Full working condition simulation modeling method and system for hydroelectric generating set

Abstract

This invention belongs to the field of water conservancy and hydropower technology, specifically a method and system for full-condition simulation modeling of hydropower units. The method includes: real-time sensing of multi-dimensional operating data of the hydropower unit through a heterogeneous sensor network; establishing a multi-scale dynamic system analysis model based on empirical mode decomposition and variational mode decomposition to deeply analyze the transient effects of water hammer, cavitation characteristics, and load disturbances on hydraulic parameters; establishing a boundary constraint model by integrating nonlinear physical mechanism constraints for boundary regions such as low head and over-rated output; constructing a unified simulation model that integrates hydraulic field, mechanical vibration field, and electromagnetic field; and establishing a nonlinear mapping network between unit structural parameters and model coefficients through a deep neural network to achieve adaptive adjustment of model parameters. Applying this invention enables high-precision simulation under all operating conditions, especially rapid transient processes and boundary conditions, significantly improving the reliability and applicability of hydropower unit simulation models.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy and hydropower technology, specifically a method and system for full-condition simulation modeling of hydropower units. Background Technology

[0002] Hydropower is an important component of renewable and clean energy. Hydropower unit simulation modeling technology is a core technical means to support unit design verification, operation optimization, fault diagnosis and control system debugging. With the continuous advancement of the construction of new power systems, hydropower units frequently participate in deep peak shaving and rapid load change operation, which puts forward higher requirements for simulation accuracy under all operating conditions, especially rapid transition process and boundary conditions.

[0003] The existing hydropower unit simulation system uses the discrete steady-state comprehensive characteristic curves obtained from hydraulic model tests as the core data foundation. It extracts operating parameters such as efficiency, flow rate, and power under the corresponding operating conditions through table lookup and interpolation. Then, it combines these parameters with the physical mechanism models of the water diversion system and the generator side, which can meet the basic simulation calculation requirements in the steady-state operating range near the design operating conditions.

[0004] However, this technical architecture suffers from unavoidable engineering bottlenecks. Discrete steady-state characteristic curves can only reflect the static hydraulic characteristics of the hydropower unit at specific operating points. In actual operation, the unit undergoes a highly nonlinear, continuous dynamic evolution process due to the combined effects of water hammer, cavitation, and load disturbances. The simulation mechanism based on table lookup interpolation cannot fundamentally capture the transient evolution of hydraulic parameters during rapid load changes and other transitional processes. This leads to significant deviations between the simulation results and the actual operating conditions. When the unit operates in boundary regions beyond the coverage of characteristic curves, such as low head and over-rated output, existing technologies can only rely on simple linear extrapolation to complete the process. Calculations are complex, but the hydraulic behavior of hydropower units in the boundary region is dominated by nonlinear factors. The linear assumption is seriously inconsistent with the actual physical mechanism. The lack of physical mechanism constraints in linear extrapolation can cause simulation errors of 20% to 30% in this type of region, which seriously weakens the credibility and guiding value of the simulation results. At the same time, the physical mechanism model and the characteristic curve data are independent of each other and lack a unified simulation framework with deep integration. When the unit is upgraded or the operating conditions change, the simulation model cannot adaptively adjust the parameters according to the new conditions. The hydraulic model test must be carried out again to update the data basis, which greatly limits the applicability and iteration efficiency of the simulation model.

[0005] Therefore, the present invention provides a method and system for full-condition simulation modeling of hydropower units. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0007] The technical solution adopted by this invention to solve its technical problem is: a full-condition simulation modeling method for hydropower units, comprising the following steps: Step 1: Real-time perception of multi-dimensional operating data under all operating conditions. The system synchronously acquires multi-dimensional operating data, including hydraulic, mechanical, and electrical characteristics, through a heterogeneous sensor network deployed at key parts of the hydropower unit. Specifically, this step uses high-precision pressure sensors to collect water pressure distribution in the turbine's flow components, flow sensors to measure the real-time flow of the water intake system, power sensors to obtain the generator's output power, speed sensors to monitor changes in the unit's speed, and efficiency sensors to collect the unit's operating efficiency. The perception process is not limited to acquiring steady-state operating parameters; it also uses vibration sensors to capture the cavitation vibration signal characteristics of the turbine runner blades and water hammer pressure sensors to monitor the transient process of water hammer pressure in the water intake system pipelines. The perceived data is transmitted in real time to the data processing unit, where multi-source data fusion and spatiotemporal alignment are performed to construct a complete information map reflecting the dynamic operating environment of the hydropower unit under all operating conditions, providing a data benchmark with absolute certainty for subsequent dynamic characteristic analysis.

[0008] Step Two: In-depth Analysis of the Dynamic Hydraulic Characteristics of the Hydropower Turbine. This invention utilizes the sensed multi-dimensional operating data to deeply analyze the nonlinear dynamic evolution mechanism of the hydropower unit across the entire operating range. This step treats the hydropower unit as a complex nonlinear dynamic system. By establishing a multi-scale dynamic system analysis model, it quantitatively describes the transient effects of water hammer, cavitation characteristics, and load disturbances on hydraulic parameters. The specific process includes: calculating the response time constant of the turbine under different guide vane openings, identifying the attenuation characteristics and frequency distribution of water pressure pulsations. The analysis step focuses on analyzing the transient evolution law of hydraulic parameters during rapid load changes, and determining the dynamic response process of the turbine head, flow rate, power, and efficiency under sudden load changes. Through this dynamic characteristic analysis, the system can accurately capture the continuous evolution process of hydraulic parameters that cannot be reflected by traditional steady-state characteristic curves, providing dynamic characteristic support for high-precision simulation under all operating conditions.

[0009] Step 3: Boundary Region Physical Mechanism Constraint Fusion Step. Based on the dynamic characteristic analysis results, this invention deeply integrates physical mechanism constraints into the characteristic curve data. This step establishes a boundary constraint model constrained by nonlinear physical mechanisms for boundary regions outside the coverage of characteristic curves, such as low head and over-rated output. In specific implementation, the boundary constraint fusion comprehensively considers the enhanced water hammer effect, cavitation coefficient variation, and cavitation characteristics of the hydropower unit in the boundary region. By introducing a nonlinear hydraulic behavior correction operator for the boundary region, it ensures that the extension of the characteristic curve in the boundary region has the rationality at the physical mechanism level. The system determines whether the simulation calculation has entered the boundary region by comparing the real-time operating data with the prediction results of the boundary constraint model. When the unit operating parameters exceed the coverage of the steady-state characteristic curve, the boundary constraint mechanism is automatically activated for correction, thereby correcting the simulation deviation caused by traditional linear extrapolation.

[0010] Step Four: Multiphysics Coupled Unified Simulation Modeling. Based on the fusion results of dynamic characteristic analysis and boundary constraints, this invention constructs a multiphysics coupled unified simulation model that integrates the hydraulic model, characteristic curve data, and boundary constraints. This step treats the entire operating process of the hydropower unit as a unified multiphysics coupled system for modeling. By establishing a set of coupled equations between the hydraulic field, mechanical vibration field, and electromagnetic field, the multiphysics interaction of the hydropower unit is quantitatively described within the entire operating range, especially during rapid transition processes and boundary conditions. Specifically, the coupled simulation model integrates the dynamic hydraulic characteristic curves of the turbine, the physical mechanism constraints of the boundary region, the water hammer transfer function of the water diversion system, and the electromagnetic dynamic equations of the generator. The coupled equations are solved using a multiphysics coupled numerical calculation method to obtain continuous dynamic simulation results of the hydropower unit within the entire operating range. This unified simulation framework eliminates the drawbacks of the independence between the physical mechanism model and characteristic curve data in traditional technologies, achieving consistency and high accuracy in full-condition simulation.

