A method and system for intelligent collaborative control of primary frequency modulation of a hydroelectric power station

Abstract

The present application relates to the technical field of automatic control of hydropower stations, and discloses a method and system for intelligent collaborative control of primary frequency modulation of a hydropower station governor, which comprises constructing a full-plant hydraulic coupling matrix according to pipe network parameters; when the rate of change of power grid frequency exceeds the limit, the pipe network execution module is divided into an active disturbance group and a pressure following group according to the coupling matrix; the active disturbance group is controlled to act to generate a hydraulic transient wave, and the actual water hammer wave speed and pipe wall friction coefficient are calculated; based on the calculated parameters and a characteristic line method calculation program, nonlinear guide vane kinematic trajectory data of the pressure following group are generated; the pressure following group is controlled to perform a phase shift delay operation and output mechanical power, and the total active power is synthesized with the power of the active disturbance group; power tracking residuals are calculated and control references are adjusted. The present application reduces the superposition degree of water hammer waves caused by synchronous action of the unit, weakens power counter-regulation caused by the water hammer effect, and improves the stability of the full-plant primary frequency modulation power output.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for hydropower stations, specifically to a method and system for intelligent coordinated control of primary frequency regulation of a hydropower station governor. Background Technology

[0002] Stable power grid frequency is fundamental to ensuring the safe operation of the power system. Hydropower stations, due to their rapid regulation speed, often participate in primary frequency regulation of the power grid as the main power source. In the construction of conventional large and medium-sized hydropower stations, in order to optimize spatial layout and reduce engineering costs, a common arrangement is to have multiple turbine generator units sharing a water diversion network.

[0003] Under the existing primary frequency regulation control mode, when the grid frequency deviates, the plant-level control system typically issues power regulation commands to all online units simultaneously according to their capacity ratio. The governors of each unit then synchronously adjust the guide vane opening. This synchronous regulation method causes hydraulic transient waves to be generated simultaneously in each branch pipeline. These transient waves intersect and directly superimpose at the nodes of the common pressure main steel pipe and the branch pipes, easily causing a sharp rise or fall in transient water pressure within the pipeline network, affecting the structural safety of the hydraulic pipeline system. Furthermore, the synchronous operation of the guide vanes of multiple units triggers a strong water hammer effect, causing the turbine's output mechanical power to change in the opposite direction to the command requirement during the initial regulation phase. Under the hydraulic coupling effect of multiple units sharing the pipeline network, the overall water hammer reverse regulation phenomenon is further amplified, resulting in hysteresis and low-frequency oscillations in the total active power response at the hydropower station's grid connection point. This prevents the hydropower station from quickly and smoothly tracking the target power command of the primary frequency regulation, reducing the quality of active power regulation and limiting the hydropower station's ability to suppress grid frequency fluctuations. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent collaborative control method and system for primary frequency regulation of hydropower station governors. This method solves the problem that when multiple hydro-generator units share a water diversion network to participate in the primary frequency regulation of the power grid, the synchronous operation of each unit will cause the superposition of hydraulic transient waves, resulting in fluctuations in water pressure in the network, which in turn leads to oscillations and hysteresis in the total active power response of the entire plant.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The first aspect of this invention provides a method for intelligent coordinated control of primary frequency regulation of a hydropower station governor, comprising the following steps:

[0007] The hydraulic coupling matrix of the entire plant is constructed based on the pipeline network parameters, and the boundary conditions and initial water hammer wave velocity are calculated using the method of characteristics.

[0008] Real-time monitoring of power grid frequency and frequency change rate; when the frequency change rate exceeds the preset grouping threshold, the pipeline execution module is divided into active disturbance group and pressure follower group according to the whole plant hydraulic coupling matrix.

[0009] The active disturbance group is controlled to generate hydraulic transient waves. The arrival time difference of pressure waves at each node is collected, and the actual water hammer wave velocity and pipe wall friction coefficient are calculated.

[0010] The actual water hammer wave velocity and pipe wall friction coefficient are input into the characteristic line method calculation program to generate the transient head time domain envelope of the volute of the unit corresponding to the pressure follower group. The nonlinear guide vane kinematic trajectory data of the pressure follower group are calculated through the transient head time domain envelope.

[0011] The pressure follower group performs a phase-shift delay operation. The proportional electro-hydraulic servo mechanism of the pressure follower group moves according to the nonlinear guide vane kinematic trajectory data and outputs mechanical power. The mechanical power output by the pressure follower group and the mechanical power output by the active disturbance group are electrically combined at the grid connection point of the hydropower station to obtain the total active power of the entire plant.

[0012] Measure the total active power and calculate the target power tracking residual. Adjust the power control reference according to the target power tracking residual until the grid frequency enters the dead zone setting range.

[0013] Furthermore, the specific steps for constructing the plant-wide hydraulic coupling matrix based on the pipeline network parameters include:

[0014] Based on the hydraulic elastic water column theory, combined with the fluid bulk elastic modulus and fluid density, the initial water hammer wave velocity of each pipe section is set.

[0015] Using the cross-sectional area of ​​the main steel pipe as a benchmark, the geometric lengths of branch pipes with different inner diameters are converted into equivalent hydraulic lengths.

[0016] The hydraulic coupling coefficient between units is calculated based on the equivalent hydraulic length between connected nodes. The hydraulic coupling coefficient between units is positively correlated with the equivalent hydraulic length of the upstream water intake pipe section shared by the units, and negatively correlated with the equivalent hydraulic length of the unit's independent branch pipe.

[0017] Based on the calculated hydraulic coupling coefficients between the units, a symmetric matrix of the corresponding dimension is constructed as the hydraulic coupling matrix of the entire plant.

[0018] Furthermore, the specific steps of dividing the pipeline network execution modules into an active disturbance group and a pressure follower group based on the overall plant hydraulic coupling matrix include:

[0019] Based on the effective active power regulation margin, rated active power, real-time operating head and rated design head of each online unit, calculate the dynamic operating potential energy coefficient of each online unit;

[0020] The dynamic operating potential energy coefficient is normalized using the range transformation method, and combined with the hydraulic coupling coefficient extracted from the hydraulic coupling matrix of the whole plant, the comprehensive topology grouping index of each online unit is calculated.

[0021] All units participating in frequency regulation online are sorted in descending order according to the comprehensive topology grouping index. Units in the sorted sequence that are at the top of the preset grouping threshold are designated as active disturbance groups, and units in the sorted sequence that are at the bottom of the preset grouping threshold are designated as pressure follower groups.

[0022] Furthermore, the specific steps for controlling the action of the active disturbance group to generate hydraulic transient waves include:

[0023] The total power compensation required for primary frequency regulation for the entire plant is calculated based on the real-time frequency deviation. The total compensation is then proportionally allocated to each unit in the active disturbance group based on the dynamic operating potential energy coefficient of each unit within the active disturbance group to obtain the target active power regulation increment.

[0024] By combining real-time working head data, the target active power adjustment increment is inversely calculated into the target guide vane opening command through bilinear interpolation algorithm;

[0025] Based on the maximum allowable increment of water hammer pressure calculated by the collaborative control module, the maximum allowable rate of change of the terminal flow is estimated, the mechanical rate limit value of the guide vane action is set, and the rate of change of the target guide vane opening command is dynamically limited.

[0026] Furthermore, the specific steps for calculating the actual water hammer wave velocity and the pipe wall friction coefficient include:

[0027] The absolute propagation time is calculated by extracting the initial action time of the mechanical displacement of the guide vanes of the active disturbance group and the recorded wavefront arrival time, and the actual water hammer wave velocity is calculated by combining the total length of the hydraulic pipeline topology path.

[0028] The real-time velocity change rate parameter is extracted from the pipeline calculation matrix. Combined with the initial steady-state friction reference value and the transient derivative of the velocity sequence with respect to time, the dynamic transient friction resistance coefficient is calculated as the pipe wall friction coefficient.

[0029] The collaborative control module calculates the dynamic Courland number in real time. When the actual water hammer wave velocity increases and causes the dynamic Courland number to exceed the limit, the calculation time step is reduced proportionally to perform grid adaptive interpolation reconstruction.

[0030] Furthermore, the specific steps for calculating the nonlinear guide vane kinematic trajectory data of the pressure-following group include:

[0031] The compatibility equations of the positive and negative characteristic lines of one-dimensional transient flow are discretized and reset using the actual water hammer wave velocity and pipe wall friction coefficient.

[0032] The active power deficit to be compensated is allocated according to the ratio of static adjustable capacity margin, and the target active power adjustment increment of each unit in the pressure follow group is obtained.

[0033] The target active power adjustment increment, equivalent transient pressure head, predicted maximum background pressure, and current guide vane initial steady-state opening are input into a pre-trained nonlinear opening inverse matching neural network model based on a multilayer perceptron. The model outputs the guide vane opening adjustment increment and, combined with the current guide vane initial steady-state opening, calculates and obtains the target guide vane opening command.

[0034] Based on the target guide vane opening command, a fifth-order polynomial S-shaped velocity curve is used to generate the guide vane position command sequence, which serves as the nonlinear guide vane kinematic trajectory data.

[0035] Furthermore, the specific steps for controlling the pressure follower group to perform the phase shift delay operation include:

[0036] Extract the sign direction of the target active power adjustment increment to determine the unit's adjustment condition. Under the load increase condition, align the target offset timestamp with the peak of the upstream pilot wave. Under the load decrease condition, align the target offset timestamp with the trough of the upstream pilot wave.

[0037] The relative waiting delay before issuing the action command to the pressure follower group is calculated by subtracting the current timestamp, the static response dead zone time, and half of the guide vane ramp time from the target hedging timestamp.

[0038] Furthermore, the specific steps of the proportional electro-hydraulic servo mechanism of the pressure follower group acting according to the nonlinear guide vane kinematic trajectory data and outputting mechanical power include:

[0039] Real-time acquisition of the actual opening feedback value from the guide vane displacement sensor; calculation of position tracking error and error change rate.

[0040] An adaptive PID control algorithm is adopted to dynamically tune the proportional gain and derivative gain in the control loop to drive the proportional electro-hydraulic servo mechanism based on the position tracking error and the error change rate.

[0041] Furthermore, the steps of measuring the total active power and calculating the target power tracking residual, and adjusting the power control benchmark based on the residual, specifically include:

[0042] A first-order digital low-pass filter algorithm is used to smooth the measured total active power.

[0043] Calculate the transient power residual between the target commanded power of the power grid and the total active power. When the absolute value of the transient power residual is greater than the set dead zone threshold, start the sliding time window to calculate the residual integral energy value.

[0044] When the residual integral energy value exceeds the preset fault tolerance energy threshold, the total active power compensation of the entire plant is calculated based on the proportional and integral adjustment gain.

[0045] Extract the current available capacity margin of each unit, dynamically weight the total active power compensation of the whole plant according to the proportion of available capacity margin, and obtain the individual unit fine-tuning instructions of each unit.

[0046] Dynamic weighted allocation is performed based on available capacity margin, and the single-machine fine-tuning command is inversely mapped to the guide vane opening compensation correction value in combination with the local micro-increase rate of power opening.

[0047] A second aspect of the present invention provides an intelligent collaborative control system for primary frequency regulation of a hydropower station governor, used to implement the above-mentioned intelligent collaborative control method for primary frequency regulation of a hydropower station governor, comprising:

[0048] The pipeline execution module acts on the upstream reservoir, surge chamber, main steel pipe of common pressure, branch pipe and spiral casing of multiple hydro-generator units, and is equipped with microcomputer digital speed governor and proportional electro-hydraulic servo mechanism;

[0049] The multi-node observation module is distributed along the spatial topology nodes of the common pressure main steel pipe and the branch pipe, and is equipped with dynamic pressure sensors;

[0050] The state sensing module is located at the grid connection point of the hydropower station and is equipped with a synchronous phasor measurement device to extract the grid connection point frequency and frequency change rate.

[0051] The collaborative control module establishes data communication connections with the pipeline execution module, the multi-node observation module, and the status perception module, and is equipped with a real-time programmable automation controller.

[0052] This invention provides a method and system for intelligent coordinated control of primary frequency regulation in a hydropower station governor. It offers the following advantages:

[0053] 1. This invention divides the pipeline network execution module into an active disturbance group and a pressure following group by constructing a whole-plant hydraulic coupling matrix. This changes the traditional control mode of multiple units operating synchronously in the same direction, and avoids the direct superposition of hydraulic transient waves caused by multiple units adjusting the guide vanes at the same time in the common water intake pipeline network. This reduces the transient extreme value of pipeline network water pressure and improves the operational safety of the hydropower station's hydraulic pipeline network system.

