Intelligent Optimization Method and System for Hydroelectric Generator Set Parameters Based on Multimodal Data Fusion

Abstract

The present invention relates to the technical field of parameter optimization of hydropower units, and particularly to an intelligent optimization method and system for hydropower unit parameters based on multi-modal data fusion. The present invention integrates multi-modal data, graph neural networks and Bayesian optimization to achieve efficient and intelligent optimization of hydropower unit parameters. First, multi-modal data in the time domain, frequency domain, acoustics, temperature and hydraulics are integrated, and feature vectors are generated through intra-modal feature extraction and hierarchical attention mechanism fusion. Then, a unit knowledge graph is constructed, and prior knowledge is extracted using graph attention networks to achieve cross-domain knowledge transfer. Finally, based on the Bayesian optimization algorithm with multi-objective constraints and combined with a Gaussian process surrogate model, the optimal PID parameter configuration is determined, and incremental optimization is carried out through closed-loop verification. This method shortens the optimization time from several days to within 15 minutes, with an efficiency improvement of more than 95%, significantly improving the efficiency and accuracy of hydropower unit parameter optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of hydropower unit parameter optimization, and particularly to an intelligent optimization method and system for hydropower unit parameters based on multi-modal data fusion. Background Art

[0002] Optimizing the power mode parameters of hydropower units is of great significance in ensuring the stable operation of the units and improving efficiency. In the prior art, for example, Chinese Patent CN 112628055 B discloses a method for on-site optimization of power mode parameters of hydropower units. This method adopts a power mode PID operation parameter optimization method based on 6 characteristic parameters, and optimizes the unit operation parameters by judging whether the adjustment performance of the parameters is excellent.

[0003] However, the prior art has the following deficiencies: First, the data source is single, only using traditional data such as active power and guide vane opening, ignoring important multi-modal data such as vibration spectrum, noise characteristics, and efficiency curve, which limits the ability to perceive the overall state of the unit; Second, it adopts a mechanical parameter adjustment logic, using fixed rules for parameter adjustment, lacking the adaptive learning ability for complex non-linear systems; Third, there is a knowledge island effect during the optimization process, with each unit optimized independently, unable to utilize the knowledge correlation and historical optimization experience between units; In addition, the prior art adopts a static optimization idea, optimizing once and using it for a long time, making it difficult to cope with the dynamic changes of working conditions such as water head and load; Finally, the existing methods rely highly on manual labor, requiring professional technicians to conduct on-site tests and parameter adjustments, which is time-consuming and the effect depends on the experience of the personnel. Summary of the Invention

[0004] The object of the present invention is to overcome the deficiencies of the prior art and provide an intelligent optimization method and system for hydropower unit parameters based on multi-modal data fusion. This method realizes the intelligent and efficient optimization of hydropower unit parameters by fusing multi-modal data, using graph neural networks to achieve knowledge transfer, and adopting the Bayesian optimization algorithm for parameter optimization.

[0005] The present invention proposes an intelligent optimization method for hydropower unit parameters based on multi-modal data fusion, including:

[0006] Obtain the multi-modal data of the hydropower unit, where the multi-modal data includes time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data;

[0007] Perform feature extraction and fusion on the multi-modal data, including:

[0008] Use the intra-modal feature extraction network to process each type of data respectively to obtain intra-modal features;

[0009] Calculate the correlation weights between different modal features based on the hierarchical attention mechanism to generate a fused feature vector;

[0010] Construct a unit knowledge graph and implement cross-domain knowledge transfer based on a graph neural network, including:

[0011] Construct a unit knowledge graph based on the fused feature vector and historical optimization experience;

[0012] Use a graph attention network to extract prior knowledge representations adapted to the current unit characteristics;

[0013] Perform parameter optimization based on a multi-objective constrained Bayesian optimization algorithm, including:

[0014] Construct a Gaussian process surrogate model according to the prior knowledge representation;

[0015] Use a multi-objective acquisition function to determine the optimal PID parameter configuration;

[0016] Perform incremental optimization on the optimized parameters through closed-loop verification to obtain the final optimized parameters.

[0017] Preferably, the acquisition of multi-modal data of the hydropower unit includes:

[0018] Deploy a distributed multi-modal sensing data acquisition network;

[0019] Implement synchronous acquisition of multi-source data based on timestamps, ensuring that the maximum time difference does not exceed 5 milliseconds;

[0020] Perform outlier detection and data standardization processing on the acquired multi-modal data;

[0021] Generate a multi-modal data tensor in a unified format.

[0022] Preferably, the use of the intra-modal feature extraction network to process various types of data respectively includes:

[0023] Use a TCN network based on causal convolution to extract time-domain data features;

[0024] Use an improved ResNet network to extract frequency-domain vibration data features;

[0025] Use Mel spectrum transformation and CNN to extract acoustic data features;

[0026] Use a spatial convolution network to extract temperature distribution data features;

[0027] Use a deep fully connected network to extract hydraulic characteristic data features.

[0028] Preferably, the calculation of the correlation weights between different modal features based on the hierarchical attention mechanism includes:

[0029] Calculate the attention weight matrix between different modal features:

[0030] ,

[0031] Where: is the feature vector of the th modality, is the feature vector of the th modality, is the similarity calculation function;

[0032] Enhance the original features based on the attention weight matrix:

[0033] ,

[0034] Integrate different scale features through a feature pyramid fusion network to generate a final fused feature vector.

