Hydropower frequency modulation capacity configuration method and device considering security boundary constraint

Abstract

The present application relates to the technical field of power control system, specifically relates to a method and device for configuring hydroelectric frequency modulation capacity considering safety boundary constraint, wherein the configuration method comprises the following steps: S1, judging whether the power system has the risk of entering the subsynchronous weak damping state based on the wide frequency data; S2, constructing a quadratic programming approximation problem for optimizing frequency modulation control; S3, solving the quadratic programming approximation problem; S4, checking and evaluating the candidate power control trajectory set; S5, selecting an optimal preselected control sequence; S6, performing a limit safety check; S7, executing corresponding control operation; the present application provides decisive safety redundancy for the frequency modulation operation of large hydroelectric generating units under complex power grid environment, systematically solves the contradiction between frequency modulation safety and regulation performance, and has significant technical progress and broad engineering application prospect.

Description

Technical Field

[0001] This invention relates to the technical field of power control systems, and more particularly to a method and apparatus for configuring hydropower frequency regulation capacity considering safety boundary constraints. Background Technology

[0002] With the large-scale grid connection of renewable energy sources such as wind power and photovoltaics, the power supply structure of the power system has undergone profound changes. The proportion of traditional synchronous power sources such as thermal power and hydropower has gradually decreased, and the system inertia and frequency regulation capability have been significantly weakened. In order to ensure the stability of the grid frequency, hydropower, as a high-quality adjustable power source, is required to undertake an increasingly heavy frequency regulation task, and its power output needs to be adjusted rapidly and frequently according to the fluctuations in grid frequency.

[0003] However, while the participation of hydropower units in rapid frequency regulation improves system frequency stability, it also brings significant safety hazards. In long-distance AC / DC hybrid transmission scenarios with series compensation, the frequency regulation action of hydropower units may induce subsynchronous resonance between the generator shaft mechanical system and the series-compensated transmission network. Once subsynchronous resonance occurs, alternating torsional stress far exceeding normal operating conditions will be generated on the generator shaft system, leading to continuous fatigue damage accumulation at the connection points of various mass blocks in the shaft system. In severe cases, this can cause shaft crack propagation or even shaft breakage within a short period of time, posing a fatal threat to megawatt-class large hydropower units.

[0004] Existing methods for configuring frequency regulation capacity in hydropower typically rely on offline tuning based on static safety boundary constraints. This means that fixed parameters such as unit output limits and ramp rates are used as constraints in the optimization model. Such methods cannot detect changes in system operating status in real time, making it difficult to anticipate and proactively avoid the risk of subsynchronous oscillations induced by sudden frequency regulation commands under specific grid operating conditions. Furthermore, existing methods often employ single classical optimization algorithms when solving frequency regulation optimization problems. Faced with complex optimization scenarios involving strong nonlinearity and multiple constraints, these algorithms are prone to getting trapped in local optima, making it difficult to achieve a global balance between ensuring absolute shaft safety and pursuing optimal frequency regulation benefits. Summary of the Invention

[0005] To address the technical problems existing in the background art, this invention proposes a method and apparatus for configuring hydropower frequency regulation capacity considering safety boundary constraints, the specific solution of which is as follows: The method for configuring hydropower frequency regulation capacity considering safety boundary constraints includes the following steps: S1. Acquire broadband data of the power system in real time, and determine whether the power system is at risk of entering a subsynchronous weakly damped state based on the broadband data; S2. If the risk is determined to exist, a quadratic programming approximation problem for optimizing frequency modulation control is constructed for a preset prediction time domain. The quadratic programming approximation problem includes dynamic safety constraints for avoiding the risk of the subsynchronous weak damping state. S3. Solve the quadratic programming approximation problem to generate a set of candidate power control trajectories containing at least one candidate power control trajectory; S4. Verify and evaluate the candidate power control trajectory set to determine a comprehensive utility evaluation value that can reflect the frequency modulation performance and shaft torsional vibration risk. S5. Select an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value; S6. Before executing the optimal pre-selected control sequence, perform a limit safety check to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold. S7. Based on the results of the extreme safety check, execute the corresponding control operations.

[0006] Furthermore, in S1, broadband data of the power system is acquired in real time, and based on the broadband data, it is determined whether the power system is at risk of entering a subsynchronous weakly damped state, as follows: Real-time acquisition of broadband data of the power system, the broadband data including at least: hydropower unit operating parameters, power grid topology information, power flow information, high voltage DC transmission power, series compensation degree, and high frequency electrical quantity data; Based on the broadband data, the damping ratio of each oscillation mode of the power system in the subsynchronous frequency band is continuously calculated using the Proni algorithm; The calculated damping ratio of each oscillation mode is compared with a preset first safety margin threshold. If the damping ratio of any of the oscillation modes decreases below the first safety margin threshold, it is determined that the power system is at risk of entering a subsynchronous weakly damped state.

[0007] Furthermore, in S2, if the risk is determined to exist, a quadratic programming approximation problem for optimizing frequency modulation control is constructed for a preset prediction time domain, as follows: At the current operating point, a local time-varying linear state-space model is obtained by real-time linearization processing through a pre-built machine-network coupled dynamic model. The state variable constraints related to the subsynchronous weak damping state risk near the current operating point and the unit's normal operation constraints are linearized to form a set of locally linearized inequality constraints, which constitute the dynamic safety constraints. Construct a quadratic performance index function with the sum of squares of the tracking deviations between the actual power output of the unit and the grid frequency regulation command over multiple consecutive time steps in the future as the optimization objective; The local time-varying linear state-space model, the local linearization inequality constraint set, and the quadratic performance index function are combined to form the quadratic programming approximation problem.

[0008] Furthermore, in S3, the quadratic programming approximation problem is solved to generate a set of candidate power control trajectories containing at least one candidate power control trajectory, as follows: The quadratic programming approximation problem is transformed into an equivalent combinatorial optimization model, and the first type of optimization algorithm is used to repeatedly solve the equivalent combinatorial optimization model to obtain the first candidate power control trajectory subset. Simultaneously, at least two second-class optimization algorithms based on different principles are used to directly solve the quadratic programming approximation problem in parallel, respectively obtaining the second candidate power control trajectory subset and the third candidate power control trajectory subset; The first candidate power control trajectory subset, the second candidate power control trajectory subset, and the third candidate power control trajectory subset are merged to form the candidate power control trajectory set; The transformation of the quadratic programming approximation problem into the equivalent combinatorial optimization model is achieved by using a discretization strategy of binary expansion encoding or one-hot encoding to map the continuous variables in the quadratic programming approximation problem into discrete binary variables, and incorporating the constraints contained in the quadratic programming approximation problem into the objective function in the form of penalty terms, thus forming an equivalent quadratic unconstrained binary optimization model. The first type of optimization algorithm is a simulated annealing algorithm or a heuristic search algorithm based on the principle of quantum annealing; the second type of optimization algorithm includes at least interior point methods and sequential quadratic programming algorithms.

