Primary frequency modulation test method and system of hydropower station governor

Abstract

The invention relates to the field of power system control, and discloses a primary frequency modulation test method and system for a hydropower station governor, and the method comprises the following steps: 1, building a digital twin model, and synchronizing the operation information of a physical hydroelectric generating set, 2, executing a virtual primary frequency modulation test, and obtaining a virtual power response and a virtual dynamic stress response; 3, quantifying the primary frequency modulation technical performance indexes of the physical hydroelectric generating set; step 4, quantizing increment health cost generated by the virtual primary frequency modulation test; and 5, updating the health state vector of the digital twin model. By establishing the digital twin model of the physical hydroelectric generating set and executing the primary frequency modulation test based on the virtual model, the test process can be transferred from a physical entity to a digital space, actual operation on the physical hydroelectric generating set is avoided, and the test efficiency is improved. Therefore, possible operation risks, economic expenditure and physical loss caused to equipment in the test process are eliminated.

Description

Technical Field

[0001] The present invention relates to the field of power system control, and in particular to a primary frequency regulation test method and system for a hydropower station speed regulator. Background Art

[0002] As a clean, renewable energy source, hydropower plays a vital role in modern power systems. Hydropower stations' turbine generators are not only the primary producers of electricity but, due to their exceptionally rapid response capabilities, they are also key resources for providing grid ancillary services and maintaining system frequency stability. Primary frequency regulation is one of the most basic and important ancillary services provided by hydropower units. It requires the units to automatically and rapidly respond to frequency deviations in the grid, balancing supply and demand in the system by increasing or decreasing output to ensure safe and stable grid operation. With the increasing proportion of intermittent renewable energy generation, such as wind power and photovoltaics, the volatility of grid frequency has significantly increased, and the performance requirements and reliance on primary frequency regulation for hydropower units have reached unprecedented levels.

[0003] To verify and evaluate whether a hydropower unit has the primary frequency regulation capability to meet grid requirements, the traditional approach is to conduct on-site primary frequency regulation tests. This test artificially introduces a simulated frequency disturbance signal into the unit's control system and directly observes and measures the power response of the physical unit.

[0004] However, this traditional on-site testing method has inherent and difficult-to-overcome drawbacks. Because the test process directly affects the high-value and complex physical hydropower units, any unexpected dynamic processes may pose a threat to the safety of the units, bringing significant operational risks. At the same time, performing such tests often requires the units to be shut down or deviate from the normal power generation plan, resulting in direct losses in power generation benefits and requiring a large investment of manpower and material resources, resulting in high economic costs. Even more insidious is that the frequency modulation test itself is a disturbance to the unit. Frequent testing operations will accelerate the accumulation of fatigue damage to key components, resulting in health costs that are difficult to quantify, thereby shortening the effective life of the unit.

[0005] To avoid the risks and costs of physical testing, existing technologies use computer simulation technology as an alternative. Although simulation methods theoretically avoid physical losses, traditional simulations usually construct mathematical models based on a set of fixed, idealized model parameters. This static model cannot reflect the degradation process of mechanical properties of hydropower units caused by factors such as fatigue and wear after long-term frequency regulation tasks. Therefore, when it is necessary to evaluate the long-term impact of a frequency regulation strategy, the accuracy of the prediction results of traditional simulation models will be greatly reduced over time due to their inability to simulate the aging of equipment, and they cannot provide a reliable basis for long-term operation decisions. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the present invention provides a primary frequency regulation test method and system for a hydropower station speed regulator, which solves the problem that traditional physical tests are high-risk and high-cost, and traditional simulations are static models and cannot simulate unit aging, resulting in inaccurate long-term evaluations, and thus making it difficult to quantify the performance and health loss of frequency regulation strategies.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A first aspect of the present invention provides a primary frequency regulation test method for a hydropower station speed regulator, the method comprising the following steps: Step S1: Based on the operation information of the physical hydropower unit, a digital twin model is established, and the operation information of the physical hydropower unit is synchronized. In a specific embodiment, the operation information of the physical hydropower unit includes: real-time operation data of the physical hydropower unit obtained by the sensor system, pre-set structure and material data, and operation safety constraint data. Synchronizing the operation information of the physical hydropower unit is specifically: obtaining the physical hydropower unit in the input signal Actual measured output under action ; and run an online parameter identification algorithm to identify the digital twin model under the same input signal Prediction output under action The actual measured output The goal is to minimize the error between the two models and continuously calibrate the model parameters of the digital twin model. The optimization objective formula is expressed as: ; in, is a preset loss function, such as the mean square error function, Step S2: performing a virtual primary frequency regulation test based on the digital twin model to obtain a virtual power response and a virtual dynamic stress response of the physical hydropower unit.