[0035] Preferably, the unit knowledge graph includes:

[0036] Unit nodes, representing the characteristic parameters of the hydropower unit;

[0037] Operating condition nodes, representing the operating parameters of the unit;

[0038] Parameter nodes, representing PID control parameters;

[0039] Physical association edges, representing the hydraulic coupling relationship between units;

[0040] Performance similarity edges, representing the similarity degree of operating characteristics between units;

[0041] Historical optimization edges, representing the transfer relationship of historical parameter optimization experience.

[0042] Preferably, the extraction of prior knowledge representation adapted to the current unit characteristics by using the graph attention network includes:

[0043] Implement a node-level attention mechanism to calculate the importance weights between nodes;

[0044] Design an edge-level message passing mechanism to realize the transfer of knowledge in the graph structure;

[0045] Through the domain alignment technology of maximum mean discrepancy, realize the knowledge transfer across units and across water heads;

[0046] Generate a prior knowledge representation adapted to the current unit characteristics.

[0047] Preferably, the multi-objective constrained Bayesian optimization algorithm includes:

[0048] Construct a Gaussian process surrogate model, adopting a multi-core function combination strategy: , where: is the basic kernel function, is the weight coefficient;

[0049] Design a multi-objective acquisition function to balance multiple optimization objectives such as response time, overshoot, and stability;

[0050] Implement a dynamic optimization strategy for condition awareness and automatically adjust the exploration intensity according to the head change;

[0051] Develop a parameter sensitivity analysis module to identify key parameters and optimize them preferentially.

[0052] Preferably, the incremental optimization of the optimized parameters through closed-loop verification includes:

[0053] Build a refined digital twin model of the unit to simulate and verify the parameter optimization effect;

[0054] Design a limit condition test framework to evaluate the performance of parameters under boundary conditions;

[0055] Implement incremental learning based on experience replay to continuously optimize the model performance;

[0056] Build a parameter degradation detection model to identify the trend of parameter performance decline;

[0057] Develop a self-triggered optimization mechanism to automatically start re-optimization when the performance decline exceeds a preset threshold.

[0058] Preferably, the method further includes:

[0059] Design a plant-level collaborative optimization interface to achieve coordinated adjustment of unit group parameters;

[0060] Develop a knowledge sharing framework based on federated learning to protect data privacy;

[0061] Realize cross-station knowledge transfer to support the rapid commissioning of new power stations.

[0062] An intelligent optimization system for hydropower unit parameters with multi-modal data fusion, including:

[0063] A multi-modal data acquisition module for acquiring multi-modal data of hydropower units, where the multi-modal data includes time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data;

[0064] A feature extraction and fusion module for extracting and fusing features from the multi-modal data to generate a fused feature vector;

[0065] A knowledge graph construction and transfer learning module for constructing a unit knowledge graph and realizing cross-domain knowledge transfer based on graph neural networks to generate prior knowledge representations;

[0066] A parameter optimization module, which is used to perform parameter optimization based on a Bayesian optimization algorithm with multi-objective constraints to determine the optimal PID parameter configuration;

[0067] A closed-loop verification and incremental optimization module, which is used to perform incremental optimization on the optimized parameters through closed-loop verification to obtain the final optimized parameters;

[0068] A collaborative optimization interface module, which is used to realize the collaborative adjustment of unit group parameters and cross-power station knowledge transfer.

[0069] The present invention has the following beneficial effects:

[0070] 1. The optimization efficiency is significantly improved: Compared with the traditional method, the optimization time is shortened from several days to within 15 minutes, the efficiency is increased by more than 95%, and the success rate of the first optimization reaches 95%, which is much higher than the 60% - 70% of the traditional method.

[0071] 2. The adaptability to all working conditions is greatly improved: The parameters can adapt within the range of ±30 meters of head change without manual intervention; it can operate stably in the full range of 10% - 100% load change, solving the problem of instability of the traditional method under load boundary conditions.

[0072] 3. The economic benefits are significant: The unit efficiency is increased by 4.2%, and the annual increased income is considerable; the vibration is reduced by 37%, and the service life of the equipment is extended by 3 - 5 years; the number of shutdown adjustments is reduced by 85%, and the equipment utilization rate is improved.

[0073] 4. The degree of intelligence is high: It realizes fully automatic parameter optimization without on-site intervention by professionals; supports remote monitoring and optimization, reducing the operation and maintenance cost; has the ability of self-learning, and the optimization effect continuously improves with the use time. Description of the Drawings

[0074] Figure 1 It is a flow chart of the intelligent optimization method for the parameters of a hydropower unit with multi-modal data fusion according to the present invention;

[0075] Figure 2 It is an architecture diagram of the multi-modal data acquisition network according to the present invention;

[0076] Figure 3 It is a structure diagram of the heterogeneous feature extraction and fusion network with a hierarchical attention mechanism according to the present invention;

[0077] Figure 4 It is a schematic diagram of the construction of the unit knowledge graph according to the present invention;

[0078] Figure 5 It is a schematic diagram of the cross-domain transfer learning mechanism enhanced by the graph neural network according to the present invention;

[0079] Figure 6Flow chart of the multi-objective constrained dynamic adaptive Bayesian optimization of the present invention;

[0080] Figure 7 Schematic diagram of the closed-loop verification and incremental optimization system of the present invention;

[0081] Figure 8 Architecture diagram of the intelligent optimization system for hydropower unit parameters with multi-modal data fusion of the present invention. Specific implementation manners

[0082] Please refer to the attached Figure 1-8 drawings, and the specific implementation manners of the present invention will be further described in detail below with reference to the drawings.