[0009] Furthermore, in S4, the candidate power control trajectory set is verified and evaluated to determine a comprehensive utility evaluation value that reflects both frequency modulation performance and shaft torsional vibration risk, as follows: A real-time simulation model of electromagnetic-electromechanical transient hybrid system that is consistent with the actual physical response characteristics of the power system is pre-constructed. Each candidate power control trajectory in the candidate power control trajectory set is input into the electromagnetic-electromechanical transient hybrid real-time simulation model for forward simulation calculation to obtain the corresponding simulation calculation results. The simulation calculation results include at least the frequency modulation tracking error and the maximum subsynchronous torsional vibration stress of the unit shaft system. Based on the simulation results and a preset comprehensive utility function, calculate the comprehensive utility evaluation value corresponding to each candidate power control trajectory; Wherein, the preset comprehensive utility function The calculation formula is: ; The frequency modulation tracking error index is... The maximum secondary synchronous torsional vibration stress index of the unit's shaft system. To adjust cost indicators; , , The weighting coefficient is dynamically adjustable; when the power system is determined to be at risk of entering a subsynchronous, weakly damped state, the weighting coefficient is increased. The value.

[0010] Furthermore, in S5, an optimal pre-selected control sequence is selected from the candidate power control trajectory set based on the comprehensive utility evaluation value, as follows: The comprehensive utility evaluation value corresponding to each candidate power control trajectory in the candidate power control trajectory set is compared one by one; The candidate power control trajectory with the highest comprehensive utility evaluation value is selected as the optimal pre-selected control sequence.

[0011] Furthermore, in S6, before executing the optimal pre-selected control sequence, a limit safety check is performed to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold, as follows: The first step control quantity in the optimal pre-selected control sequence is input into the electromagnetic-electromechanical transient hybrid real-time simulation model to perform extreme working condition simulation calculation, and a shaft torsional vibration stress check value is obtained. The shaft torsional vibration stress check value is compared with the preset limit protection threshold to determine whether executing the optimal pre-selected control sequence will cause the unit shaft torsional vibration stress to exceed the limit protection threshold.

[0012] Furthermore, in S7, based on the result of the extreme safety check, corresponding control operations are performed as follows: If the result of the limit safety check is that the shaft torsional vibration stress check value does not exceed the preset limit protection threshold, then the first step control quantity in the optimal pre-selected control sequence will be issued and executed as a frequency modulation command. If the result of the limit safety check is that the shaft system torsional vibration stress check value exceeds the preset limit protection threshold, then the optimal pre-selected control sequence is not executed, and a preset emergency safety avoidance strategy is executed.

[0013] Furthermore, the preset emergency safety avoidance strategy includes: The current power command value is locked and remains unchanged; Alternatively, the unit output can be adjusted to a predetermined safe operating power point at a preset maximum safe rate.

[0014] Hydropower frequency regulation capacity configuration devices considering safety boundary constraints include: The risk perception module is used to acquire broadband data of the power system in real time and determine whether the power system is at risk of entering a subsynchronous weakly damped state based on the broadband data. The problem construction module is used to construct a quadratic programming approximation problem for optimizing frequency modulation control for a preset prediction time domain if the risk is determined to exist. The quadratic programming approximation problem includes dynamic safety constraints for avoiding the risk of the subsynchronous weak damping state. A multi-path solver module is used to solve the quadratic programming approximation problem to generate a set of candidate power control trajectories containing at least one candidate power control trajectory. The simulation evaluation module is used to verify and evaluate the candidate power control trajectory set in order to determine a comprehensive utility evaluation value that can reflect the frequency modulation performance and shaft torsional vibration risk. The optimal selection module is used to select an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value. The limit check module is used to perform a limit safety check before executing the optimal pre-selected control sequence to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold. The instruction execution module is used to perform corresponding control operations based on the results of the extreme safety check.

[0015] Compared with the prior art, the present invention can achieve at least the following beneficial effects: 1. This invention acquires broadband data of the power system in real time and continuously monitors the damping ratio of each oscillation mode in the subsynchronous frequency band based on an online modal analysis algorithm. It can accurately identify the trend of the power system entering a subsynchronous weakly damped state in the early stage of risk formation. When a risk is detected, the state variable constraints corresponding to the high-risk area of ​​subsynchronous oscillation are actively incorporated into the optimization model in the form of a locally linearized inequality constraint set as dynamic safety constraints. This mechanism makes the frequency regulation capacity configuration no longer a passive offline parameter limitation, but a proactive avoidance of potential subsynchronous oscillation risks in each control decision, fundamentally solving the technical problem of existing methods for dynamic implicit risks. By employing a first-type optimization algorithm for global exploration and at least two second-type optimization algorithms for local deep optimization, a diverse set of candidate power control trajectories is generated through parallel solving of multiple algorithms. Based on this, a real-time simulation model of electromagnetic-electromechanical transients is used to perform high-fidelity verification on each candidate trajectory. Furthermore, quantitative evaluation and selection are performed based on a comprehensive utility function that integrates frequency modulation performance and shaft system safety indicators. This effectively offsets the computational uncertainty of a single algorithm and can approach the globally optimal frequency modulation benefit while ensuring the absolute safety of the shaft system. Dynamic safety constraints are embedded at the risk prevention level, and the torsional stress of the shaft system is directly incorporated into the comprehensive utility evaluation for scheme selection. Before execution, the first control quantity of the preferred control sequence is checked for extreme operating conditions. If the check fails, an emergency safety avoidance strategy is immediately activated. The three-layer protection is progressive, providing decisive safety redundancy for hydropower frequency regulation control and effectively ensuring the shaft system safety of megawatt-class large hydropower units in complex power grid environments.

[0016] 2. This invention achieves a leap from static passive limits to proactive avoidance by real-time sensing of subsynchronous oscillation risks in the power system and dynamically embedding risk zone constraints into the optimization model. Through a hierarchical decision-making architecture that combines parallel solution of multiple algorithms with hybrid electromagnetic-electromechanical transient simulation for optimal selection, it effectively overcomes the defect of single algorithms being prone to getting trapped in local optima, and approaches the global optimal frequency regulation efficiency while ensuring shaft system safety. Through a three-layer progressive safety protection mechanism, it provides decisive safety redundancy for the frequency regulation operation of large hydropower units in complex power grid environments, systematically solving the contradiction between frequency regulation safety and regulation performance, and has significant technological progress and broad engineering application prospects. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a block diagram illustrating the control principle of the device of the present invention. Detailed Implementation

[0019] Embodiments of the present invention are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar symbols denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0020] Example 1, please refer to Figure 1 This invention provides a method for configuring hydropower frequency regulation capacity considering safety boundary constraints, comprising the following steps: S1. Acquire broadband data of the power system in real time, and determine whether the power system is at risk of entering a subsynchronous weakly damped state based on the broadband data.

