Hydroelectric power station governor primary frequency modulation adaptive control method based on network source cooperation

Abstract

The present application relates to the technical field of automatic control of hydropower station, disclose a kind of adaptive control method of hydropower station governor primary frequency modulation based on network source cooperation, including obtaining grid frequency rate of change, active power fluctuation and bottom physical data;According to physical data, calculate dynamic water flow inertia time, combine active power fluctuation and frequency rate of change to calculate grid equivalent inertia, reconstruct anti-reverse regulation threshold boundary;Calculate the real-time module length and phase angle of the trajectory vector of frequency difference and frequency rate of change;When satisfying the threshold boundary condition of frequency drop, generate feedforward differential negative compensation based on frequency rate of change;When determining that water flow establishes positive acceleration, generate dynamic clamping coefficient using vortex belt pressure fluctuation derivative, limit the recovery trajectory of main loop proportional gain;Finally, generate comprehensive guide vane opening degree instruction by fusing each control quantity and issue execution.The present application inhibits the water hammer reverse regulation effect in the initial stage of primary frequency modulation, reduces the pressure fluctuation risk in the later stage of regulation, and guarantees the stable operation of unit.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for hydropower stations, specifically to an adaptive control method for primary frequency regulation of a hydropower station governor based on grid-source coordination. Background Technology

[0002] When a hydroelectric turbine participates in primary frequency regulation of the power grid, the water hammer effect of the water intake system causes reverse fluctuations in mechanical power during the initial regulation phase when the turbine opens its guide vanes. Existing control methods fail to establish effective boundary protection and feedforward intervention mechanisms that combine grid inertia and dynamic water flow characteristics, resulting in these reverse fluctuations in mechanical power often exacerbating the frequency drop on the grid side.

[0003] Meanwhile, in the later stages of frequency regulation, after the water flow in the intake system establishes positive acceleration, the proportional gain of the main controller in the existing control loop usually recovers according to preset parameters, lacking feedback constraints on the actual hydraulic state inside the unit. Because the low-frequency vortex pressure pulsation characteristics inside the tailrace tube are not extracted to limit the control loop, an excessively rapid gain recovery rate can easily cause a surge in pressure pulsation inside the tailrace tube, reducing the hydraulic and mechanical structural stability of the unit during frequency regulation.

[0004] The fundamental reason for the exacerbated reverse regulation in the initial stage and the surge in pulsation in the later stage of regulation is that traditional primary frequency regulation control methods mainly rely on single unit speed feedback and fixed control parameter configurations. Because a synchronous acquisition mechanism covering the status of the wide-area power grid and the underlying physical data of the unit has not been established, the existing frequency regulation strategy cannot fully perceive the real-time changes on both the grid and power source sides, making it impossible for the governor's control commands to dynamically match the current real-time inertia level of the power grid and the kinetic energy of the water flow within the water diversion system. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an adaptive control method for primary frequency regulation of hydropower station governors based on grid-source coordination. This method solves the problem of water hammer effect caused by the non-minimum phase characteristics of the water diversion system during the primary frequency regulation of the power grid, which leads to reverse fluctuations in mechanical power during the initial adjustment phase of the unit.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a primary frequency regulation adaptive control method for hydropower station governors based on grid-source coordination, which constructs a closed-loop control mechanism that integrates feedforward suppression and dynamic gain recovery by synchronously collecting and calculating underlying physical data and wide-area power grid status.

[0007] Specifically, the method performs the following control steps:

[0008] Acquire the grid frequency change rate, active power fluctuation and underlying physical data; calculate the dynamic water flow inertia time based on the underlying physical data, calculate the grid equivalent inertia by combining the active power fluctuation and grid frequency change rate, and reconstruct the anti-reverse regulation threshold boundary using the grid equivalent inertia and dynamic water flow inertia time.

[0009] Calculate the real-time magnitude and phase angle of the phase trajectory vector containing the unit frequency deviation and the grid frequency change rate in the underlying physical data;

[0010] When the phase angle and real-time modulus satisfy the frequency acceleration drop condition and the anti-reverse regulation threshold boundary, the feedforward differential negative compensation quantity is generated based on the grid frequency change rate.

[0011] When the node that establishes positive acceleration in the water flow is determined based on the underlying physical data, the time derivative data of the low-frequency vortex pressure pulsation is obtained based on the underlying physical data, a dynamic clamping coefficient is generated, and the exponential recovery trajectory of the proportional gain of the main loop regulator is restricted.

[0012] The feedforward differential negative compensation quantity is integrated with the output quantity of the main loop regulator after being limited to generate a comprehensive guide vane opening command and issue it for execution. This enables the primary frequency regulation adaptive control of the hydropower station governor and returns to the execution steps to obtain the grid frequency change rate, active power fluctuation quantity and underlying physical data to achieve cyclical operation.

[0013] The innovative principle of this invention lies in establishing a boundary assessment mechanism for grid-source coordination and a multi-stage parameter intervention mechanism. In the initial stage of control, the safety boundary for preventing reverse regulation is redefined using the equivalent inertia of the power grid and the dynamic inertial time of the water flow. When a frequency drop is detected, a feedforward differential negative compensation based on the state of the power grid system is introduced to suppress the rapid opening of the guide vanes and reduce the reverse regulation impact caused by water hammer on the power grid side. In the later stage of regulation when the positive kinetic energy of the water flow is established, a dynamic clamping mechanism based on the feedback of the pressure pulsation derivative of the tailrace vortex band is introduced. The recovery trajectory of the proportional gain in the control loop is constrained by the physical hydraulic characteristics, thereby achieving a dynamic balance between suppressing the initial reverse regulation amplitude and ensuring the hydraulic stability of the unit in the later stage of regulation.

[0014] Preferably, when acquiring underlying physical data, the distributed control system establishes data transmission channels with the wide-area measurement system, edge computing gateway, and data acquisition system. The wide-area measurement system monitors the bus voltage and bus current in real time and performs phasor operations to analyze the grid frequency change rate and active power fluctuation; the data acquisition system extracts the actual operating frequency of the unit to calculate the unit frequency deviation, measures the inlet pressure of the water intake to calculate the net head, and records the spiral casing pressure, tailrace water level elevation, guide vane opening reflecting the physical position of the mechanical guide vanes, and the raw dynamic pressure signal of the tailrace pipe.

[0015] Preferably, when calculating the dynamic flow inertia time, the diversion flow rate is extracted using the head and guide vane opening combined with the turbine flow characteristic curve; the ratio of the physical length to the cross-sectional area of ​​each pipe section in the diversion system is set as the geometric characteristic impedance ratio, and multiplied by the diversion flow rate to obtain the transient flow rate within a single pipe section; the transient flow rates of each section are accumulated to form a discrete integral of the internal flow inertia, and then this discrete integral is divided by the potential energy reference value generated by the head to output the dynamic flow inertia time. This calculation architecture maps fixed diversion geometric parameters to dynamic parameters reflecting the current hydraulic potential energy conversion.

[0016] Preferably, when calculating the equivalent inertia of the power grid, a dead zone for the rate of change of frequency is introduced to eliminate the interference of small fluctuations in the steady state of the power grid. When the absolute value of the rate of change of the power grid frequency is greater than the set dead zone for the rate of change of frequency, the absolute value of the obtained active power fluctuation is divided by the product of the absolute value of the rate of change of the power grid frequency, the rated frequency of the power grid, and a constant, to obtain the observed value of apparent inertia; then, a first-order low-pass filter is applied to the observed value of apparent inertia to obtain a smooth equivalent inertia of the power grid.

[0017] Preferably, when reconstructing the anti-reverse regulation threshold boundary, a clamping limit is applied to the dynamic water flow inertia time by setting a lower limit to prevent subsequent data overflow caused by calculating the minimum value; the grid-source inertia matching coefficient is calculated by dividing the grid's equivalent inertia by the dynamic water flow inertia time; the grid-source inertia matching coefficient is linearly calculated using an adaptive weighting coefficient to generate a dynamic correction factor, which is then multiplied by the baseline boundary constant to output the anti-reverse regulation threshold boundary. This boundary characterizes the adjustment threshold of the current power grid corresponding to the water hammer reverse regulation effect.

[0018] Preferably, when calculating the real-time magnitude and phase angle of the phase trajectory vector, the frequency difference normalization coefficient and the frequency derivative normalization coefficient are used to process the unit frequency deviation and the grid frequency change rate, respectively; the sum of squares of the normalized two values ​​is calculated and the square root operation is performed to output the real-time magnitude; the tangent ratio is obtained by dividing the normalized frequency change rate by the denominator formed by the actual normalized frequency difference or the preset zero-pole constant; and then the arctangent function is used to perform angle mapping on the tangent ratio to output the phase angle value within a specific range.

[0019] Preferably, when determining the control triggering conditions, the frequency acceleration drop condition is confirmed by determining that the unit frequency deviation and the grid frequency change rate are both in the negative range; the real-time modulus is compared with the anti-reverse regulation threshold boundary, and the anti-reverse regulation over-limit condition is confirmed when the modulus is greater than the boundary; a logical AND operation is performed on the above two conditions, and a control switch command for triggering asymmetric water hammer feedforward suppression is issued according to the logical AND operation result.