[0083] Example 1: Overall process of the intelligent optimization method for hydropower unit parameters with multi-modal data fusion

[0084] As Figure 1 shown, the intelligent optimization method for hydropower unit parameters with multi-modal data fusion provided by the present invention includes the following steps:

[0085] Obtain multi-modal data of the hydropower unit, where the multi-modal data includes time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data;

[0086] Perform feature extraction and fusion on the multi-modal data, including respectively processing various types of data using an intra-modal feature extraction network to obtain intra-modal features; calculating the correlation weights between different modal features based on a hierarchical attention mechanism to generate a fused feature vector;

[0087] Construct a unit knowledge graph and implement cross-domain knowledge transfer based on a graph neural network, including constructing a unit knowledge graph based on the fused feature vector and historical optimization experience; using a graph attention network to extract prior knowledge representations adapted to the current unit characteristics;

[0088] Perform parameter optimization based on a multi-objective constrained Bayesian optimization algorithm, including constructing a Gaussian process surrogate model according to the prior knowledge representation; determining the optimal PID parameter configuration using a multi-objective acquisition function;

[0089] Perform incremental optimization on the optimized parameters through closed-loop verification to obtain the final optimized parameters.

[0090] The overall process of this method follows a processing framework of data acquisition - feature extraction - knowledge transfer - parameter optimization - closed-loop verification. A complete data flow link is formed between each step, realizing the intelligent and efficient optimization of hydropower unit parameters.

[0091] Preferably, this method can be applied to the parameter optimization of various types of hydropower units, including but not limited to different types of units such as impulse turbines, Francis turbines, and Kaplan turbines, and is applicable to hydropower stations with different head conditions and load demands.

[0092] Embodiment 2: Multimodal Data Acquisition Method

[0093] As Figure 2 shown, this embodiment details the specific implementation method of multimodal data acquisition, mainly including:

[0094] Deploy a distributed multimodal sensing data acquisition network, and install various sensors at key parts of the hydropower unit, including power sensors, vibration sensors, acoustic sensors, temperature sensors, water pressure sensors, etc., to form a sensing network covering the entire unit.

[0095] Implement multi-source data synchronous acquisition based on timestamps, ensuring that the maximum time difference does not exceed 5 milliseconds. Preferably, use industrial Ethernet technology to build the sensing network, support Time-Sensitive Networking (TSN) technology, and ensure the time synchronization of data acquisition.

[0096] Perform outlier detection and data standardization processing on the acquired multimodal data. Preferably, use the Local Outlier Factor (LOF) algorithm to identify abnormal data points, set a dynamic threshold, and automatically mark and remove or correct the outlier points when the abnormality exceeds the threshold. Design a dedicated standardization method for different physical quantities to retain sensitive features.

[0097] Generate multimodal data tensors in a unified format. Preferably, convert different types of data into a multi-dimensional tensor format uniformly, ensure the alignment of the time dimension, and provide a standardized input for subsequent feature extraction.

[0098] The multimodal data obtained by this method includes:

[0099] Time-domain operation data: including active power, guide vane opening, water pressure at the inlet of the spiral case, unit frequency, etc., with a sampling rate of 100 Hz;

[0100] Frequency-domain vibration data: including radial vibration and axial vibration spectra, with a frequency range of 0 - 1000 Hz and an accuracy of 0.1 Hz;

[0101] Acoustic feature data: including noise spectra and sound intensity distributions, with a frequency range of 20 - 20000 Hz;

[0102] Temperature distribution data: including bearing temperature and stator temperature distribution, with an accuracy of 0.1 °C;

[0103] Hydraulic characteristic data: including cavitation characteristics, efficiency curves, head change rates, etc.

[0104] Through the acquisition and preprocessing of multimodal data, it provides a high-quality and multi-dimensional data foundation for subsequent feature extraction and fusion, overcoming the limitation of the single data source in traditional methods.

[0105] Embodiment 3: Multimodal Data Feature Extraction Method

[0106] As Figure 3 shown, this embodiment details the feature extraction method of multimodal data, specifically including:

[0107] Use a TCN network based on causal convolution to extract the time-domain data features. Preferably, the TCN network adopts a dilated convolution structure, and the convolution kernel size is set to [3, 5, 7], which can effectively capture features at different time scales. Causal convolution ensures that the output at the current moment only depends on historical data, avoiding information leakage. At the same time, the gradient propagation is improved through residual connections, enhancing the effectiveness of feature extraction.

[0108] Use an improved ResNet network to extract the frequency-domain vibration data features. Preferably, the improved ResNet network introduces an attention mechanism to enhance the ability to extract key frequency components. The network depth is 18 layers, and each layer of convolution is followed by BatchNormalization and ReLU activation functions to effectively extract the frequency-domain vibration features.