[0011] Step 5: Adaptive Adjustment of Model Parameters. This invention establishes a deep mapping relationship between simulation model parameters and unit operating conditions. This step is configured to automatically adjust the structural parameters and characteristic curve parameters of the simulation model according to the new unit parameters and operating boundary conditions when the hydropower unit undergoes capacity expansion or its operating conditions change. In specific implementation, the adaptive adjustment mechanism establishes a nonlinear mapping network between key parameters such as the unit's rated capacity, design head, and rated speed and the core coefficients of the simulation model. By comparing historical operating data with simulation results, the weight coefficients in the mapping network are dynamically corrected to achieve adaptive updates of the simulation model. The adaptive adjustment process does not require re-conducting hydraulic model tests; model updates can be completed solely through parameter calibration, significantly improving the applicability and iteration efficiency of the simulation model.

[0012] Preferably, in the real-time sensing step of multi-dimensional operation data under all operating conditions, the deployment of the heterogeneous sensor network covers key parts of the hydropower unit, such as the water intake system, pressure steel pipe, turbine casing, tailrace pipe, generator, and governor. The sampling time of the multi-source sensor data is kept consistent through a time synchronization protocol. The sampling frequency is set in stages according to the dynamic characteristics of the hydraulic parameters, wherein the sampling frequency of the water hammer pressure signal is not less than 1 kHz, and the sampling frequency of the power and efficiency signals is not less than 100 Hz.

[0013] Preferably, in the step of in-depth analysis of the dynamic hydraulic characteristics of the turbine, the multi-scale dynamic system analysis model adopts an analysis framework combining empirical mode decomposition (EMD) and variational mode decomposition (VMD). EMD extracts the intrinsic mode components from the hydraulic parameter signal, while VMD identifies the attenuation characteristics and energy distribution of different frequency components. Specifically, EMD uses cubic spline interpolation to construct upper and lower envelopes and iteratively filters the original signal into several intrinsic mode components. VMD uses the minimization of the sum of bandwidths of each mode component as the objective function and iteratively solves the problem using the alternating direction multiplier method to obtain the center frequency and bandwidth of each mode component, thereby identifying the attenuation characteristics and energy distribution of different frequency components. This analysis framework can effectively separate signal components generated by different mechanisms such as water hammer pressure waves, cavitation vibration waves, and load disturbance waves, achieving accurate characterization of the transient evolution law of hydraulic parameters.

[0014] Preferably, in the boundary region physical mechanism constraint fusion step, the boundary constraint model adopts a neural network-based adaptive boundary detection algorithm. This neural network adaptive boundary detection algorithm uses a convolutional neural network structure. The network input is a 64×1 dimensional real-time running data vector (containing parameters such as head, flow rate, power, efficiency, and rotational speed), and the network output is a boundary confidence value (a floating-point number between 0 and 1). The network structure includes three convolutional layers (with 32, 64, and 128 kernels respectively, and a kernel size of 3×1), two fully connected layers (with 256 and 64 neurons respectively), and one output layer (with one neuron). The activation function is the ReLU function, and the output layer uses the Sigmoid function. The training samples are selected from 200 sets of boundary condition samples and 200 sets of steady-state condition samples from historical operating data. The training samples are used to learn the hydraulic behavior patterns of the hydropower unit under different boundary conditions and automatically identify the boundary threshold of the characteristic curve coverage area. The boundary confidence score output by the boundary detection algorithm is used as the trigger criterion for activating the boundary constraint mechanism. When the boundary confidence score exceeds the preset threshold, the algorithm automatically switches to the boundary constraint fusion calculation mode.

[0015] Preferably, in the multiphysics coupling unified simulation modeling step, the solution of the coupling equations adopts a numerical calculation method that combines implicit time advancement and explicit spatial discretization. Implicit time advancement ensures the numerical stability of the calculation process, while explicit spatial discretization ensures the efficient processing of multiphysics coupling terms. The convergence accuracy of the numerical calculation method is controlled on the order of 10 to the power of negative 6 to ensure the reliability of the simulation results.

[0016] Preferably, in the adaptive adjustment step of the model parameters, the nonlinear mapping network adopts a deep neural network architecture. The network includes an input layer, at least one hidden layer, and an output layer. The input layer receives structural parameters such as the rated capacity of the unit, the design head, the rated speed, and the blade angle range, as well as the operating boundary conditions. The output layer outputs the core adjustment coefficients of the simulation model. The hidden layer realizes a highly nonlinear mapping from the parameter space to the coefficient space through an activation function. The network is trained using a gradient descent optimization algorithm, and the training samples are derived from a fusion dataset of historical operating data and hydraulic model test data.

[0017] The present invention also provides a full-condition simulation modeling system for hydropower units, including a multi-dimensional sensing hardware module, a dynamic characteristic analysis module, a boundary constraint fusion module, a multi-physics coupling simulation module, a model adaptive adjustment module, a data processing module, a simulation engine module, and a user interface module.

[0018] The multi-dimensional sensing hardware module is configured to construct a multi-parameter acquisition matrix at key parts of the hydropower unit. This module includes a high-frequency pressure sensor deployed in the water diversion system, a flow sensor deployed in the turbine casing, a power sensor deployed at the generator output, a speed sensor deployed on the main shaft of the unit, and a vibration sensor deployed in the turbine runner chamber. All sensing hardware is electrically coupled to the data processing module through an industrial bus to ensure the real-time performance and high concurrency processing capability of data acquisition. The sensing hardware module not only provides steady-state operating parameters but also provides water hammer pressure signals and cavitation vibration signals that reflect the dynamic process of the hydropower unit, providing raw information support for the dynamic characteristic analysis module.

[0019] The dynamic characteristic analysis module is electrically connected to the multi-dimensional sensing hardware module. It integrates a multi-scale dynamic system analysis unit. This module is responsible for receiving multi-dimensional operating data and using embedded empirical mode decomposition and variational mode decomposition algorithms to perform in-depth analysis of hydraulic parameter signals. Based on measured data, this module can automatically identify the dynamic response characteristics of the hydropower unit, extract the influence weights of water hammer effect, cavitation characteristics, and load disturbance on hydraulic parameters, thereby completing the in-depth analysis of the dynamic hydraulic characteristics of the turbine and providing high-dimensional information at the dynamic characteristic level for subsequent coupled simulation.

[0020] The boundary constraint fusion module is connected to the dynamic characteristic analysis module and is configured to perform physical mechanism constraint calculation and characteristic curve fusion in the boundary region. This module performs boundary determination on the operating status of the hydropower unit by integrating a neural network adaptive boundary detection algorithm. When the unit is detected to have entered the boundary region, the boundary constraint fusion mechanism is automatically activated. This module performs physical mechanism constraint correction on the characteristic curve data by introducing a nonlinear hydraulic behavior correction operator in the boundary region, and outputs the fused boundary constraint characteristic curve in real time, providing reliable data support for the boundary region for the coupled simulation module.

[0021] The multiphysics coupled simulation module integrates a coupled calculation engine for hydraulic field, mechanical vibration field and electromagnetic field. This module obtains the dynamic hydraulic characteristics output by the dynamic characteristic analysis module and the boundary constraint characteristic curves output by the boundary constraint fusion module. By establishing a set of coupled multiphysics equations, it performs unified simulation calculations across the entire operating range. The coupled calculation engine adopts a numerical calculation method of implicit time progression and explicit spatial discretization, which can efficiently solve the coupled equations while ensuring numerical stability, and output the dynamic simulation results of the hydropower unit in real time across the entire operating range.

[0022] The adaptive adjustment module is communicatively connected to the multiphysics coupled simulation module and is configured to perform adaptive updates of the simulation model parameters. This module integrates a deep neural network mapping network, which establishes a nonlinear mapping relationship between the unit structural parameters and the core coefficients of the simulation model. When the unit is upgraded or the operating conditions change, the module receives new unit parameter inputs, calculates the adjusted model coefficients through the mapping network, and pushes the adjusted coefficients to the coupled simulation module to achieve adaptive updates of the simulation model. The adaptive adjustment module continuously optimizes the prediction accuracy of the mapping network by continuously comparing simulation results with measured data and using a feedback mechanism.