[0020] Preferably, before generating the feedforward differential negative compensation, the proportional gain of the main loop regulator is blocked using an exponential decay algorithm. The command trigger time is recorded, the time difference at the current moment is calculated and divided by the proportional decay constant, and the negative value of the calculated result is used as the exponential parameter to obtain the dynamic decay factor. The proportional control output at the control loop freeze moment is obtained as the initial proportional gain, and these are multiplied to form the real-time proportional gain after blocking. Simultaneously, a dynamic feedforward gain is calculated based on the set feedforward differential gain constant and the grid-source inertia matching coefficient. This dynamic feedforward gain is multiplied by the absolute value of the grid frequency change rate, and the negative value is taken to form the feedforward differential negative compensation.

[0021] Preferably, during the generation of the dynamic clamping coefficient, a time integral operation is performed on the transient pressure deviation between the volute pressure and the rated steady-state volute pressure. When the accumulated pressure deviation turns into a positive value and the state maintenance time exceeds the set duration threshold, it is confirmed that the water flow establishes a positive acceleration. Subsequently, low-pass filtering and envelope detection processing are performed on the raw dynamic pressure signal of the tailrace pipe. The difference between the amplitude envelope signals of adjacent sampling periods is obtained and divided by a fixed sampling period to obtain the vortex pressure pulsation derivative. This derivative is multiplied by the sensitivity constant and the negative value is taken to generate the dynamic clamping coefficient. The time-decaying exponential transition factor is multiplied by the target proportional gain calculated from the reference proportional gain, and the result of the multiplication is used to constrain the recovery rate of the proportional gain of the main loop regulator.

[0022] Preferably, the generation of the comprehensive guide vane opening command is completed by combining the adjustment amounts calculated in each stage. In the anti-reverse control stage, the output of the main loop regulator and the feedforward differential negative compensation amount are superimposed. In the acceleration output recovery stage, the proportional gain of the recovered main regulator is used to calculate the adjustment command. The comprehensive guide vane opening commands of adjacent cycles are subtracted and divided by a fixed sampling period to obtain the command change rate. For changes that exceed the limit, clamping is performed. After double limiting of amplitude and change rate, the target opening command is output, which is converted into an analog control voltage signal and drives the mechanical guide vane of the electro-hydraulic servo system to perform the corresponding action.

[0023] This invention provides an adaptive control method for primary frequency regulation of a hydropower station governor based on grid-source coordination. It has the following beneficial effects:

[0024] 1. This invention reconstructs the anti-reverse regulation threshold boundary by combining the equivalent inertia of the power grid and the dynamic water flow inertia time, and generates a feedforward differential negative compensation amount based on the power grid frequency change rate when the phase trajectory magnitude and phase angle meet specific conditions. This suppresses the mechanical power reverse fluctuation caused by the water hammer effect in the initial regulation stage, reduces the reverse regulation amplitude caused by the opening of the guide vanes, and thus slows down the frequency drop on the power grid side.

[0025] 2. This invention generates a dynamic clamping coefficient by extracting the time derivative data of the low-frequency vortex pressure pulsation in the tailrace after determining that the water flow has established positive acceleration. This restricts the exponential recovery trajectory of the proportional gain of the main circuit regulator, constrains the gain recovery rate in the later stage of regulation, reduces the risk of pressure pulsation surge inside the tailrace, and ensures the hydraulic and mechanical structural stability of the unit during frequency regulation.

[0026] 3. This invention establishes a multi-stage parameter control mechanism that covers the states of both the grid and the source by synchronously acquiring the frequency change rate, active power fluctuation, and underlying physical data of the wide-area power grid and the generator side. This changes the limitations of traditional frequency regulation that relies on single generator speed feedback and fixed parameter configuration, and realizes dynamic matching between governor control commands and the real-time inertia level of the power grid and the kinetic energy of the water diversion system. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the overall system architecture of the present invention;

[0028] Figure 2 The flowchart is a process for the adaptive frequency regulation method of the hydropower station governor based on grid-source coordination, as described in this invention.

[0029] Figure 3 This is a flowchart of the network source dynamic parameter matching process of the present invention;

[0030] Figure 4 This is a flowchart of the asymmetric water hammer feedforward suppression control of the present invention;

[0031] Figure 5 This is a flowchart of the tailrace vortex dynamic clamping and output recovery control of the present invention;

[0032] Figure 6 This is a comparison curve of the unit frequency deviation and guide vane opening response according to the present invention;

[0033] Figure 7 This is a graph showing the relative deviation between the proportional gain of the main regulator and the volute pressure in this invention. Detailed Implementation

[0034] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] See attached document Figure 1This invention provides a primary frequency regulation adaptive control system for a hydropower station governor based on network-source collaboration, comprising: a distributed control system, a wide-area measurement system, an edge computing gateway, a data acquisition system, and an electro-hydraulic servo system.

[0036] The distributed control system connects to the wide-area measurement system via an industrial Ethernet communication bus. The distributed control system receives the grid frequency change rate and active power fluctuations from the wide-area measurement system. As the core computational hub, the distributed control system performs grid-source parameter matching and adaptive control law calculation.

[0037] The underlying physical data acquired by the data acquisition system includes raw signals of unit frequency deviation, clean water head, guide vane opening, spiral casing pressure, and tailrace pipe dynamic pressure. The data acquisition system synchronously transmits the underlying physical data to the distributed control system and edge computing gateway.

[0038] The edge computing gateway connects to both the data acquisition system and the distributed control system. It provides high-frequency signal processing computing power as a bypass for the distributed control system. The edge computing gateway receives the raw dynamic pressure signal from the tailrace pipe, performs signal processing, extracts the time derivative data of the low-frequency vortex pressure pulsation, and sends it to the distributed control system.

[0039] The electro-hydraulic servo system is connected to the distributed control system. The electro-hydraulic servo system receives the integrated guide vane opening command and drives the mechanical guide vanes of the turbine to perform opening adjustment actions.

[0040] The adaptive control system for primary frequency regulation of hydropower station governors based on grid-source collaboration provides the hardware operating environment for the adaptive control method for primary frequency regulation of hydropower station governors based on grid-source collaboration. (See attached document.) Figure 2 This invention provides a primary frequency regulation adaptive control method for a hydropower station governor based on grid-source coordination, comprising the following steps:

[0041] S10. Synchronously acquire multi-source heterogeneous data. The distributed control system acquires the grid frequency change rate and active power fluctuation through the communication link. The distributed control system synchronously acquires the raw signals of unit frequency deviation, water head, guide vane opening, volute pressure and tailrace dynamic pressure transmitted by the data acquisition system.

[0042] S20. Calculate the dynamic parameter matching boundary of the power grid. The distributed control system calculates the dynamic water flow inertia time based on the net head and guide vane opening. The distributed control system calculates the equivalent inertia of the power grid by combining the active power fluctuation and the power grid frequency change rate. The distributed control system reconstructs the anti-reverse regulation threshold boundary using the ratio of the equivalent inertia of the power grid and the dynamic water flow inertia time. The distributed control system uses the anti-reverse regulation threshold boundary as the judgment condition for executing asymmetric water hammer feedforward suppression. In this calculation stage, zero-order hold logic and low-pass filtering are arranged to avoid zero poles and suppress random data fluctuations.

[0043] S30. Construct the state space of frequency difference phase trajectory characteristics. The distributed control system extracts the vector characteristics of the unit frequency deviation and the grid frequency change rate, and calculates the real-time magnitude and phase angle of the phase trajectory vector;

[0044] S40. Execute asymmetric water hammer feedforward suppression. When the phase angle and real-time modulus satisfy the frequency acceleration drop condition and the anti-reverse regulation threshold boundary, the distributed control system blocks the proportional gain of the main circuit regulator. The distributed control system generates a feedforward differential negative compensation amount based on the grid frequency change rate and adds the feedforward differential negative compensation amount to the guide vane opening command to limit the initial opening rate of the mechanical guide vane.

[0045] S50. Execute dynamic clamping and acceleration output recovery of the tailrace vortex band. The distributed control system performs integral calculations on the volute pressure to determine the node where the water flow establishes positive acceleration. When the node where the water flow establishes positive acceleration is reached, the edge computing gateway performs low-pass filtering and envelope detection on the raw dynamic pressure signal of the tailrace, calculates the time derivative data of the low-frequency vortex band pressure pulsation, and generates a dynamic clamping coefficient based on the time derivative data of the low-frequency vortex band pressure pulsation. The dynamic clamping coefficient is used to limit the exponential recovery trajectory of the proportional gain of the main loop regulator.

