations based on the unit's current status level and historical operating data, and automatically pushes them to the operation and maintenance management system to achieve intelligent maintenance scheduling based on status assessment results.

[0130] Specifically, in some implementations, generating maintenance recommendations based on the current status level of the units and historical operating data, and automatically pushing them to the operation and maintenance management system, is a key step in achieving intelligent maintenance scheduling in this invention. This step, based on cloud model evaluation results and combined with feature weights determined by game theory combined weighting methods, dynamically analyzes and predicts the operating status of hydropower units, wind turbine units, and photovoltaic arrays, thereby generating targeted maintenance strategies.

[0131] In terms of specific technical implementation, firstly, the system collects real-time operating parameters of each unit (such as head, flow rate, speed, and power of hydropower units; wind speed, wind direction, pitch angle, and power of wind turbines; and irradiance, temperature, voltage, and current of photovoltaic arrays), and performs data fusion processing by combining these parameters with historical operating data (such as fault records, maintenance cycles, and performance degradation trends). After determining the status level, the system uses a rule-based reasoning mechanism and machine learning algorithms (such as decision trees, support vector machines, or LSTM networks) to predict the health status of the units and identify potential fault modes. Furthermore, the system generates maintenance recommendations based on preset maintenance thresholds (such as power drop rate exceeding 5% or abnormal vibration frequency duration exceeding 30 minutes), including maintenance type (such as preventive maintenance and corrective maintenance), priority (such as high, medium, and low), and recommended execution time window.

[0132] The technical benefits of this step lie in replacing traditional scheduled maintenance with a state-driven maintenance strategy, significantly reducing operation and maintenance costs and improving system availability and security. Simultaneously, its implementation based on multi-source data fusion and intelligent assessment models enhances the scientific rigor and foresight of maintenance decisions, providing solid support for building an efficient and intelligent renewable energy operation and maintenance system.

[0133] Therefore, the evaluation results are promptly fed back to system operators, providing a basis for formulating maintenance strategies. Simultaneously, based on the evaluation results and the actual operation of the system, the evaluation model is subjected to online learning and optimization, continuously updating the parameters and feature weights of the cloud model to improve its adaptability and accuracy.

[0134] This invention presents a dynamic status assessment method for hydro-wind-solar hybrid power system units based on cloud models and game theory. Through real-time analysis of massive amounts of data, it identifies and predicts potential anomalies in the system and assesses the operational status of the units. This method can evaluate the health status of hydropower units, wind turbines, and photovoltaic arrays within the hybrid system, helping the power plant to take timely maintenance or repair measures to avoid potential equipment damage and production interruptions. Furthermore, by automatically generating and pushing maintenance recommendations to the operation and maintenance management system based on the current status level and historical operating data of the units, a closed-loop linkage between status assessment and maintenance scheduling is achieved, further improving the intelligence level and response efficiency of system maintenance, effectively reducing operation and maintenance costs, and extending the service life of the units.

[0135] The beneficial effects of this invention are as follows:

[0136] Effectively collect, preprocess, and fuse massive amounts of operational data from different units and sensors to ensure data accuracy and integrity, and achieve accurate assessment of unit status.

[0137] We extract the most representative indicators for assessing unit status from complex data, and use game theory to reasonably combine and weight the features, fully considering the interaction between indicators, thereby improving the reliability and objectivity of the assessment results.

[0138] Determining the expected value, entropy, and hyperentropy of each level in the cloud model, so that it can accurately describe the different operating states of the unit, is the core of achieving accurate evaluation of the unit's status using the cloud model.

[0139] By employing game theory to assign combined weights to unit status assessment features, the interaction and influence between features are fully considered, thus improving the rationality and scientific nature of feature weight allocation. Applying cloud models to the status assessment of hydro-wind-solar hybrid systems, and using cloud models for qualitative evaluation of unit status, effectively addresses the fuzziness and randomness in the assessment process, improving the accuracy and reliability of the assessment results.

[0140] To achieve the above embodiments, such as Figure 5 As shown, this embodiment also provides a cloud-model-based hydro-wind-solar hybrid system unit status assessment device 10, including:

[0141] The data acquisition and preprocessing module 100 is used to acquire the operating data of hydropower units, wind power units and photovoltaic arrays, and to clean, filter and normalize the operating data to obtain a high-quality dataset in a unified format.

[0142] The feature extraction and filtering module 200 is used to extract key features reflecting the unit's operating status from the high-quality dataset, and to use a feature selection algorithm to filter out feature indicators that have a significant impact on the status assessment, thereby constructing a multi-dimensional unit status assessment indicator system.

[0143] The game theory weight fusion module 300 is used to perform multi-method weight fusion on the feature indicators using a game theory combination weighting method, solve for the optimal linear combination by optimizing the weight coefficients, and obtain the final combination weight of each feature indicator by normalization.