[0021] In an optional embodiment, in S1, broadband data of the power system is acquired in real time, and based on the broadband data, it is determined whether the power system is at risk of entering a subsynchronous weakly damped state, as follows: Real-time acquisition of broadband data of the power system, the broadband data including at least: hydropower unit operating parameters, power grid topology information, power flow information, high voltage DC transmission power, series compensation degree, and high frequency electrical quantity data; It should be noted that the broadband data refers to electrical and state data that can reflect the dynamic characteristics of the power system over a wide frequency range. Its frequency coverage typically includes information from the quasi-steady state to the subsynchronous frequency band and even higher frequencies. Specifically, the broadband data includes at least: operating parameters of hydropower units such as active power, reactive power, terminal voltage, and speed; topology and power flow information of the power grid such as node voltage amplitude and phase angle, and line power flow distribution; transmission power commands and actual values ​​of the high-voltage direct current transmission system; the compensation degree of the series compensation device; and high-frequency electrical data that can reflect the high-frequency dynamic process of the system, such as instantaneous three-phase voltage and current waveform data. Real-time acquisition of the aforementioned broadband data can be achieved through synchronous phasor measurement units and supporting high-speed data acquisition devices deployed at the power plant. These devices can synchronously acquire and upload data at a high sampling rate (e.g., tens to hundreds of frames per second), providing a data foundation for subsequent subsynchronous oscillation risk assessment.

[0022] Based on the broadband data, the damping ratio of each oscillation mode of the power system in the subsynchronous frequency band is continuously calculated using the Proni algorithm; It should be noted that the subsynchronous weakly damped state refers to a subsynchronous frequency band oscillation mode existing in the power system, where the damping ratio drops to a level insufficient to effectively suppress oscillation growth. At this point, the system is extremely sensitive to external disturbances or changes in operating conditions, easily inducing continuous subsynchronous oscillations and even leading to torsional vibration damage to the generator shaft system. The subsynchronous frequency band typically refers to the oscillation frequency range below the power frequency (50Hz or 60Hz), such as the 5Hz to 45Hz range. Within this frequency band, unfavorable electromechanical coupling interactions may occur between the hydropower unit's shaft mechanical system and the series-compensated transmission line or DC transmission control system, forming specific oscillation modes.

[0023] It should be noted that the Proni algorithm is a classic signal processing method that identifies system oscillation modes by fitting signals through a linear combination of exponential functions based on measured data. In this embodiment, based on real-time acquired broadband data, such as generator speed deviation signals and line power fluctuation signals, the Proni algorithm can continuously extract and calculate key parameters such as frequency, damping ratio, and amplitude of each oscillation mode in the subsynchronous frequency band of the power system. The damping ratio of each oscillation mode calculated online by this algorithm can quantitatively assess the current subsynchronous oscillation risk level of the power system in real time.

[0024] The calculated damping ratio of each oscillation mode is compared with a preset first safety margin threshold. It should be noted that the first safety margin threshold is a preset limit value used to distinguish between a power system in a safe state and a subsynchronous weakly damped state. This threshold is usually determined comprehensively based on the power system's safe and stable operation procedures, the torsional vibration fatigue characteristics of the unit's shaft system, and operational experience. For example, the first safety margin threshold can be set to a value within the range of 0.02 to 0.05 for the damping ratio. When the damping ratio of any oscillation mode is detected to drop below this threshold during online monitoring, it indicates that the power system's damping level is on the verge of danger, requiring immediate initiation of more refined frequency regulation capacity optimization control to proactively avoid potential risks. At this point, it is determined that the power system is at risk of entering a subsynchronous weakly damped state.

[0025] If the damping ratio of any of the oscillation modes decreases below the first safety margin threshold, it is determined that the power system is at risk of entering a subsynchronous weakly damped state.

[0026] It should be noted that the above design, by introducing online modal identification technology on the basis of broadband measurement information, realizes real-time quantitative perception of subsynchronous oscillation risk, providing a clear and objective criterion for whether to activate the optimization control process based on safety constraints, effectively avoiding the lag and conservatism of traditional fixed threshold or offline analysis methods.

[0027] S2. If the risk is determined to exist, a quadratic programming approximation problem for optimizing frequency modulation control is constructed for a preset prediction time domain. The quadratic programming approximation problem includes dynamic safety constraints for avoiding the risk of the subsynchronous weak damping state.

[0028] It should be noted that the prediction time domain refers to the length of the future time window considered by the optimization algorithm within the model predictive control framework. This prediction time domain is typically determined by the time step size and number of steps of the control cycle. For example, if the control cycle is 100 milliseconds and the prediction steps are 50, then the prediction time domain is 5 seconds. The purpose of setting the prediction time domain is to enable the optimization problem to proactively consider the dynamic evolution of the system over a future period, thereby planning a control trajectory that can both meet current frequency regulation requirements and actively avoid potential future risks.

[0029] In an optional embodiment, in S2, if it is determined that the risk exists, a quadratic programming approximation problem for optimizing frequency modulation control is constructed for a preset prediction time domain, as follows: At the current operating point, a local time-varying linear state-space model is obtained by real-time linearization processing through a pre-built machine-network coupled dynamic model. It should be noted that the quadratic programming approximation problem is a mathematical method that transforms a complex nonlinear optimization problem into a locally approximate form. Because the high-order nonlinear differential-algebraic equations describing the interaction between the generator, shaft system, and power grid have strong nonlinear characteristics, directly solving them globally is difficult in real-time control scenarios. Therefore, this embodiment performs local linearization on the nonlinear model at the current operating point, obtaining a locally time-varying linear state-space model that can approximate the dynamic characteristics of the system near that operating point. This model characterizes the locally linear relationship between system state variables (such as generator speed, power angle, and the speed and torsion angle of each mass block in the shaft system) and control input variables (such as power command adjustment) in the form of state-space equations.

[0030] The state variable constraints related to the subsynchronous weak damping state risk near the current operating point and the unit's normal operation constraints are linearized to form a set of locally linearized inequality constraints, which constitute the dynamic safety constraints. It should be noted that the dynamic safety constraints refer to constraints used to actively avoid the risk of subsynchronous oscillations. When the power system is in a subsynchronous, weakly damped state, certain specific combinations of state variables (e.g., shaft speed differences or torsional angle differences related to specific torsional vibration modes) constitute high-risk areas for subsynchronous oscillations. This embodiment identifies the state variables corresponding to these high-risk areas using an online modal analysis algorithm and linearizes the constraints of these areas near the current operating point, forming a set of locally linearized inequality constraints. Simultaneously, the unit's routine operating constraints (such as upper and lower output limits, ramp rate limits, etc.) are also linearized near this operating point and incorporated into the set of locally linearized inequality constraints. This set of constraints collectively constitutes the dynamic safety constraints in the quadratic programming approximation problem, ensuring that the optimized control sequence will not introduce the power system into the high-risk area of ​​subsynchronous oscillations.

[0031] Construct a quadratic performance index function with the sum of squares of the tracking deviations between the actual power output of the unit and the grid frequency regulation command over multiple consecutive time steps in the future as the optimization objective; It should be noted that the quadratic performance index function is the objective function of the optimization problem, which takes the form of the sum of squares of the tracking deviations between the actual power output of the generating unit and the frequency regulation command issued by the grid over multiple consecutive time steps. Specifically, the objective function can be expressed as the cumulative sum of the squares of the deviations at each time step, where the deviation at each step is defined as the difference between the planned output value of the generating unit and the frequency regulation command value issued by the grid dispatch at that moment. Using a quadratic function in the form of a sum of squares ensures the convexity of the optimization problem, facilitating efficient solution; furthermore, it imposes a stronger penalty on large deviations, prompting the optimization result to track the frequency regulation command as closely as possible.

