An adaptive control method and system for a hydroelectric generator set governor

Abstract

The application relates to a water turbine generator speed regulator adaptive control method and system, wavelet decomposition is performed on a comprehensive state signal fused by a first operation parameter of a water turbine generator, and a characteristic vector is extracted; a forgetting factor is combined with the second operation parameter and the characteristic vector, a recursive least square method is used to identify a hydraulic system transfer function, and a hydraulic system parameter is acquired; the hydraulic system parameter and the characteristic vector are taken as inputs, a PID control parameter correction amount is taken as output, a fuzzy controller is constructed, defuzzification is performed, and the PID control parameter correction amount is acquired; according to a control deviation of the correction amount, a reference amount and a third operation parameter, a guide vane opening degree is acquired, and adaptive control is performed on the water turbine generator speed regulator. The application can solve the problem that a PID control algorithm with fixed parameters in the prior art cannot timely adjust a control strategy, easily causes unit oscillation, and even causes a regulation instability problem.

Description

Technical Field

[0001] This application relates to the field of hydropower station automation monitoring technology, and in particular to an adaptive control method and system for a hydropower unit governor. Background Technology

[0002] During the operation of hydropower units, the governor plays a crucial role, and its performance directly affects the stability and power generation efficiency of the unit. Traditional PID control algorithms are widely used in hydropower unit governors; however, the control parameters of these algorithms are usually fixed, making it difficult to adapt to complex and changing operating conditions. For example, when the intensity of water turbulence changes or the unit oscillates, the fixed-parameter PID control algorithm cannot adjust the control strategy in a timely manner, easily leading to unit oscillations or even regulation instability, seriously affecting the safe and stable operation of the hydropower unit. Summary of the Invention

[0003] This application provides an adaptive control method and system for a hydropower unit governor, which solves the problem that the fixed-parameter PID control algorithm in related technologies cannot adjust the control strategy in time, which can easily lead to unit oscillation or even regulation instability, seriously affecting the safe and stable operation of the hydropower unit.

[0004] In a first aspect, embodiments of this application provide an adaptive control method for a hydropower unit speed governor, comprising: Multi-scale wavelet decomposition is performed on the integrated state signal after the fusion of the first operating parameters of the hydropower unit to extract feature vectors characterizing the dynamic characteristics of the hydropower unit. Combining the second operating parameters and feature vector of the hydropower unit, the recursive least squares method with forgetting factor is used to identify the hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters. Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller is constructed. The centroid method is used for defuzzification to obtain the correction values ​​of the PID control parameters. The guide vane opening is obtained based on the correction value of the PID control parameters, the reference value of the PID control parameters, and the control deviation of the third operating parameter, so as to perform adaptive control of the hydropower unit governor.

[0005] In conjunction with the first aspect, in one implementation, the comprehensive state signal is obtained by summing the signals of each of the first operating parameters after weighting them by their respective weighting coefficients.

[0006] In conjunction with the first aspect, in one implementation, the first operating parameters include the water flow turbulence intensity and the unit oscillation displacement; And / or, the weighting coefficients are obtained by dividing the reciprocal of the signal variance of the first operating parameter by the sum of the reciprocals of the signal variances of all the first operating parameters.

[0007] In conjunction with the first aspect, in one implementation, multi-scale wavelet decomposition is performed on the integrated state signal after fusing the first operating parameters of the hydropower unit to extract feature vectors characterizing the dynamic characteristics of the hydropower unit, including: Multi-scale wavelet decomposition was performed on the integrated state signal after the fusion of the first operating parameters of the hydropower unit to obtain low-frequency approximation coefficients and high-frequency detail coefficients; The energy entropy of the high-frequency detail coefficients is obtained based on the number of sampling points and the total signal energy. A feature vector is constructed using the mean amplitude, peak value, and vibration frequency of the energy entropy and the low-frequency approximation coefficient.

[0008] In conjunction with the first aspect, in one implementation, the second operating parameter includes guide vane opening, unit speed, and unit active load; Combining the second operating parameters and feature vector of the hydropower unit, a recursive least squares method with a forgetting factor is used to identify the hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters, including: Discretize the transfer function of the hydropower unit's hydraulic system; Using the guide vane opening and eigenvector as inputs, and the unit speed and active load as outputs, the recursive least squares method with a forgetting factor is used to identify the discretized hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters.