[0054] 2. This invention generates hydraulic transient waves by controlling the action of an active disturbance group, and calculates the actual water hammer wave velocity and pipe wall friction coefficient in real time using the arrival time difference of the collected nodal pressure waves. The real-time corrected hydraulic parameters are input into the characteristic line method calculation program to generate nonlinear guide vane kinematic trajectory data of the pressure following group, avoiding the calculation deviation caused by using fixed static parameters in the calculation model, and improving the accuracy of guide vane opening adjustment command calculation.

[0055] 3. This invention controls the pressure follower group to perform phase-shift delay operation, aligning the target offset timestamp with the peak or trough of the upstream pilot wave according to the regulation conditions, so that the mechanical power output of each group of units is electrically synthesized at the grid connection point. Utilizing the time-domain misalignment of the water pressure wave to form a physical state offset reduces the mechanical power reversal phenomenon caused by the water hammer effect, lowers the oscillation and hysteresis of the total active power output of the entire plant, and improves the stability of the primary frequency regulation control of the power grid. Attached Figure Description

[0056] Figure 1 This is a system framework diagram of an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram comparing the transient pressure of the volute casing according to the present invention;

[0059] Figure 4 This is a schematic diagram comparing the active power response at the grid connection point of the present invention; Detailed Implementation

[0060] 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.

[0061] See attached document Figure 1 The present invention provides an intelligent collaborative control system for primary frequency regulation of a hydropower station governor, comprising: a pipeline execution module, a multi-node observation module, a status sensing module, and a collaborative control module.

[0062] The pipeline execution module includes the upstream reservoir, surge tank, main common pressure steel pipe, branch pipes, and spiral casings of multiple hydro-turbine generator units. The pipeline execution module is equipped with a microcomputer-controlled digital speed governor and a proportional electro-hydraulic servo mechanism. The microcomputer-controlled digital speed governor and the proportional electro-hydraulic servo mechanism are used to receive control signals and drive the turbine guide vanes to perform mechanical actions.

[0063] The multi-node observation module is distributed along the spatial topology nodes of the main common pressure steel pipe and branch pipes. These nodes are located at the bottom of the surge tank, pipe bends, and branch pipe locations. Each multi-node observation module is equipped with a dynamic pressure sensor. The multi-node observation module connects to the collaborative control module via a communication network supporting a time synchronization protocol to acquire timestamped transient water pressure data.

[0064] The state sensing module is deployed at the grid connection point of the hydropower station. The state sensing module is equipped with a synchronous phasor measurement device. The state sensing module is used to acquire the three-phase voltage and three-phase current sequences at the grid connection point, and to extract the frequency data and frequency change rate at the grid connection point.

[0065] The collaborative control module establishes data communication connections with the pipeline execution module, the multi-node observation module, and the state awareness module, respectively. The collaborative control module is equipped with a real-time programmable automation controller. Internally, the collaborative control module runs a characteristic line method calculation program and an energy mapping inverse matching program.

[0066] See attached document Figure 2 This invention provides a method for intelligent coordinated control of primary frequency regulation of a hydropower station governor, comprising the following steps:

[0067] S100, the collaborative control module reads the pipeline parameters from the pipeline execution module, constructs the whole plant hydraulic coupling matrix based on the pipeline parameters, and sets the boundary conditions and initial water hammer wave velocity using the characteristic line method;

[0068] S200, the status perception module monitors the power grid frequency and frequency change rate in real time. When the frequency change rate exceeds the set threshold, the collaborative control module divides the pipeline execution module into an active disturbance group and a pressure following group according to the hydraulic coupling matrix of the whole plant and the operating status of the pipeline execution module.

[0069] S300, the collaborative control module sends the guide vane opening command to the active disturbance group, the proportional electro-hydraulic servo mechanism of the active disturbance group moves and generates hydraulic transient waves inside the pipe network, the multi-node observation module collects the pressure wave arrival time difference data at each spatial topology node, and the collaborative control module calculates the actual water hammer wave velocity and pipe wall friction coefficient based on the pressure wave arrival time difference data.

[0070] S400, the collaborative control module inputs the actual water hammer wave velocity and pipe wall friction coefficient into the characteristic line method calculation program. The collaborative control module generates the transient head time-domain envelope of the turbine casing corresponding to the pressure follower group. The collaborative control module calculates the nonlinear guide vane kinematic trajectory data of the pressure follower group based on the transient head time-domain envelope and the turbine energy mapping equation.

[0071] S500, the collaborative control module controls the pressure following group to perform phase shift delay operation. The proportional electro-hydraulic servo mechanism of the pressure following group moves according to the nonlinear guide vane kinematic trajectory data and outputs mechanical power. The mechanical power output by the pressure following group and the mechanical power output by the active disturbance group are electrically combined at the grid connection point of the hydropower station to obtain the total active power of the entire plant.

[0072] The S600's state perception module measures the total active power of the entire plant at the hydropower station's grid connection point. The collaborative control module calculates the target power tracking residual and adjusts the power control benchmark based on the target power tracking residual until the grid frequency enters the dead zone setting range.

[0073] In this embodiment, the collaborative control module and the pipeline execution module interact to obtain the basic boundary conditions required to construct the physical model of the hydraulic transient flow. The propagation characteristics of the hydraulic transient pressure wave are not only limited by the physical properties of the fluid itself, but also modulated by the mechanical response of the constraint boundary (i.e., the pipe wall). Therefore, accurately obtaining and setting the relevant parameters is a prerequisite for establishing the subsequent characteristic line method computational mesh. Specifically, this process includes the following steps:

[0074] S101, the collaborative control module accesses the local storage area or the plant's industrial-grade database to read the pipeline parameters of each physical pipe segment from the pipeline execution module. In this specific embodiment, this is further expanded to include the geometric feature data, fluid physical property data, and pipe mechanical property data of the water diversion steel pipe. As a preferred method, the collaborative control module extracts the parameters sequentially according to the pipeline spatial topology. Pipeline length of section , pipe inner diameter Pipe wall thickness Pipe elastic modulus and the initial Darcy friction coefficient Among them, the elastic modulus of the pipe The value usually depends on the specific material of the pipe section; for example, the value for conventional steel pipes is mostly in the range of 2.0 × 10. 11 Pa up to 2.1 × 10 11 Between Pa; Initial Darcy friction coefficient The absolute roughness of the pipe wall can be pre-calibrated using empirical formulas, with typical values ​​ranging from 0.01 to 0.03. This data serves as the system's basic static configuration parameters. It is written into the control system's human-machine interface before the hydropower station units are put into operation and is typically stored in the non-volatile memory area of ​​the programmable logic controller within the collaborative control module.

[0075] S102, the collaborative control module acquires the fluid medium state parameters under the current actual operating conditions. The fluid medium state parameters specifically include fluid density. With fluid bulk elastic modulus Because changes in the water temperature of the reservoir environment cause numerical drift in the water density and bulk modulus, the collaborative control module receives real-time analog temperature measurements from temperature sensors located at the inlet and performs dynamic temperature compensation and correction calculations on these two physical quantities using a pre-set water physical property mapping table within the controller. Regarding the specific implementation of the analog signal acquisition and property lookup compensation correction program for industrial temperature sensors, those skilled in the art can use conventional analog signal acquisition hardware modules combined with piecewise interpolation lookup algorithms for programming. The signal conversion and underlying code logic are well-known technologies in this field and will not be elaborated upon here.

[0076] S103, the collaborative control module calculates and sets the initial water hammer wave velocity for each pipe section based on the hydraulic elastic water column theory. The initial water hammer wave velocity is a fundamental variable determining the phase characteristics of the propagation of transient hydraulic pressure waves within a closed pipe. The collaborative control module comprehensively considers the elastic deformation effect of the metal pipe walls of the main pressure steel pipe and branch pipes in the pipeline execution module, as well as the compressibility of the water inside the pipes, and uses the elastic water column water hammer wave velocity formula for calculation: In the formula, Representing the Initial water hammer wave velocity of a section of pipeline under conditions where complex cavitation release and dynamic deformation of the pipe wall do not occur; This represents the bulk modulus of the fluid after temperature compensation correction. This represents the corrected fluid density; This represents the elastic modulus of the metal material of the corresponding spatial pipe segment in the pipeline execution module; This represents the inner diameter of that section of the pipe; This represents the wall thickness at the corresponding location.

[0077] The collaborative control module calculates the initial water hammer wave velocity for all independently divided pipe segments. Then, it is stored in a floating-point data array specified in memory. This initial water hammer wave velocity... This reflects the theoretical mechanical wave transmission capability of the hydraulic system contained in the pipeline execution module under static, undisturbed conditions. The collaborative control module uses it as the basic physical numerical basis for subsequently tracking the arrival time of the actual pressure waveform and setting the characteristic line method to calculate the boundary of the differential spatiotemporal grid.

[0078] In this embodiment, the collaborative control module further establishes a mathematical topology model reflecting the intensity of hydraulic interference between multiple units based on the read static parameters of the pipeline network. Typically, when multiple units share the same water intake pipeline system, a sudden change in flow rate caused by adjusting the guide vanes of one unit will be converted into a transient pressure wave. This pressure wave is transmitted and reflected at the branch pipe nodes, thus propagating along the common waterway and superimposed on the spiral casing boundaries of other units. The strength of this physical interference directly depends on the spatial connection topology between the units and the hydraulic impedance characteristics of the shared pipe section. By quantifying it into a matrix form, reliable data support can be provided for subsequent development of dynamic grouping and asynchronous control strategies. Specifically, this process includes the following steps.

[0079] S104, the collaborative control module parses the physical connection paths between the units in the pipeline execution module. As a preferred method, the collaborative control module uses the inlet of the spiral casing corresponding to any two generator units within the plant as endpoints, traverses their connected waterway nodes, and calculates the absolute hydraulic spatial distance between the units. Assume the pipeline execution module contains... Taiwan unit, defining the unit With the unit The relative hydraulic path lengths between them are Considering that actual hydropower station water diversion systems typically involve multiple diameter changes, the hydraulic path length here is not a simple straight-line geometric distance or merely the physical pipe length. To accurately reflect the propagation impedance of pressure waves in variable-diameter pipes, the collaborative control module, based on the principle of equivalent inertia of water flow, uses the cross-sectional area of ​​the common main steel pipe as a benchmark to convert the geometric lengths of branch pipes with different inner diameters into equivalent hydraulic lengths, and sums up the equivalent hydraulic lengths between connected nodes.

[0080] S105, the collaborative control module calculates the hydraulic coupling coefficient based on the extracted hydraulic path length. To reasonably reflect the degree of interference between units caused by sharing waterways, a dimensionless hydraulic coupling coefficient is introduced. This coefficient is positively correlated with the length of the shared main steel pipe between units and negatively correlated with the length of the independent water intake branch pipes between units. In this embodiment, the hydraulic coupling coefficient... The calculation expression is set as follows: In the formula, Representative unit With the unit The equivalent hydraulic length of the shared upstream water intake section; and These represent the flow from the shared distribution node to the generating unit. volute and unit The equivalent hydraulic length of the independent branch pipe of the volute. The calculated hydraulic coupling coefficient. The values ​​range from 0 to 1. The closer the value is to 1, the higher the proportion of the pipe section shared by the two units and the stronger the hydraulic interference between them; conversely, the closer the value is to 1, the longer the independent branch pipe and the higher the hydraulic isolation.

[0081] S106, for those containing The system of online units is presented in this matrix. The symmetric matrix form. The overall plant hydraulic coupling matrix is ​​defined as follows: Its matrix elements satisfy Specifically, the diagonal elements of the matrix This represents the self-coupling effect of the unit itself, and its value is fixed at 1.

[0082] Considering that some units are in standby or maintenance status during actual power plant operation, the collaborative control module will poll the circuit breaker closing status and guide vane limit signals of each unit in the pipeline execution module in real time. If a unit is detected to be offline, the collaborative control module will forcibly set the corresponding row and column elements of that unit in the plant-wide hydraulic coupling matrix to zero. Furthermore, if the status polling finds that the number of available units currently in grid-connected operation is less than two, the collaborative control module will automatically lock the current collaborative control logic and issue an instruction to switch to the single-unit independent primary frequency regulation mode. This fault-tolerant mechanism aims to prevent the system from falling into a matrix calculation dead zone due to a lack of follower units available for power compensation, while also preventing the target power from being incorrectly sent to the shutdown equipment. For the basic programming implementation of matrix operations and the memory allocation method of two-dimensional arrays, those skilled in the art can use conventional underlying data structures for processing, which are well-known technologies in the field and will not be elaborated here.