[0144] The cloud model construction and mapping module 400 is used to construct a cloud model based on the final combined weights, calculate the expectation, entropy and hyperentropy of each state level, and generate a cloud map using a positive cloud generator to achieve a fuzzy-random two-dimensional mapping of the unit's operating state.

[0145] The status assessment and model update module 500 is used to determine the current status level of the unit based on the correlation between the cloud model and the unit's operating status, and to feed the assessment results back to the system operation terminal to trigger the online learning mechanism, dynamically update the cloud model parameters and feature weights, and improve the adaptability and accuracy of the assessment model.

[0146] Furthermore, the data acquisition and preprocessing module is also used for:

[0147] The Z-score standardization method was used to normalize the head, flow rate, speed and power of the hydropower unit to eliminate the influence of different dimensions and magnitudes.

[0148] Wavelet filtering is applied to the wind speed, wind direction, pitch angle, and power of the wind turbine to remove high-frequency noise and retain key operating characteristics; and data on the photovoltaic array, including light intensity, temperature, voltage, and current, are processed.

[0149] Furthermore, the feature extraction and filtering module is also used for:

[0150] The vibration frequency, temperature change rate, and power fluctuation coefficient of hydropower units were extracted as key characteristics reflecting their operating status.

[0151] An information entropy-based feature selection algorithm is used to filter the features of wind turbines and photovoltaic arrays, retaining feature indicators that are highly sensitive to state changes.

[0152] Furthermore, the game theory weight fusion module is also used for:

[0153] Principal component analysis, analytic hierarchy process and entropy weight method are used as three basic weight calculation methods to construct a basic weight vector set;

[0154] Construct the optimal linear combination based on the basic weight vector set, and solve for the final combination weights.

[0155] Furthermore, the cloud model building and mapping module is also used for:

[0156] The expected value of the cloud model is calculated based on the weighted scores of each feature indicator.

[0157] A normal distribution cloud generator is used to generate cloud maps, and the distribution density and dispersion of cloud droplets can intuitively reflect the fuzziness and randomness of the unit's status.

[0158] Furthermore, it also includes:

[0159] The maintenance suggestion generation and push module is used to generate maintenance suggestions based on the current status level of the unit and historical operating data, and automatically push them to the operation and maintenance management system to realize intelligent maintenance scheduling based on status assessment results.

[0160] The cloud-based hydro-wind-solar hybrid system unit status assessment device of this invention automatically generates and pushes maintenance suggestions to the operation and maintenance management system based on the current status level of the unit and historical operating data. This achieves closed-loop linkage between status assessment and maintenance scheduling, further improving the intelligence level and response efficiency of system maintenance, effectively reducing operation and maintenance costs and extending the service life of the unit.

[0161] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is 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.

[0162] 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 assessing the condition of a hydro-wind-solar hybrid system unit based on a cloud model, characterized in that, include: S1. Collect the operating data of hydropower units, wind power units and photovoltaic arrays, and clean, filter and normalize the operating data to obtain a high-quality dataset in a unified format. S2, extract key features reflecting the unit's operating status from the high-quality dataset, and use feature selection algorithms to screen out feature indicators that have a significant impact on status assessment, and construct a multi-dimensional unit status assessment indicator system. S3. A game theory-based combination weighting method is used to fuse the weights of the feature indicators using multiple methods. The optimal linear combination is solved by optimizing the weight coefficients, and the final combination weights of each feature indicator are obtained by normalization. S4. Based on the final combined weights, construct a cloud model, calculate the expectation, entropy and hyperentropy of each state level, and use a positive cloud generator to generate a cloud map to achieve a fuzzy-random two-dimensional mapping of the unit's operating state. S5. Based on the correlation between the cloud model and the unit's operating status, determine the current status level of the unit and feed the evaluation results back to the system operation terminal to trigger the online learning mechanism and dynamically update the cloud model parameters and feature weights.

2. The method as described in claim 1, characterized in that, The process of collecting operational data from hydropower units, wind power units, and photovoltaic arrays, and cleaning, filtering, and normalizing the operational data to obtain a high-quality dataset in a unified format, also includes: S11 uses the Z-score standardization method to normalize the head, flow rate, speed and power of the hydropower unit in order to eliminate the influence of different dimensions and magnitudes. S12 performs wavelet filtering on the wind speed, wind direction, pitch angle, and power of the wind turbine to remove high-frequency noise and retain key operating characteristics; and processes the data of the photovoltaic array, including light intensity, temperature, voltage, and current.