[0009] In conjunction with the first aspect, in one implementation, a fuzzy controller is constructed using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs. The centroid method is then used for defuzzification to obtain the correction values ​​of the PID control parameters, including: Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller with two inputs and three outputs is constructed. The input and output quantities are divided into several fuzzy subsets, and a fuzzy rule set is constructed based on the experience of tuning the PID control parameters of the hydropower unit speed governor. By combining fuzzy subsets and fuzzy rule sets, the centroid method is used to defuzzify the fuzzy controller in order to obtain the correction amount of the PID control parameters.

[0010] In conjunction with the first aspect, in one implementation, the third operating parameter includes the unit speed and the unit active load; The guide vane opening is obtained based on the correction value of the PID control parameters, the reference value of the PID control parameters, and the control deviation of the third operating parameter, including: The update value of the PID control parameters is obtained based on the correction value of the PID control parameters and the reference value of the PID control parameters. The guide vane opening is obtained based on the PID control parameter update, the control deviation of the unit speed, the control deviation of the unit's active load, and the PID discrete control equation of the hydropower unit governor.

[0011] In conjunction with the first aspect, in one embodiment, before adaptive control of the hydropower unit governor, the method further includes: The updated PID control parameters, obtained from the correction and baseline values ​​of the PID control parameters, are loaded onto the digital twin platform of the hydropower unit for pre-simulation to obtain pre-simulation evaluation indicators. Based on the aforementioned pre-simulation evaluation indicators, the pre-simulation results are obtained; If the pre-simulation results are satisfactory, adaptive control will be implemented on the hydropower unit speed governor.

[0012] In conjunction with the first aspect, in one implementation, before performing multi-scale wavelet decomposition, the method further includes: obtaining an identification triggering mode based on the fourth operating parameter of the hydropower unit and the identification triggering threshold, wherein the identification triggering mode includes a steady-state condition triggering pre-adjustment mode and a transient condition triggering emergency adjustment mode. Among them, the steady-state operating condition is divided into several steady-state sub-operating conditions, the transient operating condition is divided into several transient sub-operating conditions, and at least one transient sub-operating condition is a load shedding condition; If the identification triggering mode is a steady-state condition triggering pre-tuning mode, it is loaded onto the hydropower unit digital twin platform for pre-performance, including: loading the PID control parameter update amount onto the hydropower unit digital twin platform, and pre-performance of the currently operating steady-state sub-condition and its adjacent steady-state sub-conditions. If the identification trigger mode is a transient operating condition trigger emergency adjustment mode, it is loaded onto the hydropower unit digital twin platform for pre-simulation, including: loading the PID control parameter update quantity onto the hydropower unit digital twin platform, and pre-simulating the current transient sub-operating condition and load shedding condition.

[0013] In conjunction with the first aspect, in one implementation, the fourth operating parameter includes the unit swing displacement and the unit active load; the identification trigger threshold includes the displacement peak threshold and the load fluctuation threshold. Based on the fourth operating parameter of the hydropower unit and the identification trigger threshold, the identification trigger mode is obtained as follows: If the peak value of the unit's swing displacement is less than or equal to the peak displacement threshold, and the unit's active load fluctuation is less than or equal to the load fluctuation threshold, then the identification triggering mode is the steady-state condition triggering pre-adjustment mode. Otherwise, the identification trigger mode is a transient condition trigger emergency adjustment mode.

[0014] In conjunction with the first aspect, in one implementation, the pre-performance evaluation index includes at least one of the following: adjustment time, overshoot, peak value of unit oscillation displacement, number of oscillations, and steady-state error.

[0015] In conjunction with the first aspect, in one implementation method, obtaining the pre-rehearsal results based on the pre-rehearsal evaluation indicators includes: The pre-performance evaluation indicators are compared with the corresponding qualification thresholds. If all pre-performance evaluation indicators are qualified, the pre-performance result is qualified; otherwise, the pre-performance result is unqualified.

[0016] In conjunction with the first aspect, in one implementation, when the pre-simulation result is unsatisfactory, the method further includes: The update values ​​of the PID control parameters that fail the pre-simulation results are used as the initial actions, and the current transient sub-condition is used as the initial state. The PID control parameters are then optimized using a deep deterministic strategy gradient algorithm to obtain the optimized PID control parameter update values, which are then loaded back onto the hydropower unit digital twin platform for pre-simulation.