[0083] In this embodiment, the essence of the hydraulic transient process is a continuous fluid dynamics problem described by a set of partial differential equations. Since pressure waves propagate at a finite wave speed in the pipe fluid, directly solving the nonlinear partial differential equations is extremely challenging. Therefore, in order to perform real-time prediction in the digital computing environment of the collaborative control module, it needs to be discretized into a solvable set of algebraic equations. This transformation process relies on reasonable time and space partitioning steps to ensure the convergence and stability of the numerical computation. The specific implementation includes the following steps.

[0084] S107, the cooperative control module calculates the time step and spatial partitioning step of the mesh using the characteristic line method based on the determined initial water hammer wave velocity. The stability of the numerical calculation is limited by the CFL condition. As a preferred approach, the cooperative control module pre-reads the controller's basic sampling period as the global basic time step. Considering the balance between the controller's computational load and the accuracy of the physical model, The value is typically between 10 and 50 milliseconds. Subsequently, the collaborative control module determines the initial water hammer wave velocity based on each section of the pipeline. Given compatibility conditions, calculate the corresponding spatial partitioning step size. The compatibility constraint condition is satisfied as follows: In the formula, The Courant number represents a dimensionless number; Representing the Initial water hammer wave velocity of the pipeline section; Represents the global base time step; Representing the The spatial partitioning step size for pipeline segments. In actual differential discretization processes, since the physical pipeline length is often difficult to be completely divided by the calculated spatial partitioning step size, to avoid complex spatial interpolation errors at the differential grid nodes, the collaborative control module typically fine-tunes the equivalent calculated wave velocity or local geometric parameters of the spatial pipeline segments to achieve the Coulomb number. The value is equal to 1, ensuring that the feature line accurately passes through the node positions of the spatiotemporal grid. To prevent over-adjustment from causing severe distortion of the physical model, the fine-tuning range of this wave velocity or pipeline length is usually limited to within ±5% of the original value; if the calculation finds that the required fine-tuning range exceeds this range, the cooperative control module automatically shortens the global base time step proportionally. Then perform a rematch calculation.

[0085] S108, the unsteady motion of the fluid inside the pipe is jointly described by the fluid mechanics continuity equation and the equation of motion. From the physical perspective of energy and mass conservation, the changes in head pressure and transient flow rate of a water particle propagating along a specific characteristic line follow specific algebraic constraints. The cooperative control module performs an integral transformation of the fluid state variables along the direction of the characteristic line, deriving a set of algebraic equations characterizing the evolution of nodal pressure and flow rate. This set of equations includes positive and negative characteristic line equations:

[0086] ;

[0087] ;

[0088] In the formula, and These represent the transient head and flow rate variables of the spatial node to be calculated at the current calculation time, respectively. and This represents the transient head and flow variables of the upstream adjacent spatial nodes at the previous calculation time. and This represents the transient head and flow variables of the downstream adjacent spatial nodes at the previous calculation time. The characteristic impedance of the pipe satisfies ,in Represents the gravitational acceleration constant (a common value can be set to 9.81 m / s²).2 ); This represents the cross-sectional area of ​​the k-th pipe segment; Representing the Initial water hammer wave velocity of the pipeline section; Let be the friction coefficient along the pipeline, which satisfies ,in This represents the initial Darcy friction coefficient; Represents the inner diameter of the pipe; This represents the spatial division step size of the k-th pipeline segment.

[0089] S109, the collaborative control module constructs mathematical-physical boundaries based on the actual operating modes of each physical component in the pipeline execution module. The numerical calculation boundaries of a hydraulic system typically include constant head boundaries and dynamic node boundaries. In this embodiment, the collaborative control module sets the upper free water surface of the upstream reservoir and surge tank as the constant head boundary, and configures the corresponding calculation logic to adjust the head variable at this spatial node. This is equivalent to the real-time measured static water level of the reservoir or surge tank. To avoid boundary condition calculation failures due to communication failures or data jumps of external water level sensors, the collaborative control module has a data verification and state maintenance mechanism: when the loss of the analog water level signal or the sudden change in slope exceeds the reasonable range is detected, the system automatically locks the effective static head value of the calculation cycle before the anomaly occurred as the current boundary input to ensure the continuous and reliable operation of the calculation program. Simultaneously, the turbine generator casing and guide vane end faces are set as dynamic valve boundaries. The collaborative control module extracts the real-time guide vane opening command signal from the unit's governor and, combined with the flow coefficient mapped from the turbine's comprehensive characteristic curve, constructs a nonlinear algebraic correlation between the transient head and discharge flow rate at this end node.

[0090] In this embodiment, the state awareness module is responsible for capturing real-time electrical quantity anomalies at the grid connection point to provide decision-making prerequisites for subsequent coordinated control. From the physical principle of power system power balance, the grid frequency is a direct physical scale reflecting the balance of active power supply and demand in the system. When the external load changes abruptly, the generator rotor speed will deviate, thereby causing transient changes in the grid connection point frequency and the rate of frequency change. To convert this physical change into a digital signal that the controller can recognize, the process specifically includes the following steps.

[0091] S201, the state perception module acquires high-frequency sampling data from the grid connection point and extracts frequency parameters. As a preferred method, the state perception module synchronously acquires the three-phase voltage and current sequences from the high-voltage grid-connected bus side of the hydropower station via an internally configured synchronous phasor measurement device. Based on the acquired discrete electrical sequences, the state perception module calculates the real-time frequency value and frequency change rate for the current calculation cycle. For the specific implementation of the analog-to-digital conversion, voltage and current phasor extraction, and fundamental frequency extraction based on the phase-locked loop algorithm of the synchronous phasor measurement device, those skilled in the art can develop it using conventional microcomputer protection and control device hardware combined with a digital signal processing algorithm library. The underlying data flow and conventional calculations are well-known technologies in this field and will not be elaborated upon here. To quantify the transient evolution trend of the frequency and thus establish a forward-looking control index before the frequency drop expands, the state perception module uses a first-order backward differential algorithm to calculate the frequency change rate: In the formula, Represents the rate of change of frequency at the current moment; The real-time frequency value representing the current sampling period; Represents the real-time frequency value of the previous sampling period; This represents the basic measurement and control sampling period of the state awareness module. To ensure the real-time performance of transient feature capture matches the time scale of subsequent feature line method calculations, this basic measurement and control sampling period... It is typically set to 10 to 20 milliseconds.

[0092] S202. Considering the electromagnetic interference and measurement noise of voltage transformers in industrial settings, the originally calculated frequency and rate of change often contain high-frequency noise. Directly introducing this noise as a trigger criterion can cause mechanical malfunctions in the speed governor actuator. Therefore, the state perception module is equipped with a moving average filtering algorithm to smooth the original sequence parameters. As a preferred approach, the length of this moving window is typically tuned based on the power grid frequency cycle, for example, selecting 5 to 10 calculation cycles to achieve a reasonable balance between filtering out high-frequency harmonics and retaining effective transient trends. Simultaneously, to prevent communication interruptions or hardware damage from causing control system logic collapse, the state perception module establishes an over-limit blocking fault-tolerant mechanism. When a real-time frequency value is detected... When the frequency exceeds the physical interference immunity limit set by the power grid (e.g., drops below 45Hz or spikes above 55Hz), or when a check error is detected in the internal digital communication frame, the state awareness module will automatically block the subsequent frequency modulation trigger judgment logic and output an invalid data flag.

[0093] S203: Normal, minor load fluctuations in the power grid can be absorbed by the system's own rotational inertia, without requiring mechanical power intervention from the hydro-generator units. The state-aware module calculates the real-time frequency deviation. Its satisfaction ,in, The real-time frequency value representing the current sampling period; The reference frequency is the power grid's rated frequency, typically 50Hz in conventional systems. Subsequently, the state-aware module uses the smoothed frequency variable to construct a comprehensive triggering criterion model: In the formula, This represents the frequency modulation trigger status bit. A value of 1 indicates that the trigger condition is met, and a value of 0 indicates that it is not triggered. This represents the calculated real-time frequency deviation. This represents the system's preset frequency dead zone threshold. This represents the smoothed rate of change of frequency. This represents the set frequency change rate trigger threshold, used to initiate frequency regulation in advance under transient conditions where the frequency deviation has not yet exceeded the limit but the change trend is rapid. The specific settings of the above parameters are usually uniformly issued by the dispatch center of the relevant power grid based on the overall rotational inertia level of the regional power grid and the installed capacity of generating units. In conventional applications... The range of values ​​is mostly within to between, Multiple settings to .

[0094] Furthermore, to avoid frequent switching of the trigger signal caused by minor fluctuations in the grid frequency near the dead zone threshold, the state awareness module incorporates anti-jitter delay and hysteresis interval logic. When the aforementioned comprehensive trigger criterion model outputs 1 for the first time, the state awareness module starts its internal high-precision timer. Only when the out-of-limit state persists for a set time (e.g., 0.1 seconds) and does not fall back to the safe hysteresis dead zone is the generated frequency modulation trigger command finally confirmed. Subsequently, the state awareness module pushes this command, along with real-time frequency deviation data with an absolute timestamp, to the collaborative control module via a high-speed communication bus supporting a time synchronization protocol. Upon receiving the interrupt command, the collaborative control module transitions from standby monitoring to transient collaborative control calculation, preparing to allocate the action sequence of the pipeline execution modules based on the plant-wide hydraulic coupling matrix.

[0095] In this embodiment, after receiving the frequency modulation trigger command, the collaborative control module needs to dynamically assign task roles to multiple online generator units within the plant. Traditional same-frequency and same-modulation modes often easily lead to strong hydraulic interference and transient pressure superposition in multi-unit shared piping networks, thus threatening system safety. By decoupling the units and dividing them into control clusters of different tiers, the physical delay effect of the spatial piping network can be utilized to mitigate this mutual interference. This process specifically includes the following steps.

[0096] S204, the collaborative control module calculates the dynamic operating potential energy coefficients of each online generating unit participating in frequency regulation. Operating potential energy is a physical margin indicator characterizing the generator unit's ability to quickly and safely respond to active power regulation demands under current transient operating conditions. As a preferred method, the collaborative control module reads the data in real time from the status sensing module. The rated active power, current active power output, and real-time operating head data of the generator unit are used to construct a dynamic operating potential energy coefficient model. In the formula, Representing the Dynamic operating potential energy coefficient of the unit; This represents the unit's current effective active power regulation margin in response to the direction of grid frequency deviation. This represents the rated active power of the unit; and These represent the unit's current real-time operating head and rated design head, respectively. and This represents the preset dimensionless weighting factor. The specific value of the weighting factor is usually determined by the actual operating strategy of the hydropower station: if the station focuses on rapid response to the grid's active power dispatching, then the weighting factor should be appropriately increased. The value of is typically between 0.6 and 0.7; if the system prioritizes maintaining the stability margin of the reservoir and hydraulic network, then the value should be increased accordingly. The proportion of active power (its typical range is between 0.5 and 0.6), in engineering applications, is usually limited to a sum of 1. In engineering implementation, the effective active power regulation margin... The value of this margin depends on the direction of frequency regulation: when the grid frequency drops and the unit needs to increase its output, this margin is equal to the difference between the unit's maximum allowable operating power and the current power; when the grid frequency spikes and the unit needs to reduce its output, this margin is equal to the difference between the current power and the unit's minimum stable operating power.

[0097] S205, the collaborative control module extracts the plant-wide hydraulic coupling matrix constructed in the previous step and, combined with the dynamic operating potential energy coefficient, calculates the comprehensive topology grouping index for each unit. Since the directly calculated dynamic potential energy coefficient and the hydraulic coupling coefficient inherently differ in magnitude and distribution range, to ensure the rationality of the multi-objective evaluation, the collaborative control module introduces the range transformation method to normalize the potential energy coefficient. Its mapping logic satisfies... ,in and These represent the maximum and minimum values ​​of the potential energy coefficients of all currently online generating units, respectively. Representing the The dynamic operating potential energy coefficient of the unit.