[0023] The data processing module interacts with the multi-dimensional sensing hardware module, the dynamic characteristic analysis module, the boundary constraint fusion module, and the simulation engine module. This module is responsible for receiving raw sensing data, performing multi-source data fusion, noise filtering, and spatiotemporal alignment processing, and distributing the processed standardized data to the corresponding functional modules. The data processing module also undertakes the tasks of storing intermediate results and caching data during the simulation process to ensure the continuity of data supply for multi-physics coupling simulation.

[0024] The simulation engine module is electrically connected to the multiphysics coupled simulation module and the model adaptive adjustment module. It integrates a high-performance numerical calculation unit and a result output unit. The module receives real-time simulation data calculated by the coupled simulation module, combines it with the adaptively adjusted model parameters, executes the final solution of the multiphysics coupled simulation, and converts the simulation results into a standardized output format. The simulation engine module also supports real-time monitoring and breakpoint resume functions of the simulation process to ensure the integrity and reliability of the full-condition simulation calculation.

[0025] The user interface module is communicatively connected to the simulation engine module and the data processing module, and is configured to provide human-computer interaction and result visualization. This module supports the configuration and modification of simulation parameters, including setting operating conditions, inputting boundary conditions, and adjusting model coefficients. This module also provides a graphical display function for simulation results, including drawing the turbine's operating trajectory, presenting efficiency characteristic surfaces, drawing dynamic curves of the transition process, and displaying boundary area warning information. The user interface module assists engineering technicians in analyzing simulation results and optimizing operating strategies through intuitive visualization.

[0026] The beneficial effects of this invention are as follows: 1. The present invention provides a full-condition simulation modeling method and system for hydropower units. By establishing a multi-scale dynamic system analysis model, it deeply analyzes the nonlinear dynamic evolution mechanism of hydropower units within the full operating condition range. Through an analysis framework combining empirical mode decomposition and variational mode decomposition, it accurately captures the transient effects of water hammer, cavitation characteristics, and load disturbances on hydraulic parameters. This solves the technical problem in the prior art that discrete steady-state characteristic curves cannot reflect the true dynamic evolution law of rapid load change processes, and achieves a significant improvement in the simulation accuracy of transient processes.

[0027] 2. The full-condition simulation modeling method and system for hydropower units described in this invention integrates nonlinear physical mechanism constraints into characteristic curve data through a boundary region physical mechanism constraint fusion mechanism. It automatically identifies the boundary threshold of the characteristic curve coverage area through a neural network adaptive boundary detection algorithm and introduces a nonlinear hydraulic behavior correction operator in the boundary region to perform physical mechanism-level constraint correction on the characteristic curve data. This solves the simulation deviation caused by linear extrapolation of the boundary region in the prior art, and reduces the boundary region simulation error from 20% to 30% to a reasonable range.

[0028] 3. The present invention provides a full-condition simulation modeling method and system for hydropower units. Through a multi-physics field coupled unified simulation modeling framework, the hydraulic model, characteristic curve data, and boundary constraints are deeply integrated. A set of coupled equations between the hydraulic field, mechanical vibration field, and electromagnetic field is established. The unified calculation of full-condition simulation is realized through the multi-physics field coupled numerical calculation method. This eliminates the error superposition effect caused by the independence of the physical mechanism model and characteristic curve data in traditional technology, and establishes a unified simulation framework.

[0029] 4. The full-condition simulation modeling method and system for hydropower units described in this invention establishes a deep neural network mapping relationship between the unit structural parameters and the core coefficients of the simulation model through an adaptive adjustment mechanism for model parameters. When the unit is upgraded or the operating conditions change, the simulation model can automatically adjust the model coefficients according to the new unit parameters. The model can be updated without having to conduct hydraulic model tests again, which significantly improves the applicability and iteration efficiency of the simulation model and solves the engineering bottleneck of the inability of the model to adaptively adjust in the prior art. Attached Figure Description

[0030] The invention will now be further described with reference to the accompanying drawings.

[0031] Figure 1 This is a structural block diagram of a full-condition simulation modeling system for hydropower units according to the present invention; Figure 2 This is a flowchart of a full-condition simulation modeling method for hydropower units in this invention. Detailed Implementation

[0032] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0033] Example 1: like Figure 1 As shown in the embodiment of this application, a full-condition simulation modeling system for hydropower units is provided, including a multi-dimensional sensing hardware module, a dynamic characteristic analysis module, a boundary constraint fusion module, a multi-physics field coupling simulation module, a model adaptive adjustment module, a data processing module, a simulation engine module, and a user interface module.

[0034] The multi-dimensional sensing hardware module is used to collect multi-dimensional operating data containing hydraulic, mechanical and electrical characteristics in real time through a sensor network deployed in key parts of the hydropower unit. The collected data is then fused from multiple sources and spatiotemporally aligned to construct an information map that reflects the dynamic operating environment under all working conditions. The dynamic characteristic analysis module is used to establish a multi-scale dynamic system analysis model based on multi-dimensional operating data, calculate the turbine response time constant, identify the water pressure pulsation attenuation characteristics and frequency distribution, analyze the dynamic response process of head, flow, power and efficiency under sudden load changes, and accurately capture the continuous evolution characteristics of hydraulic parameters. The boundary constraint fusion module is used to establish a boundary constraint model constrained by nonlinear physical mechanisms based on the dynamic characteristic analysis results. By introducing a nonlinear hydraulic behavior correction operator in the boundary region, the characteristic curve data is constrained and corrected to solve the simulation deviation caused by linear extrapolation in the boundary region. The multiphysics coupling simulation module is used to model the entire operating process of a hydropower unit as a unified multiphysics coupling system, establish a set of coupling equations between the hydraulic field, mechanical vibration field and electromagnetic field, and solve them through the multiphysics coupling numerical calculation method to obtain the continuous dynamic simulation results of the entire operating condition. The model adaptive adjustment module is used to establish a nonlinear mapping network between unit parameters and simulation model coefficients. When the hydropower unit is upgraded or the operating conditions change, the model coefficients are automatically adjusted according to the new unit parameters, and the model can be updated without having to conduct hydraulic model tests again.

[0035] The full-condition simulation modeling system for hydropower units provided in this embodiment includes the following components: a multi-dimensional sensing hardware module is configured to construct a multi-parameter acquisition matrix at key parts of the hydropower unit. This module includes a high-frequency pressure sensor deployed in the water diversion system, a flow sensor deployed in the turbine casing, a power sensor deployed at the generator output end, a speed sensor deployed on the main shaft of the unit, and a vibration sensor deployed in the turbine runner chamber.

[0036] All sensing hardware is electrically coupled to the data processing module via an industrial bus to ensure real-time data acquisition and high-concurrency processing capabilities. The sensing hardware module not only provides steady-state operating parameters, but also provides water hammer pressure signals and cavitation vibration signals that reflect the dynamic process of the hydropower unit, providing raw information support for the dynamic characteristic analysis module.

[0037] Specifically, high-frequency pressure sensors are used to collect water pressure distribution data from the turbine's flow components, acquiring pressure data at key locations such as the volute inlet pressure, guide vane area pressure, and tailrace outlet pressure. Flow sensors are used to measure the real-time flow rate of the water intake system, obtaining the water flow rate within the intake pipe through ultrasonic or electromagnetic flow meters.

[0038] Power sensors are used to obtain the output power of the generator and monitor the active and reactive power on the stator side of the generator in real time through power measurement devices. Speed ​​sensors are used to monitor the speed changes of the unit and obtain the real-time speed signal of the main shaft of the unit through speed probes. Vibration sensors are used to capture the cavitation vibration signal characteristics of the turbine runner blades and obtain the vibration signal in the runner chamber through acceleration sensors or displacement sensors.