[0032] The local time-varying linear state-space model, the local linearization inequality constraint set, and the quadratic performance index function are combined to form the quadratic programming approximation problem.

[0033] It should be noted that the locally time-varying linear state-space model, the locally linearized inequality constraint set, and the quadratic performance index function together constitute the quadratic programming approximation problem. This problem is a standard convex quadratic programming problem, where the decision variables are the power control increment sequences at each time step in the future prediction time domain, and the optimization objective is to minimize the tracking deviation while satisfying the constraints of the power system dynamic equations and various inequality constraints. By constructing and solving this quadratic programming approximation problem in each control cycle, an approximately optimal plan for the future frequency regulation response can be obtained under real-time requirements.

[0034] It should be noted that the real-time linearization process refers to a numerical method that performs a local linear approximation of the aforementioned high-order nonlinear differential-algebraic equations at the current operating point of the system. Since this equation system contains a large number of state variables and nonlinear terms, analytical differentiation is difficult. Therefore, in practical implementation, numerical perturbation or automatic differentiation techniques can be used. The numerical perturbation method approximates the Jacobian matrix by applying small perturbations to each state variable and control variable and observing the rate of change of the system response; the automatic differentiation technique utilizes the principle of computational graph differentiation to automatically calculate the partial derivatives of each variable in a computer program. Both methods can complete the linearization calculation within the real-time control cycle, providing the necessary linear model for the subsequent construction of the quadratic programming problem.

[0035] In an optional embodiment, the real-time linearization process is a local linear approximation of the high-order nonlinear differential-algebraic equations at the current running point using numerical perturbation or automatic differentiation techniques; the future consecutive time steps constitute a preset short-time prediction time domain, the length of which is determined by the time step size and number of steps of a control cycle.

[0036] It should be noted that the above technique cleverly transforms a complex nonlinear optimization problem into a standard convex optimization problem that can be efficiently solved within the real-time control cycle by constructing a local quadratic programming approximation problem at the current operating point. Simultaneously, by incorporating the subsynchronous oscillation high-risk region constraint into the optimization problem in the form of a linearized inequality, it achieves a forward-looking consideration of dynamic safety constraints, providing a clear and efficient mathematical foundation for subsequent multi-algorithm parallel solutions.

[0037] In an optional embodiment, the pre-built machine-grid coupled dynamic model is a set of high-order nonlinear differential-algebraic equations describing the interaction between the generator shaft system and the power grid. The steps for constructing the higher-order nonlinear differential-algebraic equation system include: Based on electromechanical coupling dynamics theory and circuit theory, the mechanical torsional vibration equation of the generator set is established; Establish the electromagnetic transient equations of the synchronous generator; Establish the torque distribution and speed control equations for the prime mover; Establish the excitation regulation control equation; Establish the series capacitor equation for the grid-connected line; The mechanical torsional vibration equation, electromagnetic transient equation, torque distribution and speed control equation, excitation regulation control equation, and series compensation capacitor equation are combined to form a unified set of high-order nonlinear differential algebraic equations.

[0038] It should be noted that the aforementioned generator-grid coupling dynamic model is the fundamental model for real-time linearization. This model is a high-order nonlinear differential-algebraic equation system describing the interaction between the generator, shaft system, and power grid, and its construction is based on electromechanical coupling dynamics theory and circuit theory. Specifically, this equation system consists of the following five parts: First, the mechanical torsional vibration equation of the generator set, used to describe the torque transmission and torsional vibration dynamics between the mass blocks of the steam turbine or hydro turbine, usually using a lumped mass-spring model to represent the shaft system as several mass blocks and connecting springs; second, the electromagnetic transient equation of the synchronous generator, used to describe the electromagnetic transient processes of the generator stator and rotor windings; third, the torque distribution and speed control equation of the prime mover, used to describe the regulation characteristics of the prime mover's output mechanical power; fourth, the excitation regulation control equation, used to describe the regulation effect of the excitation system on the generator terminal voltage; and fifth, the series compensation capacitor equation of the grid-connected line, used to describe the characteristics of the subsynchronous resonant circuit that may be formed between the series compensation capacitor and the line inductance. The above five parts of the equation are coupled together to describe the complete dynamic behavior of the hydropower unit in a complex power grid environment.

[0039] S3. Solve the quadratic programming approximation problem to generate a set of candidate power control trajectories containing at least one candidate power control trajectory.

[0040] It should be noted that the process of solving the quadratic programming approximation problem employs an integrated decision-making framework of multi-algorithm parallel optimization. Its core idea is to obtain a high-quality control solution while meeting real-time requirements by integrating the advantages of different types of optimization algorithms. Due to the complex and variable operating states of power systems, a single optimization algorithm may suffer from problems such as convergence to local optima, uncontrollable computation time, or sensitivity to initial values ​​when facing different operating conditions. Adopting a multi-algorithm parallel solution strategy can effectively mitigate the uncertainties of a single algorithm and improve the robustness and globality of the optimization results.

[0041] In an optional embodiment, in S3, the quadratic programming approximation problem is solved to generate a set of candidate power control trajectories containing at least one candidate power control trajectory, as follows: The quadratic programming approximation problem is transformed into an equivalent combinatorial optimization model, and the first type of optimization algorithm is used to repeatedly solve the equivalent combinatorial optimization model to obtain the first candidate power control trajectory subset. The purpose of transforming the quadratic programming approximation problem into an equivalent combinatorial optimization model is to enable the problem to be effectively solved by first-class optimization algorithms (such as simulated annealing or heuristic search algorithms based on quantum annealing). Since first-class optimization algorithms typically require the decision variables to be discrete, while the decision variables in the original quadratic programming approximation problem (such as power adjustments at each time step) are continuous variables, discretization is necessary. Binary expansion encoding or one-hot encoding are commonly used strategies for discretizing continuous variables. Taking binary expansion encoding as an example, for continuous variables with values ​​within a certain range, they can be represented as a combination of several binary bits, each bit being 0 or 1, thus achieving the mapping from continuous variables to discrete binary variables. The number of bits in the encoding determines the precision of the discretization; more bits result in higher precision, but the solution space also grows exponentially.

[0042] It should be noted that incorporating constraints into the objective function as penalty terms is a classic method for handling constrained optimization problems. After transforming the constraints in the original quadratic programming approximation problem (including the set of locally linearized inequalities in the dynamic safety constraints) into penalty terms, the original problem is transformed into a quadratic unconstrained bivariate optimization model. When a candidate solution violates the constraints, the penalty term will generate a large positive contribution to the objective function, thereby reducing the overall evaluation value of the solution and causing it to be naturally eliminated during the optimization process. This approach allows the first type of optimization algorithm to find feasible solutions that satisfy the constraints without explicitly handling complex constraint logic, guided by the objective function value.