[0017] In conjunction with the first aspect, in one implementation, if the pre-test results are unqualified after multiple consecutive optimizations of the PID control parameters, a safety protection mechanism is triggered.

[0018] Secondly, embodiments of this application provide an adaptive control system for a hydropower unit speed governor, comprising: The feature extraction module is used to perform multi-scale wavelet decomposition on the comprehensive state signal after the fusion of the first operating parameters of the hydropower unit in order to extract feature vectors that characterize the dynamic characteristics of the hydropower unit. The identification module is used to identify the hydraulic system transfer function of the hydropower unit by combining the second operating parameters and feature vector of the hydropower unit and using the recursive least squares method with a forgetting factor, so as to obtain the hydraulic system parameters. The correction quantity acquisition module is used to construct a fuzzy controller by taking the hydraulic system parameters and feature vector as input quantities and the correction quantity of the PID control parameters as output quantities, and to perform defuzzification using the centroid method to obtain the correction quantity of the PID control parameters. The control module is used to obtain the guide vane opening based on the correction amount of the PID control parameters, the reference amount of the PID control parameters, and the control deviation of the third operating parameter, so as to perform adaptive control of the hydropower unit governor.

[0019] The beneficial effects of the technical solution provided in this application include: This application addresses the problem of unit oscillation and instability caused by the inability of fixed PID parameters to adapt to the time-varying characteristics of hydraulic systems, proposing an adaptive control method based on multi-source data fusion. First, using multi-source data fusion technology, the comprehensive state signal after fusing the first operating parameters is decomposed into multi-scale wavelet decomposition to extract feature vectors representing dynamic characteristics. Combined with the second operating parameter, a recursive least squares method with a forgetting factor is used to identify hydraulic system parameters in real time, solving the problem of lag in sensing changes in operating conditions. Second, a fuzzy controller is constructed, using the identified hydraulic system parameters and feature vectors as inputs to calculate the correction amount of the PID control parameters, achieving dynamic matching of the control strategy with hydraulic characteristics. Finally, the guide vane opening is obtained by combining the correction amount, the reference amount, and the control deviation of the third operating parameter. This scheme, through multi-source data fusion and online identification, enables real-time adaptive adjustment of PID control parameters, avoiding oscillations caused by parameter mismatch, improving the regulation quality and stability of the unit under complex operating conditions, effectively suppressing regulation instability problems, and ensuring the safe and stable operation of the hydropower unit. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of the adaptive control method for a hydropower unit speed governor provided in an embodiment of this application; Figure 2 This is a block diagram of the adaptive control system module for the hydropower unit speed governor provided in an embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] While the MGC-5000 system, as a new generation speed controller, has improved in response speed and control precision, the applicant, after conducting extensive research on the existing MGC-5000 system, discovered that it still has many problems, such as: (1) Insufficient adaptability to complex hydraulic environments, especially under high head and large fluctuation conditions.

[0024] (2) Lack of control strategy optimization methods based on multi-physics coupling.

[0025] (3) Limited ability to predict equipment status and provide early warning of faults.

[0026] To solve these problems, see Figure 1 As shown in the figure, this application provides an adaptive control method for a hydropower unit speed governor, which includes the following steps: 101: Perform multi-scale wavelet decomposition on the integrated state signal after fusing the first operating parameters of the hydropower unit to extract feature vectors characterizing the dynamic characteristics of the hydropower unit.

[0027] 102: Combining the second operating parameters and feature vector of the hydropower unit, the recursive least squares method with forgetting factor is used to identify the hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters.

[0028] 103: Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller is constructed, and the centroid method is used for defuzzification to obtain the correction values ​​of the PID control parameters.

[0029] 104: Based on the correction amount of the PID control parameters, the reference amount of the PID control parameters, and the control deviation of the third operating parameter, the guide vane opening is obtained to perform adaptive control of the hydropower unit governor.