3. The method as described in claim 1, characterized in that, The process of extracting key features reflecting the unit's operating status from the high-quality dataset, and using feature selection algorithms to screen out feature indicators that have a significant impact on status assessment, thereby constructing a multi-dimensional unit status assessment indicator system, also includes: S21, extract the vibration frequency, temperature change rate and power fluctuation coefficient of the hydropower unit as key features reflecting its operating status; S22 uses an information entropy-based feature selection algorithm to filter the features of wind turbines and photovoltaic arrays, retaining feature indicators that are highly sensitive to state changes.

4. The method as described in claim 1, characterized in that, The method employs a game theory-based combinatorial weighting approach to fuse the feature indicators using multiple weighting methods. It then optimizes the weight coefficients to find the optimal linear combination and normalizes the results to obtain the final combined weights for each feature indicator. The method further includes: S31, using principal component analysis, analytic hierarchy process and entropy weight method as three basic weight calculation methods, construct a basic weight vector set; S32 constructs the optimal linear combination based on the basic weight vector set and solves for the final combination weights.

5. The method as described in claim 1, characterized in that, The process of constructing a cloud model based on the final combined weights, calculating the expected value, entropy, and hyperentropy of each state level, and generating a cloud map using a forward cloud generator to achieve a fuzzy-random two-dimensional mapping of the unit's operating state also includes: S41, Calculate the expectation of the cloud model based on the weighted scores of each feature indicator; S42 uses a normal distribution cloud generator to generate cloud maps, and the distribution density and dispersion of cloud droplets intuitively reflect the fuzziness and randomness of the unit's status.

6. The method as described in claim 1, characterized in that, Also includes: S6 generates maintenance recommendations based on the unit's current status level and historical operating data, and automatically pushes them to the operation and maintenance management system to achieve intelligent maintenance scheduling based on status assessment results.

7. A cloud-model-based hydro-wind-solar hybrid system unit condition assessment device, characterized in that, include: The data acquisition and preprocessing module is used to collect the operating data of hydropower units, wind power units and photovoltaic arrays, and to clean, filter and normalize the operating data to obtain a high-quality dataset in a unified format. The feature extraction and filtering module is used to extract key features reflecting the unit's operating status from the high-quality dataset, and to use a feature selection algorithm to filter out feature indicators that have a significant impact on the status assessment, thereby constructing a multi-dimensional unit status assessment indicator system. The game theory weight fusion module is used to perform multi-method weight fusion on the feature indicators using a game theory combinatorial weighting method. It solves the optimal linear combination by optimizing the weight coefficients and normalizes the result to obtain the final combined weight of each feature indicator. The cloud model construction and mapping module is used to construct a cloud model based on the final combined weights, calculate the expectation, entropy and hyperentropy of each state level, and generate a cloud map using a positive cloud generator to achieve a fuzzy-random two-dimensional mapping of the unit's operating state. The status assessment and model update module is used to determine the current status level of the unit based on the correlation between the cloud model and the unit's operating status, and to feed the assessment results back to the system operation terminal to trigger the online learning mechanism, dynamically update the cloud model parameters and feature weights, and improve the adaptability and accuracy of the assessment model.

8. The apparatus as claimed in claim 7, characterized in that, The data acquisition and preprocessing module is also used for: The Z-score standardization method was used to normalize the head, flow rate, speed and power of the hydropower unit to eliminate the influence of different dimensions and magnitudes. Wavelet filtering is applied to the wind speed, wind direction, pitch angle, and power of the wind turbine to remove high-frequency noise and retain key operating characteristics; and data on the photovoltaic array, including light intensity, temperature, voltage, and current, are processed.

9. The apparatus as claimed in claim 7, characterized in that, The feature extraction and filtering module is also used for: The vibration frequency, temperature change rate, and power fluctuation coefficient of hydropower units were extracted as key characteristics reflecting their operating status. An information entropy-based feature selection algorithm is used to filter the features of wind turbines and photovoltaic arrays, retaining feature indicators that are highly sensitive to state changes.

10. The apparatus as claimed in claim 7, characterized in that, The game theory weight fusion module is also used for: Principal component analysis, analytic hierarchy process and entropy weight method are used as three basic weight calculation methods to construct a basic weight vector set; Construct the optimal linear combination based on the basic weight vector set, and solve for the final combination weights.

11. The apparatus as claimed in claim 7, characterized in that, The cloud model construction and mapping module is also used for: The expected value of the cloud model is calculated based on the weighted scores of each feature indicator. A normal distribution cloud generator is used to generate cloud maps, and the distribution density and dispersion of cloud droplets can intuitively reflect the fuzziness and randomness of the unit's status.

12. The apparatus as claimed in claim 7, characterized in that, Also includes: The maintenance suggestion generation and push module is used to generate maintenance suggestions based on the current status level of the unit and historical operating data, and automatically push them to the operation and maintenance management system to realize intelligent maintenance scheduling based on status assessment results.

Images