[0008] Step S3: quantifying the primary frequency regulation technical performance index of the physical hydropower unit based on the virtual power response.

[0009] Step S4: Based on the virtual dynamic stress response, quantify the incremental health cost generated by the virtual primary frequency modulation test. In a specific embodiment, this step is completed by a dynamic stress analysis model coupled with the digital twin model and a fatigue damage accumulation model. The dynamic stress analysis model extracts the amplitude and number of stress cycles based on the virtual dynamic stress response using methods such as rain flow counting. Subsequently, the fatigue damage accumulation model calculates the incremental fatigue damage based on the material's SN curve and the Palmgren-Miner linear cumulative damage criterion. The calculation process is expressed as follows: ; Where, It is in The actual number of stress cycles that occur at each stress level is It is in The maximum number of cycles allowed for a material under a stress level. is defined as the incremental fatigue damage function.

[0010] Step S5: Based on the incremental health cost, update the health state vector of the digital twin model and perform self-evolution on the digital twin model. It is a vector containing multiple variables that characterize the health status of the unit, including at least the cumulative fatigue damage value of the key components of the physical hydropower unit The incremental health cost results in an update of the cumulative fatigue damage value. The model parameters of the digital twin model The health state vector is defined as The function of In one embodiment, the model parameters Includes model parameters related to the mechanical properties of the physical hydropower unit. After updating, the model parameters It is updated accordingly, thus achieving self-evolution of the model.

[0011] Furthermore, the method of the present invention may further comprise the following steps: At least one primary frequency regulation strategy to be evaluated is defined, wherein each primary frequency regulation strategy to be evaluated is characterized by a set of speed regulator control parameters applied in the digital twin model.

[0012] In an optional embodiment, the above steps S2 to S5 are repeated for a plurality of different primary frequency modulation strategies to be evaluated. Based on the obtained incremental health cost and primary frequency modulation technical performance index corresponding to each primary frequency modulation strategy to be evaluated, a multi-objective optimization is performed to generate a Pareto optimal frontier. The multi-objective optimization problem can be expressed as: finding a set of frequency modulation strategies , so that the objective function vector The optimal expression is: ; Where, To adopt a strategy Primary frequency modulation technical performance indicators, To adopt a strategy The incremental health costs of

[0013] In another optional embodiment, the method of the present invention may further include the following steps: generating an ordered disturbance sequence representing a preset operating cycle based on historical power grid data, the sequence containing multiple virtual frequency disturbance signals. According to the primary frequency regulation strategy to be evaluated, the virtual frequency disturbance signals in the ordered disturbance sequence are sequentially executed, and steps S2 to S5 are repeated in each execution to simulate the evolution process of the digital twin model within the operating cycle. And based on the final result of the evolution process, the long-term risk-benefit of the primary frequency regulation strategy to be evaluated is evaluated. The long-term risk-benefit specifically includes: the accumulated health cost at the end of the preset long-term operating cycle, and the probability that the primary frequency regulation technical performance indicators always meet the preset performance requirements within the cycle.

[0014] A second aspect of the present invention provides a system for a primary frequency regulation test of a hydropower station speed regulator, the system comprising: The digital twin model module is used to build a digital twin model based on the operating information of the physical hydropower unit and synchronize the operating information of the physical hydropower unit; a virtual test module, connected to the digital twin model module, for performing a virtual primary frequency regulation test based on the digital twin model module to obtain a virtual power response and a virtual dynamic stress response of the physical hydropower unit; a performance quantification module, connected to the virtual test module, for quantifying a primary frequency regulation technical performance index of the physical hydropower unit based on the virtual power response; a health cost quantification module, connected to the virtual test module, for quantifying the incremental health cost caused by the virtual primary frequency modulation test based on the virtual dynamic stress response; A model evolution module is connected to the health cost quantification module and the digital twin model module, and is used to update the health status vector of the digital twin model module according to the incremental health cost, and update the model parameters of the digital twin model module based on the updated health status vector to self-evolve the digital twin model module.

[0015] The present invention provides a primary frequency regulation test method and system for a hydropower station speed regulator, which has the following beneficial effects: 1. The present invention establishes a digital twin model of the physical hydropower unit and performs a frequency modulation test based on the virtual model. This method transfers the test process from the physical entity to the digital space, fundamentally avoiding actual operations on the physical unit, thereby eliminating the operational risks, economic expenses and physical damage to the equipment that may exist during the test.