[0043] Simultaneously, at least two second-class optimization algorithms based on different principles are used to directly solve the quadratic programming approximation problem in parallel, respectively obtaining the second candidate power control trajectory subset and the third candidate power control trajectory subset; The first candidate power control trajectory subset, the second candidate power control trajectory subset, and the third candidate power control trajectory subset are merged to form the candidate power control trajectory set; It should be noted that merging the first subset of candidate power control trajectories, the second subset of candidate power control trajectories, and the third subset of candidate power control trajectories to form the candidate power control trajectory set aims to combine the solution advantages of different algorithms. The first subset focuses on global exploration and can provide candidate solutions that can escape local optima; the second and third subsets focus on local depth optimization and can provide theoretically optimal or near-optimal solutions under the current linearization approximation. After merging the three sets of candidate subsets, they are sent to the subsequent simulation verification and evaluation stage. The high-fidelity simulation model verifies the actual effect of each candidate power control trajectory, which can effectively identify theoretical deviations caused by model approximation or algorithm characteristics, and finally select the control sequence that performs best in the actual physical system.

[0044] The transformation of the quadratic programming approximation problem into the equivalent combinatorial optimization model is achieved by using a discretization strategy of binary expansion encoding or one-hot encoding to map the continuous variables in the quadratic programming approximation problem into discrete binary variables, and incorporating the constraints contained in the quadratic programming approximation problem into the objective function in the form of penalty terms, thus forming an equivalent quadratic unconstrained binary optimization model. The first type of optimization algorithm is a simulated annealing algorithm or a heuristic search algorithm based on the principle of quantum annealing; the second type of optimization algorithm includes at least interior point methods and sequential quadratic programming algorithms.

[0045] It should be noted that the first type of optimization algorithm refers to heuristic optimization algorithms with global stochastic exploration capabilities, such as simulated annealing or heuristic search algorithms based on the principle of quantum annealing. Simulated annealing simulates the physical process of solid annealing, and by accepting inferior solutions with a certain probability during the search process, it can effectively escape local optimum traps and gradually approach the global optimum. Heuristic search algorithms based on the principle of quantum annealing utilize the quantum tunneling effect to efficiently explore the solution space, possessing unique advantages in solving combinatorial optimization problems with a large number of local extrema. This embodiment repeatedly solves the quadratic unconstrained binary optimization model, using the stochastic exploration mechanism of the first type of optimization algorithm to obtain multiple candidate solutions with different optimization paths, forming a subset of the first candidate power control trajectory. Repeated solving further enriches the diversity of candidate solutions and reduces the risk of getting trapped in poor local solutions in a single solution.

[0046] It should be noted that the second type of optimization algorithm refers to deterministic optimization algorithms based on gradient information, including at least the interior-point method and the sequential quadratic programming algorithm. The interior-point method transforms inequality constraints into part of the objective function by introducing a barrier function, iteratively approximating the optimal solution within the feasible region, making it particularly suitable for solving large-scale convex quadratic programming problems. The sequential quadratic programming algorithm approximates the original problem as a quadratic programming subproblem at the current iteration point and solves it step by step, exhibiting good convergence performance for nonlinear constraint problems. The aforementioned second type of optimization algorithms are all mature mathematical programming methods, characterized by fast convergence speed and predictable solution quality. This embodiment simultaneously runs at least two second-type optimization algorithms based on different principles, directly solving the quadratic programming approximation problem to generate a second subset of candidate power control trajectories and a third subset of candidate power control trajectories.

[0047] It should be noted that the above technology constructs a highly fault-tolerant optimization decision-making front-end through a mechanism of parallel solving of multiple algorithms and merging of candidate solution sets. This mechanism does not rely on the absolute reliability or absolute optimality of any single algorithm, but rather ensures the quality and safety of the final control decision through algorithm diversity and simulation verification mechanisms, providing a solid technical guarantee for the real-time optimization control of hydropower frequency regulation in complex power grid environments.

[0048] S4. Verify and evaluate the candidate power control trajectory set to determine a comprehensive utility evaluation value that can reflect the frequency modulation performance and shaft torsional vibration risk.

[0049] In an optional embodiment, in S4, the candidate power control trajectory set is verified and evaluated to determine a comprehensive utility evaluation value that reflects the frequency modulation performance and shaft torsional vibration risk, as follows: A real-time simulation model of electromagnetic-electromechanical transient hybrid system that is consistent with the actual physical response characteristics of the power system is pre-constructed. It should be noted that the electromagnetic-electromechanical transient hybrid real-time simulation model is a simulation tool capable of highly faithfully reproducing the dynamic response characteristics of power systems. This simulation model simultaneously considers the dynamic characteristics of both electromagnetic and electromechanical transient processes at two time scales: the electromagnetic transient simulation part accurately simulates the rapid dynamics such as the electromagnetic relationships of the generator stator and rotor windings, the characteristics of line distributed parameters, and the switching processes of power electronic equipment with time steps ranging from microseconds to milliseconds; the electromechanical transient simulation part simulates the slow dynamics of the generator rotor motion equations, the prime mover speed control system, and the excitation regulation system with time steps ranging from milliseconds to seconds. Since subsynchronous oscillation phenomena essentially involve the coupled interaction between the generator shaft mechanical system and the power grid electrical system, a single-scale simulation model cannot accurately capture the dynamic evolution of shaft torsional vibration. Therefore, an electromagnetic-electromechanical transient hybrid real-time simulation model must be used to accurately reproduce the torsional stress response of the unit shaft system during frequency regulation.

[0050] Each candidate power control trajectory in the candidate power control trajectory set is input into the electromagnetic-electromechanical transient hybrid real-time simulation model for forward simulation calculation to obtain the corresponding simulation calculation results. The simulation calculation results include at least the frequency modulation tracking error and the maximum subsynchronous torsional vibration stress of the unit shaft system. It should be noted that inputting each candidate power control trajectory into the simulation model for forward simulation calculation means using the power command values ​​at each time step specified in the candidate power control trajectory as the input excitation of the simulation model, driving the simulation model to advance the calculation along the time axis, and obtaining the complete trajectory of the system state variables changing over time. Through forward simulation calculation, detailed system response information after the execution of the candidate power control trajectory can be obtained, which includes at least two key indicators: one is the frequency regulation tracking error index, which measures the degree of deviation between the actual power output of the unit and the frequency regulation command of the grid under the power control trajectory. The smaller the deviation, the more accurate the frequency regulation response; the other is the maximum subsynchronous torsional stress index of the unit shaft system, which measures the maximum torsional stress value borne by each mass block of the unit shaft system during the frequency regulation process. This value directly reflects the degree of mechanical damage risk caused by subsynchronous oscillation to the unit shaft system.

[0051] Based on the simulation results and a preset comprehensive utility function, calculate the comprehensive utility evaluation value corresponding to each candidate power control trajectory; Wherein, the preset comprehensive utility function The calculation formula is: ; The frequency modulation tracking error index is... The maximum secondary synchronous torsional vibration stress index of the unit's shaft system. To adjust cost indicators; , , The weighting coefficient is dynamically adjustable; when the power system is determined to be at risk of entering a subsynchronous, weakly damped state, the weighting coefficient is increased. The value.