[0098] In the physical mechanism of fluid conduction, the larger the sum of the hydraulic coupling coefficients of a unit with other units, the higher the proportion of their shared main pipeline network and the shorter their independent water intake branches. Spatially, this is equivalent to the unit being relatively close to the upstream common diversion node. After achieving dimensional unification, the collaborative control module uses the following formula to calculate the comprehensive topology grouping index: In the formula, Representing the The overall topology grouping index of the Taiwanese unit; This represents the dynamic operating potential energy coefficient after normalization. The units extracted from the overall plant hydraulic coupling matrix With the unit The hydraulic coupling coefficient between them; This represents the total number of units currently online. and This represents the positive adjustment weight set by the system, and the sum of the two values ​​is usually set to 1. As a preferred method, if the common main section of the hydropower station's water diversion pipeline is long while the independent branch pipes of each generating unit are extremely short, then this is often used... The value is set in the range of 0.6 to 0.8 to highlight the dominant role of pipeline topology location in grouping decisions. This indicator mathematically unifies the pre-existing hydraulic location attributes of the unit with its power regulation potential.

[0099] S206, the coordinated control module groups all online units participating in frequency regulation according to the calculated comprehensive topology index. Sort in descending order from largest to smallest. To avoid the sorting algorithm outputting an uncertain physical array due to multiple identical units having completely equal calculated indicators, the system presets a secondary sorting criterion at this stage: when the comprehensive indicators are the same, priority is given to sorting by the physical equipment number of the unit or the cumulative fault-free operating time of the unit, thereby ensuring the uniqueness of the optimization result of each grouping and the stability of the algorithm execution.

[0100] After obtaining a stable descending sequence, the collaborative control module, based on a preset grouping threshold (a proportional constant to the total number of participating frequency regulation units), multiplies the total number of online units by this proportional constant and rounds it down. Units at the top of the sequence are designated as the active disturbance group, while the remaining units at the bottom are designated as the pressure follower group. From a physical perspective, units in the active disturbance group have a greater regulation margin and are hydraulically closer to the upstream main pipeline, allowing them to trigger initial pilot transient waves within the pipeline network. Units in the pressure follower group, located at the end of the network, are used to offset and smooth pressure oscillations caused by the active disturbance group in subsequent control stages.

[0101] To ensure the continuity of the control system's operation under extreme electrical conditions, the cooperative control module is equipped with degradation takeover logic. When the status polling mechanism detects that the total number of currently available units that have not reached their mechanical limits is less than two, the cooperative control module will automatically suspend the aforementioned dynamic grouping strategy and instruct the remaining units to revert to the single-unit independent primary frequency regulation mode. This exception handling mechanism effectively prevents the control algorithm from falling into a computational dead zone due to an empty array or insufficient dimensions. Regarding the aforementioned sorting algorithm and the data flow method of the internal multidimensional array, those skilled in the art can compile it using conventional quicksort code logic combined with the register mapping function of a programmable logic controller. The underlying instruction calls are well-known techniques in this field and will not be elaborated upon here.

[0102] In this embodiment, after completing the control cluster division of the online units, the cooperative control module prioritizes mobilizing the generator units within the active disturbance group to perform active power compensation tasks for primary frequency regulation. From the perspective of pipeline network hydraulics, the active disturbance group is located at the front-end node of the pipeline network hydraulic topology. The initial displacement of its guide vane actuator will change the local flow pattern, thereby generating initial pressure fluctuations, i.e., leader transient waves, in the common water intake system. The physical evolution process of this transient wave in the pipeline network will directly serve as the boundary constraint condition for subsequently calculating the action delay of the pressure follower group. The specific implementation includes the following steps.

[0103] S301, the collaborative control module calculates the total target regulation power of the active disturbance group and issues independent power compensation commands to each unit within the group. As a preferred method, the collaborative control module calculates the total primary frequency regulation power compensation required for the entire plant based on the real-time frequency deviation transmitted by the state sensing module. Subsequently, based on the dynamic operating potential coefficient of each unit within the group, the total compensation is proportionally allocated to each unit in the active disturbance group. The mathematical expression corresponding to this allocation logic is set as follows:

[0104] In the formula, The representative issued the information to the active disturbance group. The target active power regulation increment for the unit; The system represents the overall frequency regulation coefficient of the entire plant as set by the system. Its physical meaning is the power regulation amount corresponding to a unit frequency deviation. According to conventional power grid dispatching guidelines, it is usually taken as three to six percent of the rated capacity of the entire plant per hertz. This represents the real-time frequency deviation calculated in the previous stage; The set of online units representing the active disturbance group; and These represent the dynamic operating potential energy coefficients of the corresponding units within the group under current operating conditions, after quantitative evaluation. To prevent the assigned commands from exceeding the mechanical limits of the unit's physical system, the collaborative control module adds hard-limiting logic after calculation: if the calculated... If the unit's target output exceeds its nameplate rated upper limit or falls below the vibration zone lower limit, the command increment will be forcibly truncated to the corresponding safety boundary value to ensure that the power generation equipment operates within the safety boundary. Furthermore, to ensure the achievement of the overall primary frequency regulation target for the entire plant, the collaborative control module will calculate in real time the active power deficit that could not be executed due to triggered limiting, and accumulate this deficit as an additional control benchmark, which will be passed on to the subsequent pressure follower group control logic. This avoids deadlocks caused by insufficient plant-wide response when the algorithm is locally constrained.

[0105] In S302, the turbine output power and guide vane opening exhibit a nonlinear coupling relationship. The collaborative control module utilizes a pre-set three-dimensional surface representing the turbine's comprehensive characteristics within the pipeline execution module, combined with the current real-time operating head data, to apply a bilinear interpolation algorithm to the active power regulation increment. Reverse calculation to obtain the target guide vane opening command To prevent rapid guide vane movement from causing drastic changes in local flow and thus triggering excessive transient water hammer pressure, the collaborative control module dynamically limits the rate of change of the guide vane opening, thereby restricting the guide vane movement speed. satisfy: In the formula, Representing the The current guide vane operating rate of the generator unit; Represents a variable that has been continuously adjusted over time; This represents the maximum allowable guide vane opening and closing rate threshold for the pipeline execution module. This safety threshold is not a fixed constant, but rather derived in real-time by the collaborative control module based on physical pressure-bearing boundaries. Specifically, the collaborative control module calculates the maximum allowable increment of water hammer pressure that the corresponding spatial pipe section can physically withstand, based on previously read pipe wall thickness, pipe material elastic modulus, and allowable material stress. Subsequently, based on the Jukowski fundamental equation of water hammer theory, the pressure increment limit is inverted and calculated into the maximum allowable rate of change of the terminal flow, and finally mapped to the mechanical rate limit of the guide vane movement by combining the aforementioned turbine comprehensive characteristic curve. For the specific implementation of the proportional-integral-derivative control algorithm and displacement sensor feedback closed loop at the bottom layer of the guide vane servo valve, those skilled in the art can deploy it using a conventional electro-hydraulic servo control system. Its hydraulic cylinder drive and error tracking mechanism are well-known technologies in the field and will not be elaborated upon here.

[0106] In S303, the continuous change in the cross-sectional area of ​​the guide vanes forces a step change in the water flow rate at the turbine casing. This local flow imbalance rapidly transforms into transient pressure fluctuations under the influence of fluid inertia, forming a leading transient wave that propagates in the reverse direction along the main intake pipe. During this fluid dynamic evolution, the collaborative control module uses the real-time guide vane opening signal measured by the active disturbance group as a dynamic node boundary and forcibly substitutes it into the aforementioned discretized one-dimensional transient flow compatibility equation set. By continuously iterating along the characteristic line over the time step, the collaborative control module can predict the spatial sweep path and pressure amplitude evolution trend of this leading transient wave in the bifurcation pipe and the common main pipe network. The hydraulic transient field data obtained from this real-time calculation not only reflects the water flow energy state excited by the active disturbance group itself, but also constitutes the core physical boundary constraint for the subsequent control action sequence issued to the pressure follower group, thereby effectively suppressing the unfavorable superposition of pressure waves caused by the synchronous operation of multiple units.

[0107] In this embodiment, after deriving the global evolution path of the leading transient wave, the collaborative control module needs to extract its waveform characteristics at specific spatial nodes. From the perspective of fluid dynamics, pressure waves in a fluid pipeline network exhibit a physical evolution law of dual attenuation with time and space and multi-end reflection. To provide numerical basis for the counter-attack action of the subsequent pressure follower group, the system reduces the complexity of the spatiotemporal fluid field into discrete key feature parameters. This process specifically includes the following steps.

[0108] S304, the cooperative control module locates the spatial boundary nodes corresponding to the pressure follower units and extracts time-series data within a discretized one-dimensional transient flow characteristic line grid. As a preferred method, the system extracts the pressure follower group set based on a previously established physical grouping array. The physical topological coordinates of the inlet pipe sections of each generating unit's spiral casing are then determined. Subsequently, the collaborative control module extracts the pressure head time series for the corresponding spatial coordinate nodes within the future prediction period from the pipeline transient field calculation matrix. During this mapping process, since the actual physical location of the generating units typically deviates from the discrete nodes of the pipeline differential grid, the collaborative control module uses an inverse spatial distance weighting method to interpolate the calculated head values ​​of adjacent nodes.

[0109] In the formula, The extracted pressure represents the number of people in the group. Taiwanese unit in The equivalent transient pressure head for each time measurement and control step, within this mathematical constraint, is the set of units in this group. It belongs to the physical subset of the aforementioned plant-wide frequency regulation units and has no element intersection with the active disturbance group set; and The characteristic lines representing the head values ​​of adjacent differential grid nodes on both sides of the unit at the same time are used to solve for the head values. and This represents the dimensionless distance weighting coefficient calculated based on the physical distance between the units. To accurately achieve the numerical mapping from discrete grids to physical coordinates, it is assumed that the distance between the corresponding unit and its upstream adjacent grid node is... The actual pipe section length is Distance from downstream adjacent grid nodes The actual pipe section length is Then the above weighting coefficients satisfy the calculation rules. as well as This spatial interpolation mechanism effectively corrects the local truncation error caused by grid discretization.

[0110] S305, the key time points of the transient waveform include the wavefront arrival time and the extreme value occurrence time. To capture the wavefront arrival time, the cooperative control module uses adaptive threshold discrimination logic to scan the instantaneous rate of change of the pressure sequence. When the rate of change of pressure exceeds the steady-state background noise threshold for multiple consecutive calculation steps, the moment of the first exceedance is recorded as the wavefront arrival time. In conventional industrial applications, this steady-state background noise threshold is typically dynamically adjusted by the system by extracting the pressure fluctuation range within a preset time window (e.g., the past 5 seconds) under undisturbed, stable operating conditions. Its typical value is between 0.5% and 1.0% of the steady-state operating pressure. Furthermore, the collaborative control module uses a central difference algorithm to capture the extreme moments of the waveform, and the system calculates the first and second derivatives of the time series:

[0111] ;

[0112] ;

[0113] In the formula, and These represent the pressure time series at the 1st, 2nd, and 3rd respectively. The first-order central difference and the second-order central difference approximation for each calculation step size; This refers to the global fundamental computation time step during the characteristic line method operation. (This is determined by the collaborative control module.) The value enters the tolerance band close to zero and When the value is less than zero, mark the predicted future time as the peak arrival time. ;like If the value is greater than zero, it is marked as the trough time. To avoid harmonic interference from multi-terminal hydraulic reflections in the pipeline network, the collaborative control module is equipped with a first-wave locking function, which only captures the wavefront arrival time. The first extreme value is then taken as the valid output. Based on engineering experience and the dissipation characteristics of fluid viscosity, the tolerance band for this extreme value determination is usually set within the range of 0.1% to 0.5% of the steady-state operating pressure. For the specific implementation of the search for extreme points in a continuous discrete sequence, those skilled in the art can use conventional digital signal processing algorithms combined with the loop comparison instructions of a programmable logic controller for configuration. The digital derivative calculation is a well-known technique in this field and will not be elaborated here.