[0046] S60: Output comprehensive control command and execute adjustment action. The distributed control system integrates the feedforward differential negative compensation quantity with the output quantity of the main loop regulator that has undergone exponential recovery after being limited by the dynamic clamping coefficient, and generates a comprehensive guide vane opening command. The distributed control system sends the comprehensive guide vane opening command to the electro-hydraulic servo system to drive the mechanical guide vane to perform the action. The distributed control system cyclically executes S10 to S60 according to the set control cycle to realize the primary frequency regulation adaptive control of the hydropower station governor.

[0047] In conjunction with step S10, after initializing the system, the distributed control system executes S10 to synchronously acquire multi-source heterogeneous data in order to obtain the underlying physical data of the system operation and achieve time alignment of multi-source data. This specifically includes the following sub-steps:

[0048] S11. Establish a communication link and perform clock synchronization. The distributed control system establishes a data transmission channel with the wide-area measurement system, edge computing gateway, and data acquisition system. To ensure the time reference consistency between grid-side data and unit-side physical data, the distributed control system uses a precision time protocol to send synchronization messages to the wide-area measurement system, edge computing gateway, and data acquisition system for clock synchronization. For the specific time synchronization mechanism of the precision time protocol, those skilled in the art can consult relevant communication standards and specifications; its underlying message interaction logic is well-known in the field and will not be elaborated here.

[0049] In this embodiment, the wide-area measurement system, edge computing gateway, and data acquisition system all attach a high-precision timestamp based on a time reference when generating data packets. When receiving data packets, the distributed control system sets a time tolerance window determined by the maximum transmission delay and clock jitter margin of the field communication network. Multi-source data packets falling within the same time tolerance window are then sequence-aligned, and expired or incomplete data frames are discarded. Sequence alignment and data frame discarding ensure the timing matching of cross-domain data. The communication link adopts the Industrial Ethernet network communication protocol, which specifically covers substation event mechanisms or sampled measurement value message mechanisms for general-purpose applications, to meet the millisecond-level real-time requirements of data interaction.

[0050] S12. Acquire macroscopic operational data of the power grid. The wide-area measurement system is deployed at the grid connection point substation of the hydropower station. The wide-area measurement system monitors the bus voltage and bus current at the grid connection point in real time through synchronous phasor measurement technology. For the underlying phasor extraction algorithm of synchronous phasor measurement technology, those skilled in the art can use standard methods such as discrete Fourier transform. Its phasor extraction operation is a well-known technology in this field and will not be elaborated here. Based on the principle of synchronous phasor operation, the wide-area measurement system obtains the power grid frequency change rate and active power fluctuation by performing phasor operations on the bus voltage and bus current. The power grid frequency change rate characterizes the gradient of the power grid system frequency change per unit time, and the active power fluctuation characterizes the absolute value of power fluctuation caused by load changes or power source disconnection at the grid connection point or tie line.

[0051] Because the measurement basis noise is amplified during differential operations, leading to signal distortion, the wide-area measurement system performs first-order low-pass filtering and sets a frequency deviation dead-zone threshold based on the allowable fluctuation range of the grid's daily steady-state operating frequency before outputting the grid frequency change rate. When the grid frequency fluctuation amplitude is lower than the set frequency deviation dead-zone threshold, the wide-area measurement system forces the grid frequency change rate output to zero, avoiding the transmission of irrelevant high-frequency disturbance signals to the control side. The wide-area measurement system packages the grid frequency change rate and active power fluctuation into high-speed Ethernet packets and sends them to the distributed control system.

[0052] S13. Acquire steady-state and quasi-steady-state physical data of the generator unit. The data acquisition system collects the operating parameters of the hydro-generator unit in real time through a local sensor array. The local sensor array includes a toothed disc speed probe, a pressure transmitter, and a displacement sensor. The data acquisition system uses the toothed disc speed probe to obtain the actual operating frequency of the unit and subtracts the rated frequency of the power grid to obtain the unit frequency deviation. At the same time, the data acquisition system uses the pressure transmitter to measure the inlet pressure and tailrace water level, uses the pressure transmitter located at the inlet of the turbine spiral casing to obtain the spiral casing pressure, and uses the displacement sensor on the servo to obtain the guide vane opening, which reflects the physical position of the mechanical guide vanes.

[0053] Because the dimensions and reference surfaces of pressure and elevation parameters differ, the data acquisition system performs head conversion based on the principle of fluid energy conservation to obtain the net head. The corresponding conversion formula is as follows:

[0054] ;

[0055] in, The current water head; This represents the current water inlet pressure. This represents the density constant of water, with a value of 1000 kg / m³. 3 ; This represents the acceleration due to gravity, with a value of 9.81 m / s². 2 ; The installation reference elevation represents the imported pressure transmitter. Represents the current moment; This represents the tailwater level elevation measured at the current moment. This represents the conversion of the inlet pressure of the water intake system into an equivalent pressure liquid column head height; This represents the head difference between the imported pressure transmitter and the tailwater level. The water density constant is also included. With gravitational acceleration All values ​​are fixed approximations based on conventional engineering standards and are well-known fundamental physical constants in this field, serving as fixed factors for physical quantity conversion. Through a unified mathematical conversion of physical dimensions, the physical consistency of the subsequent hydraulic calculation benchmarks is ensured.

[0056] As a preferred approach, the data acquisition system performs amplitude change rate limit verification after acquiring operational parameters. To define the boundary conditions for outlier cleaning in the digital control system, the data acquisition system performs data cleaning using the discrete difference condition formula:

[0057] when If the conditions are met, determine that the data is reasonable and output the result. ;

[0058] when If the condition is met, determine that the data is abnormal and output the result. ;

[0059] in, This represents the valid physical parameters confirmed in the output at the current moment; This represents the original physical parameters collected at the current moment; Represents the historical valid physical parameters of the previous sampling period; Represents the current moment; This represents the fixed sampling period set by the system. This represents the physical boundary threshold set based on the physical range of the field sensors and the maximum response rate of the mechanical equipment. This represents the transient jump deviation between the raw physical parameters acquired at the current moment and the historical valid physical parameters from the previous sampling period. The amplitude change rate limit verification uses historical valid data across periods to replace the current distorted data, isolating random pulse outliers caused by electromagnetic interference in complex industrial environments. The unit frequency deviation, head, volute pressure, and guide vane opening, after validity verification, together constitute the unit's steady-state physical data. The data acquisition system transmits the unit's steady-state physical data to the distributed control system.

[0060] S14. Acquire high-frequency dynamic physical data from the generator side. Based on the evolution characteristics of the water flow vortex bands inside the turbine's draft tube, the data acquisition system uses high-frequency dynamic pressure sensors deployed in the conical section of the draft tube to acquire the raw dynamic pressure signal of the draft tube. Since the raw dynamic pressure signal of the draft tube contains high-frequency noise components from fluid dynamics, the data acquisition system independently transmits the raw dynamic pressure signal of the draft tube to the edge computing gateway through a high-frequency analog input channel.

[0061] To address the aliasing distortion during the digitization of high-frequency analog signals, a hardware anti-aliasing low-pass filter is configured at the front end of the input channel of the data acquisition system. The sampling frequency of the data acquisition system is set to be more than twice the cutoff frequency of the anti-aliasing low-pass filter, based on the Nyquist sampling theorem. This allows the data acquisition system to retain the true low-frequency envelope characteristics of the water flow vortex band. For the parameter configuration of the hardware anti-aliasing low-pass filter and the sampling frequency setting based on the Nyquist sampling theorem, those skilled in the art can consult relevant signal processing manuals; the hardware anti-aliasing filter settings are well-known in the field and will not be elaborated further here. The edge computing gateway is configured with a high-frequency signal processor independent of the main CPU of the distributed control system. This high-frequency signal processor performs independent buffering and preprocessing of the high-frequency pressure signal, preventing high-frequency signal acquisition from consuming the core control computing power of the distributed control system.

[0062] See attached document Figure 3 After synchronously acquiring heterogeneous data from multiple sources, the distributed control system performs the following specific sub-steps to calculate the matching boundary of the dynamic parameters of the power grid and to resolve the matching conflict between the water hammer hysteresis effect of the hydraulic system and the dynamic inertia support capacity of the power grid:

[0063] S21. Estimate the dynamic flow inertia time. The flow inertia characteristics of the water intake system are affected by the actual operating conditions of the unit. The distributed control system obtains the net head and guide vane opening, and combines this with the turbine flow characteristic curve to obtain the water intake flow rate. The net head and guide vane opening together determine the real-time flow capacity of the turbine. For the three-dimensional interpolation calculation of the turbine flow characteristic curve, those skilled in the art can use the Lagrange interpolation method. The Lagrange interpolation algorithm is a well-known technology in this field and will not be described in detail here.

[0064] To mitigate the risk of computational overflow caused by sensor malfunctions or extreme dead water levels, the distributed control system sets a lower limit for the net head before performing calculations. When the acquired net head is lower than this lower limit, the lower limit is used instead of the actual acquired net head in the calculations. The lower limit is set based on the minimum operating head designed for the hydro-generator unit.