[0052] It should be noted that the comprehensive utility function is an evaluation criterion used to quantitatively assess the overall merits of candidate power control trajectories. This function comprehensively considers three core evaluation dimensions: frequency modulation performance (using frequency modulation tracking error as an indicator). Characterization), operational safety dimension (based on the maximum secondary synchronous torsional vibration stress index of the unit shaft system) (Characteristics) and economic dimensions (to adjust cost indicators) Characterization). Specifically, frequency modulation tracking error index It can be defined as the root mean square value or absolute value integral of the deviation between the actual power and the commanded power at each time step within the simulation period; the maximum secondary synchronous torsional vibration stress index of the unit shaft system. The maximum torsional stress at each section of the shaft system during the simulation period; adjusting the cost index. It can comprehensively reflect the costs of mechanical wear and turbine cavitation losses during power regulation. (Comprehensive utility function) The calculation formula is: ;because , and These are all indicators where smaller is better, therefore they all take a negative sign in the function, making the overall utility value... The larger the value, the better the solution.

[0053] It should be noted that the weighting coefficients , , Weighting coefficients are used to adjust the relative importance of the three evaluation dimensions and can be dynamically adjusted according to the operating status of the power system. The larger the value, the higher the requirement for frequency modulation tracking accuracy, and the more likely the optimization results will favor the control trajectory with smaller tracking error; weighting coefficient A larger value indicates a higher level of emphasis on shaft torsional vibration safety, and the optimization results will be more inclined to avoid control trajectories with high torsional stress; weighting coefficient A larger value indicates a greater emphasis on the economic efficiency of regulation. In particular, when a power system is deemed at risk of entering a subsynchronous, weakly damped state, system operational safety becomes the primary consideration, and in this case, the weighting coefficient is adaptively increased. The value of the value makes the comprehensive utility function impose a stronger penalty on the torsional vibration stress index, thereby guiding the subsequent selection process to prioritize the control sequence that has the least impact on shaft torsional vibration, and achieving proactive avoidance of subsynchronous oscillation risk.

[0054] It should be noted that the above technology, by introducing a high-fidelity electromagnetic-electromechanical transient hybrid real-time simulation model, performs refined simulation verification on each candidate power control trajectory set generated by the parallel solution of multiple algorithms, effectively compensating for the theoretical deviations caused by local linearization of the model and approximation processing in S2 and S3. Based on the simulation calculation results and the quantitative evaluation mechanism of the comprehensive utility function, it can objectively and comprehensively measure the comprehensive performance of each candidate scheme in the three dimensions of frequency modulation performance, shaft safety, and economy, providing a reliable data foundation for the subsequent selection of the optimal pre-selected control sequence.

[0055] S5. Select an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value.

[0056] It should be noted that the step of selecting an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value is a decision-making step performed after the simulation verification and quantitative evaluation of each candidate power control trajectory is completed in the aforementioned S4. Since the design of the comprehensive utility function U ensures that a higher evaluation value represents a better overall performance of the scheme in terms of frequency regulation performance, shaft safety, and economy, the problem of selecting the optimal scheme is transformed into finding the candidate power control trajectory with the largest comprehensive utility evaluation value from the candidate set.

[0057] In an optional embodiment, in S5, an optimal pre-selected control sequence is selected from the candidate power control trajectory set based on the comprehensive utility evaluation value, as follows: The comprehensive utility evaluation value corresponding to each candidate power control trajectory in the candidate power control trajectory set is compared one by one; It should be noted that the step-by-step comparison refers to the process of comparing the comprehensive utility evaluation values ​​corresponding to each candidate power control trajectory in the candidate power control trajectory set. Since the candidate power control trajectory set includes a first candidate subset generated from repeated solutions using the first type of optimization algorithm, and second and third candidate subsets generated from solutions using at least two second type of optimization algorithms, the number of candidate schemes is large and their sources are diverse. A comprehensive comparison ensures that no potentially high-quality schemes are overlooked. In practical implementation, a simple sorting algorithm or a maximum value search algorithm can be used to efficiently complete the comparison process.

[0058] The candidate power control trajectory with the highest comprehensive utility evaluation value is selected as the optimal pre-selected control sequence.

[0059] It should be noted that the candidate power control trajectory with the highest comprehensive utility evaluation value was selected as the optimal pre-selected control sequence because this candidate scheme demonstrated the best frequency regulation tracking capability, the lowest shaft torsional vibration risk, and reasonable regulation economy in simulation verification, thus maximizing the balance between safety and efficiency in hydropower frequency regulation. It is particularly important to point out that the sequence selected here is called the pre-selected control sequence because it undergoes a final safety verification in the S6 limit safety check stage before execution, ensuring that even under extreme operating conditions, the sequence will not cause unacceptable damage to the unit's shaft system.

[0060] It should be noted that the above steps automatically select the control sequence with the best overall performance from a variety of candidate schemes through an objective and quantitative utility evaluation value comparison mechanism, avoiding the subjectivity and uncertainty of human experience judgment, and providing a reliable technical means for realizing intelligent decision-making in hydropower frequency regulation control.

[0061] S6. Before executing the optimal pre-selected control sequence, perform a limit safety check to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold.

[0062] It should be noted that the extreme safety check is the final safety verification step performed before the optimal pre-selected control sequence is executed. Although S4 has performed a forward simulation evaluation of each candidate power control trajectory using an electromagnetic-electromechanical transient hybrid real-time simulation model and calculated the maximum subsynchronous torsional vibration stress index of the unit shaft system as part of the comprehensive utility evaluation, this evaluation was conducted under the assumption of normal operating conditions of the power system. However, in actual power system operation, there may be various uncertainties and potential disturbances, such as sudden drops in grid voltage, transient processes after clearing line short-circuit faults, and emergency DC power modulation, among other extreme conditions. These extreme conditions may cause the torsional vibration stress of the shaft system to increase sharply in a short period of time, exceeding the stress level under normal operating conditions. Therefore, it is necessary to perform a dedicated extreme condition safety check on the first control quantity of the optimal pre-selected control sequence before executing the control command.

[0063] In an optional embodiment, in S6, before executing the optimal pre-selected control sequence, a limit safety check is performed to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold, as follows: The first step control quantity in the optimal pre-selected control sequence is input into the electromagnetic-electromechanical transient hybrid real-time simulation model to perform extreme working condition simulation calculation, and a shaft torsional vibration stress check value is obtained. It should be noted that the first-step control quantity refers to the power command adjustment quantity in the optimal pre-selected control sequence that should be executed immediately in the current control cycle. Under the rolling optimization framework of model predictive control, although each solution generates a complete control sequence in the future prediction time domain, only the first-step control command of that sequence is issued during actual execution. After the system state is updated, the optimization solution is re-performed in the next control cycle. Therefore, the limit safety check only needs to be performed on the first-step control quantity that is about to be executed, without needing to check the entire control sequence, thus reducing the computational burden while ensuring safety.