[0114] S306, the collaborative control module extracts the transient pressure head values ​​corresponding to the peak and trough times based on the aforementioned time markers, and defines them as the maximum pressure amplitude of the node, respectively. Minimum bearing capacity of the node To prevent water column separation and cavitation caused by transient negative pressure under low-load conditions, a dynamic differential pressure safety closed-loop verification mechanism is established within the collaborative control module. The system calculates the minimum pressure-bearing amplitude. The safety margin is calculated by subtracting the fluid vaporization pressure head from the local atmospheric pressure. If this margin is lower than the set lower threshold (often 3 to 5 meters of water column height in large run-of-river hydropower stations), the system determines that the current pilot transient wave poses a risk of vacuum interruption. To prevent the look-ahead algorithm from deviating from physical reality and falling into a control dead zone, the cooperative control module triggers an internal loop correction command under this condition, automatically reducing the target guide vane opening command limit value of the preceding active disturbance group by a preset step size ratio (e.g., 5%), and restarting the transient characteristic field calculation. To ensure that this recursive safety verification does not cause program deadlock, the system synchronously monitors the command margin after reduction. When the target power regulation command of the active disturbance group is reduced to below the unit's minimum mechanical regulation dead zone and still cannot meet the above safety margin requirements, the cooperative control module will forcibly terminate the internal loop calculation, directly output a hydraulic over-limit alarm signal, and suspend the current cooperative frequency regulation task, returning to steady-state monitoring mode, thereby fundamentally ensuring the safety of subsequent frequency regulation strategy implementation.

[0115] In this embodiment, after capturing the key spatiotemporal characteristics of the leading transient wave, the collaborative control module needs to dynamically correct the physical parameters of the underlying pipeline network based on measured data. From the perspective of the physical evolution characteristics of the fluid medium, the theoretical wave velocity and initial friction coefficient set in the preliminary steps are mostly based on an ideal bubble-free pure water model and steady-state flow assumptions. In the actual transient evolution process, with the drastic fluctuations in local pressure, trace amounts of dissolved gas in the water are released, and the rapid changes in flow velocity trigger strong additional shear stress, causing the actual physical parameters to deviate from the initial static setpoints. To compensate for the calculation phase and amplitude errors caused by parameter drift, the system performs dynamic reconstruction of physical parameters based on measured boundaries. This process specifically includes the following steps.

[0116] S307, the collaborative control module calculates the actual transient water hammer wave velocity of the pipeline network based on the captured time stamp. As a preferred method, the system extracts the initial action time of the substantial mechanical displacement of the guide vanes in the active disturbance group and, combined with the wavefront arrival time recorded in previous steps, calculates the absolute propagation time of the pressure wave in the actual physical space of the pipeline network. Subsequently, the collaborative control module extracts the total length of the hydraulic pipeline topology path between the active disturbance group unit nodes and the pressure following group unit nodes, and constructs an inversion model of the actual water hammer wave velocity based on this. In the formula, The inversion-derived transmission to pressure follows the group's first... The actual transient water hammer wave velocity of the unit; Represents the active perturbation group The node where the unit is located extends along the water diversion pipeline to the aforementioned section. The total axial length of the physical piping between the nodes where the unit is located; The wavefront arrival time locked by the feature capture stage; This represents the timestamp at which the guide vanes actually begin to move after the corresponding unit of the active disturbance group receives the power command. To avoid abnormal minimum values ​​in the denominator caused by delays in telemetry and control communication or synchronization clock drift, the coordinated control module has a time difference lower limit verification threshold. This verification threshold is usually set to the theoretical shortest conduction time, i.e., it satisfies... ,in The initial water hammer wave velocity is the reference value. This is a preset upper limit coefficient for wave velocity distortion (often taken as 1.1 to 1.2 in engineering). When the calculated wavefront propagation time is extremely short (i.e.... If the results do not conform to the fundamental laws of fluid physics, the system will trigger a safety backoff mechanism, temporarily discarding the inversion results for that cycle and continuing to use the initial water hammer wave velocity read from the previous steps. Participate in subsequent calculations.

[0117] S308. Traditional Darcy's steady-state frictional resistance model struggles to characterize the additional energy dissipation induced by dramatic changes in the boundary layer velocity gradient under transient flow conditions. The collaborative control module introduces a one-dimensional unsteady transient frictional resistance attenuation model, dynamically calculating the transient frictional coefficient using measured discrete sequences of pressure and velocity. The system extracts real-time velocity change rate parameters from the pipeline calculation matrix and combines them with initial steady-state frictional baseline values ​​for superposition calculation. In the formula, Representing the The dynamic transient frictional resistance coefficient for each calculation step; This represents the initial Darcy friction coefficient of the pipe section; Representing the dimensionless attenuation coefficient based on Bruno's transient friction theory, its value is usually related to the Reynolds number of the fluid and the absolute roughness of the pipe wall. In conventional industrial large-diameter steel pipe applications, it is mostly calibrated in the range of 0.01 to 0.05. It is the gravitational acceleration constant; This represents the transient velocity of the corresponding spatial node at the current step size; The transient derivative of the velocity sequence with respect to time is calculated by the collaborative control module using first-order differential logic. The minimum positive bias constant preset for the system (e.g., 10) -4 The introduction of the function (m / s) is intended to prevent the denominator from being zero when the fluid is stationary or when the flow rate crosses zero in the reverse direction, which would cause the underlying floating-point division-to-zero overflow to crash.

[0118] S309, in the difference operation framework of the method of characteristics, the time step and the spatial partitioning step must satisfy the CFL condition. When the actual water hammer wave velocity... When drift occurs, the original static mesh faces the risk of divergence. The cooperative control module calculates the real-time Courland number: In the formula, Represents the Courant number; This represents the actual wave velocity after the inversion update; Represents the global fundamental computation time step of the system; Representing the The spatial division step size of the pipeline segment. The collaborative control module determines in real time whether the Courland number is satisfied. The system imposes a mandatory constraint. Once the wave velocity increases to the point that this value exceeds the limit, the system will automatically trigger a mesh adaptive interpolation reconstruction procedure to proportionally reduce the global basic calculation time step. To ensure that the numerical calculation process remains within the convergence region, a safety lower limit threshold is set for the time step, considering the upper limit of the single-task scan cycle of the industrial controller processor (typically aligned with its underlying hardware clock cycle, such as 1ms to 5ms). If the reduction in time still fails to meet the stability condition even when it reaches the lower limit, the collaborative control module will lock the time step and instead increase the spatial partitioning step of the relevant computational grid. Perform a spatial dimension relaxation reconstruction to prevent the controller from triggering a watchdog reset anomaly due to excessive computational load.

[0119] In this embodiment, after completing the real-time dynamic inversion of physical parameters, the collaborative control module resets the parameters of the underlying characteristic line method calculation model using the corrected actual water hammer wave velocity and transient friction coefficient. In the hydraulic calculation of the hydropower station's transient process, advanced prediction based on high-precision boundary conditions can provide a forward-looking decision-making benchmark for the subsequent action sequence of the pressure follower group, thereby avoiding the deterioration of hydraulic conditions caused by non-coordinated actions of multiple units. This process specifically includes the following steps.

[0120] S401, the cooperative control module extracts the updated actual water hammer wave velocity and transient frictional resistance coefficient from the previous step, and discretizes and resets the compatibility equations of the positive and negative characteristic lines of the one-dimensional transient flow. Conventional fixed-parameter MOC algorithms, when dealing with complex disturbances, often underestimate the attenuation effect due to the lack of consideration for unsteady frictional resistance, leading to predicted phase distortion. From a hydraulic perspective, drastic changes in flow velocity during transient processes alter the shear stress distribution of the pipe wall boundary layer, thus dynamically affecting the wave impedance of the water flow. Therefore, the cooperative control module establishes a set of discrete difference equations containing a dynamic damping attenuation factor on the reconstructed adaptive spatiotemporal difference grid:

[0121] ;

[0122] ;

[0123] ;

[0124] ;

[0125] In the formula, and These represent known numerical values ​​passed to constants corresponding to the positive and negative characteristic lines, respectively. and Representing the At each computational step, the upstream adjacent spatial grid nodes Pressure head and water flow rate; and Represents the corresponding downstream adjacent node Water head and flow rate; and This represents the dynamic hydraulic impedance coefficient that incorporates transient physical variables; This represents the actual wave velocity after the inversion update; The dynamic transient frictional resistance coefficient represents the current step size; and Representing the first The cross-sectional area and inner diameter of the water diversion pipeline section; This represents the corresponding spatial grid step size; This is the acceleration due to gravity.

[0126] Based on the updated transport constant and impedance coefficient, the collaborative control module calculates the target node. Physical state parameters at future moments:

[0127] ;

[0128] ;

[0129] In the formula, and Representing the first Target nodes on the pipeline In the The predicted pressure head and transient flow rate are calculated in steps. The resetting of this equation set allows the computational model to more objectively reflect the energy dissipation law of fluid under complex forced vibration.

[0130] S402, to quantify this impact, the system uses a reduced global base computation time step. As an iterative benchmark, the fluid state is rapidly extrapolated within a virtual spatiotemporal domain. During this extrapolation process, the cooperative control module actively disturbs the unit (set of units). The actual guide vane action sequence is substituted as the upstream forced boundary condition, and the pressure is forced to follow all units within the group (set). The guide vane opening remains constant at the current steady-state value in the prediction model. Through this asymmetrical boundary constraint setting, the cooperative control module can separate the background disturbance pressure component and calculate a predicted time window. The background predicted pressure sequence within ,in This refers to the discrete prediction time steps within the prediction time window. As a preferred approach, this prediction time window... The span is usually set to the longest water hammer phase period of the pipeline network (i.e., satisfying expression 4). ,in The length of the pipeline is 2 to 3 times the physical length of the longest continuous pipe segment along the wave propagation path in the pipeline network, to completely cover the initial reflection envelope of the transient wave. For the ultra-real-time accelerated execution of the underlying matrix operations, those skilled in the art can use a conventional multi-core digital signal processor combined with a field-programmable gate array (FPGA) hardware pipeline mechanism for parallel compilation. The underlying register allocation and multiply-accumulate operation acceleration are well-known technologies in the field and will not be described in detail here.

[0131] S403, The collaborative control module scans the predicted sequence within this time domain. Extraction pressure follows the group's first The predicted maximum background pressure that the Taiwanese generator set will soon experience With minimum background pressure The system compares these extreme values ​​with the structural yield strength limits of the corresponding spatial nodes in the pipeline execution module and the lower limit of vacuum to prevent water column separation. If the prediction verification finds that, without any counterbalancing action from the pressure following group, the background transient wave generated and transmitted solely by the active disturbance group is sufficient to cause local node pressure to exceed the limit, it indicates that the system's pressure-bearing capacity under the current dynamic pipeline parameters has reached its physical limit.

[0132] To prevent the system from falling into control deadlock under such extreme conditions due to the inability of normal frequency regulation logic to converge, the collaborative control module is equipped with hardware-level extreme condition protection logic. Once the predicted over-limit state is identified, the collaborative control module will immediately block subsequent conventional offset calculations and issue an emergency collaborative control command to the generator unit corresponding to the pressure over-limit node. This will force the unit's bypass drain valve or guide vane actuator to return to a safe opening range, artificially disrupting the superposition energy of the water hammer reflection wave. Simultaneously, the system sends a primary frequency regulation capability degradation alarm frame to the higher-level power grid dispatch center, switching the control objective from adequately responding to the power grid frequency deviation to ensuring the mechanical safety of the hydraulic system. This ensures the safe operation of the power generation equipment under severe hydrological conditions at the underlying logic level.

[0133] In this embodiment, after confirming that the background hydraulic environment is within a safe boundary, the collaborative control module needs to calculate the mechanical opening command for the pressure following group to execute specific actions. Because the relationship between turbine flow rate, head, and power is a complex spatial surface, the hydro-mechanical coupling link of the turbine-generator unit exhibits nonlinear characteristics. Conventional linear proportional gain allocation cannot adapt to power approximation under conditions of severe head fluctuations during transient processes. Therefore, the system introduces an inverse matching mechanism based on energy mapping. This process specifically includes the following steps.