[0016] 2. This invention quantifies the incremental health cost generated by virtual testing and uses this cost to update the health state vector of the digital twin model, thereby adjusting the model parameters related to this vector function. This self-optimization mechanism enables the model to simulate the health degradation of the unit due to continuous frequency regulation. This overcomes the limitation of traditional fixed-parameter simulations that cannot reflect the aging process of equipment, thereby significantly improving the accuracy of long-term performance predictions of the unit.

[0017] 3. The present invention calculates the technical performance indicators and incremental health costs of multiple different primary frequency regulation strategies to be evaluated, and performs multi-objective optimization to generate the Pareto optimal frontier. It can intuitively display the quantitative constraint relationship between frequency regulation performance and unit health loss, and provide objective and quantitative decision support for operators to weigh and choose between different strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Schematic diagram of the method flow of the present invention; Figure 2 Schematic diagram of the system architecture of the present invention.

[0019] Among them, 10. Digital twin model module; 20. Virtual test module; 30. Performance quantification module; 40. Health cost quantification module; 50. Model evolution module. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the specification of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] In order to better understand the present invention, the above contents are described in detail below in conjunction with specific embodiments.

[0022] Please see the attached Figure 1 and attached Figure 2 , attached Figure 1 Attached is a flow chart of a method for testing a primary frequency regulation of a hydropower station speed regulator according to an embodiment of the present invention. Figure 2 It is a schematic diagram of the system architecture of a primary frequency regulation test system for a hydropower station speed regulator according to another embodiment of the present invention.

[0023] Specifically, during step S1, the digital twin model module 10 performs the operation. The digital twin model module 10 first establishes a digital twin model based on preset physical hydropower unit operating information. The physical hydropower unit operating information includes real-time operating data of the physical hydropower unit acquired through a sensor system, pre-set structural and material data, and operational safety constraint data.

[0024] The digital twin model module 10 also performs synchronization operations, which are achieved through an online parameter identification algorithm. Specifically, the physical hydropower unit is obtained when the input signal Actual measured output under action ; and use the digital twin model to Prediction output under action The actual measured output The goal is to minimize the error between the two models and continuously calibrate the model parameters of the digital twin model. The optimization objective can be expressed as: ; Where, is a preset loss function, such as the mean square error function; The digital twin model is the input signal Predicted output under action; The physical hydropower unit is input signal The actual measured output under action; is a set of adjustable model parameters of the digital twin model.

[0025] During step S2, the virtual test module 20 performs the operation. This module is connected to the digital twin module 10 and performs a virtual primary frequency regulation test based on the synchronized digital twin model within the digital twin module 10. This test is performed by applying a virtual frequency disturbance signal to the model, thereby obtaining the virtual power response and virtual dynamic stress response of the physical hydropower unit. These two response data are then output to the performance quantification module 30 and the health cost quantification module 40, respectively.

[0026] When executing step S3, the operation is performed by the performance quantification module 30. The performance quantification module 30 is connected to the virtual test module 20, receives the virtual power response data, and quantifies the primary frequency regulation technical performance index of the physical hydropower unit according to the preset calculation rules.

[0027] In a specific embodiment, the primary frequency modulation technical performance index includes at least one primary frequency modulation adjustment rate. The rate is calculated according to the following formula: ; Where: is the primary frequency modulation rate; is the maximum change of the unit response power in the virtual power response time series; It is the time from the start of the disturbance to the time when the power reaches the maximum change.

[0028] When executing step S4, the health cost quantification module 40 performs the operation. The health cost quantification module 40 is connected to the virtual test module 20, receives the virtual dynamic stress response data, and quantifies the incremental health cost generated by the virtual primary frequency modulation test through the dynamic stress analysis model and the fatigue damage accumulation model. Specifically, the fatigue damage accumulation model is used to calculate the incremental fatigue damage based on the Palmgren-Miner linear cumulative damage criterion. , its calculation process can be expressed as: ; Where, It is in The actual number of stress cycles that occur at each stress level is It is in The maximum number of cycles allowed for a material under a stress level. is defined as the incremental fatigue damage function.

[0029] When executing step S5, the operation is performed by the model evolution module 50. The model evolution module 50 is connected to the health cost quantification module 40 and the digital twin model module 10. The model evolution module 50 receives the incremental health cost calculated by the health cost quantification module 40 and updates the health state vector of the digital twin model based on the cost. The health status vector At least the cumulative fatigue damage value of the key components of the physical hydropower unit ; Then, the model evolution module 50 is based on the updated health state vector Model parameters of the digital twin model The model parameters are adjusted. The health state vector is defined as The function relationship can be expressed as: ; Where, is a preset functional relationship used to convert the health state vector Mapping to model parameters ; The model parameters The parameters related to the mechanical properties of the physical hydroelectric unit are included. It is transmitted back to the digital twin model module 10 for updating the digital twin model therein, thereby completing the self-evolution of the model.