[0030] This application addresses the problem of unit oscillation and instability caused by the inability of fixed PID parameters to adapt to the time-varying characteristics of hydraulic systems, proposing an adaptive control method based on multi-source data fusion. First, using multi-source data fusion technology, the comprehensive state signal after fusing the first operating parameters is decomposed into multi-scale wavelet decomposition to extract feature vectors representing dynamic characteristics. Combined with the second operating parameter, a recursive least squares method with a forgetting factor is used to identify hydraulic system parameters in real time, solving the problem of lag in sensing changes in operating conditions. Second, a fuzzy controller is constructed, using the identified hydraulic system parameters and feature vectors as inputs to calculate the correction amount of the PID control parameters, achieving dynamic matching of the control strategy with hydraulic characteristics. Finally, the guide vane opening is obtained by combining the correction amount, the reference amount, and the control deviation of the third operating parameter. This scheme, through multi-source data fusion and online identification, enables real-time adaptive adjustment of PID control parameters, avoiding oscillations caused by parameter mismatch, improving the regulation quality and stability of the unit under complex operating conditions, effectively suppressing regulation instability problems, and ensuring the safe and stable operation of the hydropower unit.

[0031] Compared with the original fixed PID control of the MGC-5000 system, this application reduces the unit's settling time by more than 60%, the overshoot by more than 70%, and the peak-to-peak value of the unit's oscillation by more than 50%. It effectively suppresses the unit's oscillation caused by hydraulic turbulence and operating condition fluctuations, and completely solves the risk of regulation instability under high head and large fluctuation conditions.

[0032] The method provided in this application embodiment can be applied to existing MGC-4000 / 5000 series speed controllers. For example, a multi-source data acquisition module and a protocol conversion module can be added to the existing MGC-4000 / 5000 series speed controller hardware. The specific design is as follows: Data Collection Targets: The core data collection parameters include two types of data sources, forming a multi-source data system: ① Water flow turbulence intensity: By installing high-frequency pressure pulsation sensors at the turbine casing inlet, guide vane outlet, and tailrace inlet, water flow pressure pulsation signals are collected, and the real-time water flow turbulence intensity is calculated. ② Unit swing displacement: By installing eddy current displacement sensors at the upper guide, lower guide, and water guide bearings of the turbine main shaft, real-time displacement data of the radial / axial swing of the main shaft is collected; Hardware and Protocol Design: For the distributed architecture of the MGC-5000 system, a protocol conversion module is adopted to support bidirectional conversion of mainstream industrial protocols such as Modbus TCP, IEC 61850, and PROFINET, so as to achieve multi-channel synchronous acquisition with a sampling frequency of ≥10kHz and a time synchronization accuracy between channels of ≤1μs. At the same time, it supports redundancy backup and fault self-diagnosis functions to ensure the real-time performance, synchronization and reliability of the acquired data.

[0033] This application has specifically optimized the architecture of the MGC-5000 system, and developed a dedicated protocol conversion module and driver module. It is fully compatible with the existing hardware and software architecture of the MGC-4000 / 5000 series speed governors, eliminating the need for large-scale modifications to the original system and reducing modification costs by more than 80%. At the same time, it can be adapted to hydropower units of different capacities and water heads, and can be widely used in the upgrade and transformation of speed governors in major hydropower stations across the country, possessing extremely high promotional value.

[0034] The first set of collected operating parameters are preprocessed before data fusion is performed to address the shortcomings of existing technologies where data is used independently without fusion.

[0035] Data preprocessing: The 3σ criterion is used to remove outliers in the first operating parameter signal, and a Butterworth low-pass filter is used to remove high-frequency noise to obtain the noise-reduced effective signal; Data spatiotemporal registration: Based on a high-precision time synchronization benchmark, the signals of the first operating parameters are timestamped and aligned. Spatial registration of the sensor installation position is achieved through spatial coordinate mapping to ensure the spatiotemporal consistency of each signal.

[0036] Furthermore, in step 101 above, the comprehensive state signal is obtained by summing the signals of each of the first operating parameters after weighting them with their respective weighting coefficients.

[0037] For example, taking two signals as an example, the two signals are the water flow turbulence intensity signal and the unit swing displacement signal.

[0038] For the two spatiotemporally registered signals, an adaptive weighted fusion algorithm is used to calculate the fused comprehensive state signal. The fusion calculation formula is as follows:

[0039] In the formula: The fused integrated state signal at time k; The signal representing the turbulence intensity of the water flow at time k; The unit's oscillation displacement signal at time k; , These are the weighting coefficients for the two signals, satisfying... The values ​​range from [0,1] and are dynamically adjusted based on the real-time signal-to-noise ratio of the two signals. The higher the signal-to-noise ratio, the larger the weighting coefficient. The weighting coefficient is obtained by dividing the reciprocal of the signal variance of the first operating parameter by the sum of the reciprocals of the signal variances of all the first operating parameters. The calculation formula is as follows:

[0040] In the formula, For the first The weighting coefficients of the road signal, For the first Real-time variance of the road signal, For the first The real-time variance of the signal is used to characterize the noise level of the signal; the smaller the variance, the higher the signal-to-noise ratio. .