[0039] The data processing module interacts with the multi-dimensional sensing hardware module, dynamic characteristic analysis module, boundary constraint fusion module, and simulation engine module. This module is responsible for receiving raw sensing data, performing multi-source data fusion, noise filtering, and spatiotemporal alignment processing, and distributing the processed standardized data to the corresponding functional modules.

[0040] The data processing module also undertakes the tasks of storing intermediate results and caching data during the simulation process, ensuring the continuity of data supply for multi-physics coupled simulation. During the multi-source data fusion process, the data processing module ensures the consistency of sampling time of multi-source sensor data through time synchronization protocol. The sampling frequency is set in stages according to the dynamic characteristics of hydraulic parameters, with the sampling frequency of water hammer pressure signal not less than 1 kilohertz and the sampling frequency of power and efficiency signals not less than 100 kilohertz.

[0041] The dynamic characteristic analysis module is electrically connected to the multi-dimensional sensing hardware module. It integrates a multi-scale dynamic system analysis unit. This module is responsible for receiving multi-dimensional operating data and using embedded empirical mode decomposition and variational mode decomposition algorithms to perform in-depth analysis of hydraulic parameter signals.

[0042] This module can automatically identify the dynamic response characteristics of hydropower units based on measured data, extract the influence weights of water hammer effect, cavitation characteristics and load disturbance on hydraulic parameters, thereby completing an in-depth analysis of the dynamic hydraulic characteristics of the turbine and providing high-dimensional information on dynamic characteristics for subsequent coupled simulation.

[0043] The boundary constraint fusion module is connected to the dynamic characteristic analysis module and is configured to perform physical mechanism constraint calculation and characteristic curve fusion of the boundary region. This module performs boundary determination on the operating status of the hydropower unit by integrating a neural network adaptive boundary detection algorithm. When the unit is detected to have entered the boundary region, the boundary constraint fusion mechanism is automatically activated.

[0044] The boundary constraint fusion module performs constraint correction on the characteristic curve data by introducing a nonlinear hydraulic behavior correction operator in the boundary region. Specifically, the boundary constraint fusion module enhances the correction coefficient by introducing the water hammer effect. cavitation coefficient variation correction factor and cavitation characteristic correction coefficient Constraint corrections are applied to the characteristic curve data; under typical rated operating conditions, the water hammer effect enhancement correction factor is... The reference value range is 1.0~1.3, and the correction factor for the variation of the cavitation coefficient is... The reference value range is 1.0~1.5, cavitation characteristic correction coefficient. The reference value range is 0.9~1.2; the fitting process uses measured boundary condition data as a benchmark, and employs the least squares method for nonlinear regression fitting, with the goodness of fit... It needs to be greater than 0.95.

[0045] The correction coefficients are expressed as follows: The correction factor for the water hammer effect is given, where Flow rate (unit: m³ / s or equivalent engineering unit). Water head (unit: m). Rotational speed (in r / min); This is the correction factor for the change in cavitation coefficient, where The cavitation coefficient (dimensionless, defined as follows) , To suck in saliva, (for vaporization pressure head); These are correction coefficients for cavitation characteristics. The specific analytical form of each correction coefficient is determined through fitting experimental data, specifically including the following steps: S1: Collect experimental data under different boundary conditions, including parameters such as head, flow rate, rotational speed, and cavitation coefficient; S2: Choose the function form of the correction coefficient, such as a polynomial function, exponential function, or logarithmic function; S3: Use the least squares method or nonlinear regression method to fit the experimental data and establish a quantitative relationship model between the correction coefficient and each physical quantity; S4: Verify the prediction accuracy of the correction coefficient model. If the accuracy does not meet the requirements, return to S2 to reselect the function form.

[0046] The multiphysics coupling simulation module integrates the coupling calculation engine of hydraulic field, mechanical vibration field and electromagnetic field. This module obtains the dynamic hydraulic characteristics output by the dynamic characteristic analysis module and the boundary constraint characteristic curves output by the boundary constraint fusion module. By establishing a multiphysics coupling equation set, it performs unified simulation calculations across the entire working condition range.

[0047] The coupled computation engine employs a numerical computation method that combines implicit time progression with explicit spatial discretization. This method can efficiently solve coupled equations while ensuring numerical stability, and output dynamic simulation results of hydropower units in real time across the entire operating range.

[0048] The simulation engine module is electrically connected to the multiphysics coupled simulation module and the model adaptive adjustment module. It integrates a high-performance numerical calculation unit and a result output unit. This module receives real-time simulation data calculated by the coupled simulation module, combines it with the adaptively adjusted model parameters, executes the final solution of the multiphysics coupled simulation, and converts the simulation results into a standardized output format.

[0049] The simulation engine module also supports real-time monitoring and breakpoint resume functions during the simulation process, ensuring the integrity and reliability of full-condition simulation calculations.

[0050] The model adaptive adjustment module is communicatively connected to the multiphysics coupled simulation module and is configured to perform adaptive updates of the simulation model parameters. This module integrates a deep neural network mapping network, which establishes a nonlinear mapping relationship between the unit structural parameters and the core coefficients of the simulation model.

[0051] When the unit is upgraded or the operating conditions change, this module receives new unit parameter inputs, calculates the adjusted model coefficients through the mapping network, and pushes the adjusted coefficients to the coupled simulation module to realize the adaptive update of the simulation model. The adaptive adjustment module continuously optimizes the prediction accuracy of the mapping network by continuously comparing the simulation results with the measured data and using the feedback mechanism.

[0052] The user interface module communicates with the simulation engine module and the data processing module, and is configured to provide human-computer interaction and result visualization. This module supports the configuration and modification of simulation parameters, including setting operating conditions, inputting boundary conditions, and adjusting model coefficients.

[0053] This module also provides graphical display functions for simulation results, including turbine running trajectory plotting, efficiency characteristic surface presentation, transient process dynamic curve plotting, and boundary area early warning information display. The user interface module assists engineers in analyzing simulation results and optimizing operation strategies through intuitive visualization.

[0054] like Figure 2 As shown, the full-condition simulation modeling method for hydropower units described in this invention includes the following steps: Step 1: Collect multi-dimensional operational data containing hydraulic, mechanical and electrical characteristics in real time through a sensor network deployed at key parts of the hydropower unit. Perform multi-source fusion and spatiotemporal alignment processing on the collected data to construct an information map reflecting the dynamic operating environment under all working conditions.

[0055] The multi-dimensional sensing hardware module collects water pressure distribution of turbine flow components through high-frequency pressure sensors deployed in the water diversion system, measures real-time flow of the water diversion system through flow sensors, obtains generator output power through power sensors, and monitors unit speed changes through speed sensors.

[0056] The sensing process is not limited to acquiring steady-state operating parameters, but also captures the cavitation vibration signal characteristics of turbine runner blades through vibration sensors and monitors the transient process of water hammer pressure in the water intake system pipeline through water hammer pressure sensors.

[0057] The data processing module performs multi-source data fusion processing on the collected raw data, performs noise filtering to remove measurement noise and interference signals, and performs spatiotemporal alignment processing to calibrate the time deviation caused by inconsistent sampling times of different sensors.

[0058] The time synchronization protocol ensures the consistency of sampling time for multi-source sensor data. The sampling frequency is set in stages according to the dynamic characteristics of hydraulic parameters. The sampling frequency of water hammer pressure signal is set to 2 kHz, the sampling frequency of power and efficiency signal is set to 200 Hz, and the sampling frequency of other signals is set to 1 kHz.

[0059] The processed multidimensional operating data is used to construct a complete information map reflecting the dynamic operating environment of the hydropower unit under all operating conditions, providing a data benchmark with absolute certainty for subsequent dynamic characteristic analysis.