[0030] Specifically, in step S1, the physical hydropower unit operation information based on which the digital twin model is established is a comprehensive description of the multi-dimensional characteristics of the physical hydropower unit. In a specific embodiment, this information is divided into the following three categories: The first category is real-time operational data. This data is collected and transmitted in real time by sensor networks and supervisory control and data acquisition (SCADA) systems deployed on the physical hydropower units. This data includes, but is not limited to, unit speed, active power, reactive power, generator stator and rotor currents and voltages, turbine guide vane opening, the head generated by the upstream and downstream water levels, and vibration and temperature measurements of key components such as the main shaft and bearings.

[0031] The second category is structural and material data. This data is static or quasi-static information describing the inherent properties of the physical hydropower unit. Specifically, it includes: detailed 3D geometric model data and dimensional parameters of the unit's key pressure-bearing or load-bearing components (such as the runner blades, main shaft, and volute); the material grades and corresponding mechanical properties of these components, such as elastic modulus and Poisson's ratio; and the material's SN curve, used for fatigue life calculations, which defines the number of cycles to failure under different stress amplitudes.

[0032] The third category is operational safety constraint data. This data defines the safety boundary conditions that must not be exceeded during the operation of the physical hydropower unit. Specifically, it includes: preset vibration alarm and shutdown thresholds at each measurement point; the maximum allowable operating temperature of components such as bearings and generator coils; the permissible pressure fluctuation range within the turbine and water diversion system; and the maximum and minimum active power output limits specified by the power grid or the equipment itself. This data is used during virtual testing and evaluation to determine whether the system behavior is within the safe range.

[0033] The digital twin model is a mathematical formalization of the dynamic characteristics of a physical hydroelectric unit. The specific structure of the model can be implemented using different technical solutions based on the requirements for modeling accuracy and computational efficiency.

[0034] In one embodiment, the digital twin model is constructed using a mechanism-driven modeling approach. The model consists of multiple sub-models, each describing a physical component or process within the hydropower unit. For example, the model may include a governor model describing the governor control logic, a servo system model describing the response characteristics of the guide vane servo, a turbine model describing the conversion of water flow energy into mechanical energy, and a generator model describing the conversion of mechanical energy into electrical energy. These sub-models are coupled to each other through pre-defined physical equations (such as differential or algebraic equations) to form the model.

[0035] In a specific implementation, some or all of the sub-models can be expressed using transfer functions in linear system theory. The general transfer function model formula can be expressed as: ; Where, is the transfer function of the sub-model; Laplace transform of the input signal to the sub-model; Laplace transform of the sub-model output signal; is a complex frequency variable; , ,…, and , ,…, are the coefficients of the transfer function.

[0036] In another embodiment, the digital twin model is constructed using a data-driven modeling approach. This model does not rely on precise physical equations, but instead learns from historical operating data to fit the input-output relationship of the system. For example, a recurrent neural network (RNN) or its variants, such as a long short-term memory (LSTM) network, can be used to construct the model. The state update and output of a general recurrent neural network can be expressed as: ; ; Where, is a discrete time step; For the time step The input vector of For the time step The hidden state vector of The previous time step The hidden state vector of For the time step The output vector of , , is the weight matrix of the network; , is the bias vector of the network; , is a preset activation function, such as Sigmoid function or Tanh function. In this embodiment, all weight matrices and bias vectors in the network together constitute the model parameters to be identified and evolved. .

[0037] In one embodiment, a hybrid modeling approach can be employed. This approach combines mechanism-driven models with data-driven models. For example, a mechanism model is used for components of the system with clear physical laws (such as the generator swing equation), while a data-driven model is used to compensate and correct for nonlinear aspects that are difficult to accurately model (such as hydraulic disturbances or friction losses). This approach combines the advantages of both approaches.

[0038] To ensure that the digital twin model accurately reflects the current state of the physical hydropower unit, particularly when unit characteristics drift due to changes in operating conditions or minor degradation, the present invention further includes an online synchronization and parameter identification process in step S1. This process runs continuously to minimize the error between the predicted output of the digital twin model and the actual measured output of the physical hydropower unit.