[0041] After fusion, the integrated state signal can be decomposed into multi-scale wavelet decomposition to extract feature vectors that characterize the dynamic properties of hydropower units.

[0042] This application achieves feature-level deep fusion of multi-source data such as water flow turbulence intensity and unit oscillation, solving the problems of independent application and poor coordination of multi-source data in the prior art. Through multi-scale wavelet decomposition and intelligent algorithms, it comprehensively captures the dynamic characteristics of hydraulic-mechanical-electrical multi-physics fields, realizes global optimization control of the governor, and significantly improves the system's intelligence level.

[0043] Furthermore, before proceeding to step 101, to address the shortcomings of the original triggering logic, this application adopts a multi-threshold triggered identification program startup logic, that is, designs a multi-threshold triggering mechanism for steady-state pre-adjustment and transient emergency adjustment as the startup condition for the real-time hydraulic parameter identification program, as follows: Before performing multi-scale wavelet decomposition, the method further includes: obtaining an identification trigger mode based on the fourth operating parameter of the hydropower unit and the identification trigger threshold, wherein the identification trigger mode includes a steady-state condition trigger pre-adjustment mode and a transient condition trigger emergency adjustment mode.

[0044] As an example, the fourth operating parameter includes the unit swing displacement and the unit active load; the identification trigger thresholds include the displacement peak threshold and the load fluctuation threshold.

[0045] Based on the fourth operating parameter of the hydropower unit and the identification trigger threshold, the identification trigger mode is obtained as follows: if the peak value of the unit's swing displacement is less than or equal to the peak displacement threshold, and the active power load fluctuation of the unit is less than or equal to the load fluctuation threshold, then the identification trigger mode is a steady-state condition trigger pre-adjustment mode; otherwise, the identification trigger mode is a transient condition trigger emergency adjustment mode.

[0046] The peak displacement threshold and load fluctuation threshold can be set according to the acceptable limits for shaft oscillation in GB / T 7894-2023 Basic Technical Requirements for Hydropower Generators.

[0047] Taking a displacement peak threshold of 0.1 mm and a load fluctuation threshold of ±2% of rated load as an example, the startup logic of the dual-threshold trigger identification program is as follows: (1) Steady-state operating condition triggering pre-tuning mode ① Triggering conditions: The unit is in grid-connected steady-state operation, the measured peak-to-peak value of the unit's swing displacement is ≤0.1mm (steady-state qualified range), and the unit's active load fluctuation is ≤±2% of the rated load; ② Startup logic: When the triggering conditions are met, the real-time hydraulic parameter identification program is started every 5 minutes to complete the steady-state pre-adjustment of the PID control parameters, adapt to the slow changes in hydraulic characteristics in advance, and prevent the risk of unit oscillation.

[0048] (2) Transient operating conditions trigger emergency adjustment mode ① Triggering conditions: The measured peak-to-peak value of the unit's swing displacement is greater than 0.1 mm (exceeding the steady-state acceptable range), or the unit's active load fluctuation is greater than ±2% of the rated load (including transitional conditions such as load shedding and sudden load changes). ② Startup logic: When the triggering condition is met, the current control flow is immediately interrupted, the real-time hydraulic parameter identification program and PID control parameter emergency adjustment are started to quickly suppress unit oscillation and restore stable operation.

[0049] This embodiment divides the steady-state pre-adjustment and transient emergency adjustment modes based on the unit's oscillation displacement and active load. In steady state, fine optimization improves control accuracy; in transient state, rapid response suppresses oscillations. This reduces invalid identifications, ensuring the governor balances accuracy and response speed, preventing false triggering, and improving unit reliability.

[0050] Furthermore, for the fused integrated state signal, this application employs multi-scale wavelet decomposition to extract features, accurately capturing the dynamic characteristics of the hydraulic system. Specifically, in step 101 above, multi-scale wavelet decomposition is performed on the integrated state signal after fusing the first operating parameters of the hydropower unit to extract feature vectors characterizing the dynamic characteristics of the hydropower unit, including: 201: Perform multi-scale wavelet decomposition on the integrated state signal after fusing the first operating parameters of the hydropower unit to obtain low-frequency approximation coefficients and high-frequency detail coefficients.