[0060] Step Two: Based on multi-dimensional operating data, establish a multi-scale dynamic system analysis model to calculate the turbine response time constant, identify the water pressure pulsation attenuation characteristics and frequency distribution, and analyze the dynamic response process of head, flow, power, and efficiency under sudden load changes, accurately capturing the continuous evolution characteristics of hydraulic parameters. The dynamic characteristic analysis module treats the hydropower unit as a complex nonlinear dynamic system. By establishing a multi-scale dynamic system analysis model, it quantitatively describes the water hammer effect, cavitation characteristics, and transient effects of load disturbances on hydraulic parameters. The multi-scale dynamic system analysis model adopts an analysis framework combining empirical mode decomposition and variational mode decomposition. The specific analysis process is divided into the following three stages.

[0061] The first stage is the empirical mode decomposition extraction stage. In this stage, the hydraulic parameter signal is processed by empirical mode decomposition, which decomposes the original signal into several intrinsic mode components. Each intrinsic mode component satisfies the condition that the number of signal extrema points is equal to or differs from the number of zero crossings by no more than one, and its upper and lower envelopes are locally symmetrical about the time axis. Through empirical mode decomposition, the water hammer pressure signal, cavitation vibration signal and load disturbance signal are separated into different intrinsic mode components.

[0062] The second stage is the variational mode decomposition identification stage. Based on the empirical mode decomposition results, a variational constraint problem is established. Each intrinsic mode component is used as a variable to be solved, and a variational constraint problem is constructed that includes two constraints: minimizing the sum of the bandwidths of each intrinsic mode component and minimizing the sum of the center frequencies of each intrinsic mode component.

[0063] The variational constrained problem is solved iteratively using the alternating direction multiplier algorithm to obtain the center frequency and bandwidth of each intrinsic mode component, thereby identifying the attenuation characteristics and energy distribution of different frequency components. Variational mode decomposition can effectively separate signal components generated by different mechanisms such as water hammer pressure waves, cavitation vibration waves, and load disturbance waves, achieving accurate characterization of the transient evolution of hydraulic parameters. Based on the decomposition results, the system calculates the dominant frequency and its attenuation time constant of the water hammer pressure wave, calculates the frequency distribution and energy proportion of the cavitation vibration wave, and obtains the energy proportion of each frequency component.

[0064] The third stage is the dynamic response analysis stage. Based on the results of the multi-scale analysis, this stage further analyzes the dynamic response process of head, flow, power, and efficiency under load abrupt changes. The specific analysis process includes calculating the response time constant of the turbine at different guide vane openings, identifying the attenuation characteristics and frequency distribution of water pressure pulsations, quantitatively describing the rate of change of head, flow, power, and efficiency at the moment of load abrupt changes, and calculating the dynamic response time and overshoot of head, flow, power, and efficiency during the load abrupt changes.

[0065] Through this dynamic characteristic analysis, the system can accurately capture the continuous evolution of hydraulic parameters that cannot be reflected by traditional steady-state characteristic curves, and output dynamic characteristic analysis results such as turbine response time constant, water pressure pulsation attenuation characteristics and frequency distribution, and dynamic response process of head, flow and power efficiency under load change conditions, providing dynamic characteristic support for subsequent coupled simulation.

[0066] Step 3: Based on the dynamic characteristic analysis results, establish a boundary constraint model constrained by nonlinear physical mechanisms. By introducing a nonlinear hydraulic behavior correction operator in the boundary region, constrain and correct the characteristic curve data to solve the simulation deviation caused by linear extrapolation in the boundary region.

[0067] The boundary constraint fusion module receives the dynamic characteristic analysis results output by the dynamic characteristic analysis module and begins to execute the boundary constraint model establishment and fusion process. The boundary constraint model establishment process is divided into the following three sub-steps.

[0068] The first sub-step is the boundary determination sub-step. This sub-step uses a neural network adaptive boundary detection algorithm to determine the boundary of the hydropower unit's operating state. Specifically, real-time operating data and dynamic characteristic analysis results are input into the neural network adaptive boundary detection algorithm. This algorithm learns the hydraulic behavior pattern of the hydropower unit under different boundary conditions through training samples, automatically identifies the boundary threshold of the characteristic curve coverage area, and outputs the boundary confidence score as the criterion for triggering boundary constraint fusion. When the boundary confidence score exceeds the preset threshold, the system determines that the unit has entered the boundary region and automatically switches to the boundary constraint fusion calculation mode. When the boundary confidence score does not exceed the preset threshold, the system determines that the unit is operating within the coverage area of ​​the steady-state characteristic curve and continues to use standard characteristic curve data for simulation calculation.

[0069] The second sub-step is the physical mechanism fusion sub-step. When the system determines that the unit has entered the boundary region, the boundary constraint fusion mechanism is automatically activated. This sub-step introduces a nonlinear hydraulic behavior correction operator for the boundary region to perform physical mechanism-level constraint correction on the characteristic curve data. The nonlinear hydraulic behavior correction operator comprehensively considers the enhanced water hammer effect, cavitation coefficient variation, and flow cavitation characteristics of the hydropower unit in the boundary region. The enhanced water hammer effect is manifested in the nonlinear changes in the propagation velocity and reflection characteristics of the water hammer pressure wave under low head or over-rated output conditions. The correction operator calculates the correction amount of the water hammer pressure based on the enhanced water hammer effect. The cavitation coefficient variation is manifested in the nonlinear increase trend of the cavitation coefficient as the head decreases under low head conditions. The correction operator calculates the correction range of the cavitation region based on the cavitation coefficient variation characteristics.

[0070] The cavitation characteristics of water flow are characterized by the intensification of cavitation in the boundary region. The generation and collapse of cavitation bubbles have a significant impact on hydraulic parameters. The correction operator calculates the correction coefficient of cavitation on flow rate and efficiency based on the cavitation characteristics of water flow. The hydraulic behavior offset in the boundary region is predicted by a neural network. This offset is then superimposed on the original characteristic curve data as a correction amount to obtain the boundary constraint characteristic curve that integrates physical mechanism constraints.

[0071] The third sub-step is the constraint correction sub-step. This sub-step compares the real-time running data with the prediction results of the boundary constraint model to determine whether the simulation calculation needs further correction. When the deviation between the real-time running data and the prediction results of the boundary constraint model exceeds the preset tolerance, the system adjusts the parameters of the nonlinear hydraulic behavior correction operator and re-performs the constraint correction iterative calculation until the deviation meets the convergence condition. Through this boundary constraint fusion process, the extension of the characteristic curve in the boundary region is ensured to have physical mechanism rationality, effectively solving the simulation deviation problem caused by traditional linear extrapolation.

[0072] Step 4: The entire operating process of the hydropower unit is modeled as a unified multi-physics coupled system. A set of coupled equations between the hydraulic field, mechanical vibration field and electromagnetic field is established and solved by the multi-physics coupled numerical calculation method to obtain the continuous dynamic simulation results of the entire operating condition. The multi-physics coupled simulation module obtains the dynamic hydraulic characteristics output by the dynamic characteristic analysis module and the boundary constraint characteristic curves output by the boundary constraint fusion module, and starts the multi-physics coupled unified simulation modeling process.

[0073] The establishment of the coupled equation set is divided into four stages. The first stage is the establishment of the hydraulic field equation set, which describes the dynamic behavior of water flow in the turbine's flow passage components. The hydraulic field equation set includes a continuity equation to describe the conservation of water flow mass, a momentum equation to describe the laws governing water flow motion, and an energy equation to describe the energy conversion relationships of the water flow. The inputs to the hydraulic field equation set include hydraulic parameters such as real-time flow rate, pressure distribution, and head height. These parameters are derived from real-time operating data provided by the data processing module and dynamic hydraulic characteristics provided by the dynamic characteristic analysis module.

[0074] The second stage is the establishment stage of the mechanical vibration field equation set. In this stage, the mechanical vibration field equation set is established to describe the vibration characteristics of the turbine runner blades and main shaft. The mechanical vibration field equation set includes vibration equations to describe the vibration response of the runner blades and main shaft under external forces, and damping equations to describe the dissipation characteristics of vibration energy. The hydraulic excitation terms in the mechanical vibration field equation set are provided by the hydraulic field equation set, including the force and torque of the water flow on the runner blades.