[0039] In one embodiment, the digital twin model module 10 performs this online synchronization operation. This step is achieved by an online parameter identification algorithm. Specifically, the digital twin model module 10 obtains the physical hydropower unit under the input signal The actual measured output under At the same time, the digital twin model has the same input signal Next, according to the current model parameters , produces the predicted output, recorded as The online synchronization process aims to minimize the error between the predicted output and the actual measured output, and continuously corrects the model parameters of the digital twin model. The optimization objective can be expressed as: ; Where, is a preset loss function used to quantify the difference between the predicted output and the actual measured output, for example, it can be the square of the difference between the two; is the model parameter vector to be corrected; is the input signal; Predict output for the model; It is the actual measured output of the physical unit.

[0040] In a specific embodiment, the online parameter identification process is implemented using a recursive least squares algorithm. This algorithm is suitable for linear or linearizable system models near the operating point, and updates the model parameters online in an iterative manner. .

[0041] First, the system model to be identified is expressed as the following linear regression form: ; Where, is a discrete time step; For the time step The actual measured output of the system; For the time step The information vector, is the transpose operator; is the parameter vector to be identified, which contains the unknown or uncertain parameters of the model; is the system noise or model error; The recursive least squares algorithm uses the following set of equations at each time step Estimates of the parameter vector To update: ; ; ; Where, For the time step The parameter vector estimated value of; is the estimated value of the parameter vector at the previous time step; For the time step The gain matrix of For the time step The covariance matrix reflects the uncertainty of the parameter estimates; For the time step The covariance matrix of is the forgetting factor, a constant ranging from (0,1), which is used to adjust the weight of the influence of historical data on the current parameter estimation; is the identity matrix; For the time step The actual measured output of the system; At each time step , the digital twin model module 10 obtains the real-time input and output data of the physical unit and constructs the information vector and measurement output , and perform the above recursive calculation to obtain the updated parameter vector Subsequently, the digital twin model module 10 uses the updated Adjust the parameters within the digital twin model to complete a synchronization operation, so that the model output is close to the actual output of the physical unit.

[0042] In another embodiment, for nonlinear systems or scenarios with significant process noise and measurement noise, an extended Kalman filter or an unscented Kalman filter algorithm may be used to simultaneously estimate the state and parameters of the system.

[0043] In step S2, the virtual test module 20 performs a virtual primary frequency modulation test. The purpose of this step is to obtain the performance and stress data of the unit when responding to a standard frequency disturbance without affecting the operation of the physical unit.

[0044] Specifically, the virtual test module 20 first receives the online synchronized data containing the latest model parameters from the digital twin model module 10. The module then defines the standard virtual frequency disturbance signal In one embodiment, the virtual frequency disturbance signal is a step signal, and its mathematical expression is: ; Where, For time; is the starting time of the disturbance application; It is a preset frequency deviation amplitude, such as -0.1 Hz, used to simulate a sudden drop in grid frequency.

[0045] The virtual test module 20 uses this virtual frequency disturbance signal As input, it is applied to the speed regulator sub-model in the digital twin model. Then, the module dynamically simulates the entire digital twin model through numerical integration and other methods. The simulation time covers the preset time interval, such as from arrive ,in is the total simulation time sufficient to observe the system reaching a new steady state.

[0046] During the simulation process, the virtual test module 20 records the time series data of specific variables within the model, thereby obtaining two sets of key outputs. The first set of outputs is the virtual power response This is the data sequence of the active power output by the generator sub-model in the digital twin model over time. This sequence fully characterizes the power regulation behavior of the unit during the virtual primary frequency regulation process.

[0047] The second set of outputs is the virtual dynamic stress response This is the preset first A data series of the time-varying equivalent stress of a key component or location is generated. This stress response is calculated using a stress analysis sub-model coupled to the digital twin. This sub-model takes dynamic variables from the simulation (such as the guide vane opening rate, water pressure pulsation, and speed fluctuations) as input and calculates and outputs the dynamic stress values ​​at the corresponding locations.

[0048] After the simulation is completed, the virtual test module 20 will respond to the virtual power The complete data is sent to the performance quantification module 30, and the virtual dynamic stress response The data is sent to the health cost quantification module 40 for use in subsequent steps S3 and S4.

[0049] In step S3, the performance quantification module 30 receives the virtual power response data sequence generated by the virtual test module 20. , and calculate the primary frequency regulation technical performance indicators based on this data. In a specific embodiment, these indicators include but are not limited to regulation dead zone, regulation rate and regulation accuracy. For example, the regulation dead zone is calculated based on the minimum frequency deviation that triggers the unit power response; the regulation rate is calculated based on the slope of a specific segment on the power response curve, that is, The regulation accuracy is calculated based on the deviation between the stable value of virtual power and the theoretical target value after the frequency modulation process is completed.