[0051] 202: Obtain the energy entropy of high-frequency detail coefficients based on the number of sampling points and the total signal energy.

[0052] 203: Construct a feature vector using the mean amplitude, peak value, and vibration frequency of the energy entropy and the low-frequency approximation coefficient.

[0053] For steps 201 to 203, the complete processing procedure will be explained in detail below with specific examples.

[0054] First, the wavelet basis and decomposition level are selected: the db4 wavelet is chosen as the mother wavelet because it is suitable for feature extraction of hydraulic non-stationary signals. Other wavelets can also be selected according to actual needs. The fused signal is decomposed into 5 levels of multi-scale wavelet decomposition, and the number of levels can be selected according to actual needs, resulting in one low-frequency approximation coefficient. and 5 high-frequency detail coefficients The decomposition formula is:

[0055] In the formula, j0 is the scale of wavelet decomposition. Let be the high-frequency detail coefficients at the j0-th scale, where j0 = 1, 2, ..., 5.

[0056] Next, feature vector construction is performed: taking the first operating parameters, including water flow turbulence intensity and unit oscillation displacement, as an example, the following feature parameters are extracted for each scale coefficient to form a feature vector for hydraulic parameter identification and PID adjustment. : Turbulence intensity: The energy entropy of each high-frequency detail coefficient, characterizing the turbulence energy distribution in different frequency bands, is calculated using the following formula:

[0057] In the formula, The energy entropy of the high-frequency detail coefficients at the j0-th scale. , Let N be the nth sampling point in the discrete sequence of high-frequency detail coefficients at scale j0, and N be the number of sampling points in the discrete sequence of high-frequency detail coefficients at scale j0. The total energy of the integrated state signal after the fusion of the single-segment signal acquisition time participating in wavelet decomposition is the total energy of the integrated state signal. The single-segment signal acquisition time is determined according to actual needs. For example, in this application, the single-segment signal acquisition time adopts a standardized fixed value (such as a fixed 1s / 5s, which matches the identification triggering period). Unit oscillation displacement: Low-frequency approximation coefficient The average amplitude Peak Vibration frequency This characterizes the overall trend and low-frequency characteristics of the unit's oscillation. Among them, the low-frequency approximation coefficient The average amplitude It is the arithmetic mean of the absolute values ​​of the amplitudes of all sampling points in the discrete sequence of low-frequency approximation coefficients.

[0058] Final feature vector: A total of 8 feature vectors are used to comprehensively characterize the dynamic characteristics of the hydraulic system and the unit's mechanical system.

[0059] It is understandable that hydropower units are complex systems with strong coupling of multiple physical fields, including hydraulic, mechanical, electro-hydraulic, and electrical systems. Their core components mainly include four subsystems: hydraulic system, unit mechanical system, electro-hydraulic servo system, and electrical control system. These four subsystems work together to complete the conversion of water energy into electrical energy and ensure the safe and stable operation of the unit.

[0060] This application employs the recursive least squares method with a forgetting factor (FF-RLS) to achieve online real-time identification of hydraulic system transfer function parameters. Specifically, the second operating parameters include guide vane opening, unit speed, and unit active load; combining the second operating parameters and feature vector of the hydropower unit, the recursive least squares method with a forgetting factor is used to identify the hydraulic system transfer function of the hydropower unit to obtain hydraulic system parameters, including: 301: Discretize the transfer function of the hydraulic system of the hydropower unit.

[0061] In step 301, the hydraulic system model is first constructed: The hydraulic system is simplified into a second-order linear transfer function, which is then used as the object of identification.

[0062] In the formula: For the transfer function of the hydraulic system; This is the hydraulic system gain coefficient, with a value range of [0.5, 2.0]. Its physical meaning is the degree of influence of the guide vane opening change on the unit output. The inertial time constant of the water flow has a value range of [0.1s, 5.0s], and its physical meaning is the inertial effect of water flow in a pressure pipe; The damping ratio of the hydraulic system is defined as the value range of [0.1, 1.0], which physically represents the oscillation damping capability of the hydraulic system. It is the Laplace transform operator (also called the complex frequency variable), which is a core mathematical tool in classical control theory used to describe the dynamic characteristics of linear time-invariant systems.

[0063] The transfer function of the hydraulic system is then discretized into a difference equation.