[0075] The third stage is the establishment of the electromagnetic field equations, which describes the internal electromagnetic dynamics of the generator. The electromagnetic field equations include voltage equations to describe the voltage balance between the stator and rotor windings, torque equations to describe the balance between electromagnetic torque and mechanical torque, and flux linkage equations to describe the relationship between flux linkage and current in each winding. The mechanical speed term in the electromagnetic field equations is provided by the mechanical vibration field equations.

[0076] The fourth stage is the establishment and solution of the coupled equations. In this stage, the hydraulic field equations, the mechanical vibration field equations, and the electromagnetic field equations are coupled and integrated to establish a multi-physics field coupled equation set. The coupling relationship between the fields is reflected through coupling terms. The hydraulic excitation term in the mechanical vibration field equations comes from the calculation results of the hydraulic field equations, and the mechanical rotation speed term in the electromagnetic field equations comes from the calculation results of the mechanical vibration field equations.

[0077] The solution of the coupled equations employs a numerical computation method combining implicit time-propagation and explicit spatial discretization. The implicit time-propagation process first implicitly discretizes the coupled equations with respect to the time variable, expressing the time derivative in terms of the value at the next moment, forming a nonlinear equation system. Then, the Newton-Raphson iteration method is used to iteratively solve this nonlinear equation system. At each iteration step, the state variables of each physical field are calculated and the residuals are checked to see if they meet the convergence accuracy requirements.

[0078] The explicit spatial discretization process explicitly calculates the spatial derivatives of the coupled equations using known values ​​at the current time, achieving efficient handling of multiphysics coupling terms. Implicit time propagation ensures the numerical stability of the computation process, while explicit spatial discretization ensures efficient handling of multiphysics coupling terms.

[0079] The convergence accuracy is controlled to the order of 10 to the power of negative 6 to ensure the reliability of the simulation results. Through multiphysics coupled numerical calculation, continuous dynamic simulation results of the hydropower unit are obtained within the full operating range, including the dynamic changes of parameters such as head, flow rate, power, efficiency, speed, and vibration.

[0080] Step 5: Establish a nonlinear mapping network between unit parameters and simulation model coefficients. When the hydropower unit is upgraded or the operating conditions change, the model coefficients are automatically adjusted according to the new unit parameters, and the model can be updated without having to conduct hydraulic model tests again.

[0081] The model adaptive adjustment module establishes a deep neural network mapping network between the unit's structural parameters and the core coefficients of the simulation model. This network establishes a nonlinear mapping relationship between key parameters such as the unit's rated capacity, design head, and rated speed and the core coefficients of the simulation model.

[0082] The architecture of the deep neural network mapping network includes an input layer, at least one hidden layer, and an output layer. The input layer receives structural parameters such as the rated capacity, design head, rated speed, and blade angle range of the receiver unit, as well as operational boundary conditions. The network training uses the mean square error loss function, with the optimization objective being to minimize the deviation between the model simulation values ​​and the measured values. The training convergence criterion is that the loss decreases by less than 10% after 100 consecutive training rounds. -4 Or the number of training rounds reaches the preset limit.

[0083] The number of hidden layers is set to five, with each layer containing 128 neurons. Adjacent hidden layers are connected by a weight matrix, and the hidden layers achieve a highly nonlinear mapping from the parameter space to the coefficient space through activation functions.

[0084] The output layer outputs the core adjustment coefficients of the simulation model, including the parameter adjustment coefficients for the turbine's comprehensive characteristic curve, efficiency correction coefficients, water hammer effect parameters, cavitation parameters, etc. The network training employs a gradient descent optimization algorithm, with a learning rate set to 0.001 and a momentum factor set to 0.9.

[0085] The training samples are derived from a fusion dataset of historical operating data and hydraulic model test data. This dataset contains structural parameters and corresponding simulation model coefficients for different types of hydropower units under various operating conditions. When hydropower units undergo capacity expansion or their operating conditions change, the model adaptive adjustment module receives new unit parameter inputs, inputs these new unit parameters into a deep neural network mapping network, and calculates the adjusted model coefficients through forward propagation.

[0086] The module pushes the adjustment coefficients to the multiphysics coupled simulation module. The coupled simulation module recalculates the parameters of the coupled equations based on the new model coefficients. The updated simulation model can accurately reflect the characteristics of the hydropower unit after capacity expansion or under new operating conditions.

[0087] The adaptive model adjustment module continuously compares simulation results with measured data, dynamically corrects the weight coefficients in the mapping network using a feedback mechanism, and continuously optimizes the network's prediction accuracy, thus achieving adaptive updates to the simulation model. This adaptive adjustment process does not require conducting new hydraulic model tests; model updates can be completed solely through parameter calibration, significantly improving the applicability and iteration efficiency of the simulation model.

[0088] Taking the full-condition simulation of a 300 MW mixed-flow turbine generator unit in a hydropower station as an example, the technical solution of this embodiment is described in detail. The turbine of this hydropower station is a mixed-flow type with a rated head of 120 meters, a rated speed of 100 rpm, and a turbine runner diameter of 7.5 meters.

[0089] In this application scenario, the multi-dimensional sensing hardware module deploys high-frequency pressure sensors at key parts of the water diversion system, installs vibration sensors at the turbine casing and tailrace, configures water hammer pressure sensors in the runner chamber, configures power sensors at the generator output, and configures speed sensors on the main shaft of the unit. All sensors are connected to the data processing module via an industrial bus. The sampling frequency is set in stages: the water hammer pressure signal sampling frequency is 2 kHz, the power and efficiency signal sampling frequency is 200 Hz, and the sampling frequency of other signals is 1 kHz. The system performs a full-condition simulation modeling process under the operating conditions of a rated head of 120 meters and a rated speed of 100 rpm.

[0090] When the unit rapidly increases from 70% of its rated load to 100% of its rated load, the system collects multi-dimensional operating data in real time through the sensor network during the load change process. The data processing module performs multi-source fusion and spatiotemporal alignment processing to construct an information map reflecting this dynamic process.

[0091] The dynamic characteristic analysis module extracts the intrinsic mode components of the water hammer pressure signal. Through variational mode decomposition, it identifies the attenuation characteristics and energy distribution of different frequency components, determining the dominant frequency of the water hammer pressure wave to be 12 Hz, the attenuation time constant to be 0.5 seconds, and the dominant frequency of the cavitation vibration wave to be 45 Hz. The dynamic characteristic analysis module calculates the dynamic response process after a sudden load change, determining the instantaneous head response time to be 0.8 seconds, the flow response time to be 0.6 seconds, the power response time to be 0.4 seconds, and the efficiency response time to be 0.3 seconds.

[0092] The multiphysics coupled simulation module establishes a set of coupled equations based on the dynamic characteristic analysis results and the fusion results of boundary constraints. It solves the coupled equations using a combination of implicit time-progression and explicit spatial discretization. The implicit time-progression time step is set to 0.01 seconds. The nonlinear equations are solved using the Newton-Raphson iteration method. At each iteration step, the residuals of the coupled terms are checked to see if they meet the convergence accuracy requirement of the order of 10 to the power of negative 6. Through time-progression calculations, the dynamic change curves of head, flow rate, power, and efficiency during the load change process, as well as the dynamic change curves of rotational speed and vibration displacement, are obtained. The simulation results are compared with the measured data, showing that the simulation error of the transient process is controlled within 3%.

[0093] When the unit operates in the low head region, the neural network adaptive boundary detection algorithm of the boundary constraint fusion module outputs a boundary confidence of 85%, which exceeds the preset threshold of 70%, and the system automatically activates the boundary constraint fusion mechanism.