[0050] In parallel, in step S4, the health cost quantification module 40 receives the virtual dynamic stress response data sequence for the preset key components generated by the virtual test module 20. , and quantify the incremental health cost generated by the virtual test accordingly. In a specific implementation, this quantification process is accomplished by collaboratively completing the dynamic stress analysis model and the fatigue damage accumulation model through two coupled models.

[0051] Specifically, first, the dynamic stress analysis model responds to the input virtual dynamic stress time series The model uses the rainflow counting method to decompose the irregular stress-time history into a series of independent stress reversal cycles with constant amplitude. The output of this step is a set of data pairs. ,in is the total number of decomposed stress amplitude levels, , For the The stress amplitude of each level, is the number of cycles counted at this stress amplitude level.

[0052] The fatigue damage accumulation model then receives the output of the rainflow counting method and calculates the incremental fatigue damage value based on the Palmgren-Miner linear cumulative damage criterion. The calculation formula of this criterion is as follows: ; Where, It is in The actual number of stress cycles that occur at each stress level is It is in The maximum number of cycles allowed for a material under a stress level. is defined as the incremental fatigue damage function.

[0053] The maximum number of cycles The value is determined by the SN curve (stress-life curve) of the key component material. The SN curve is part of the pre-set structure and material data, which is used to define the specific material at different stress amplitudes. Fatigue life under For a given stress amplitude , , the corresponding SN can be obtained by querying the SN curve or using its fitting equation (such as the Basquin equation). The calculated incremental fatigue damage value Used as the incremental health cost of this virtual experiment The incremental health cost is then output to the model evolution module 50 for subsequent step S5.

[0054] In step S5, the model evolution module 50 executes the model self-evolution process. The model evolution module 50 is connected to the health cost quantification module 40 and the digital twin model module 10. It first receives the incremental health cost generated by a single virtual test calculated in step S4 from the health cost quantification module 40. The cost is quantified as an incremental fatigue damage value. .

[0055] The model evolution module 50 maintains the health state vector The vector is a multi-dimensional vector used to store and characterize the cumulative health status of key components of a physical hydropower unit. In a specific embodiment, the health status vector At least one or more key components of the cumulative fatigue damage value .

[0056] Upon receiving the incremental fatigue damage value After that, the model evolution module 50 pairs of health state vectors Specifically, the incremental fatigue damage value is added to the corresponding cumulative fatigue damage value in the vector. If the virtual test is performed times, the updating process of the accumulated fatigue damage can be expressed as: ; Where, For the The cumulative fatigue damage value updated after the virtual test; For the The cumulative fatigue damage value after the virtual test; For the The incremental fatigue damage value generated by the virtual test. Through this step, the health state vector Get updated, recorded as .

[0057] Subsequently, the model parameters of the digital twin model Defined as the health state vector This allows the model parameters to be adaptively adjusted as the unit's health deteriorates. The functional relationship can be expressed as: ; Where, For the The updated model parameter vector after the virtual test; For the The updated health status vector after the virtual test; is a preset mapping function used to transform the health state vector Mapping to model parameters .

[0058] In a specific embodiment, the model parameters Includes parameters related to the mechanical properties of the physical hydroelectric unit. For example, the parameter vector The parameter representing the effective stiffness of a spindle component , which is related to the cumulative fatigue damage The functional relationship can be defined as: ; in, is the initial stiffness value of the component, is the preset material damage degradation coefficient. Similarly, other mechanical characteristic parameters such as the system's damping coefficient and friction coefficient can also be defined as the health state vector functions of the corresponding variables in .

[0059] According to the above functional relationship Calculate the updated model parameter vector After that, the model evolution module 50 converts this new parameter vector The digital twin model module 10 receives the new model parameter vector After that, the original parameters within the model are replaced with the updated ones. This completes a complete model self-evolution process. The updated digital twin model reflects the cumulative damage caused by the virtual test and serves as the basis for the next virtual test or evaluation.

[0060] In another application embodiment, the present invention provides a further application embodiment for performing a long-term risk-benefit assessment of a specific primary frequency regulation strategy. The first step in this embodiment is to generate an ordered disturbance sequence. This sequence is used in subsequent simulations to represent all frequency disturbance events experienced by the physical hydropower unit within a predetermined operating period (e.g., a year or a quarter).

[0061] The specific generation process starts with obtaining a data set that contains high-resolution time series records of historical power grid frequency, denoted as Based on this data, we first calculate the historical frequency deviation sequence , its expression formula is: ; Where, is the nominal frequency of the grid (e.g. 50 Hz).