[0065] The distributed control system uses the water diversion flow rate to calculate the dynamic water flow inertia time. The corresponding calculation formula is:

[0066] ;

[0067] in, Represents the dynamic inertial time of the water flow at the current moment; Representing the water diversion system The physical length of a section of pipe; Representing the water diversion system The cross-sectional area of ​​the pipe section, the physical length and the cross-sectional area data are obtained from the fixed configuration database pre-entered inside the distributed control system; Represents the water diversion flow rate obtained at the current moment; Represents the current moment; Represents gravitational acceleration; The current water head; This represents the total number of pipeline segments in the water diversion system. The ratio of geometric characteristic impedance of each section of the pipeline; This represents the transient water flow rate within a single section of the water diversion system. The discrete integral quantity representing the inertia of the water flow within the entire water diversion system; The potential energy reference quantity representing the net head; the dynamic flow inertia time characterizes the time scale by which the mechanical output of the turbine lags behind the guide vane adjustment action due to the water hammer effect.

[0068] S22. Calculate the equivalent inertia of the power grid and perform pole avoidance. Since the rate of change of the power grid frequency approaches zero during steady-state operation, direct division will lead to pole calculation errors. Therefore, the distributed control system sets a dead zone for the rate of change of frequency. The dead zone is set to 0.01Hz / s to 0.05Hz / s based on the baseline noise level of the power grid's steady-state operation monitoring and control. The distributed control system compares the absolute value of the power grid's rate of change of frequency with the dead zone. When the absolute value of the power grid's rate of change of frequency is greater than the dead zone, the apparent inertia observation value is calculated.

[0069] The specific calculation logic is as follows: The distributed control system takes the absolute value of the acquired active power fluctuation as the dividend, multiplies the absolute value of the grid frequency change rate, the grid rated frequency, and constant 2 to obtain the divisor, and divides the absolute value of the active power fluctuation by the divisor obtained by the above multiplication to obtain the apparent inertia observation value.

[0070] In the calculation logic of apparent inertia observation, the product of the absolute value of the grid frequency change rate, the grid rated frequency and constant 2 represents the gradient parameter of kinetic energy change caused by the frequency change of the grid system. The ratio of the absolute value of active power fluctuation to the gradient parameter of kinetic energy change represents the ratio of the overall inertia of the grid calculated by local power fluctuation and frequency gradient.

[0071] When the absolute value of the grid frequency change rate is less than or equal to the frequency change rate dead zone, the distributed control system triggers zero-order hold logic, freezes the observation update mechanism, and forcibly sets the current apparent inertia observation value to the apparent inertia observation value of the previous sampling period. To suppress data divergence interference caused by abnormal fluctuations in active power, the distributed control system performs first-order low-pass filtering on the apparent inertia observation value to obtain the grid equivalent inertia. The corresponding calculation formula is as follows:

[0072] ;

[0073] in, The equivalent inertia of the power grid at the current moment; Represents the equivalent inertia of the power grid in the previous sampling period; Represents the current moment; This represents the fixed sampling period set by the system. This represents the filter time, which is set to 2 to 3 times the dynamic water flow inertia time reference value. The apparent inertia observation value at the current moment; The parameter representing the difference between the current apparent inertia observation and the power grid equivalent inertia of the previous sampling period; This represents the smoothing adjustment coefficient of a first-order low-pass filter; This represents the inertia filtering smoothing correction amount for the current cycle.

[0074] For the digital discretization of the first-order low-pass filtering algorithm, those skilled in the art can use difference equation transformation. The digital discretization of the first-order low-pass filtering algorithm is a well-known technique in this field and will not be elaborated further here. Distributed control systems combine zero-order hold logic and low-pass filtering to avoid poles at zeros and suppress random data fluctuations.

[0075] S23. Reconstructing the Anti-Reverse Regulation Threshold Boundary. To prevent numerical oscillations caused by an excessively small denominator in division operations, the distributed control system sets a lower limit for the dynamic water flow inertia time. This lower limit is set to 0.1 to 0.5 seconds based on the minimum physical inertia of the hydro-generator unit under no-load conditions. When the dynamic water flow inertia time falls below this lower limit, the distributed control system forcibly clamps it to the lower limit. The distributed control system divides the grid's equivalent inertia by the dynamic water flow inertia time to calculate the grid-source inertia matching coefficient. The grid-source inertia matching coefficient represents the relative ratio of the grid inertia support strength to the water flow inertia hysteresis effect.

[0076] The distributed control system calculates the anti-back-tuning threshold boundary using the network-source inertial matching coefficient. The specific calculation logic is as follows: the distributed control system multiplies the adaptive weight coefficient with the network-source inertial matching coefficient, and adds the product result to the constant 1 to obtain the dynamic correction ratio; subsequently, the distributed control system multiplies the system-set reference boundary constant with the dynamic correction ratio to obtain the anti-back-tuning threshold boundary.

[0077] In the calculation logic of the anti-reverse regulation threshold boundary, the product of the adaptive weight coefficient and the grid-source inertia matching coefficient represents the boundary dynamic adjustment amount obtained based on the grid-source parameter matching status; the benchmark boundary constant and the adaptive weight coefficient are established based on the historical load shedding test data of the hydro-generator unit. The benchmark boundary constant ranges from 0.5 to 1.5, and the adaptive weight coefficient ranges from 0.1 to 0.5.

[0078] The distributed control system uses the anti-reverse regulation threshold boundary as the criterion for determining whether to implement asymmetric water hammer feedforward suppression. When the equivalent inertia of the power grid is low and the dynamic water flow inertia time is large, resulting in a decrease in the grid-source inertia matching coefficient, the anti-reverse regulation threshold boundary shrinks synchronously based on the dynamic correction ratio.

[0079] In conjunction with step S30, after synchronously acquiring multi-source heterogeneous data and calculating the matching boundary of grid-source dynamic parameters, the distributed control system performs the following specific sub-steps in order to accurately identify the dynamic evolution trend of grid disturbances and solve the phase trajectory distortion problem caused by asynchronous sampling of heterogeneous data:

[0080] S31. Perform data dimension normalization and vector feature extraction. To convert parameters with different physical attributes to a unified scale for geometric calculations, the distributed control system first extracts the hardware timestamps of the unit frequency deviation and the grid frequency change rate, determining data with a timestamp deviation within 1ms as valid synchronization data of the same period. Since the physical dimension of the unit frequency deviation is Hz and the physical dimension of the grid frequency change rate is Hz / s, their physical dimensions are inconsistent, making direct vector synthesis impossible. Therefore, the distributed control system sets frequency difference normalization coefficients and frequency derivative normalization coefficients in its internal memory.

[0081] As a preferred approach, to avoid geometric distortion in the two-dimensional phase plane, the per-unit values ​​of frequency deviation and frequency change rate need to be mapped to similar numerical ranges. The distributed control system, based on the unit's primary frequency regulation dead zone parameters and the grid standard disturbance rejection test curve, inverts and sets a frequency difference normalization coefficient ranging from 10 to 50, and a frequency derivative normalization coefficient ranging from 100 to 500.

[0082] Subsequently, the distributed control system multiplies the unit frequency deviation by the frequency difference normalization coefficient to obtain the normalized frequency difference, and multiplies the grid frequency change rate by the frequency derivative normalization coefficient to obtain the normalized frequency change rate. After obtaining the normalized frequency difference and the normalized frequency change rate, the distributed control system constructs a phase trajectory plane with the normalized frequency difference as the horizontal axis and the normalized frequency change rate as the vertical axis, and then extracts the phase trajectory vector containing two-dimensional features.

[0083] S32. Calculate the real-time magnitude and phase angle of the phase trajectory vector. The distributed control system calculates the real-time magnitude based on the extracted two-dimensional phase trajectory vector. The corresponding calculation formula is:

[0084] ;

[0085] in, The real-time module length represents the current moment; Represents the frequency difference normalization coefficient; This represents the unit frequency deviation acquired at the current moment; Represents the current moment; Represents the normalized coefficient of the frequency derivative; Represents the rate of change of the power grid frequency at the current moment; Represents the normalized frequency difference; Represents the normalized rate of change; The term representing the square of the normalized frequency difference; The term representing the square of the normalized rate of change of frequency; The energy distance element representing the phase trajectory vector in a two-dimensional plane; The Euclidean distance represents the orthogonal synthesis of the normalized frequency difference and the normalized rate of frequency change. The real-time modulus characterizes the overall disturbance degree of the power grid deviating from its steady-state operating point due to external shocks.

[0086] The distributed control system employs a four-quadrant arctangent algorithm with zero-pole protection logic to calculate the phase angle of the phase trajectory vector, thus avoiding the pole collapse problem in division operations caused by the normalized frequency difference approaching zero. The specific calculation logic is as follows: the distributed control system sets a zero-pole protection constant, which is set to 10 based on the lower limit of the floating-point precision of the underlying control unit. -6 When the absolute value of the normalized frequency difference is less than the zero-pole constant, the distributed control system uses the zero-pole constant with the same sign to replace the actual normalized frequency difference as the denominator in the division operation.