[0109] Use Mel spectrum transformation and CNN to extract the acoustic data features. Preferably, first perform Mel spectrum transformation on the acoustic signal to map the linear frequency to the Mel frequency that is more in line with human ear perception, and then extract the acoustic features through a 3-layer CNN network. The convolution kernel sizes are 5×5, 3×3, and 3×3 respectively, and the activation function is LeakyReLU.

[0110] Use a spatial convolution network to extract the temperature distribution data features. Preferably, regard the temperature distribution data as a two-dimensional image and extract the spatial distribution features through a 3-layer spatial convolution network to effectively capture the temperature gradient and distribution pattern.

[0111] Use a deep fully connected network to extract the hydraulic characteristic data features. Preferably, design a 3-layer fully connected network, and the number of hidden layer nodes is 128, 64, and 32 respectively. The activation function is ReLU to extract the non-linear relationship features of the hydraulic characteristics.

[0112] Through intra-modal feature extraction, various types of data are converted into feature vectors with a fixed dimension, providing a standardized input for subsequent feature fusion. Intra-modal feature extraction makes full use of the characteristics of various types of data, selects a network structure suitable for its characteristics, and ensures the effectiveness and accuracy of feature extraction.

[0113] Embodiment 4: Feature Fusion Method Based on Hierarchical Attention Mechanism

[0114] This embodiment details a method for calculating the correlation weights between different modality features based on a hierarchical attention mechanism, which is the core of feature fusion and specifically includes:

[0115] Calculate the attention weight matrix between different modality features:

[0116] ,

[0117] Where: is the feature vector of the th modality, is the feature vector of the th modality, is the similarity calculation function, and the scaled dot-product attention mechanism can be adopted: , and are learnable transformation matrices, is the feature dimension.

[0118] Enhance the original features based on the attention weight matrix:

[0119] ,

[0120] Integrate features of different scales through a feature pyramid fusion network to generate the final fused feature vector. Preferably, the feature pyramid network adopts a combination of top-down and lateral connections to construct a multi-scale feature fusion structure. The top-down is achieved through upsampling, and the lateral connection is achieved after adjusting the number of channels through 1×1 convolution.

[0121] In addition, introduce a channel attention mechanism to adaptively adjust the weights of different feature channels:

[0122] ,

[0123] Where: AvgPool and MaxPool are average pooling and max pooling operations respectively, MLP is a multi-layer perceptron, is the sigmoid activation function. Channel attention can dynamically adjust the weights according to the importance of different feature channels.

[0124] Finally, develop a feature redundancy suppression module to reduce the dimension through feature distillation technology. Preferably, an autoencoder structure is adopted to achieve feature compression. Both the encoder and decoder are 3-layer fully connected networks, and a compact representation is learned by minimizing the reconstruction error.

[0125] Through the feature fusion of the hierarchical attention mechanism, this method can effectively integrate the complementary information in multi-modal data, generate a compact and information-rich fused feature vector, and provide a high-quality feature representation for subsequent knowledge transfer and parameter optimization.

[0126] Example 5: Method for Constructing Unit Knowledge Graph

[0127] As Figure 4 shown, this example details the method for constructing a unit knowledge graph, which is the basis for realizing cross-domain knowledge transfer and specifically includes:

[0128] Construct unit nodes to represent the characteristic parameters of hydropower units. Preferably, the attributes included in the unit nodes are: basic parameters such as unit type, rated capacity, design head, rotational speed, number of blades, etc., as well as statistical characteristics of historical operation data, such as average load rate, start-stop frequency, vibration intensity, etc.

[0129] Construct operating condition nodes to represent the operating parameters of the units. Preferably, the attributes included in the operating condition nodes are: head range, load range, water temperature, characteristics of the water diversion system, etc., as well as current operating state parameters, such as real-time head, load, efficiency, etc.

[0130] Construct parameter nodes to represent PID control parameters. Preferably, the attributes included in the parameter nodes are: PID parameter combination, applicable operating condition range of the parameters, parameter performance evaluation indicators (response time, overshoot, stability, etc.), parameter acquisition time, etc.

[0131] Construct physical association edges to represent the hydraulic coupling relationship between units. Preferably, the attributes included in the physical association edges are: common system type (tail water system, water diversion system, etc.), coupling strength coefficient, relative position relationship, etc.

[0132] Construct performance similarity edges to represent the similarity degree of operating characteristics between units. Preferably, the performance similarity edges are obtained by calculating the cosine similarity of the operating characteristic vectors of the units, and the threshold is set to 0.85. Edge connections are established between unit pairs that exceed the threshold.

[0133] Construct historical optimization edges to represent the transfer relationship of historical parameter optimization experience. Preferably, the historical optimization edges record the transfer history of parameter optimization, including information such as source unit, target unit, transfer time, change in optimized parameters, etc.

[0134] The unit knowledge graph adopts a dynamic update mechanism to update the graph structure in real time according to new optimization experience, and the weights are adjusted adaptively. Preferably, after each successful parameter optimization, the optimization experience is added to the knowledge graph, and at the same time, the weights of the relevant edges are updated according to the optimization effect to ensure that the knowledge graph always reflects the latest optimization experience.

[0135] By constructing a multi-level and multi-relationship unit knowledge graph, this method can effectively capture various association relationships between units and provide a structured knowledge representation for subsequent knowledge transfer based on graph neural networks.