[0062] Then, the historical frequency deviation series Processing is performed to identify and extract a series of independent, significant frequency disturbance events. In a specific embodiment, by setting a frequency deviation threshold , all the time series that satisfy The consecutive time periods are identified as independent frequency disturbance events.

[0063] For each identified disturbance event, the corresponding frequency deviation time segment is completely extracted to form an independent virtual frequency disturbance signal. By traversing the entire historical frequency data set, a A library of independent virtual frequency perturbation signals.

[0064] Finally, all independent virtual frequency disturbance signals in the library are arranged according to the natural time sequence in which they occur in the original historical data, thus forming the final ordered disturbance sequence. . This sequence can be expressed as: ; Where, It is A virtual frequency disturbance signal is extracted from historical data. This ordered disturbance sequence is used as the standard input for subsequent long-term evolution simulations, and its statistical characteristics are consistent with the disturbance environment faced by physical units in real power grids.

[0065] Specifically, after generating the ordered disturbance sequence, the present invention further provides a long-term evolution simulation process to evaluate the performance and impact of the "primary frequency modulation strategy to be evaluated" during the entire operation cycle.

[0066] First, the "primary frequency regulation strategy to be evaluated" needs to be defined. This strategy is characterized by a specific set of initial control parameters, such as proportional, integral, and differential gains, as well as transient slip coefficients. These initial parameters are configured in the speed regulator submodel in the digital twin model. At the same time, the health state vector of the digital twin model is is initialized. In a specific embodiment, its initial accumulated fatigue damage value is is set to zero.

[0067] Then, the system starts the long-term evolution simulation. This process is a sequential execution cycle, which follows an ordered perturbation sequence. Each virtual frequency disturbance signal is applied to the digital twin model in sequence.

[0068] For the first (Among them, From 1 to ) Virtual frequency disturbance signal , the system completely executes steps S2 to S5 of the present invention once.

[0069] Specifically, in the In the second cycle: the virtual test module 20 will disturb the signal Applied to the model parameters updated after the previous cycle The digital twin model of the virtual test is performed (step S2). The performance quantification module 30 and the health cost quantification module 40 respectively calculate the technical performance indicators and incremental health costs corresponding to the disturbance. (Steps S3 and S4).

[0070] Next, the model evolution module 50 performs the evolution operation (step S5). It calculates the incremental health cost Accumulate to the health state vector Then, based on the functional relationship , calculate the new model parameter vector and use it to update the digital twin model for the next cycle ( times) to prepare.

[0071] During the entire long-term evolution simulation process, the system records the technical performance indicators generated by each cycle (i.e. each disturbance event) and the health status vector after each update. .

[0072] When the last disturbance signal in the ordered disturbance sequence After the processing is completed, the long-term evolution simulation process ends. The final results of the evolution process include: a complete sequence of health state evolution trajectories , and the corresponding technical performance indicator sequence. These output data are used in subsequent long-term risk-benefit assessments. In one specific embodiment, the long-term risk-benefit assessment specifically includes: evaluating the cumulative health costs of the physical hydropower units at the end of a preset long-term operation cycle; and evaluating the probability that the primary frequency regulation technical performance indicators will always meet performance requirements during the preset long-term operation cycle.

[0073] In another application embodiment, the goal is to identify a set of primary frequency modulation strategies to be evaluated that form an optimal trade-off between performance and cost. This problem is formalized as a multi-objective optimization problem.

[0074] The decision variables of this optimization problem are vectors , which contains all adjustable control parameters in the primary frequency modulation strategy to be evaluated.

[0075] In one embodiment, the multi-objective optimization problem is based on the results of a single virtual experiment, and its objective function is expressed as: ; Where, To adopt a strategy The primary frequency modulation technical performance index calculated by step S3 is To adopt a strategy is the incremental health cost calculated by step S4, T is the transposition operator; In another more comprehensive embodiment, the multi-objective optimization problem is based on the results of long-term evolution simulation, and its objective function is expressed as: ; Where, and Adoption strategies The long-term return and long-term risk when T is the transposition operator; In order to solve the multi-objective optimization problem, a multi-strategy optimization module 60 can be set. In a specific embodiment, the module adopts a non-dominated sorting genetic algorithm. (NSGA-II) is used to solve the problem. The algorithm performs iterative selection, crossover, and mutation operations on a population of different strategies, and ultimately obtains a set of strategies called the Pareto optimal solution set.

[0076] The objective function values ​​corresponding to the Pareto optimal solution are then plotted in a two-dimensional coordinate system to obtain the Pareto frontier, which serves as an auxiliary decision-making tool. A decision support module 70 is then established. This module receives external inputs, such as a maximum acceptable risk threshold and a minimum performance requirement threshold, and selects a subset of strategies that meet the constraints from the Pareto frontier. Based on pre-defined secondary criteria (e.g., risk minimization), a final strategy is selected for actual deployment.