[0087] The distributed control system obtains the tangent ratio by dividing the normalized rate of frequency change by the replaced anti-zero pole constant, and then uses the arctangent function for angle mapping. Combining the positive and negative signs of the normalized frequency difference and the normalized rate of frequency change, the distributed control system determines the phase angle value within the range of 0 to 360 degrees. The phase angle intuitively represents the dynamic trend direction of the frequency evolution of the power grid system. The underlying numerical calculation logic of the arctangent function can be implemented by a coordinate rotation digital computer algorithm by those skilled in the art; its angle approximation iterative calculation is a well-known technique in the field and will not be elaborated upon here.

[0088] S33. Establish operating condition judgment criteria. In this embodiment, the distributed control system classifies and judges the grid disturbance conditions based on the quadrant position of the phase angle in the two-dimensional phase trajectory plane. When the unit frequency deviation is negative and the grid frequency change rate is negative, the phase angle is in the third quadrant. The phase angle being in the third quadrant indicates that there is a load deficit in the grid active power, and the system frequency is not only lower than the rated operating frequency but is also in a state of accelerating decline. The distributed control system defines the phase angle being in the third quadrant as the frequency acceleration drop condition.

[0089] The distributed control system reads the anti-reverse regulation threshold boundary, which is obtained by inverting the maximum envelope of the phase trajectory from the unit's historical load shedding tests, with a value range set from 2.0 to 5.0. The distributed control system compares the real-time magnitude with the anti-reverse regulation threshold boundary. The anti-reverse regulation threshold boundary serves as a safety warning line to distinguish between normal load fluctuations and large-scale grid faults. When the real-time magnitude exceeds the anti-reverse regulation threshold boundary, it indicates that the power surge currently experienced by the grid exceeds the physical response capability of the hydro-generator unit under normal control logic. The distributed control system defines a real-time magnitude exceeding the anti-reverse regulation threshold boundary as an anti-reverse regulation over-limit condition.

[0090] To avoid false triggering caused by relying solely on a single numerical limit exceeding the limit, the distributed control system performs a logical AND operation on the frequency acceleration drop condition and the anti-reverse adjustment limit exceeding condition, and uses the result of the logical AND operation as the control switch command to trigger the asymmetric water hammer feedforward suppression.

[0091] See attached document Figure 4 In conjunction with step S40, after establishing the operating condition judgment benchmark, the distributed control system performs the following specific sub-steps to limit the reverse drop in mechanical output during the non-minimum phase caused by water hammer effect and to solve the problem of sudden drop in water flow potential energy in the initial stage of unit response:

[0092] S41. Block the proportional gain of the main loop regulator. To freeze the control state of the continuous system and achieve a smooth transition, in this embodiment, the main loop regulator adopts a closed-loop control architecture including proportional and integral elements. The distributed control system receives the generated control switch command that triggers asymmetric water hammer feedforward suppression. When it detects that the frequency acceleration drop condition and the anti-reverse regulation over-limit condition are simultaneously met, the distributed control system blocks the fast adjustment channel of the main loop regulator based on the unit frequency deviation, freezes the response output of the proportional control element, and records the moment when the frequency acceleration drop condition and the anti-reverse regulation over-limit condition are simultaneously met as the trigger moment.

[0093] As a preferred approach, to prevent hydraulic shocks and mechanical oscillations in the electro-hydraulic servo system caused by a sudden step change in the proportional gain, the distributed control system utilizes an exponential decay algorithm to smoothly reduce the proportional gain of the main loop regulator. Before executing the exponential decay algorithm, the system pre-sets a proportional decay constant. This constant is obtained through a step response test of the guide vanes of the hydro-generator unit and is set to 0.1 to 0.5 seconds based on the physical execution dead time of the electro-hydraulic servo system. Before performing the division operation, the distributed control system determines whether the proportional decay constant is greater than the minimum precision threshold of the system's underlying floating-point arithmetic, thus avoiding the risk of computational overflow when the denominator approaches zero during the division operation.

[0094] The distributed control system divides the time difference between the current moment and the trigger moment by the proportional decay constant, and takes the negative value of the calculation result as the exponent of the natural logarithm to obtain the dynamic decay factor. The proportional control output at the moment of freeze is obtained as the initial proportional gain, and this initial proportional gain is multiplied by the dynamic decay factor to obtain the real-time proportional gain after the blockage. Through the exponential decay algorithm, the proportional gain of the main loop regulator approaches zero within a millisecond period. For the discretized programming implementation of the exponential decay algorithm, those skilled in the art can use the difference equation transformation method. The discretization calculation of the difference equation is a well-known technique in this field and will not be elaborated further here.

[0095] S42. Generate feedforward negative compensation and superimpose guide vane commands. Due to the inertial hysteresis effect of the water column in the water diversion system, even after blocking the proportional gain of the main loop regulator, the integral link inside the main loop regulator will still accumulate mechanical guide vane opening commands. In order to actively suppress the mechanical guide vane opening trend caused by the integral link, the distributed control system generates feedforward negative compensation based on the grid frequency change rate. The grid frequency change rate can directly reflect the deterioration gradient of the active power deficit of the external grid, and has a dynamic forward prediction attribute compared with the unit frequency deviation. Combined with the previously calculated grid-source inertial matching coefficient, the feedforward negative compensation is calculated, and the corresponding calculation formula is:

[0096] ;

[0097] in, This represents the current feedforward negative compensation amount; The feedforward differential gain constant, representing the system setting, is tuned by inversion of the peak time of non-minimum phase back-adjustment during historical load shedding tests of the turbine, with a value range of 1.5 to 3.0. Represents the network source inertia matching coefficient acquired at the current moment; This represents the absolute value of the rate of change of the power grid frequency at the current moment; Represents the current moment; Represents the dynamic feedforward gain established based on the network source parameter matching state; This represents the product of the dynamic feedforward gain and the absolute value of the rate of change of the grid frequency. This represents the reverse guide vane suppression component generated by combining the grid frequency sag rate. Because the calculation logic contains a negative sign and the absolute value of the grid frequency change rate is always positive, the feedforward negative compensation is always negative.

[0098] The distributed control system superimposes the feedforward negative compensation onto the guide vane opening command. This negative compensation offsets the opening increment output of the main loop regulator's integral stage, thus limiting the initial opening rate of the mechanical guide vanes. By limiting the initial opening rate of the mechanical guide vanes, the abrupt change in water flow acceleration within the diversion system is mitigated, reducing the pressure drop in the volute casing, ensuring a smooth transition of turbine mechanical output, and avoiding water hammer and backflow. The output of the feedforward negative compensation is maintained until the system detects a state transition in the hydrodynamic state that crosses an inflection point.

[0099] See attached document Figure 5 In conjunction with step S50, during the period of maintaining the feedforward negative compensation output state, the distributed control system performs the following specific sub-steps to accurately determine the physical transformation of the hydrodynamic state and accelerate the recovery of the turbine's mechanical output while ensuring the hydraulic stability of the tailrace:

[0100] S51. Determine the node where the water flow establishes positive acceleration. To quantify the evolution of fluid momentum during the water hammer elimination process of the water diversion system, in this embodiment, the distributed control system acquires the volute pressure and subtracts it from the preset rated steady-state volute pressure to obtain the transient pressure deviation. Subsequently, a time integration operation is performed on the transient pressure deviation to obtain the cumulative pressure deviation. By calculating the cumulative pressure deviation, the system can quantify the evolution of the total momentum of the water flow inside the water diversion pipeline.

[0101] Considering that transient hydraulic noise from sensors can easily lead to zero-crossing detection failure, the distributed control system introduces a dual confirmation mechanism of time and amplitude. The distributed control system sets an integral dead zone threshold based on the amplitude of the turbine's normal no-load water pressure pulsation, ranging from 0.01 MPa to 0.05 MPa, and a duration threshold based on the half-wave time of the water acoustic propagation period in the intake pipe, ranging from 0.1 seconds to 0.3 seconds. When the cumulative pressure deviation changes from negative to positive and its positive value is greater than the integral dead zone threshold, and the duration of this state exceeds the duration threshold, the distributed control system determines that the water flow within the intake system has crossed the water hammer hysteresis interval and established positive acceleration. Reaching the point where the water flow establishes positive acceleration indicates that the hydraulic system possesses the physical conditions to increase the turbine's mechanical output. The distributed control system then cancels the feedforward negative compensation and initiates the recovery procedure for the main controller's proportional gain.

[0102] S52. Extracting the derivative of vortex band pressure pulsation. As a preferred method, during the acceleration and recovery of power output of the hydro-generator unit, changes in the guide vane opening state easily excite eccentric vortex bands inside the draft tube, inducing low-frequency hydraulic vibrations. The physical reason for choosing the draft tube dynamic pressure signal as the monitoring parameter is that the rotational sweep of the vortex band cavity directly maps to periodic pressure changes on the inner wall of the cone section. To monitor the evolution trend of the vortex band, the edge computing gateway acquires the raw dynamic pressure signal of the draft tube.