[0136] Example 6: Knowledge Transfer Method Based on Graph Neural Network

[0137] As Figure 5 shown, this example details a knowledge transfer method based on graph neural network, specifically including:

[0138] Implement a node-level attention mechanism to calculate the importance weights between nodes:

[0139] ,

[0140] Among them: and are the feature representations of node i and node respectively, W is the weight matrix, is the attention vector, || represents the vector concatenation operation, is node 's neighbor set.

[0141] Design an edge-level message passing mechanism to realize the transfer of knowledge in the graph structure:

[0142] ,

[0143] Among them: and represent the feature representations of node i at the l-th layer and the (l + 1)-th layer respectively, σ is the activation function, and preferably the ReLU function is adopted. Edge-level message passing realizes the propagation of knowledge in the graph structure by iteratively updating node representations and gradually integrating neighborhood information.

[0144] Through the domain alignment technology of maximum mean discrepancy ( ), realize knowledge transfer across units and heads:

[0145] ,

[0146] Among them: and represent the data sets of the source domain and the target domain respectively, φ is the feature mapping function, and H is the reproducing kernel Hilbert space. By minimizing the MMD distance, the feature distributions of the source domain and the target domain are made to tend to be consistent, thereby realizing cross-domain knowledge transfer.

[0147] In addition, design a progressive transfer strategy to retain the effective knowledge of the source domain through weight regularization. Preferably, L2 regularization is used to limit the change range of weights during the transfer process to ensure that the transferred model retains the knowledge of the source domain while adapting to the characteristics of the target domain.

[0148] Implement a parameter soft sharing mechanism to dynamically adjust the knowledge transfer intensity according to similarity. Preferably, calculate the knowledge transfer coefficient λ based on the similarity between the source domain and the target domain. The higher the similarity, the larger the value of λ, and the greater the knowledge transfer intensity. The knowledge transfer coefficient is used to adjust the sharing degree of parameters between the source domain and the target domain: θt=(1-λ)θt+λθs, where θt and θs are the parameters of the target domain and the source domain respectively.

[0149] Through the cross-domain transfer learning mechanism enhanced by the graph neural network, this method can effectively utilize historical optimization experience, achieve cross-unit and cross-condition transfer of knowledge, provide high-quality prior knowledge for parameter optimization under new units or new conditions, and greatly improve the optimization efficiency and success rate.

[0150] Example 7: Bayesian optimization algorithm with multi-objective constraints

[0151] As Figure 6 shown, this example details the Bayesian optimization algorithm with multi-objective constraints, specifically including:

[0152] Construct a Gaussian process surrogate model and adopt a multi-kernel function combination strategy:

[0153] ,

[0154] where: is the base kernel function, including the radial basis function (RBF), the Matérn kernel function, the periodic kernel function, etc., is the weight coefficient, which is automatically determined by maximizing the marginal likelihood. The multi-kernel function combination strategy can adapt to the characteristics of different parameter spaces and improve the expression ability and generalization performance of the model.

[0155] Design a multi-objective acquisition function to balance multiple optimization objectives such as response time, overshoot, and stability. Preferably, adopt a combination of three acquisition functions: expected improvement (EI), upper confidence bound (UCB), and expected entropy reduction (EES):

[0156] ,

[0157] where: 、 and are the weight coefficients, which are dynamically adjusted according to the optimization stage. The weight of the exploration term is increased in the initial stage, and the weight of the exploitation term is increased in the later stage.

[0158] Introduce a prediction variance weighting mechanism:

[0159] ,

[0160] where: is the predicted mean, is the standard deviation of prediction, is the exploration-exploitation balance coefficient, initially set to 2.0, and gradually reduced to 0.5 as the optimization progresses to achieve a smooth transition from exploration to exploitation.

[0161] Implement constraint handling techniques to ensure that the parameters meet the safe operating boundary conditions. Preferably, the penalty function method is used to handle constraints, and penalties are given to solutions that violate the constraints. The degree of penalty is proportional to the degree of violation to ensure that the optimization results meet all constraint conditions.

[0162] Design a dynamic exploration strategy based on Thompson sampling to automatically enhance the exploration intensity when the water head changes. Preferably, when it is detected that the water head change exceeds a predetermined threshold (default is 5 meters), increase the exploration coefficient to strengthen the exploration of the parameter space and adapt to new operating conditions.

[0163] Develop a parameter sensitivity analysis module to identify key parameters and optimize them preferentially. Preferably, estimate the sensitivity by calculating the partial derivative of the parameter with respect to the objective function, sort the parameters according to sensitivity, and preferentially optimize the parameters with high sensitivity to improve the optimization efficiency.

[0164] Implement progressive search space reduction to improve the optimization efficiency. Preferably, in the initial stage, a global search is carried out within a large range. As the optimization progresses, the search space is gradually reduced according to the historical evaluation results, and resources are concentrated on fine-grained search in promising areas.

[0165] In addition, introduce the idea of simulated annealing to prevent falling into local optima; develop a safety exploration area definition based on model uncertainty; design a perturbation analysis module to evaluate the robustness of parameters to perturbations.

[0166] Through the multi-objective constrained dynamic adaptive Bayesian optimization algorithm, this method can efficiently and accurately achieve parameter optimization, balance multiple optimization objectives, adapt to dynamically changing operating conditions, and ensure the effectiveness and robustness of the parameter optimization results.