[0134] S404, the collaborative control module calculates the target active power regulation increment that each generator unit in the pressure-following group needs to undertake. The system extracts the total primary frequency regulation power compensation amount issued by the grid for the entire plant, and deducts the total regulation power already allocated by the preceding active disturbance group to obtain the remaining active power deficit to be compensated. As a preferred method, the collaborative control module uses the pressure-following group set... The static adjustable capacity margin of each internal unit is allocated proportionally to the power deficit: In the formula, The representative distributed the information to the pressure-following group. The target active power regulation increment for the unit; This represents the total primary frequency regulation power compensation amount calculated by the system for the entire plant; This represents the actual amount of regulation undertaken by each unit within the active disturbance group; and These represent the current static adjustable capacity margin of the corresponding generating units. Considering that primary frequency regulation has bidirectional adjustment attributes, this capacity margin is dynamically defined according to the direction of frequency deviation: when the grid frequency decreases and active power generation needs to be increased, Take the unit's operating upper limit threshold and subtract the current real-time output; when the grid frequency increases and active power generation needs to be reduced. Take the current real-time output minus the lower limit threshold for stable unit operation; in the formula... For a preset small positive real constant to prevent division by zero (e.g., 10) -5 This is used to prevent processor-level division-by-zero errors caused by a lack of adjustment margin in the pressure following group. When the total margin of the denominator is detected to be less than the set dead zone, the cooperative control module determines that the pressure following group's capacity is exhausted, at which point... Automatically reset to zero and report the frequency modulation limitation status to the host computer.

[0135] After clarifying the target active power regulation increments for each unit, and considering that transient head fluctuations will change the turbine's work capacity in real time, the system utilizes a pre-set nonlinear opening degree inverse matching neural network model for mapping calculation. This model employs a multilayer perceptron architecture, with an internal hierarchical structure including an input layer, two fully connected hidden layers, and an output layer. Modules are fully connected via weighted neural synapses, and data flows forward. To achieve nonlinear fitting of hydraulic characteristics, the activation functions of both fully connected hidden layers are configured as linear rectified functions, while the output layer uses a linear activation function to output continuous physical opening degree values.

[0136] In the context of frequency regulation in hydropower stations, the collaborative control module defines the input data as a four-dimensional feature vector. .in, The target active power adjustment increment calculated above; The equivalent transient pressure head at the current moment is extracted; The predicted maximum background pressure obtained from the advanced forecasting process is used to characterize the extreme value of head energy distortion in the near future. This represents the initial steady-state opening of the guide vane. The preprocessing logic uses a maximum-minimum normalization algorithm to map the physical quantities of each dimension to... Numerical range. The model output is a one-dimensional numerical value. In its specific business sense, it represents the additional guide vane opening adjustment increment necessary to achieve the target power increment.

[0137] To construct and apply this neural network model, the system needs to complete the training step offline. Training samples and labels are derived from a historical dynamic operating condition database of power regulation of similar generator sets, and a transient process dataset simulated using one-dimensional feature line method fluid dynamics software. The collaborative control module extracts the synchronously recorded discrete sequences of head, power, and opening degree as the training set. Regarding key parameters, the loss function uses the mean squared error function to measure the residual between the predicted opening degree and the actual action label. The optimizer uses an adaptive moment estimation algorithm to update the inter-layer weights of the network, with an initial learning rate set to 0.001 and supplemented by a learning rate decay strategy. This engineered training mechanism enables those skilled in the art to reproduce the parameter solidification process of this mapping model based on the unit's factory measured characteristic curves and historical waveform data.

[0138] After outputting the guide vane opening adjustment increment through the aforementioned neural network model, the system performs instruction synthesis and execution boundary protection verification to ensure the mechanical safety of the actuator. The cooperative control module then calculates the current steady-state opening. The opening increment output by the above model The absolute value of the target guide vane opening command is obtained by superimposing the numerical values. To prevent the neural network model from outputting abnormal extreme values ​​when processing singular samples outside the training set boundaries, the collaborative control module is equipped with mechanical limit hard verification logic. The system judges... Is it within the safe physical travel range of the guide vane actuator? Inside, among which and These correspond to the lower limit of the unit's no-load operation and the physical extreme value of its full-load operation, respectively. Once this calculated value exceeds the limit, the system will forcibly restrict it to the corresponding upper and lower limit boundary values.

[0139] In this embodiment, after calculating the mechanical opening command of the pressure following group, the collaborative control module needs to determine the specific timing of the command's issuance and physical execution. From the perspective of the superposition mechanism of the hydraulic transient process, when the turbine generator unit adjusts the guide vane displacement, its own actions inevitably generate secondary transient pressure waves. If the timing of the actions of the master and slave units is not constrained, the secondary waveform is very likely to overlap with the leading transient wave transmitted from upstream, leading to pressure exceeding limits at local pressure-bearing nodes and endangering the safety of the pipeline network structure. Therefore, the system implements a phase-shift delay control strategy based on transient waveforms, aiming to effectively offset and reduce the peak of water flow energy by utilizing the phase difference between the two physical waveforms. This process specifically includes the following steps.

[0140] S501, the collaborative control module formulates a phase shift matching strategy based on the physical antagonistic relationship between the generator set's adjustment direction and waveform extreme values. As a preferred approach, the system extracts the previously calculated target active power adjustment increment. The direction of the sign. Specifically, when When this occurs, it indicates that the unit is in a load-increase regulation condition. Opening the guide vanes will increase the water flow area, thereby generating a momentary pressure-reducing reflected wave along the pipeline network. To achieve reverse energy counteraction, the timing of this pressure-reducing effect should coincide as closely as possible with the peak of the upstream pilot wave (i.e., the maximum positive pressure). When the unit is operating under reduced load, the closed guide vanes will compress the fluid, generating a pressure-boosting reflected wave. The corresponding physical counter-impact target point will then be converted into the trough of the pilot wave (i.e., the minimum negative pressure). Through this polarity determination mechanism based on physical flow direction, the collaborative control module locks the target counter-impact timestamp in the time-domain sequence. That is, to meet the requirements under increased load conditions Under reduced load conditions, it meets the requirements. In the formula, and These are the arrival times of the extreme values ​​of the leading waveforms identified and saved by the preceding feature capture stage.

[0141] S502, due to the inherent mechanical static dead zone of the guide vane actuator and the fact that changes in water inertia require a certain accumulation time, the control command must be issued before the expected physical counter-current moment. The collaborative control module constructs the equation for calculating the action delay time:

[0142] ;

[0143] ;

[0144] In the formula, The representative followed the pressure and followed the group's first The relative waiting delay before the machine unit issues an action command; The timestamp for the aforementioned locked physical targets; The system's absolute timestamp represents the current control cycle; This represents the combined static response dead time of the unit's governor system and electro-hydraulic servo valve. This value can be determined by the unit's factory load rejection test and pre-fixed in the system. Conventional large hydropower units are mostly calibrated between 0.1s and 0.3s. The theoretical ramping time representing the guide vane reaching the target opening command; This is the absolute value of the guide vane opening adjustment increment calculated by the preceding stage; This is the limit value for the mechanical action rate of the guide vanes set by the system under the current head conditions. Subtracting half of the ramp time on the time axis has the physical meaning of aligning the center point of the maximum water flow disturbance energy generated during continuous operation approximately with the extreme position of the pilot waveform to obtain the best crushing and suppression effect.

[0145] Considering the timing overlap issues that can arise from communication network delays or excessively fast computation speeds in practical engineering, the system synchronously performs algorithm boundary checks after calculating the delay time. If the calculated... This indicates that the system's command issuance time is delayed, physically missing the optimal offset phase of the initial waveform. Under such conditions, the cooperative control module will trigger phase delay logic, postponing the target offset timestamp to the next water hammer reflection waveform cycle based on the inherent oscillation and reflection characteristics of the hydraulic network. The system will then perform a reset update. ,in This refers to the length of the longest continuous physical pipe segment along the pipeline network. To retrieve the updated actual wave velocity, the new value is substituted back into the above calculation equation. To prevent the control program from endlessly iterating phase shifts after the transient wave energy has completely decayed, the cooperative control module has a maximum delay period threshold (e.g., set to delay 3 times). When the delay number reaches this upper limit threshold and the output delay still does not meet the positive value constraint, the system determines that the effective interference waveform has dissipated and then forcibly... Assigning a value of zero automatically terminates the delay process. This exit mechanism effectively prevents the algorithm from falling into a logical dead zone of infinite loop.

[0146] S503, the cooperative control module will include the absolute value of the target guide vane opening command. Relative waiting delay time The coordinated action data frames are sent to the corresponding programmable controller of the pipeline execution module via the industrial Ethernet bus. After receiving the instruction, the underlying controller of each unit starts the local hardware timing interrupt service, and triggers the electro-hydraulic converter to drive the guide vane servo cylinder at the moment the countdown reaches zero.

[0147] In this embodiment, after the countdown of the action delay time issued by the collaborative control module reaches zero, the underlying controller formally triggers the mechanical action of the pressure-following unit. Analyzing from the perspective of fluid mechanics operation, during hydraulic transients, the guide vane hydraulic torque is not a constant value, but rather exhibits strong nonlinear time-varying characteristics due to the dynamic evolution of Bernoulli effect and wake vortex shedding, caused by the drastic fluctuations in flow rate and head within the blade channel. Traditional constant-speed or linear trajectory servo tracking is prone to mechanical jamming or control overshoot due to sudden force changes, thereby inducing secondary hydraulic excitation. To ensure smooth operation and trajectory matching, the system executes a nonlinear guide vane trajectory servo tracking strategy, which specifically includes the following steps.

[0148] The S504's underlying controller reconstructs a smooth, nonlinear motion trajectory based on the target guide vane opening command. As a preferred approach, the system uses a fifth-order polynomial S-shaped velocity curve to plan the guide vane position command sequence, ensuring the continuity of opening, velocity, and acceleration at the start and end times and avoiding mechanical shocks caused by sudden acceleration jumps. The constructed nonlinear position reference trajectory equation is as follows:

[0149] In the formula, Representative at Reference position of guide vane opening generated at any time; This represents the initial steady-state opening of the guide vane. This represents the absolute value of the target guide vane opening command calculated and issued by the front-end collaborative control module. This represents the total time taken for the guide vane to climb the slope. This represents the relative execution time of the action after the delay ends, and its numerical range is strictly limited to within [a certain range]. Inside.

[0150] S505, the underlying controller collects the actual opening feedback value from the guide vane displacement sensor in real time. And calculate the current position tracking error. and the rate of change of error To address the time-varying parameter problem of the controlled object caused by nonlinear hydraulic torque, an adaptive PID control algorithm is introduced into the underlying controller. The system dynamically tunes the proportional and derivative gains in the control loop based on the absolute magnitude of the error.

[0151] ;

[0152] ;

[0153] ;

[0154] In the formula, This represents the integrated control voltage signal output by the controller to the electro-hydraulic converter; and , These represent the basic proportional gain, integral gain, and differential gain constants, respectively. Represents the dynamic proportional gain that varies nonlinearly with position error; The dynamic differential gain represents the nonlinear variation of the error rate. and These represent the nonlinear compensation coefficients for proportional gain and differential gain, respectively. and These represent the shape factor constants that determine the nonlinear adjustment rates of the two, respectively. This represents the cumulative amount of error over time. The aforementioned compensation coefficients and shape factor constants are typically engineered through unit no-load disturbance tests to obtain empirical values ​​that match the characteristics of the current governor hydraulic system.

[0155] Based on the above Taking the tuning rule as an example, when the position error When the value is large, the exponent term tends to zero. Rapidly increased to nearly To provide a larger control output to overcome static frictional resistance and sudden changes in hydraulic torque; when the error shrinks to near zero, Smoothly fall back to the base value This prevents mechanical overshoot or high-frequency oscillations from occurring near the target opening.

[0156] The S506 servo valve's permissible drive current and the guide vane servo's operating speed are both limited by physical constraints imposed by the hardware manufacturing process. When the control signal... When abnormally high amplitude spikes occur due to severe transient hydraulic disturbances, the underlying controller limits them to the saturation limit range of the electro-hydraulic servo valve. Internally, to prevent the integrator inside the controller from continuously accumulating errors under output limiting conditions, which could lead to severe desaturation delay when the error reverses, the system is equipped with a conditional integral separation mechanism. When the calculated absolute value of the control command... Reaching the upper limit boundary And error and When the signs are consistent, the controller automatically freezes the accumulation operation of the integral term, retains the current integral state value, until the error sign is reversed or the control signal autonomously exits the saturation limiting region.

[0157] In this embodiment, after the aforementioned phase-shift delay triggering and nonlinear trajectory following control, all the hydro-generator units participating in primary frequency regulation in the plant achieve staggered actions in both physical space and time. This asynchronous guide vane adjustment causes the mechanical power generated by each unit to exhibit differentiated time-domain characteristics, i.e., forming heterogeneous mechanical power responses. To verify whether this asynchronous regulation action based on internal hydraulic transient counterbalancing meets the overall primary frequency regulation power assessment requirements of the external power grid, the collaborative control module needs to perform simulation and closed-loop evaluation of the aggregation process of electrical power from multiple units within the plant. This process specifically includes the following steps.