[0103] The edge computing gateway utilizes a low-pass filter to remove high-frequency fluid noise interference from the raw dynamic pressure signal of the tailrace pipe, retaining low-frequency characteristic components. Then, it performs envelope detection processing on these low-frequency characteristic components to obtain the amplitude envelope signal reflecting the intensity of pressure pulsations. For the specific algorithm of envelope detection processing, those skilled in the art can use the Hilbert transform method; the signal envelope extraction operation of the Hilbert transform is a well-known technique in the field and will not be elaborated upon here.

[0104] After acquiring the amplitude envelope signal, the edge computing gateway subtracts the amplitude envelope signal of the previous sampling period from the current sampling period's amplitude envelope signal, and divides the difference by a set fixed sampling period to obtain the vortex pressure pulsation derivative. To avoid pole anomalies caused by a zero denominator during the difference operation, the fixed sampling period is constantly set to a real number between 5 and 20 milliseconds, based on the main loop cycle of the control system. The vortex pressure pulsation derivative characterizes the divergence or convergence rate of the vortex energy inside the tailrace pipe. The edge computing gateway packages and sends the vortex pressure pulsation derivative to the distributed control system.

[0105] S53. Generate dynamic clamping coefficients and execute accelerated output recovery. To dynamically adjust the closed-loop response agility based on the real-time oscillation intensity of the tailrace pipe, the distributed control system receives the derivative of the vortex belt pressure pulsation, calculates the dynamic clamping coefficients, and limits the exponential recovery trajectory of the main controller's proportional gain. The corresponding calculation formula is:

[0106] ;

[0107] in, This represents the proportional gain of the master regulator being restored at the current moment; This represents the baseline proportional gain, which is set to 2.0 to 5.0 based on the optimization results of the PID parameters under rated load. The base constant representing the natural logarithm; Represents the current moment; This represents the trigger moment when the water flow establishes a positive acceleration node; The proportional gain recovery time is set to 1.0 to 3.0 seconds based on the maximum response rate of the relay. The sensitivity to vortex clamping is set to 0.5 to 1.5 based on the allowable pressure pulsation critical amplitude in the tailrace tube. The derivative of the vortex pressure pulsation at the current moment;

[0108] This represents the time difference between the current moment and the trigger moment; Represents the relative ratio of the time difference to the recovery time constant; The negative exponent term representing the ratio of time differences; Represents an exponential transition factor that decays over time; The underlying exponential recovery trajectory parameter; This represents the target proportional gain under unrestricted conditions. This represents the product of the sensitivity constant and the vortex zone derivative. This represents the negative product term used to suppress gain recovery; This represents the dynamic clamping coefficient generated based on the deterioration trend of the vortex zone.

[0109] When the derivative of the vortex band pressure pulsation is positive, it indicates that the water pressure pulsation inside the draft tube is showing a divergent and worsening trend. At this time, the negative product term in the dynamic clamping coefficient calculation process decreases, causing the dynamic clamping coefficient to contract accordingly, and the system forcibly suppresses the recovery rate of the main controller's proportional gain. When the vortex band evolution trend is gradual or the derivative of the vortex band pressure pulsation is negative, the dynamic clamping coefficient approaches a constant of 1 or increases slightly. The distributed control system uses the dynamic clamping coefficient multiplied by the target proportional gain to obtain the current recovered main controller proportional gain. The controller increases the gain according to the set exponential recovery trajectory, accelerating the recovery of turbine output while suppressing draft tube hydraulic vibration.

[0110] In conjunction with step S50, after generating the dynamic clamping coefficient and performing acceleration output recovery, the distributed control system executes S60—in order to construct a physical drive model characterizing the mechanical output of the turbine and solve the dynamic response matching problem between the digital control signal and the hydraulic actuator—specifically including the following sub-steps:

[0111] S61. Synthesize and limit the target opening command. The distributed control system integrates the regulation components of each control stage. In this embodiment, during the anti-reverse regulation control stage, the distributed control system algebraically superimposes the conventional control command output by the main loop regulator with the feedforward negative compensation; during the acceleration output recovery stage, the distributed control system calculates the regulation command using the proportional gain of the main regulator recovered at the current moment. To prevent the step change of the integrated control command from causing mechanical shock to the water diversion system and mechanical structure, the distributed control system performs amplitude limiting and rate of change limiting processing on the synthesized integrated command.

[0112] The distributed control system reads the internally set upper and lower limits of the turbine opening, constraining the amplitude of the comprehensive command between these limits. The upper and lower limits are set as a 100% fully open position and a 0% to 5% no-load opening range, respectively, based on the turbine's maximum flow capacity and the unit's no-load operation requirements. Simultaneously, the distributed control system reads the maximum response speed parameter of the relay. The distributed control system subtracts the comprehensive command of the current cycle from the comprehensive command of the previous cycle to obtain the command change, and then divides the command change by the system's fixed sampling period to obtain the command change rate.

[0113] Before performing the division operation, the distributed control system checks whether the fixed sampling period is greater than the precision of the underlying system clock to avoid division pole anomalies caused by the denominator approaching zero. The fixed sampling period is set to 5ms to 20ms based on the task scheduling cycle of the control system. The distributed control system compares the calculated absolute value of the command change rate with the maximum response speed parameter of the relay. If the absolute value of the command change rate exceeds the limit, the command change is forcibly clamped to the product of the maximum response speed parameter of the relay and the fixed sampling period. The maximum response speed parameter of the relay is obtained by inverting the fastest action time specified on the speed controller nameplate, and its value ranges from 0.1 / s to 0.2 / s. After amplitude limiting and rate of change limiting processing, the distributed control system outputs the target opening command.

[0114] S62. Perform electro-hydraulic servo position closed-loop tracking. As a preferred method, the distributed control system converts the target opening command into an analog control voltage signal via a high-speed digital-to-analog converter channel and sends it to the electro-hydraulic servo system. During the digital-to-analog conversion, the distributed control system uses industrial standard electrical signals of 4mA to 20mA or -10V to +10V to proportionally map the 0% to 100% digital opening command into a continuous physical electrical signal. The electro-hydraulic servo system is equipped with a servo amplifier, an electro-hydraulic servo valve, a main pressure regulating valve, and a relay.

[0115] The servo amplifier inside the electro-hydraulic servo system receives the target opening command and calculates the difference between it and the actual guide vane opening fed back in real time by the servo displacement sensor, obtaining the opening position deviation signal. The servo amplifier amplifies the opening position deviation signal and outputs a drive current. The drive current output by the servo amplifier drives the baffle inside the electro-hydraulic servo valve to shift, changing the hydraulic balance of the control oil circuit, and thus pushing the valve core of the main pressure regulating valve to move. The movement of the main pressure regulating valve core controls the connection status of the high-pressure oil circuit and the return oil circuit, guiding high-pressure oil into the opening or closing chamber of the servo, pushing the servo piston to undergo physical displacement, thereby changing the actual mechanical opening of the turbine guide vanes.

[0116] For the position closed-loop control algorithm at the bottom layer of the electro-hydraulic servo system, those skilled in the art can use the standard incremental proportional-integral-derivative control algorithm. Its position closed-loop tracking calculation and hydraulic amplification logic are well-known technologies in the field and will not be described in detail here.

[0117] S63. Operation Status Monitoring and Control Logic Reset. During the execution of hydraulic mechanical actions by the electro-hydraulic servo system, the distributed control system continuously monitors the frequency recovery status on the grid side and the hydraulic stability status on the unit side. The distributed control system compares the normalized frequency difference, normalized frequency change rate, and vortex pressure pulsation derivative with their respective steady-state dead zone thresholds in real time.

[0118] To avoid erroneous control state cutoff caused by single-point extrema or transient electromagnetic noise, the distributed control system identifies steady-state characteristics based on multi-dimensional time-domain confirmation. Internally, the system sets the normalized frequency difference dead zone threshold to 0.05% to 0.1% of the unit's rated frequency, and the normalized frequency change rate dead zone threshold to 0.01 Hz / s to 0.02 Hz / s. Simultaneously, the reset delay time window is set to 1.0 s to 3.0 s.

[0119] When the normalized frequency difference and the normalized frequency change rate both fall back and stabilize within the aforementioned steady-state dead zone threshold range, and the vortex belt pressure pulsation derivative is in a convergent state and the cumulative pressure deviation remains in the positive range, and the duration of the normalized frequency difference, normalized frequency change rate, vortex belt pressure pulsation derivative, and cumulative pressure deviation meeting the conditions exceeds the reversion delay time window, it indicates that the local power grid active power deficit has been filled, the trend of grid frequency drop has been curbed, and the internal hydrodynamic state of the turbine has re-established a new equilibrium.

[0120] The distributed control system then determines that the system's disturbance rejection control process has ended. The distributed control system smoothly exits the tailrace vortex dynamic clamping logic and the asymmetric water hammer feedforward suppression logic, and reverts to the conventional frequency closed-loop proportional-integral control mode, waiting to respond to the next power grid system frequency fluctuation.