[0167] Example 8: Closed-loop Verification and Incremental Optimization Method

[0168] As Figure 7 shown, this example details the closed-loop verification and incremental optimization method, specifically including:

[0169] Construct a refined digital twin model of the unit to simulate and verify the parameter optimization effect. Preferably, the digital twin model includes a hydraulic model, a mechanical model, and an electrical model, which can simulate the dynamic response of the unit under different operating conditions and provide a safe and reliable verification environment for parameter optimization.

[0170] Design the extreme operating condition test framework to evaluate the performance of parameters under boundary conditions. Preferably, the extreme operating conditions include typical conditions such as the highest water head, the lowest water head, full load, low load, and rapid load change. By simulating these conditions in the digital twin model, evaluate the adaptability and stability of parameters under extreme conditions.

[0171] Implement incremental learning based on experience replay to continuously optimize the model performance. Preferably, maintain an experience pool to store historical optimization experiences. Each time the model is updated, randomly select a batch of historical experiences and mix them with new experiences for training to avoid catastrophic forgetting and ensure the steady improvement of the model performance.

[0172] Develop knowledge distillation technology to compress and integrate new experiences into the existing model. Preferably, use the optimized model as the teacher model and the newly trained model as the student model. By minimizing the KL divergence between the outputs of the two models, transfer the knowledge of the teacher model to the student model to achieve efficient model update.

[0173] Design a forgetting control mechanism to retain key historical experiences. Preferably, perform importance scoring on the samples in the experience pool. Samples with high importance are retained for a longer time, reducing the probability of being replaced to ensure that key experiences are not forgotten.

[0174] Build a parameter degradation detection model to identify the trend of parameter performance degradation. Preferably, by monitoring the change trends of key performance indicators (response time, overshoot, stability, etc.), when the deterioration trend of the indicators exceeds the preset threshold, trigger a parameter degradation alarm.

[0175] Develop a self-triggered optimization mechanism to automatically start re-optimization when the performance degradation exceeds the preset threshold. Preferably, set a performance degradation threshold. When it is detected that the parameter performance degradation exceeds the threshold (default is 10%), automatically start the parameter re-optimization process without manual intervention.

[0176] Implement fault traceability analysis to identify the root causes affecting parameter performance. Preferably, analyze the correlation between parameter performance degradation and system state changes through a causal inference model to find out the root causes of performance degradation and provide accurate guidance for parameter adjustment.

[0177] Through the closed-loop verification and incremental optimization system, this method realizes the closed-loop control of parameter optimization, forms a continuous improvement mechanism of optimization - verification - re-optimization, and ensures that the unit parameters always maintain the optimal state and adapt to the changing operating environment.

[0178] Example 9: Cooperative Optimization Interface Method

[0179] This example details the cooperative optimization interface method, which specifically includes:

[0180] Design the plant-level collaborative optimization interface to achieve coordinated adjustment of unit group parameters. Preferably, the plant-level collaborative optimization interface considers the mutual influence between units, especially those sharing a common tailrace system or intake system, and coordinates the adjustment of the parameters of each unit to optimize the overall operating performance.

[0181] Develop a knowledge sharing framework based on federated learning to protect data privacy. Preferably, each unit trains the model locally, only sharing the model parameters without sharing the original data. The central server aggregates the model parameters of each unit, generates a global model, and then distributes it to each unit to achieve knowledge sharing while protecting data privacy.

[0182] Implement cross-power station knowledge transfer to support the rapid commissioning of newly built power stations. Preferably, through transfer learning technology, transfer the optimization experience of existing power stations to newly built power stations, accelerate the commissioning process of new power stations, and improve the commissioning efficiency and quality.

[0183] Design a multi-scale optimization strategy to achieve hierarchical optimization from single units to unit groups and then to the entire power station. Preferably, first optimize the parameters of single units, then conduct collaborative optimization of unit groups based on the optimization results of single units, and finally conduct collaborative optimization at the power station level to gradually improve the optimization effect.

[0184] Develop an asynchronous collaborative update mechanism to support the flexible switching between independent optimization and collaborative optimization of different units. Preferably, each unit can independently optimize parameters according to its own needs, and at the same time can access the collaborative optimization framework when needed to achieve flexible adjustment of optimization strategies.

[0185] Implement an optimization experience sharing and traceability mechanism based on blockchain. Preferably, record the optimization experience on the blockchain to ensure the credibility and integrity of the data, and at the same time support the traceability of optimization experience to understand the source and evolution process of optimization experience.

[0186] Through the collaborative optimization interface method, this system has achieved the expansion from single unit optimization to unit group collaborative optimization and then to cross-power station knowledge sharing, greatly improving the utilization efficiency of optimization experience and providing an effective solution for the overall optimization of hydropower station groups.

[0187] Example 10: Intelligent Optimization System for Hydroelectric Unit Parameters with Multi-Modal Data Fusion

[0188] As Figure 8 shown, the present invention also provides an intelligent optimization system for hydroelectric unit parameters with multi-modal data fusion, and this system includes:

[0189] The multimodal data acquisition module is used to obtain the multimodal data of the hydropower unit. The multimodal data includes time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data. This module adopts a distributed design, including a variety of sensors and data acquisition units, to achieve high-precision synchronous acquisition of the unit's multimodal data.