[0158] The S507 collaborative control module constructs the dynamic calculation logic for the electrical energy conversion of a single-unit water turbine. The mechanical power generated by the turbine runner under dynamic water flow impact exhibits complex nonlinear changes due to the real-time influence of drastic fluctuations in head and flow rate. Simultaneously, the generator rotor, due to its large moment of inertia, experiences an inherent physical inertial delay during the conversion of mechanical power to electromagnetic power, resulting in a first-order inertial lag characteristic in the electrical power output. This provides a low-pass filtering effect for high-frequency power fluctuations. The collaborative control module establishes the discrete difference equations for single-unit power conversion:

[0159] ;

[0160] ;

[0161] In the formula, Representing the Taiwanese crew The transient mechanical power calculated at time t; 9.81 is the product constant of the gravitational density of water and gravitational acceleration; and These represent the actual flow rate and working head flowing through the runner at that moment, respectively. In engineering practice, the above two high-frequency transient values ​​can be calculated in real time through the high-speed sampling sequence of the volute differential pressure sensor and the tailrace pressure sensor. This represents the overall conversion efficiency of the corresponding turbine under the current operating conditions. This efficiency value can be obtained by interpolation of the steady-state comprehensive characteristic curve matrix of the unit at the factory. and These represent the stator-side output electrical power of the single machine at the current moment and at the previous calculation step, respectively. The step size for discrete calculation of the system; This represents the comprehensive rotor inertia time constant of the generator unit. Through the above calculations, the system can quantify the actual electrical active power response trajectory of a single unit under complex water hammer disturbances.

[0162] S508: The electrical energy generated by multiple generator units within the power plant needs to be transmitted to the common connection point (CUP) connecting to the external power grid via the plant's AC bus, main transformer, and transmission circuit. During this aggregation process, the heterogeneous electrical power generated by each unit based on different phase shift delay commands is superimposed on the same physical node in a time sequence. The coordinated control module, based on Kirchhoff's current law and the power flow distribution of the electrical network, establishes the comprehensive active power synthesis equation at the grid connection point: In the formula, Representative at The total electrical active power actually delivered to the power grid at the grid connection point at any given time; Represents the total number of generator units connected to the grid throughout the plant; This refers to the electrical power output of a single unit. Representing the The sum of real-time active copper losses and iron losses on the generator set, its corresponding step-up transformer, and connecting cables can be dynamically calculated based on the product of the real-time current square and the equivalent impedance collected by the electrical measuring devices of each branch. This represents the total active load consumed by the plant's public power system at that moment. The algebraic superposition of this heterogeneous power at spatial nodes allows the power dips caused by reverse regulation or delays in some units to be filled by the power increments of other units in the main regulation phase. By utilizing the complementary characteristics of peaks and troughs, the total power response that meets the grid dispatch requirements is output.

[0163] S509, the collaborative control module extracts the target power command for primary frequency regulation of the entire plant issued by the power grid. And calculate its combined power with the grid connection point. The real-time power residual between the two. If the absolute value of this power residual is continuously greater than the dead zone threshold allowed by the primary frequency regulation assessment within the set steady-state judgment time window (preferably, the window length can be set to 10s to 15s), the collaborative control module determines that the current phase shift delay offset strategy has resulted in insufficient power response depth. At this time, the system will generate a fine-tuning compensation command and allocate it to the units not in the extreme limit state according to the available capacity margin for secondary opening correction. At the same time, if it is detected that all online units in the plant have reached the physical limit of the guide vanes or the power output limit, the system will automatically terminate the issuance of the fine-tuning command and report the frequency regulation capacity depletion warning status to the automatic generation control station of the power grid to prevent the controller from entering the dead zone of invalid integral saturation.

[0164] In this embodiment, after the aforementioned heterogeneous hydroelectric and mechanical actions are executed, the system needs to perform a closed-loop verification of the actual power response of the entire hydropower station. Due to measurement noise and unmodeled dynamic interference present in actual power grid operation, the system executes a real-time monitoring and residual calculation program for the total power of the entire plant to provide a quantitative basis for subsequent power compensation. This process specifically includes the following steps.

[0165] Step S601: The system synchronously collects and processes electrical quantity data at the grid connection point. The collaborative control module acquires real-time instantaneous values ​​of three-phase voltage and current through a high-precision phasor measurement device deployed at the substation's common connection point. Based on instantaneous power theory, the system multiplies and sums the voltage of each phase with the corresponding phase current to calculate the original active power measurement value. Considering the electromagnetic interference on site and the high-order harmonics generated by power electronic equipment such as frequency converters, the original measurement signal is usually accompanied by high-frequency jitter. To separate the power envelope that truly reflects the low-frequency characteristics of the turbine's mechanical regulation, as a preferred method, the system introduces a first-order digital low-pass filter algorithm to smooth the original data:

[0166] ;

[0167] In the formula, Representative at The actual effective power value at the grid connection point after filtering at any given moment; This represents the calculated value of the raw active power obtained in the current sampling period; This represents the effective power value of the previous sampling period; The step size for discrete calculation of the system; The filter weighting coefficient ranges from (0,1) and is typically tuned to around 0.1 to 0.3 based on a trade-off between the power grid noise spectrum characteristics and control response requirements. This filtering mechanism helps to reduce interference from high-frequency measurement noise while preserving the low-frequency response characteristics of primary frequency modulation.

[0168] S602, the collaborative control module extracts the total target power for the entire plant in the current frequency regulation cycle. This target power is composed of the initial steady-state reference power before frequency regulation and the total active power compensation required by the system for primary frequency regulation. The system constructs the power residual evaluation equation: In the formula, Representative at The residual power response of the entire plant calculated at each moment; This represents the current steady-state reference output issued by the automatic generation control system of the power grid; This represents the total primary frequency regulation power compensation amount for the entire plant, as determined by previous calculations. Representative at The actual effective power value at the grid connection point after filtering at any given time. The positive and negative polarities and absolute values ​​of the value can intuitively reflect whether the overall active power response of the hydropower station is in a state of under-generation or over-generation, as well as the specific physical difference from the target command.

[0169] S603: The system needs to perform dead zone determination and time window integration evaluation based on residuals. Simple transient residuals are prone to misjudgment due to hydraulic delays. The collaborative control module is configured with steady-state determination logic based on energy integration. The system pre-sets the power residual assessment dead zone threshold. This threshold is typically taken as 0.5% to 1% of the plant's rated installed capacity. When the absolute value of the transient residual... When the system determines that the current actual power has met the grid connection assessment indicators, it clears the internal accumulator and maintains the current mechanical opening of each unit unchanged. At that time, the system initiates the sliding time window integration function: In the formula, Representative in the past The power residual integral energy value within the time window; The set evaluation time window length is usually taken as 1 to 2 times the decay period of the dominant water hammer wave to avoid interference from the initial fluctuations of the water hammer. For integral infinitesimal elements The transient power residual at time t. When the calculated integral energy value... Exceeding the preset fault tolerance energy threshold upper limit At that time, the collaborative control module determined that there was an unrecoverable steady-state power deviation throughout the plant. Among these, The tuning is based on the dead zone threshold and the time window length, and its calculation formula is as follows: The tolerance margin coefficient is typically set to be between 1.2 and 1.5.

[0170] Once an out-of-limit condition is detected, the collaborative control module immediately generates an out-of-limit trigger signal and latches the current residual value for subsequent compensation. Simultaneously, to ensure system operational safety, the system has a built-in timeout circuit breaker mechanism: if the out-of-limit state lasts longer than the set maximum fault waiting time (e.g., 30 seconds), the system will determine that there is a serious fault in the field physical actuator or power grid measurement circuit, automatically terminate the residual integration program, and issue the highest-level frequency modulation failure alarm to the central control room, thereby preventing the algorithm from falling into a logical dead zone of continuously requesting compensation.

[0171] In this embodiment, when the total power monitoring system determines that there is a persistent and non-recoverable over-limit residual between the actual output power and the target command, the system triggers a dynamic fine-tuning and frequency modulation closed-loop mechanism for the power reference. This mechanism eliminates steady-state power deviation by performing secondary corrections on the mechanical opening of some units without disrupting the previous hydraulic counterbalancing transient equilibrium. This process specifically includes the following steps.

[0172] S604, the collaborative control module calculates the total active power compensation for the entire plant based on the previously latched deviation data. The system extracts the transient power residuals latched by the upstream monitoring link and the corresponding time window integral energy values ​​to construct the comprehensive compensation command calculation equation: In the formula, This represents the total plant power compensation increment command generated for the current frequency regulation cycle; This represents the transient power residual value latched by the previous monitoring link when the over-limit state was triggered; This represents the residual integral energy value latched at the corresponding time point; and These represent the proportional gain and integral gain preset by the compensation controller, respectively. As a preferred method, to avoid overshoot oscillation caused by secondary regulation, The value typically ranges from 0.1 to 0.3, while The delay time is then set between 0.01 and 0.05 based on the allowable delay time of the power plant. By integrating transient deviations and integral trends, this compensation calculation logic helps generate relatively smooth power fine-tuning commands.

[0173] S605. Considering that after the initial phase-shift delay and nonlinear trajectory following adjustments, the physical opening positions of the guide vane servo mechanisms of each unit in the plant are different, with some units approaching the adjustment dead zone or full-load limit, the collaborative control module executes a dynamic weighted allocation strategy based on the current online status and available capacity margin of the units.

[0174] To ensure the logical completeness of the allocation algorithm and prevent errors in division by zero calculations, the system first iterates through all available units and calculates the total plant regulation margin. For the first... The unit calculates its effective capacity margin in the current adjustment direction. Under increased load conditions, it equals the difference between the unit's maximum rated output and the current actual output power; under decreased load conditions, it equals the difference between the current actual output power and the minimum steady-state output limit. If the total plant margin is detected... Less than the required absolute compensation amount Or the number of units currently capable of regulation This indicates that the plant's remaining adjustable capacity has been exhausted. At this point, the coordinated control module automatically stops the execution of the allocation program, maintains the current command status of each unit, and simultaneously reports the frequency regulation capability limitation flag to the external power grid dispatching master station.

[0175] If the total margin verification passes, the system constructs a weight allocation equation to proportionally split the total instructions:

[0176] ;

[0177] ;

[0178] In the formula, Representative assigned to the first Compensation power weighting coefficient for the unit; This represents the total number of online generating units that currently have the ability to continue adjusting and have not yet reached their mechanical limits; The calculated result is sent to the first The single unit active power fine-tuning command of the generator set; This represents the total plant power compensation increment command generated for the current frequency regulation cycle.

[0179] S606: Due to the highly nonlinear characteristics of the turbine's flow and power transmission, a fixed proportional conversion would lead to operational distortion under low load or high head conditions. The underlying controller receives individual unit active power fine-tuning commands and maps them to the guide vane servo motor's displacement increment based on the turbine's nonlinear transmission characteristics.

[0180] ;

[0181] ;

[0182] In the formula, This represents the calculated guide vane opening compensation correction value; This represents the local incremental rate of power opening of the turbine under the current head and guide vane opening conditions. This value reflects the sensitivity of the change in mechanical opening to power output at the current specific operating point. It can be obtained in real time by referring to the partial derivative table of the turbine comprehensive characteristic surface stored in the controller memory. The current actual steady-state opening is fed back by the guide vane displacement sensor; This is the final superimposed correction command for the guide vane target. Through the above compensation calculation, weighted allocation, and inverse mapping of physical opening, the system can take into account the closed-loop requirements of both internal hydraulic conditions and external electrical response evaluation indicators.

[0183] Specific application examples:

[0184] A large-scale diversion-type hydroelectric power station adopts a one-pipe, four-unit layout. A main 800m long common pressure steel pipe connects to branch pipes, each linking to one of the four turbine-generator units. Each unit has a rated power of 50MW, with a total installed capacity of 200MW and a rated head of 100m. In terms of pipe network topology, the branch pipes are relatively close to units 1 and 2, resulting in strong hydraulic coupling; while the branches to units 3 and 4 are farther away, located at the end of the pipe network.

[0185] At a certain moment (t=0), a large-capacity load surge occurs in the external power grid, causing the system frequency to spike from 50.00Hz to 50.15Hz (exceeding the dead zone threshold of 0.05Hz).