[0077] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A primary frequency regulation test method for a hydropower station speed regulator, characterized in that: The following steps are involved: Step 1: Based on the operation information of the physical hydropower unit, a digital twin model is established and the operation information of the physical hydropower unit is synchronized; Step 2: Perform a virtual primary frequency regulation test based on the digital twin model to obtain the virtual power response and virtual dynamic stress response of the physical hydropower unit; Step 3: quantifying the primary frequency regulation technical performance index of the physical hydropower unit based on the virtual power response; Step 4: quantifying the incremental health cost generated by the virtual primary frequency modulation test based on the virtual dynamic stress response; Step 5: Based on the incremental health cost, update the health state vector of the digital twin model, wherein the model parameters of the digital twin model are functions of the health state vector and are used to self-evolve the digital twin model.

2. A primary frequency regulation test method for a hydropower station speed regulator according to claim 1, characterized in that: In step 1, the operation information of the synchronous physical hydropower unit specifically includes: Acquire real-time operating data of the physical hydropower unit; and continuously calibrate model parameters of the digital twin model based on an online parameter identification algorithm to minimize the error between the predicted output of the digital twin model and the actual measured output of the physical hydropower unit.

3. A primary frequency regulation test method for a hydropower station speed regulator according to claim 2, characterized in that: The health state vector includes the accumulated fatigue damage value of the components of the physical hydropower unit, and the operation information of the physical hydropower unit includes: real-time operation data, structure and material data, and operation safety constraint data.

4. A primary frequency regulation test method for a hydropower station speed regulator according to claim 1, characterized in that: The method further includes: generating an ordered disturbance sequence based on historical power grid data, the ordered disturbance sequence including a virtual frequency disturbance signal; According to the primary frequency modulation strategy to be evaluated, the virtual frequency disturbance signal in the ordered disturbance sequence is sequentially executed to simulate the evolution process of the digital twin model during the operation cycle, and the simulation process includes: repeatedly executing steps 2 to 5; Based on the final result of the evolution process, the long-term risk-return of the primary frequency modulation strategy to be evaluated is evaluated.

5. A primary frequency regulation test method for a hydropower station speed regulator according to claim 4, characterized in that: The long-term risk-benefit includes: the accumulated health cost at the end of the preset long-term operation cycle, and the probability that the primary frequency modulation technical performance indicators always meet the performance requirements within the preset long-term operation cycle.

6. A primary frequency regulation test method for a hydropower station speed regulator according to claim 1, characterized in that: The method further includes defining at least one primary frequency regulation strategy to be evaluated, each of the primary frequency regulation strategies to be evaluated being characterized by a set of speed regulator control parameters applied in the digital twin model.

7. A primary frequency regulation test method for a hydropower station speed regulator according to claim 6, characterized in that: For multiple different primary frequency modulation strategies to be evaluated, steps 2 to 5 are repeatedly executed respectively; and based on the obtained incremental health cost and primary frequency modulation technical performance index corresponding to each primary frequency modulation strategy to be evaluated, multi-objective optimization is performed to generate a Pareto optimal frontier.

8. The primary frequency regulation test method of a hydropower station speed regulator according to claim 1, characterized in that: The quantified incremental health cost in step 4 is accomplished collaboratively by coupling the dynamic stress analysis model and the fatigue damage accumulation model in the digital twin model.

9. A primary frequency regulation test method for a hydropower station speed regulator according to claim 1, characterized in that: The model parameters in step 5 include model parameters of the mechanical characteristics of the physical hydroelectric generator set.

10. A system for a primary frequency regulation test of a hydropower station speed regulator, based on a primary frequency regulation test method of a hydropower station speed regulator according to any one of claims 1 to 9, characterized in that: include: The digital twin model module is used to build a digital twin model based on the operating information of the physical hydropower unit and synchronize the operating information of the physical hydropower unit; A virtual test module performs a virtual primary frequency regulation test based on the digital twin model module to obtain a virtual power response and a virtual dynamic stress response of the physical hydropower unit; a performance quantification module, which quantifies the primary frequency regulation technical performance index of the physical hydropower unit based on the virtual power response; a health cost quantification module, which quantifies the incremental health cost caused by the virtual primary frequency modulation test according to the virtual dynamic stress response; A model evolution module is used to update the health state vector of the digital twin model module according to the incremental health cost. The model parameters of the digital twin model module are functions of the health state vector and are used to self-evolve the digital twin model module.

Images