[0064] 302: Using the guide vane opening and eigenvector as inputs, and the unit speed and active load as outputs, the discretized hydropower unit hydraulic system transfer function is identified using the recursive least squares method with a forgetting factor to obtain the hydraulic system parameters.

[0065] The transfer function is discretized into a difference equation, and the parameters are identified online using a recursive least squares method with a forgetting factor. The core recursive formula is as follows:

[0066]

[0067]

[0068] In the formula: The hydraulic system parameters identified at time k; The hydraulic system output at time k includes the unit speed and active load, both of which are measured values. Let be the input observation vector at time k, including the measured guide vane opening and the eigenvector F; it should be understood that the guide vane opening here is the input observation vector, not the guide vane opening control command to be calculated. Instead, it refers to the historical actual guide vane opening data obtained by sensor measurements prior to time k, which has already been executed. This is the gain matrix; It is the covariance matrix; The forgetting factor has a value range of [0.95, 0.999]. In this application, it can be set to 0.98 to weaken the influence of historical data and improve the ability to track time-varying parameters. It is an identity matrix.

[0069] Real-time output of identified hydraulic system parameters This serves as the core basis for the adaptive adjustment of PID control parameters.

[0070] Furthermore, this application employs a fuzzy adaptive PID control algorithm as the core adaptive control algorithm. Specifically, in step 103 above, the hydraulic system parameters and feature vectors are used as inputs, and the correction values ​​of the PID control parameters are used as outputs to construct a fuzzy controller. The centroid method is used for defuzzification to obtain the correction values ​​of the PID control parameters, including: 401: Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller with two inputs and three outputs is constructed.

[0071] For example, as an example, in step 401, the identified hydraulic system parameters are... As the first input, and with the feature vector As the second input, the correction amount of the PID control parameters As three outputs, a two-input, three-output fuzzy controller is constructed. These are the PID proportional coefficient correction, PID integral time constant correction, and PID derivative time constant correction, respectively.

[0072] 402: Divide the input and output quantities into several fuzzy subsets respectively, and construct a fuzzy rule set based on the experience of tuning the PID control parameters of the hydropower unit speed governor.

[0073] For example, input fuzzification: divide the two inputs into 7 fuzzy subsets: {NB (negative large), NM (negative medium), NS (negative small), ZO (zero), PS (positive small), PM (positive medium), PB (positive large)}, all with a universe of discourse of [-6, 6].

[0074] Output blurring: Each of the three output quantities is divided into 7 fuzzy subsets: {NB (negative large), NM (negative medium), NS (negative small), ZO (zero), PS (positive small), PM (positive medium), PB (positive large)}, with universes of discourse of [-3,3], [-2,2], and [-1,1], respectively.

[0075] Fuzzy rule base: Based on the experience of tuning the PID control parameters of hydropower unit speed governor, 49 fuzzy rules were constructed, as shown in the table below.

[0076] As an example, the tuning experience is listed below: Regarding the inertial characteristics of water flow: when the inertial time constant of water flow When increasing, the proportional gain needs to be reduced. Increase integration time To avoid system oscillations caused by water hammer; when When decreasing, it is necessary to increase , reduce This improves system response speed.

[0077] Regarding the system damping characteristics: when the damping ratio of the hydraulic system is... When decreasing, it is necessary to lower Significantly increase the differential time To improve system damping and suppress oscillations; when When increasing, it is necessary to improve , reduce To expedite system response.

[0078] Regarding load / speed deviation characteristics: When the unit experiences significant load changes and large deviations, it is necessary to increase... , reduce To achieve rapid adjustment; when the system is in steady state and the deviation is small, it is necessary to improve... , reduce Increase This improves steady-state accuracy and anti-interference capabilities.

[0079]

[0080] Example of core rule: If Increase (increase the inertia of the water flow), then For PS, For PB, To reduce the proportional gain and increase the integral time for PS, avoid system oscillation; if Decrease (damping ratio decreases, system is prone to oscillation), then For NS, For PS, To achieve PB, reduce the proportional gain, increase the derivative time, and improve system damping; if the turbulence intensity energy entropy Increase (intensify turbulence), then For NS, For ZO, To reduce the proportional gain to PS, improve the system's anti-interference capability.

[0081] 403: Combining fuzzy subsets and fuzzy rule sets, the centroid method is used to defuzzify the fuzzy controller to obtain the correction amount of the PID control parameters.