[0186] According to the control method of the present invention, the system operates as follows:

[0187] The state-aware module detected a frequency increase to 50.15Hz and used the formula... The real-time frequency deviation was calculated. =0.15Hz, and the frequency change rate exceeded the limit based on the differential algorithm. The collaborative control module calculated that the plant needs to reduce active power by 12MW. Based on the plant's hydraulic coupling matrix and the current potential energy of the units, the system comprehensively calculated the topology grouping index. Units 1 and 2, which are closer to the branch pipe, are classified as the active disturbance group, while Units 3 and 4, which are at the end of the pipeline network, are classified as the pressure following group.

[0188] The controller immediately issued shutdown commands to Units 1 and 2, reducing power output by 3MW each. The guide vanes of Units 1 and 2 quickly closed, generating a pressure-boosting water hammer wave (pilot transient wave) within the main steel pipe. The system calculated the actual water hammer wave velocity of this waveform to be approximately 1050 m / s.

[0189] Using the method of characteristics to predict in advance, the rising pressure wave will... The energy reaches the spiral casings of Units 3 and 4 at a time of seconds. To achieve energy offsetting, the controller calculates the phase shift delay of Units 3 and 4 (assumed to be 0.15 seconds). At 10:00, Units 3 and 4 began executing shutdown actions according to the fifth-order polynomial nonlinear trajectory calculated by the neural network. At this time, the local pressure rise generated by the closure of the guide vanes of Units 3 and 4 precisely filled the trough of the attenuation of the pilot pressure wave, avoiding the superposition of pressure wave peaks caused by the simultaneous shutdown of multiple units.

[0190] After the operation was completed, the measured combined electrical power output at the grid connection point decreased by 11.5MW, and the system passed the equation. The calculated residual is 0.5MW. The controller is evaluated through integration. The steady-state limit was exceeded. The controller triggered fine-tuning through integral evaluation, issued fine-tuning instructions to Unit 1, and finally reduced the total power of the plant to the target value, and the grid frequency smoothly returned to the dead zone.

[0191] Experimental verification and effect comparison:

[0192] To verify the effectiveness of the present invention, a comparative experiment was conducted between the intelligent collaborative control mode of the present invention and the traditional synchronous control mode with the same frequency and modulation in the same one-pipe four-machine hydraulic transient process simulation platform.

[0193] See attached document Figure 3 Comparison of transient pressures on the volute:

[0194] In the traditional synchronous mode, the guide vanes of all four units close rapidly at the same time, resulting in four water hammer waves superimposed in the same direction at the common branch pipe. The peak value of the maximum transient pressure in the spiral casing soars to 135% of the rated head, approaching the pipeline's pressure limit and posing a significant risk of pipe burst.

[0195] The collaborative mode of this invention: Due to the implementation of phase shift delay and nonlinear trajectory following, Units 1 and 2 move first, followed by Units 3 and 4 with smooth trajectories. Transient waveforms are spatially crushed and offset, and the maximum peak pressure of the spiral casing is only 115% of the rated head. The peak pressure is reduced by 20%, improving the safety of the hydraulic network.

[0196] See attached document Figure 4 Comparison of active power response at grid connection points:

[0197] Traditional synchronous mode: Due to the severe water hammer effect (water hammer reverse regulation) caused by the simultaneous operation of multiple units, the power does not immediately decrease due to the sudden increase in water pressure at the beginning of the guide vane closure. Instead, there is a reverse surge lasting up to 1.5 seconds (power reverse regulation phenomenon), resulting in a serious lag in the primary frequency regulation response.

[0198] The cooperative mode of this invention: Due to the electrical synthesis effect of heterogeneous mechanical power, the power reversal period of Units 1 and 2 is neutralized by the delayed initial state of Units 3 and 4; when Units 3 and 4 enter the reversal period, the power of Units 1 and 2 has substantially decreased. The total active power of the entire plant achieves a smooth decrease, eliminating the reversal phenomenon and shortening the response time by approximately 1.2 seconds.

Claims

1. A method for intelligent collaborative control of primary frequency regulation of a hydropower station governor, characterized in that, Includes the following steps: The hydraulic coupling matrix of the entire plant is constructed based on the pipeline network parameters, and the boundary conditions and initial water hammer wave velocity are calculated using the method of characteristics. Real-time monitoring of power grid frequency and frequency change rate; when the frequency change rate exceeds a preset grouping threshold, the pipeline execution module is divided into an active disturbance group and a pressure following group according to the whole plant hydraulic coupling matrix. The active disturbance group is controlled to generate hydraulic transient waves, the arrival time difference of pressure waves at each node is collected, and the actual water hammer wave velocity and pipe wall friction coefficient are calculated. The actual water hammer wave velocity and the pipe wall friction coefficient are input into the characteristic line method calculation program to generate the transient head time domain envelope of the volute of the unit corresponding to the pressure follower group. The nonlinear guide vane kinematic trajectory data of the pressure follower group are calculated through the transient head time domain envelope. The pressure following group is controlled to perform a phase shift delay operation. The proportional electro-hydraulic servo mechanism of the pressure following group moves according to the nonlinear guide vane kinematic trajectory data and outputs mechanical power. The mechanical power output by the pressure following group and the mechanical power output by the active disturbance group are electrically combined at the grid connection point of the hydropower station to obtain the total active power of the entire plant. The total active power is measured and the target power tracking residual is calculated. The power control reference is adjusted according to the target power tracking residual until the grid frequency enters the dead zone setting range.

2. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, The specific steps for constructing the plant-wide hydraulic coupling matrix based on pipeline network parameters include: Based on the hydraulic elastic water column theory, combined with the fluid bulk elastic modulus and fluid density, the initial water hammer wave velocity of each pipe section is set. Using the cross-sectional area of ​​the main steel pipe as a benchmark, the geometric lengths of branch pipes with different inner diameters are converted into equivalent hydraulic lengths. Calculate the hydraulic coupling coefficient between each unit based on the equivalent hydraulic length between the connected nodes; The hydraulic coupling coefficient between the units is positively correlated with the equivalent hydraulic length of the upstream water intake pipe section shared by the units, and negatively correlated with the equivalent hydraulic length of the independent branch pipe of the unit. Based on the calculated hydraulic coupling coefficients between the units, a symmetric matrix of the corresponding dimension is constructed as the hydraulic coupling matrix of the entire plant.

3. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, Based on the plant-wide hydraulic coupling matrix, the pipeline execution module is divided into an active disturbance group and a pressure follower group, including the following steps: Based on the effective active power regulation margin, rated active power, real-time operating head and rated design head of each online unit, calculate the dynamic operating potential energy coefficient of each online unit; The dynamic operating potential energy coefficient is normalized using the range transformation method, and combined with the hydraulic coupling coefficient extracted from the whole plant hydraulic coupling matrix, the comprehensive topology grouping index of each online unit is calculated. All online frequency modulation units are sorted in descending order according to the comprehensive topology grouping index. Units in the sorted sequence that are at the top of the preset grouping threshold are designated as the active disturbance group, and units in the sorted sequence that are at the bottom of the preset grouping threshold are designated as the pressure follower group.

4. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, Controlling the active disturbance group to generate hydraulic transient waves includes the following steps: The total power compensation required for primary frequency regulation of the entire plant is calculated based on the real-time frequency deviation. The total compensation is then proportionally allocated to each unit in the active disturbance group based on the dynamic operating potential energy coefficient of each unit in the active disturbance group to obtain the target active power regulation increment. By combining real-time working head data, the target active power adjustment increment is inversely calculated into the target guide vane opening command using a bilinear interpolation algorithm; Based on the maximum allowable increment of water hammer pressure calculated by the collaborative control module, the maximum allowable rate of change of the terminal flow is estimated, the mechanical rate limit value of the guide vane action is set, and the rate of change of the target guide vane opening command is dynamically limited.

5. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, The calculation of the actual water hammer wave velocity and pipe wall friction coefficient includes the following steps: The absolute propagation time is calculated by extracting the starting action time of the mechanical displacement of the guide vanes of the active disturbance group and the recorded wavefront arrival time, and the actual water hammer wave velocity is calculated by combining the total length of the hydraulic pipeline topology path. The real-time velocity change rate parameter is extracted from the pipeline calculation matrix. Combined with the initial steady-state friction reference value and the transient derivative of the velocity sequence with respect to time, the dynamic transient friction resistance coefficient is calculated as the pipe wall friction coefficient. The collaborative control module calculates the dynamic Courland number in real time. When the actual water hammer wave velocity increases and causes the dynamic Courland number to exceed the limit, the calculation time step is reduced proportionally to perform grid adaptive interpolation reconstruction.

6. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, The calculation of the nonlinear guide vane kinematic trajectory data of the pressure-following group includes the following steps: Using the actual water hammer wave velocity and the pipe wall friction coefficient, the compatibility equation of the positive and negative characteristic lines of the one-dimensional transient flow is discretized and reset. The active power deficit to be compensated is allocated according to the static adjustable capacity margin ratio to obtain the target active power adjustment increment of each unit in the pressure following group. The target active power adjustment increment, equivalent transient pressure head, predicted maximum background pressure, and current guide vane initial steady-state opening are input into a pre-trained nonlinear opening inverse matching neural network model based on a multilayer perceptron. The model outputs the guide vane opening adjustment increment and combines it with the current guide vane initial steady-state opening to calculate and obtain the target guide vane opening command. Based on the target guide vane opening command, a fifth-order polynomial S-shaped velocity curve is used to generate the guide vane position command sequence, which serves as the nonlinear guide vane kinematic trajectory data.

7. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 6, characterized in that, The control of the pressure follower group to perform a phase shift delay operation includes the following steps: Extract the sign direction of the target active power regulation increment to determine the unit's regulation condition. Under the load increase condition, align the target offset timestamp with the peak of the upstream leader wave. Under the load decrease condition, align the target offset timestamp with the trough of the upstream leader wave. The relative waiting delay before issuing the action command to the pressure follower group is calculated by subtracting the current timestamp, the static response dead zone time, and half of the guide vane ramp time from the target hedging timestamp.

8. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, The proportional electro-hydraulic servo mechanism of the pressure following group operates and outputs mechanical power according to the nonlinear guide vane kinematic trajectory data, including the following steps: Real-time acquisition of the actual opening feedback value from the guide vane displacement sensor; calculation of position tracking error and error change rate. An adaptive PID control algorithm is used to dynamically tune the proportional gain and derivative gain in the control loop based on the position tracking error and the rate of change of error to drive the proportional electro-hydraulic servo mechanism.

9. The intelligent collaborative control method for primary frequency regulation of a hydropower station governor according to claim 1, characterized in that, Measuring the total active power and calculating the target power tracking residual, and adjusting the power control benchmark based on the residual, includes the following steps: The measured total active power is smoothed using a first-order digital low-pass filter algorithm; Calculate the transient power residual between the target commanded power of the power grid and the total active power. When the absolute value of the transient power residual is greater than the set dead zone threshold, start the sliding time window to calculate the residual integral energy value. When the residual integral energy value exceeds the preset fault-tolerant energy threshold upper limit, the total active power compensation of the entire plant is calculated based on the proportional and integral adjustment gain. Extract the current available capacity margin of each unit, and dynamically weight the total active power compensation of the whole plant according to the proportion of the available capacity margin to obtain the individual unit fine-tuning instructions of each unit; Dynamic weighted allocation is performed based on available capacity margin, and the single-machine fine-tuning command is inversely mapped to the guide vane opening compensation correction value in combination with the local micro-increase rate of power opening.

10. A smart collaborative control system for primary frequency regulation of a hydropower station governor, used to implement the smart collaborative control method for primary frequency regulation of a hydropower station governor as described in any one of claims 1-9, characterized in that, include: The pipeline execution module acts on the upstream reservoir, surge chamber, main steel pipe of common pressure, branch pipe and spiral casing of multiple hydro-generator units, and is equipped with microcomputer digital speed governor and proportional electro-hydraulic servo mechanism; The multi-node observation module is distributed along the spatial topology nodes of the common pressure main steel pipe and the branch pipe, and is equipped with dynamic pressure sensors; The state sensing module is located at the grid connection point of the hydropower station and is equipped with a synchronous phasor measurement device to extract the grid connection point frequency and frequency change rate. The collaborative control module establishes data communication connections with the pipeline execution module, the multi-node observation module, and the status perception module, and is equipped with a real-time programmable automation controller.

Images