[0094] The module introduces a nonlinear hydraulic behavior correction operator to perform physical mechanism-level constraint correction on the characteristic curve data. Taking into account the enhanced characteristics of water hammer effect in the low head region, the nonlinear variation characteristics of cavitation coefficient, and the cavitation characteristics of water flow, the calculated flow correction is -12% and the efficiency correction is -8%. The corrected characteristic curve data reduces the simulation error in the boundary region from 25% to less than 8%.

[0095] In the capacity expansion scenario, the rated capacity of the unit is increased from 300 MW to 350 MW. The model adaptive adjustment module receives the new unit parameters, such as the rated capacity of 350 MW, the design head of 120 meters, and the rated speed of 100 rpm, and inputs them into the deep neural network mapping network. Through forward propagation, the adjusted model coefficients are obtained, including parameter adjustment coefficients for the turbine's comprehensive characteristic curve, efficiency correction coefficients, and water hammer effect parameter adjustment coefficients. The module pushes the adjustment coefficients to the coupled simulation module, which recalculates the parameters of the coupled equations based on the new model coefficients. No new hydraulic model tests are required; the model update is completed solely through parameter calibration. The updated simulation model accurately reflects the operating characteristics of the unit after the capacity expansion.

[0096] Comparative example: To verify the technical effect of Embodiment 1 of the present invention, a comparative analysis was conducted using a comparative example. The comparative example employs a traditional method to construct a hydroelectric unit simulation model. This method uses discrete steady-state comprehensive characteristic curves obtained from hydraulic model tests as the core data foundation. Operating parameters such as efficiency, flow rate, and power under corresponding operating conditions are extracted through table lookup interpolation, and then combined with the physical mechanism models of the water diversion system and the generator side. A simple linear extrapolation method is used in the boundary region. In capacity expansion and renovation scenarios, it is necessary to conduct new hydraulic model tests to update the data foundation. The comparative example does not establish a multi-scale dynamic system analysis model for dynamic characteristic analysis, nor does it establish a boundary constraint fusion mechanism for physical mechanism constraints; it only independently combines the physical mechanism model and characteristic curve data.

[0097] Simulation comparison test results show that in the simulation of rapid load change transition process, the comparative model, lacking a dynamic characteristic analysis mechanism, can only rely on table lookup interpolation to obtain steady-state parameters, resulting in an error of 15% to 20% between the simulation results and the measured data. In the boundary region simulation, the comparative model uses linear extrapolation, leading to a simulation error of 20% to 30%. Regarding the adaptability of capacity expansion and modification, the comparative model lacks an adaptive adjustment mechanism, necessitating the conduct of new hydraulic model tests to update the data foundation. In terms of model integration, the physical mechanism model and characteristic curve data in the comparative model are independent, lacking a unified simulation framework.

[0098] Table 1: Comparison of Embodiments of the Invention with Comparative Examples

[0099] Example 2: To verify the application effect of this invention in different types of hydropower units, a full-condition simulation modeling method was further verified using an axial-flow propeller turbine generator unit of a hydropower station as an example. The unit has a single-unit capacity of 80 MW, a rated head of 18 meters, a rated speed of 136.4 rpm, a turbine runner diameter of 6.5 meters, and a blade angle adjustment range of -15 degrees to +25 degrees.

[0100] This second embodiment optimizes the configuration based on the characteristics of axial-flow propeller turbine generators, building upon the first embodiment. The multi-dimensional sensing hardware module adds the following sensors to the configuration of the first embodiment: a blade angle sensor is added to the runner chamber to collect real-time blade angle signals, with a sampling frequency of 50 Hz; a hydraulic pressure sensor is added to the blade operating mechanism to monitor the operating status of the blade adjustment mechanism, with a sampling frequency of 1 kHz; and a cavitation sensor is added to the volute inlet to monitor the degree of water cavitation, with a sampling frequency of 500 Hz. Other sensor configurations remain consistent with the first embodiment.

[0101] The full-condition simulation modeling system provided in this embodiment two is executed according to the following steps: Step one involves real-time acquisition of multi-dimensional operational data, including hydraulic, mechanical, and electrical characteristics, through a sensor network deployed at key components of the hydropower unit. The acquired data undergoes multi-source fusion and spatiotemporal alignment processing to construct an information map reflecting the dynamic operating environment under all conditions. The multi-dimensional sensing hardware module adds blade angle sensors, oil pressure sensors, and cavitation sensors to the sensor configuration described in Example 1. The data processing module receives all sensor data and performs multi-source data fusion, noise filtering, and spatiotemporal alignment processing to construct an information map of the dynamic operating environment under all conditions, including runner blade angle information.

[0102] Step two establishes a multi-scale dynamic system analysis model based on multi-dimensional operating data, calculates the turbine response time constant, identifies the water pressure pulsation attenuation characteristics and frequency distribution, analyzes the dynamic response process of head, flow, power, and efficiency under sudden load changes, and accurately captures the continuous evolution characteristics of hydraulic parameters. The dynamic characteristic analysis module adds dynamic characteristic analysis of the runner blade angle adjustment process to the multi-scale dynamic system analysis model of Example 1. Specifically, the analysis process includes introducing the runner blade angle as an analysis variable into the multi-scale dynamic system analysis model, analyzing the influence weight of runner blade angle adjustment on turbine hydraulic parameters, and extracting the response characteristics of hydraulic parameters during blade adjustment. The empirical mode decomposition process adds decomposition processing of the blade angle adjustment signal, and the variational mode decomposition process adds identification processing of the components influencing blade angle adjustment. Based on the decomposition results, the system calculates the turbine response time constant under different blade angles, identifies the influence of blade angle changes on water pressure pulsation attenuation characteristics, and analyzes the coupling influence of blade angle adjustment and load changes on hydraulic parameters.

[0103] Step 3 establishes a boundary constraint model constrained by nonlinear physical mechanisms based on the dynamic characteristic analysis results. A nonlinear hydraulic behavior correction operator is introduced to constrain and correct the characteristic curve data, resolving simulation deviations caused by linear extrapolation in the boundary region. The boundary constraint fusion module optimizes the hydraulic behavior characteristics of axial-flow propeller turbines in the low-head region. The optimized boundary constraint model comprehensively considers the following factors: the nonlinear influence of the runner blade angle on hydraulic performance; the influence weight of the runner blade angle on efficiency and flow rate changes in the low-head region; the nonlinear variation of cavitation characteristics with decreasing head; the accelerated increase of the cavitation coefficient with decreasing head in the low-head region; and the difference in water hammer effect at different blade angles, with the water hammer effect being more pronounced in the positive angle region. The neural network adaptive boundary detection algorithm adds input processing for the runner blade angle; when the unit is detected to have entered the low-head boundary region and the runner blade angle is in the high-angle region, a higher boundary confidence is output. The nonlinear hydraulic behavior correction operator adds correction terms related to the runner blade angle, calculating the corresponding flow rate and efficiency corrections based on the blade angle.

[0104] Step four involves modeling the entire operating process of the hydropower unit as a unified multiphysics coupled system. A set of coupled equations is established between the hydraulic field, mechanical vibration field, and electromagnetic field. These equations are then solved using a multiphysics coupled numerical calculation method to obtain continuous dynamic simulation results for the entire operating condition. The multiphysics coupled simulation module adds the runner blade angle as a state variable to the coupled equations in Example 1, establishing the coupling relationship between the blade angle and the hydraulic and mechanical vibration fields. A blade angle influence term is added to the hydraulic field equations, describing the effect of runner blade angle changes on water flow velocity and direction. A blade angle adjustment force term is added to the mechanical vibration field equations, describing the force exerted by the blade operating mechanism on the runner blades. The solution process for the coupled equations incorporates iterative calculations of the blade angle state variable. When the runner blade angle changes, the hydraulic excitation term and mechanical vibration response in the coupled equations change accordingly. The system obtains full-condition dynamic simulation results containing blade angle information through multiphysics coupled numerical calculation.