[0190] The feature extraction and fusion module is used to extract and fuse features from the multimodal data to generate a fused feature vector. This module contains multiple dedicated feature extraction networks and a hierarchical attention fusion network, which can extract and fuse valuable feature information from multimodal data.

[0191] The knowledge graph construction and transfer learning module is used to construct a unit knowledge graph and achieve cross-domain knowledge transfer based on graph neural networks to generate prior knowledge representations. This module contains a knowledge graph construction unit and a graph neural network processing unit, which can capture the knowledge associations between units and achieve effective knowledge transfer.

[0192] The parameter optimization module is used to perform parameter optimization based on the Bayesian optimization algorithm with multi-objective constraints to determine the optimal PID parameter configuration. This module contains a Gaussian process modeling unit, a multi-objective acquisition function unit, and a parameter sensitivity analysis unit, which can efficiently and accurately achieve parameter optimization.

[0193] The closed-loop verification and incremental optimization module is used to perform incremental optimization of the optimized parameters through closed-loop verification to obtain the final optimized parameters. This module contains a digital twin verification unit, an incremental learning unit, and a parameter degradation detection unit, which can achieve closed-loop control and continuous improvement of parameter optimization.

[0194] The collaborative optimization interface module is used to achieve coordinated adjustment of unit group parameters and cross-station knowledge transfer. This module contains a plant-level collaborative interface unit, a federated learning unit, and a cross-station knowledge transfer unit, which can expand the optimization scope and improve the utilization efficiency of optimization experience.

[0195] Each module of the system is connected through a data bus to form a complete data flow link. The multimodal data acquisition module transports the collected data to the feature extraction and fusion module, the feature extraction and fusion module transfers the fused feature vector to the knowledge graph construction and transfer learning module, the knowledge graph construction and transfer learning module transports the prior knowledge representation to the parameter optimization module, and the parameter optimization module transfers the optimized parameters to the closed-loop verification and incremental optimization module. Finally, the optimized parameters are obtained and applied to the hydropower unit.

[0196] This system adopts a hierarchical architecture design, including a perception layer, a feature layer, a cognitive layer, a decision layer, and an execution layer. The information transfer between layers is clear, the function division is clear, and they work together to achieve intelligent and efficient optimization of hydropower unit parameters.

[0197] Example 11: Application Case

[0198] This example demonstrates the application effect of the present invention through a practical application case.

[0199] In the parameter optimization project of 4 Francis hydro-generator units in a certain hydropower station, the intelligent optimization method for hydro-generator unit parameters based on multi-modal data fusion of the present invention has achieved remarkable results.

[0200] The operating head range of the units in this hydropower station is 40 - 80 meters, and the rated power is 180 MW. The PID control parameters have not been optimized for a long time, resulting in large differences in the response performance of the units under different head and load conditions. Especially under low head and small load conditions, the stability is poor, and oscillation phenomena frequently occur.

[0201] First, deploy a multi-modal data acquisition network, install various sensors at key parts of the unit to collect time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data.

[0202] Then, perform feature extraction and fusion on the collected multi-modal data to generate a fusion feature vector reflecting the overall state of the unit. Based on the fusion feature vector and historical optimization experience, construct a knowledge graph of the unit, and use graph neural network to achieve cross-unit knowledge transfer to generate a prior knowledge representation suitable for the current unit characteristics.

[0203] Next, perform parameter optimization based on the Bayesian optimization algorithm with multi-objective constraints, comprehensively consider multiple optimization objectives such as response time, overshoot, and stability, and determine the optimal PID parameter configuration. Conduct closed-loop verification through a digital twin model, and perform incremental optimization according to the verification results to obtain the final optimized parameters.

[0204] Finally, apply the optimized parameters to the actual unit, and achieve coordinated adjustment of the parameters of 4 units through a coordinated optimization interface.

[0205] After applying the method of the present invention, the response performance of the unit has been significantly improved: the response time has been shortened from an average of 4.5 seconds to 2.8 seconds, a 37.8% improvement; the overshoot has been reduced from an average of 6.2% to 3.1%, a 50% reduction; the number of oscillations has been reduced from an average of 1.5 times to 0.6 times, a 60% reduction; the adjustment time has been shortened from an average of 38 seconds to 22 seconds, a 42.1% shortening.

[0206] The overall efficiency of the unit has been increased by 4.2%, and the annual increased income is about 4.3 million yuan. The vibration intensity has been reduced by 37%, and the service life of the equipment has been extended by about 4 years. The optimization process only takes 15 minutes, which is more than 95% shorter than the traditional method. The unit can maintain stable operation within the range of ±30 meters of head change without manual intervention, comprehensively improving the adaptability and stability of the unit.

[0207] In addition, through the collaborative optimization of the unit group, the overall operation of the four units is more coordinated. Especially when the water head and load change rapidly, they can quickly adjust and maintain stability, greatly improving the regulating ability and operation reliability of the power station.

[0208] This application case fully demonstrates the technical advantages and practical application value of the present invention, verifying the effectiveness and advancement of multi-modal data fusion, graph neural network knowledge transfer, and Bayesian optimization in the parameter optimization of hydro-generating units.