[0064] The shaft torsional vibration stress check value is compared with the preset limit protection threshold to determine whether executing the optimal pre-selected control sequence will cause the unit shaft torsional vibration stress to exceed the limit protection threshold.

[0065] It should be noted that the extreme operating condition simulation calculation refers to setting more stringent boundary conditions than normal operating conditions in the simulation model. For example, a grid voltage disturbance of preset amplitude can be superimposed in the simulation, a transient fault in a nearby line can be simulated, or the response characteristics of the speed governor and excitation system under the most unfavorable parameter combination can be considered. Through simulation calculation under extreme operating conditions, the maximum value of shaft torsional vibration stress that the control command may induce under the worst scenario can be obtained, i.e., the shaft torsional vibration stress check value.

[0066] It should be noted that the preset limit protection threshold is the upper limit of torsional vibration stress that the unit's shaft system can withstand, and it is usually determined comprehensively based on the fatigue characteristic data of the shaft system material, design specifications, and operating procedures provided by the unit manufacturer. This threshold has a different function and numerical level from the first safety margin threshold mentioned in S1: the first safety margin threshold is used for risk warning and activation of optimized control processes, and is a preventive threshold; the limit protection threshold is used to determine whether to allow the execution of control commands, and is a protective threshold. Generally speaking, the value of the limit protection threshold is higher than the stress level corresponding to the first safety margin threshold, and directly corresponds to the allowable stress limit or fatigue damage critical value of the shaft system material.

[0067] S7. Based on the results of the extreme safety check, execute the corresponding control operations.

[0068] It should be noted that the implementation of corresponding control operations based on the results of the limit safety check reflects the layered safety control concept of this method, which prioritizes prevention and provides a safety net. Two different control response methods are adopted based on different results of the limit safety check.

[0069] In an optional embodiment, in S7, based on the result of the extreme safety check, a corresponding control operation is performed as follows: If the result of the limit safety check is that the shaft torsional vibration stress check value does not exceed the preset limit protection threshold, then the first step control quantity in the optimal pre-selected control sequence will be issued and executed as a frequency modulation command. It should be noted that when the result of the limit safety check shows that the shaft torsional vibration stress check value does not exceed the preset limit protection threshold, it indicates that even under the set extreme operating conditions, executing the first step of the control will not cause unbearable mechanical damage to the unit shaft. At this time, the control is deemed safe and feasible, and it is formally issued to the hydropower unit controller as a frequency regulation command for execution. After receiving the command, the unit controller adjusts the guide vane opening or blade angle through the speed governor to change the unit output, thereby responding to the grid frequency regulation requirements.

[0070] If the result of the limit safety check is that the shaft system torsional vibration stress check value exceeds the preset limit protection threshold, then the optimal pre-selected control sequence is not executed, and a preset emergency safety avoidance strategy is executed.

[0071] It should be noted that when the result of the limit safety check shows that the shaft torsional vibration stress check value exceeds the preset limit protection threshold, it indicates that if this control quantity is executed, under the worst operating conditions, the unit shaft may be subjected to torsional vibration stress exceeding the safety limit, posing a significant safety hazard of shaft fatigue damage or even fracture. In this case, the execution of the optimal pre-selected control sequence must be refused, and the preset emergency safety avoidance strategy must be activated immediately.

[0072] It should be noted that the preset emergency safety avoidance strategy is a pre-established safety response measure to address situations where the extreme safety check fails, with its primary objective being to ensure the safety of the unit's shaft system. This strategy includes at least two options: First, locking the current power command value unchanged, i.e., temporarily suspending responses to changes in grid frequency regulation commands, maintaining the unit output at the current safe level, and waiting for the risk of subsynchronous oscillations in the power system to decrease before resuming normal frequency regulation; Second, adjusting the unit output to a predetermined safe operating power point at a preset maximum safe rate. The predetermined safe operating power point refers to the output level determined based on offline analysis and operational experience, where the torsional stress of the unit's shaft system is within a safe range under this operating mode. Regardless of the method adopted, the execution of the emergency safety avoidance strategy aims to minimize the impact on the grid frequency regulation function while ensuring the safety of the unit.

[0073] It should be noted that after completing the control operation in S7, regardless of whether a normal frequency regulation command or an emergency safety avoidance strategy is executed, the power system state has changed. At this point, the system will return to S1 to reacquire the updated power system broadband data and enter the risk assessment and optimization control process of the next control cycle. This cycle repeats continuously, forming a continuous rolling optimization closed loop under the model predictive control architecture, ensuring that hydropower frequency regulation control always operates under the premise of meeting dynamic safety boundary constraints.

[0074] Example 2, please refer to Figure 2 A hydropower frequency regulation capacity configuration device considering safety boundary constraints includes: The risk perception module is used to acquire broadband data of the power system in real time and determine whether the power system is at risk of entering a subsynchronous weakly damped state based on the broadband data. The problem construction module is used to construct a quadratic programming approximation problem for optimizing frequency modulation control for a preset prediction time domain if the risk is determined to exist. The quadratic programming approximation problem includes dynamic safety constraints for avoiding the risk of the subsynchronous weak damping state. A multi-path solver module is used to solve the quadratic programming approximation problem to generate a set of candidate power control trajectories containing at least one candidate power control trajectory. The simulation evaluation module is used to verify and evaluate the candidate power control trajectory set in order to determine a comprehensive utility evaluation value that can reflect the frequency modulation performance and shaft torsional vibration risk. The optimal selection module is used to select an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value. The limit check module is used to perform a limit safety check before executing the optimal pre-selected control sequence to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold. The instruction execution module is used to perform corresponding control operations based on the results of the extreme safety check.

[0075] In summary, this invention provides a systematic solution to the contradiction between safety boundary constraints and frequency regulation performance in hydropower frequency regulation, and has significant technological advancements and broad engineering application prospects.

[0076] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0077] In the embodiments provided by this invention, it should be understood that the disclosed system or method can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for instance, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0078] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in a combination of hardware and software functional modules.

[0080] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the basic characteristics of the present invention.

[0081] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for configuring hydropower frequency regulation capacity considering safety boundary constraints, characterized in that, Includes the following steps: S1. Acquire broadband data of the power system in real time, and determine whether the power system is at risk of entering a subsynchronous weakly damped state based on the broadband data; S2. If the risk is determined to exist, a quadratic programming approximation problem for optimizing frequency modulation control is constructed for a preset prediction time domain. The quadratic programming approximation problem includes dynamic safety constraints for avoiding the risk of the subsynchronous weak damping state. S3. Solve the quadratic programming approximation problem to generate a set of candidate power control trajectories containing at least one candidate power control trajectory; S4. Verify and evaluate the candidate power control trajectory set to determine a comprehensive utility evaluation value that can reflect the frequency modulation performance and shaft torsional vibration risk. S5. Select an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value; S6. Before executing the optimal pre-selected control sequence, perform a limit safety check to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold. S7. Based on the results of the extreme safety check, execute the corresponding control operations.

2. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 1, characterized in that: In S1, broadband data of the power system is acquired in real time, and based on the broadband data, it is determined whether the power system is at risk of entering a subsynchronous weakly damped state, as follows: Real-time acquisition of broadband data of the power system, the broadband data including at least: hydropower unit operating parameters, power grid topology information, power flow information, high voltage DC transmission power, series compensation degree, and high frequency electrical quantity data; Based on the broadband data, the damping ratio of each oscillation mode of the power system in the subsynchronous frequency band is continuously calculated using the Proni algorithm; The calculated damping ratio of each oscillation mode is compared with a preset first safety margin threshold. If the damping ratio of any of the oscillation modes decreases below the first safety margin threshold, it is determined that the power system is at risk of entering a subsynchronous weakly damped state.

3. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 1, characterized in that: In S2, if the risk is determined to exist, a quadratic programming approximation problem for optimizing frequency modulation control is constructed for a preset prediction time domain, as follows: At the current operating point, a local time-varying linear state-space model is obtained by real-time linearization processing through a pre-built machine-network coupled dynamic model. The state variable constraints related to the subsynchronous weak damping state risk near the current operating point and the unit's normal operation constraints are linearized to form a set of locally linearized inequality constraints, which constitute the dynamic safety constraints. Construct a quadratic performance index function with the sum of squares of the tracking deviations between the actual power output of the unit and the grid frequency regulation command over multiple consecutive time steps in the future as the optimization objective; The local time-varying linear state-space model, the local linearization inequality constraint set, and the quadratic performance index function are combined to form the quadratic programming approximation problem.

4. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 1, characterized in that: In S3, the quadratic programming approximation problem is solved to generate a set of candidate power control trajectories containing at least one candidate power control trajectory, as follows: The quadratic programming approximation problem is transformed into an equivalent combinatorial optimization model, and the first type of optimization algorithm is used to repeatedly solve the equivalent combinatorial optimization model to obtain the first candidate power control trajectory subset. Simultaneously, at least two second-class optimization algorithms based on different principles are used to directly solve the quadratic programming approximation problem in parallel, respectively obtaining the second candidate power control trajectory subset and the third candidate power control trajectory subset; The first candidate power control trajectory subset, the second candidate power control trajectory subset, and the third candidate power control trajectory subset are merged to form the candidate power control trajectory set; The transformation of the quadratic programming approximation problem into the equivalent combinatorial optimization model is achieved by using a discretization strategy of binary expansion encoding or one-hot encoding to map the continuous variables in the quadratic programming approximation problem into discrete binary variables, and incorporating the constraints contained in the quadratic programming approximation problem into the objective function in the form of penalty terms, thus forming an equivalent quadratic unconstrained binary optimization model. The first type of optimization algorithm is a simulated annealing algorithm or a heuristic search algorithm based on the principle of quantum annealing; the second type of optimization algorithm includes at least interior point methods and sequential quadratic programming algorithms.

5. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 1, characterized in that: In S4, the candidate power control trajectory set is verified and evaluated to determine a comprehensive utility evaluation value that reflects both frequency modulation performance and shaft torsional vibration risk, as follows: A real-time simulation model of electromagnetic-electromechanical transient hybrid system that is consistent with the actual physical response characteristics of the power system is pre-constructed. Each candidate power control trajectory in the candidate power control trajectory set is input into the electromagnetic-electromechanical transient hybrid real-time simulation model for forward simulation calculation to obtain the corresponding simulation calculation results. The simulation calculation results include at least the frequency modulation tracking error and the maximum subsynchronous torsional vibration stress of the unit shaft system. Based on the simulation results and a preset comprehensive utility function, calculate the comprehensive utility evaluation value corresponding to each candidate power control trajectory; Wherein, the preset comprehensive utility function The calculation formula is: ； The frequency modulation tracking error index is... The maximum secondary synchronous torsional vibration stress index of the unit's shaft system. To adjust cost indicators; , , The weighting coefficient is dynamically adjustable; when the power system is determined to be at risk of entering a subsynchronous, weakly damped state, the weighting coefficient is increased. The value.

6. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 1, characterized in that: In S5, an optimal pre-selected control sequence is selected from the candidate power control trajectory set based on the comprehensive utility evaluation value, as follows: The comprehensive utility evaluation value corresponding to each candidate power control trajectory in the candidate power control trajectory set is compared one by one; The candidate power control trajectory with the highest comprehensive utility evaluation value is selected as the optimal pre-selected control sequence.

7. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 5, characterized in that: In S6, before executing the optimal pre-selected control sequence, a limit safety check is performed to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold, as follows: The first step control quantity in the optimal pre-selected control sequence is input into the electromagnetic-electromechanical transient hybrid real-time simulation model to perform extreme working condition simulation calculation, and a shaft torsional vibration stress check value is obtained. The shaft torsional vibration stress check value is compared with the preset limit protection threshold to determine whether executing the optimal pre-selected control sequence will cause the unit shaft torsional vibration stress to exceed the limit protection threshold.

8. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 7, characterized in that: In S7, based on the result of the extreme safety check, the corresponding control operation is performed as follows: If the result of the limit safety check is that the shaft torsional vibration stress check value does not exceed the preset limit protection threshold, then the first step control quantity in the optimal pre-selected control sequence will be issued and executed as a frequency modulation command. If the result of the limit safety check is that the shaft system torsional vibration stress check value exceeds the preset limit protection threshold, then the optimal pre-selected control sequence is not executed, and a preset emergency safety avoidance strategy is executed.

9. The hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in claim 8, characterized in that: The preset emergency safety avoidance strategies include: The current power command value is locked and remains unchanged; Alternatively, the unit output can be adjusted to a predetermined safe operating power point at a preset maximum safe rate.

10. A hydropower frequency regulation capacity configuration device considering safety boundary constraints, employing the hydropower frequency regulation capacity configuration method considering safety boundary constraints as described in any one of claims 1-9, characterized in that, include: The risk perception module is used to acquire broadband data of the power system in real time and determine whether the power system is at risk of entering a subsynchronous weakly damped state based on the broadband data. The problem construction module is used to construct a quadratic programming approximation problem for optimizing frequency modulation control for a preset prediction time domain if the risk is determined to exist. The quadratic programming approximation problem includes dynamic safety constraints for avoiding the risk of the subsynchronous weak damping state. A multi-path solver module is used to solve the quadratic programming approximation problem to generate a set of candidate power control trajectories containing at least one candidate power control trajectory. The simulation evaluation module is used to verify and evaluate the candidate power control trajectory set in order to determine a comprehensive utility evaluation value that can reflect the frequency modulation performance and shaft torsional vibration risk. The optimal selection module is used to select an optimal pre-selected control sequence from the candidate power control trajectory set based on the comprehensive utility evaluation value. The limit check module is used to perform a limit safety check before executing the optimal pre-selected control sequence to determine whether executing the optimal pre-selected control sequence will cause the torsional vibration stress of the unit shaft system to exceed a preset limit protection threshold. The instruction execution module is used to perform corresponding control operations based on the results of the extreme safety check.

Images