[0105] Step five establishes a nonlinear mapping network between unit parameters and simulation model coefficients. When the hydropower unit undergoes capacity expansion or changes in operating conditions, the model coefficients are automatically adjusted based on the new unit parameters, allowing for model updates without the need for re-conducting hydraulic model tests. The adaptive model adjustment module adds blade angle range parameters as input variables to the deep neural network mapping network in Example 1. The network input layer adds a blade angle range parameter input port, receiving the adjustable angle range of the runner blades as input. The network output layer adds adjustment coefficients related to blade angles, including blade efficiency correction coefficients and blade flow correction coefficients. When the unit undergoes runner blade modification or blade adjustment mechanism upgrades, the adaptive model adjustment module automatically adjusts the relevant parameters of the simulation model based on the new blade angle range.

[0106] The technical solution of Embodiment 2 is verified using this axial-flow propeller turbine generator unit as an example. During the rapid change from 60% load to 100% load, the system collects multi-dimensional operating data in real time, including the runner blade angle. The dynamic characteristic analysis module analyzes the transient impact of runner blade angle adjustment on hydraulic parameters, determining that the influence weight of blade angle on flow rate is 35%, and its influence weight on efficiency is 45%. The coupled simulation module calculates the dynamic changes in head, flow rate, power, efficiency, and runner blade angle during load variation, with the simulation results and measured data error controlled within 4%. During operation in the low-head region, the boundary constraint fusion module's neural network adaptive boundary detection algorithm outputs a boundary confidence level of 80%, exceeding a preset threshold. The system automatically activates the boundary constraint fusion mechanism, performing constraint correction on the characteristic curve data including blade angle factors, keeping the boundary region simulation error within 10%. In the scenario of capacity expansion and renovation, when the rated capacity of the unit is increased from 80 MW to 100 MW, the model adaptive adjustment module receives the new unit parameter input and automatically calculates the model parameter update scheme including the blade angle adjustment coefficient, thus completing the model update without having to conduct a new hydraulic model test.

[0107] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A full-condition simulation modeling method for hydropower units, characterized in that, Includes the following steps: Step 1: Collect multi-dimensional operational data containing hydraulic, mechanical and electrical characteristics in real time through a sensor network deployed in key parts of the hydropower unit. Perform multi-source fusion and spatiotemporal alignment processing on the collected data to construct an information map reflecting the dynamic operating environment under all working conditions. Step 2: Establish a multi-scale dynamic system analysis model based on multi-dimensional operating data, calculate the turbine response time constant, identify the water pressure pulsation attenuation characteristics and frequency distribution, analyze the dynamic response process of head, flow, power and efficiency under sudden load changes, and accurately capture the continuous evolution characteristics of hydraulic parameters. Step 3: Based on the dynamic characteristic analysis results, establish a boundary constraint model constrained by nonlinear physical mechanisms. By introducing a nonlinear hydraulic behavior correction operator in the boundary region, constrain and correct the characteristic curve data to solve the simulation deviation caused by linear extrapolation in the boundary region. Step 4: Model the entire operating process of the hydropower unit as a unified multi-physics coupled system, establish the coupling equation set between the hydraulic field, mechanical vibration field and electromagnetic field, solve it through the multi-physics coupled numerical calculation method, and obtain the continuous dynamic simulation results of the entire operating condition. Step 5: Establish a nonlinear mapping network between unit parameters and simulation model coefficients. When the hydropower unit is upgraded or the operating conditions change, the model coefficients are automatically adjusted according to the new unit parameters, and the model can be updated without having to conduct hydraulic model tests again.

2. The full-condition simulation modeling method for hydropower units according to claim 1, characterized in that, In step two, the multi-scale dynamic system analysis model adopts an analytical framework that combines empirical mode decomposition and variational mode decomposition. Empirical mode decomposition extracts the intrinsic mode components in the hydraulic parameter signal, while variational mode decomposition identifies the attenuation characteristics and energy distribution of different frequency components.

3. The full-condition simulation modeling method for hydropower units according to claim 1, characterized in that, In step three, the boundary constraint model adopts an adaptive boundary detection algorithm based on neural networks. It learns the hydraulic behavior pattern of the hydropower unit under different boundary conditions through training samples, automatically identifies the boundary threshold of the characteristic curve coverage area, and outputs the boundary confidence as the criterion for triggering boundary constraint fusion.

4. The full-condition simulation modeling method for hydropower units according to claim 1, characterized in that, In step four, the solution of the coupled equations adopts a numerical calculation method that combines implicit time propagation and explicit spatial discretization. Implicit time propagation ensures numerical stability, while explicit spatial discretization enables efficient processing of multi-physics coupling terms.

5. The full-condition simulation modeling method for hydropower units according to claim 1, characterized in that, In step five, the nonlinear mapping network adopts a deep neural network architecture, which includes an input layer, at least one hidden layer and an output layer. It achieves a highly nonlinear mapping from the parameter space to the coefficient space through an activation function. The training adopts a gradient descent optimization algorithm, and the training samples are derived from a fusion dataset of historical operating data and hydraulic model test data.

6. A full-condition simulation modeling system for hydropower units, applicable to the full-condition simulation modeling method for hydropower units as described in any one of claims 1-5, characterized in that, This includes a processing link, which collects multi-dimensional operational data containing hydraulic, mechanical and electrical characteristics in real time through a sensor network deployed at key parts of the hydropower unit. The collected data is then fused from multiple sources and spatiotemporally aligned to construct an information map that reflects the dynamic operating environment under all working conditions. The processing link establishes a multi-scale dynamic system analysis model based on multi-dimensional operating data, calculates the turbine response time constant, identifies the water pressure pulsation attenuation characteristics and frequency distribution, and analyzes the dynamic response process of head, flow, power and efficiency under sudden load changes. The processing link establishes a boundary constraint model based on the dynamic characteristic analysis results, compares the real-time operating data with the prediction results of the boundary constraint model, and automatically activates the boundary constraint mechanism when the unit operating parameters exceed the coverage range of the steady-state characteristic curve. It introduces a nonlinear hydraulic behavior correction operator to perform physical mechanism-level constraint correction on the characteristic curve data. The processing link establishes a set of coupled equations between the hydraulic field, mechanical vibration field and electromagnetic field based on the dynamic characteristic analysis and boundary constraint fusion results. It performs unified simulation calculations for all working conditions and solves the problem using a numerical calculation method that combines implicit time progression and explicit spatial discretization. The result is a continuous dynamic simulation of all working conditions. The processing link establishes a nonlinear mapping relationship between the unit structure parameters and the core coefficients of the simulation model. When a new unit parameter input is received, the adjusted model coefficients are calculated through the mapping network to achieve adaptive updating of the simulation model.

7. The full-condition simulation modeling system for hydropower units according to claim 6, characterized in that, The processing link employs a neural network adaptive boundary detection algorithm in the boundary constraint model establishment. It learns hydraulic behavior patterns under different boundary conditions through training samples and outputs boundary confidence as a criterion for activating boundary constraint fusion.

8. The full-condition simulation modeling system for hydropower units according to claim 6, characterized in that, The processing link employs implicit time advancement to ensure numerical stability in solving the coupled equations, and uses explicit spatial discretization to efficiently process multi-physics coupling terms, with convergence accuracy controlled within 10. -6 Magnitude.

9. The full-condition simulation modeling system for hydropower units according to claim 6, characterized in that, The processing link adopts a deep neural network architecture in establishing the mapping relationship, which includes an input layer, at least one hidden layer and an output layer. It achieves a highly nonlinear mapping from the parameter space to the coefficient space through activation functions. The training samples are derived from a fusion dataset of historical operating data and hydraulic model test data.

10. The full-condition simulation modeling system for hydropower units according to claim 6, characterized in that, The processing link also interacts with the data processing module to receive raw sensing data, perform multi-source data fusion, noise filtering and spatiotemporal alignment processing, store simulation intermediate results and undertake data caching tasks. The processing link also receives the final solution results from the simulation engine module and outputs simulation data in a standardized format.

Images