[0121] See attached document Figure 6 and attached Figure 7 In a specific application scenario, a hydroelectric power station's turbine generator unit is operating at its rated grid connection. A large-capacity load trip fault occurs in the receiving-end power grid, resulting in a significant active power deficit and an accelerated frequency drop in the power grid system. The distributed control system is constructed using a dual-redundant industrial control computer, with the edge computing gateway employing a digital signal processing chip with a floating-point unit.

[0122] The distributed control system establishes communication links with the wide-area measurement system, edge computing gateway, and data acquisition system. The communication network adopts a substation event message mechanism oriented towards general objects. The distributed control system uses a precise time protocol to send synchronization messages to each measurement node to perform clock synchronization. The wide-area measurement system extracts the bus voltage and current phasors through discrete Fourier transform and calculates the grid frequency change rate and active power fluctuation. For the underlying phasor extraction algorithm of the synchronization phasor measurement technology, those skilled in the art can use standard methods such as discrete Fourier transform. The phasor extraction operation is a well-known technology in this field and will not be elaborated here.

[0123] The data acquisition system uses an absolute displacement sensor to obtain the guide vane opening and a silicon piezoresistive transmitter to obtain the inlet pressure and volute pressure. Based on the formula... Calculate the net head of water. Among them, The current water head; This represents the current water inlet pressure. Represents the density constant of water; Represents the current moment; Represents gravitational acceleration; The installation reference elevation represents the imported pressure transmitter. This represents the tailrace water level elevation. The data acquisition system performs amplitude change rate limit verification of the filter pulse threshold and synchronously sends data such as unit frequency deviation, clean water head, and spiral casing pressure to the distributed control system.

[0124] The distributed control system combines the acquired head and guide vane opening to perform three-dimensional interpolation calculations to obtain the water flow rate, based on the formula... Estimate the dynamic flow inertia time. Among them, and Representing the water diversion system The physical length and cross-sectional area of ​​a pipe section; This represents the water diversion flow rate obtained at the previous moment; This represents the total number of pipe segments in the water diversion system. In this embodiment, the calculated dynamic water flow inertia time is 2.4 seconds.

[0125] The distributed control system calculates the apparent inertia observation and obtains the grid equivalent inertia after first-order low-pass filtering. The system divides the grid equivalent inertia by the dynamic water flow inertia time to obtain the grid-source inertia matching coefficient. The distributed control system multiplies the adaptive weighting coefficient by the grid-source inertia matching coefficient and adds it to a constant 1 to obtain the dynamic correction factor. This factor is then multiplied by the reference boundary constant to obtain the anti-reverse regulation threshold boundary, which is calculated to be 3.2.

[0126] The distributed control system performs dimensional normalization on the unit frequency deviation and the grid frequency change rate, with the frequency deviation normalization coefficient set to 20 and the frequency derivative normalization coefficient set to 200. The system calculates the real-time modulus by the orthogonal Euclidean distance between the normalized frequency deviation and the normalized frequency change rate, achieving a modulus of 3.8 under this operating condition. The system uses a four-quadrant arctangent algorithm with zero-pole protection logic to calculate the phase angle. At this point, both the unit frequency deviation and the grid frequency change rate are negative, and the phase angle falls into the third quadrant.

[0127] The phase angle is in the third quadrant, satisfying the frequency acceleration drop condition. Simultaneously, the real-time modulus length of 3.8 is greater than the anti-reverse modulation threshold boundary of 3.2, satisfying the anti-reverse modulation over-limit condition. The distributed control system triggers the asymmetric water hammer feedforward suppression control switch command. The distributed control system uses an exponential decay algorithm to smoothly reduce the proportional gain of the main loop regulator from the initial 3.0 to near zero. The distributed control system follows the formula... Generate the feedforward negative compensation amount. Among them, Represents the feedforward differential gain constant; Represents the source inertia matching coefficient; This represents the absolute value of the rate of change of the power grid frequency. The generated feedforward negative compensation is added to the guide vane opening command to actively offset the opening increment caused by the integral stage and limit the initial opening rate of the mechanical guide vane.

[0128] The distributed control system performs time integration on the acquired volute pressure. When the cumulative pressure deviation changes from negative to positive and the duration exceeds a threshold of 0.2 seconds, it is determined that the water flow inside the water intake system has crossed the water hammer hysteresis interval and established positive acceleration. The edge computing gateway performs low-pass filtering and envelope detection processing on the raw dynamic pressure signal of the tailrace pipe, and obtains the discrete differential derivative of the amplitude envelope signal as the vortex pressure pulsation derivative.

[0129] Distributed control system based on formula Generate the current moment's recovered main regulator proportional gain and execute accelerated output recovery. Wherein, Represents the baseline proportional gain; This represents the trigger moment when the water flow establishes a positive acceleration node; Represents the proportional gain recovery time; Represents the sensitivity of the vortex clamp; This represents the derivative of the vortex pressure pulsation. The system utilizes the dynamic clamping coefficient within this derivative to limit the exponential recovery trajectory of the main controller's proportional gain, thereby suppressing the deterioration of low-frequency vibrations in the tailrace pipe.

[0130] The distributed control system superimposes the output of the main controller with the feedforward negative compensation, performs amplitude and rate-of-change limiting processing, and generates a target opening command. The command is output to the servo amplifier of the electro-hydraulic servo system via a digital-to-analog converter. The servo amplifier drives the electro-hydraulic servo valve and the main pressure regulating valve, guiding high-pressure oil into the servo actuator to actuate the guide vanes. During the electro-hydraulic servo closed-loop execution, if the normalized frequency difference and normalized frequency change rate fall back to the steady-state dead zone and the duration exceeds the recovery delay time window, the system smoothly exits the dynamic clamping and asymmetric water hammer feedforward suppression logic, resuming the normal closed-loop mode.

[0131] See attached document Figure 6 This figure shows a comparison between the unit's frequency deviation and the guide vane opening response. The horizontal axis of the attached figure represents time (s), the left vertical axis represents the unit's frequency deviation (Hz), and the right vertical axis represents the guide vane opening command (pu). Figure 6 The legend includes unit frequency deviation, guide vane opening (traditional PID), and guide vane opening (the method of this invention).

[0132] See attached document Figure 7This figure shows the relative deviation between the proportional gain of the main controller and the volute pressure. The horizontal axis of the figure represents time (s), the left vertical axis represents the proportional gain of the main controller, and the right vertical axis represents the relative deviation of the volute pressure. Figure 7 The graph includes the main controller proportional gain and the relative deviation of the volute pressure.

[0133] Based on the control scheme of this invention and the comparison with the data in the above figures, the following conclusions can be drawn: In the initial stage (0 to 3 seconds) of the frequency acceleration drop caused by a severe active power deficit in the power grid, combined with... Figure 6 As can be seen, the unit frequency deviation increases rapidly. Traditional PID control methods rapidly increase the guide vane opening command during this stage, leading to severe water hammer and adverse adjustment phenomena. The sudden drop in water flow potential energy worsens the initial support capacity of the mechanical output. Correspondingly, the system of this invention, upon identifying that the phase trajectory magnitude exceeds the limit and the phase angle is in the third quadrant, quickly blocks the proportional gain and applies a feedforward negative compensation, maintaining the guide vane opening at an extremely low rate in the initial stage, thus avoiding the non-minimum phase reverse drop effect.

[0134] according to Figure 7 Data indicates that the proportional gain of this invention decays exponentially to near zero after a disturbance. Accompanying the time integral of the volute pressure deviation to determine the positive acceleration of the water flow (around the 4th second), the system initiates an exponential recovery procedure. Because the vortex pressure pulsation derivative extracted by the edge computing gateway is used as a clamping factor in the control law calculation, the recovery trajectory of the proportional gain is subject to smoothing constraints, preventing the rapid opening of the guide vanes from triggering low-frequency hydraulic oscillations in the tailrace. Overall, this embodiment suppresses the negative effects of water hammer while ensuring hydraulic dynamic stability, making the frequency modulation response process more consistent with the underlying physical evolution.

Claims

1. A method for adaptive frequency regulation of a primary frequency regulator in a hydropower station based on grid-source coordination, characterized in that, Includes the following steps: Acquire power grid frequency change rate, active power fluctuation, and underlying physical data; The dynamic water flow inertia time is calculated based on the underlying physical data, and the equivalent inertia of the power grid is calculated by combining the active power fluctuation and the power grid frequency change rate. The anti-reverse regulation threshold boundary is reconstructed using the equivalent inertia of the power grid and the dynamic water flow inertia time. Calculate the real-time magnitude and phase angle of the phase trajectory vector of the generator frequency deviation and the grid frequency change rate in the underlying physical data; When the phase angle and the real-time modulus satisfy the frequency acceleration drop condition and the anti-reverse regulation threshold boundary, a feedforward differential negative compensation amount is generated based on the power grid frequency change rate. When the node for establishing positive acceleration of the water flow is determined based on the underlying physical data, the time derivative data of the low-frequency vortex pressure pulsation is obtained based on the underlying physical data, a dynamic clamping coefficient is generated, and the exponential recovery trajectory of the proportional gain of the main loop regulator is restricted. The feedforward differential negative compensation quantity is combined with the limited output quantity of the main circuit regulator to generate a comprehensive guide vane opening command and issue it for execution, thereby completing the primary frequency regulation adaptive control of the hydropower station governor, and then returning to execute the steps of obtaining the grid frequency change rate, active power fluctuation quantity and underlying physical data to achieve a loop.

2. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 1, characterized in that, The underlying physical data includes the original signals of unit frequency deviation, water head, guide vane opening, volute pressure, and tailrace pipe dynamic pressure. When acquiring power grid frequency change rate, active power fluctuation, and underlying physical data: The distributed control system establishes a data transmission channel with the wide-area measurement system, edge computing gateway, and data acquisition system; The wide-area measurement system monitors the bus voltage and bus current at the grid connection point in real time, and performs phasor calculations on the bus voltage and bus current to obtain the grid frequency change rate and the active power fluctuation. The data acquisition system obtains the actual operating frequency of the unit and subtracts the actual operating frequency of the unit from the rated frequency of the power grid to obtain the frequency deviation of the unit. It measures the inlet pressure and tailwater level elevation, obtains the volute pressure and the guide vane opening that reflects the physical position of the mechanical guide vane, calculates the net water head using the inlet pressure, and obtains the raw dynamic pressure signal of the tailwater pipe using a high-frequency dynamic pressure sensor.

3. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 2, characterized in that, The specific steps for calculating the dynamic water flow inertia time based on the underlying physical data include: The distributed control system uses the head and guide vane opening, combined with the turbine flow characteristic curve, to obtain the water intake flow rate. The ratio of the physical length to the cross-sectional area of ​​the water intake system pipeline is used as the geometric characteristic impedance ratio of each pipeline segment. The geometric characteristic impedance ratio of each pipeline segment is multiplied by the water intake flow rate to obtain the transient water flow rate in a single pipeline segment of the water intake system. The transient water flow rate in a single pipeline segment of the water intake system is accumulated to obtain the discrete integral of the water flow inertia inside the water intake system. The potential energy reference value generated by the purified water head is obtained by multiplying the purified water head by the constant gravitational acceleration. The discrete integral value of the water flow inertia inside the water diversion system is divided by the potential energy reference value generated by the purified water head to obtain the dynamic water flow inertia time.

4. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 2, characterized in that, The steps for calculating the equivalent inertia of the power grid by combining the active power fluctuation and the power grid frequency change rate specifically include: The distributed control system sets a frequency change rate dead zone, obtains the absolute value of the power grid frequency change rate by acquiring the absolute value of the power grid frequency change rate, and compares the absolute value of the power grid frequency change rate with the frequency change rate dead zone. When the absolute value of the power grid frequency change rate is greater than the dead zone of the frequency change rate, the absolute value of the active power fluctuation is obtained by taking the absolute value of the active power fluctuation. The absolute value of the power grid frequency change rate, the rated frequency of the power grid, and constant 2 are multiplied together to obtain the divisor. The absolute value of the active power fluctuation is divided by the divisor obtained by multiplication to obtain the apparent inertia observation value. The apparent inertia observation value is then subjected to first-order low-pass filtering to obtain the equivalent inertia of the power grid.

5. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 2, characterized in that, The step of reconstructing the anti-reverse regulation threshold boundary using the equivalent inertia of the power grid and the dynamic water flow inertial time specifically includes: The distributed control system sets a lower limit for the inertial time of the dynamic water flow. When the inertial time of the dynamic water flow is lower than the lower limit, the inertial time of the dynamic water flow is forcibly clamped to the lower limit. The grid-source inertia matching coefficient is obtained by dividing the equivalent inertia of the power grid by the dynamic water flow inertia time. The adaptive weighting coefficient is multiplied by the network source inertia matching coefficient, and the product is added to a constant 1 to obtain the dynamic correction ratio. The reference boundary constant is multiplied by the dynamic correction ratio to obtain the anti-reverse modulation threshold boundary.

6. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 2, characterized in that, The steps for calculating the real-time magnitude and phase angle of the phase trajectory vector of the generator frequency deviation and the grid frequency change rate in the underlying physical data specifically include: The distributed control system multiplies the unit frequency deviation by the frequency difference normalization coefficient to obtain the normalized frequency difference, and multiplies the grid frequency change rate by the frequency derivative normalization coefficient to obtain the normalized frequency change rate. The square term of the normalized frequency difference is added to the square term of the normalized frequency change rate, and the square root of the sum is taken to obtain the real-time modulus. Set a zero-pole constant, and use the zero-pole constant or the actual normalized frequency difference as the denominator of the division operation. Divide the normalized frequency change rate by the denominator to obtain the tangent ratio. Use the arctangent function to perform angle mapping on the tangent ratio to obtain the phase angle value in the range of 0 to 360 degrees as the phase angle.

7. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 2, characterized in that, Determining whether the frequency acceleration drop condition and the anti-reverse modulation threshold boundary are met includes: When the frequency deviation of the generator unit is negative and the rate of change of the grid frequency is negative, the distributed control system determines that the frequency acceleration drop condition is met. The real-time modulus is compared with the anti-reverse modulation threshold boundary. When the real-time modulus is greater than the anti-reverse modulation threshold boundary, it is determined that the anti-reverse modulation over-limit condition is met. Perform a logical AND operation between the frequency acceleration drop condition and the anti-reverse modulation limit condition, and confirm the control switch command to trigger asymmetric water hammer feedforward suppression based on the result of the logical AND operation.

8. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 5, characterized in that, Before generating the feedforward differential negative compensation amount, the distributed control system performs the step of blocking the proportional gain of the main loop regulator. The step of generating the feedforward differential negative compensation amount based on the grid frequency change rate specifically includes: The distributed control system records the trigger time, uses an exponential decay algorithm to control the proportional gain of the main loop regulator to decrease smoothly, subtracts the trigger time from the current time to obtain the time difference, divides the time difference by the proportional decay constant to obtain the calculation result, takes the negative value of the calculation result as the exponent of the natural logarithm to obtain the dynamic decay factor, and obtains the proportional control output at the moment of freezing as the initial proportional gain. Multiplies the initial proportional gain by the dynamic decay factor to obtain the real-time proportional gain after blocking. A dynamic feedforward gain is established based on the set feedforward differential gain constant and the network source inertia matching coefficient; The dynamic feedforward gain is multiplied by the absolute value of the grid frequency change rate, and the negative value of the calculation result is taken as the negative feedforward differential compensation amount.

9. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination as described in claim 2, characterized in that, The steps of determining the node where the water flow establishes positive acceleration based on the underlying physical data and generating dynamic clamping coefficients by acquiring the time derivative data of low-frequency vortex pressure pulsations specifically include: The distributed control system subtracts the volute pressure from the preset rated steady-state volute pressure to obtain the transient pressure deviation, and performs time integration on the transient pressure deviation to obtain the cumulative pressure deviation. Set an integral dead zone threshold and a duration threshold. When the cumulative pressure deviation changes from negative to positive and the positive value is greater than the integral dead zone threshold, and the state in which the cumulative pressure deviation is greater than the integral dead zone threshold is maintained for a longer time than the duration threshold, it is determined that the node of the water flow establishing positive acceleration has been reached. The edge computing gateway performs low-pass filtering and envelope detection on the raw dynamic pressure signal of the tailrace pipe to obtain the amplitude envelope signal reflecting the intensity of the pressure pulsation. The amplitude envelope signal of the current sampling period is subtracted from the amplitude envelope signal of the previous sampling period to obtain the difference. The difference is divided by the set fixed sampling period to obtain the vortex pressure pulsation derivative as the time derivative data of the low-frequency vortex pressure pulsation. The distributed control system obtains a product term by multiplying the sensitivity constant by the derivative of the vortex pressure pulsation, takes a negative value for the product term to generate the dynamic clamping coefficient, and multiplies the time-decreasing exponential transition factor by the reference proportional gain to obtain the target proportional gain. The result of multiplying the dynamic clamping coefficient by the target proportional gain is used to limit the exponential recovery trajectory of the proportional gain of the main loop regulator.

10. The adaptive frequency control method for primary frequency regulation of a hydropower station governor based on grid-source coordination according to claim 2, characterized in that, The steps of generating and issuing the integrated guide vane opening command specifically include: During the anti-reverse control phase, the distributed control system algebraically superimposes the output of the main loop regulator with the feedforward differential negative compensation. During the acceleration output recovery phase, the distributed control system uses the proportional gain of the main regulator recovered at the current moment to calculate the adjustment command and generate the comprehensive guide vane opening command. The command change is obtained by subtracting the comprehensive guide vane opening command of the current cycle from the comprehensive guide vane opening command of the previous cycle. The command change is obtained by dividing the command change by the fixed sampling period set by the system. If the absolute value of the command change rate exceeds the limit, the command change is clamped. After amplitude limiting and change rate limiting processing, the target opening command is output. The target opening command is converted into an analog control voltage signal and sent to the electro-hydraulic servo system to drive the mechanical guide vane to perform the action.

Images