[0209] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various changes and modifications can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent optimization method for hydropower unit parameters based on multimodal data fusion, characterized in that, Including: Obtain multimodal data of the hydropower unit, where the multimodal data includes time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data; Perform feature extraction and fusion on the multimodal data, including: Use the intra-modal feature extraction network to process various types of data respectively to obtain intra-modal features; The use of the intra-modal feature extraction network to process various types of data respectively includes: Adopt a TCN network based on causal convolution to extract time-domain data features; Adopt an improved ResNet network to extract frequency-domain vibration data features; an attention mechanism is introduced into the improved ResNet network; Adopt Mel spectrum transformation and CNN to extract acoustic data features; Adopt a spatial convolution network to extract temperature distribution data features; Adopt a deep fully connected network to extract hydraulic characteristic data features; Calculate the correlation weights between different modal features based on a hierarchical attention mechanism to generate a fused feature vector; The calculation of the correlation weights between different modal features based on the hierarchical attention mechanism includes: Calculate the attention weight matrix between different modal features: , Wherein: is the eigenvector of the th mode, is the eigenvector of the th mode, is the similarity calculation function; Enhance the original features based on the attention weight matrix: ， Integrate different-scale features through a feature pyramid fusion network to generate a final fused feature vector; Construct a unit knowledge graph and achieve cross-domain knowledge transfer based on a graph neural network, including: Construct a unit knowledge graph based on the fused feature vector and historical optimization experience; Use a graph attention network to extract prior knowledge representations adapted to the current unit characteristics; The use of a graph attention network to extract prior knowledge representations adapted to the current unit characteristics includes: Implement a node-level attention mechanism to calculate the importance weights between nodes; Design an edge-level message passing mechanism to achieve the transfer of knowledge in the graph structure; Through the domain alignment technology of maximum mean discrepancy, achieve cross-unit and cross-headwater knowledge transfer; Generate prior knowledge representations adapted to the current unit characteristics; Perform parameter optimization based on a multi-objective constrained Bayesian optimization algorithm, including: Construct a Gaussian process surrogate model according to the prior knowledge representation; Adopt a multi-objective acquisition function to determine the optimal PID parameter configuration; Perform incremental optimization on the optimized parameters through closed-loop verification to obtain the final optimized parameters.

2. The method according to claim 1, characterized in that, The obtaining of the multimodal data of the hydropower unit includes: Deploy a distributed multimodal sensing data acquisition network; Implement multi-source data synchronous acquisition based on timestamps, ensuring that the maximum time difference does not exceed 5 milliseconds; Perform outlier detection and data standardization processing on the acquired multimodal data; Generate a multimodal data tensor in a unified format.

3. The method according to claim 1, characterized in that The unit knowledge graph includes: Unit nodes, representing the characteristic parameters of the hydropower unit; Operating condition nodes, representing the operating parameters of the unit; Parameter nodes, representing PID control parameters; Physical association edges, representing the hydraulic coupling relationship between units; Performance similarity edges, representing the similarity degree of operating characteristics between units; Historical optimization edges, representing the transfer relationship of historical parameter optimization experience.

4. The method according to claim 1, wherein The multi-objective constrained Bayesian optimization algorithm includes: Construct a Gaussian process surrogate model and adopt a multi-kernel function combination strategy: , where: is the basic kernel function, is the weight coefficient; Design a multi-objective acquisition function to balance multiple optimization objectives such as response time, overshoot, and stability; Implement a dynamic optimization strategy with operating condition perception, and automatically adjust the exploration intensity according to the head change; Develop a parameter sensitivity analysis module to identify key parameters and prioritize optimization.

5. The method according to claim 1, wherein The incremental optimization of the optimized parameters through closed-loop verification includes: Construct a refined digital twin model of the unit to simulate and verify the effect of parameter optimization; Design an extreme condition test framework to evaluate the performance of parameters under boundary conditions; Implement incremental learning based on experience replay to continuously optimize the model performance; Construct a parameter degradation detection model to identify the trend of parameter performance decline; Develop a self-triggered optimization mechanism to automatically start re-optimization when the performance decline exceeds a preset threshold.

6. The method according to claim 1, wherein The method further includes: Design a plant-level collaborative optimization interface to achieve coordinated adjustment of unit group parameters; Develop a knowledge sharing framework based on federated learning to protect data privacy; Implement cross-power station knowledge transfer to support the rapid commissioning of new power stations.

7. An intelligent optimization system for hydro-generator unit parameters with multi-modal data fusion, which is used to execute the method according to any one of claims 1-6, and is characterized in that, It includes: A multi-modal data acquisition module for obtaining multi-modal data of the hydropower unit, where the multi-modal data includes time-domain operation data, frequency-domain vibration data, acoustic feature data, temperature distribution data, and hydraulic characteristic data; A feature extraction and fusion module for extracting and fusing features from the multi-modal data to generate a fused feature vector; A knowledge graph construction and transfer learning module for constructing a unit knowledge graph and implementing cross-domain knowledge transfer based on a graph neural network to generate a prior knowledge representation; A parameter optimization module for parameter optimization based on a multi-objective constrained Bayesian optimization algorithm to determine the optimal PID parameter configuration; A closed-loop verification and incremental optimization module for incrementally optimizing the optimized parameters through closed-loop verification to obtain the final optimized parameters; A collaborative optimization interface module for achieving coordinated adjustment of unit group parameters and cross-power station knowledge transfer.

Images