[0082] In step 403, after obtaining the correction value of the PID control parameters, the update value of the PID control parameters can be obtained based on the correction value and the reference value of the PID control parameters, as follows:

[0083]

[0084]

[0085] In the formula, , , These represent the update amounts of the PID proportional coefficient, the PID integral time constant, and the PID derivative time constant at time k, respectively. The reference values ​​for the original PID control parameters of the speed controller, such as the MGC-5000 system, are specifically the reference values ​​for the PID proportional coefficient, the PID integral time constant, and the PID derivative time constant. Through the above calculations, the adaptive dynamic adjustment of the PID control parameters can be achieved as the operating conditions and hydraulic characteristics change.

[0086] Furthermore, the third operating parameters include unit speed and unit active load; obtaining the guide vane opening includes: The guide vane opening is obtained based on the PID control parameter update, the control deviation of the unit speed, the control deviation of the unit's active load, and the PID discrete control equation of the hydropower unit governor.

[0087] The PID discrete control equations for the hydropower unit speed governor are as follows:

[0088] In the formula: Let be the guide vane opening output by the PID controller at time k, and be the control variable to be calculated. The control deviation at time k includes the control deviation of the unit speed and the control deviation of the unit active load. The control deviation of the unit speed is the difference between the setpoint and the measured value of the unit speed, and the control deviation of the unit active load is the difference between the setpoint and the measured value of the unit active load. The setpoint of the unit speed is the rated speed of the unit, and the setpoint of the unit active load is the target load command issued by the power grid dispatch or the target load value set by the on-site operators.

[0089] This is the proportionality coefficient. The integral time constant is... The differential time constant; For the control cycle, this application can use 1ms to adapt to the control cycle of the MGC-5000 system.

[0090] Currently, there is a lack of effective pre-testing methods for optimizing the control strategy of hydropower unit speed governors. In actual operation, the effectiveness of different control strategies can only be verified through actual operation. If problems occur, they may damage the unit equipment, and the adjustment process is time-consuming and costly.

[0091] To address this issue, this application constructs a digital twin platform for hydropower units, and uses this platform to perform parameter simulation of PID control parameter updates.

[0092] For example, as a case study, for the MGC-5000 speed control system, a multi-physics coupled digital twin platform comprising four major subsystems is constructed to achieve real-time state synchronization with the physical entity, as detailed below: System composition: (1) Electro-hydraulic servo system model: includes an electro-hydraulic converter, main pressure valve, and servo motor electro-hydraulic coupling model to accurately simulate the dynamic characteristics of electro-hydraulic conversion and hydraulic execution; (2) Mechanical transmission system model: includes the mechanical dynamics model of the unit's main shaft, bearings, and water guide mechanism, where bearing temperature data is used to correct the stiffness and damping parameters of the bearing oil film in real time (the stiffness of the oil film decreases nonlinearly with increasing temperature), accurately simulating the influence of temperature changes on the unit's oscillation characteristics, and solving the problem that the original bearing temperature data had no clear function; (3) Hydraulic system model: includes a one-dimensional / three-dimensional coupled hydraulic model of the volute, water guide mechanism, impeller, and tailpipe, linked with the real-time hydraulic parameter identification results, and updating the model parameters in real time; (4) Electrical control system model: completely replicates the PID control logic, protection logic, and communication protocol of the MGC-5000 system to achieve program-level compatibility with the physical speed controller.

[0093] Real-time synchronization mechanism: Real-time operating data (guide vane opening, unit speed, load, PID control parameters, bearing temperature, vibration, pressure pulsation, etc.) of the MGC-5000 system are connected to the digital twin platform via 5G / industrial Ethernet, with a data update frequency of ≥1kHz; Extended Kalman filter algorithm is adopted to correct the twin model parameters online based on real-time operating data, ensuring that the state deviation between the twin and the physical entity is ≤2%, and achieving accurate mapping throughout the entire life cycle.

[0094] Before performing adaptive control of the hydropower unit governor by obtaining the guide vane opening based on the correction amount of the PID control parameters, the reference amount of the PID control parameters, and the control deviation of the third operating parameter, the method further includes: 501: Load the updated PID control parameters onto the digital twin platform of the hydropower unit for pre-simulation to obtain pre-simulation evaluation indicators.

[0095] 502: Based on the aforementioned pre-performance evaluation indicators, obtain the pre-performance results.

[0096] 503: If the pre-simulation results are satisfactory, adaptive control shall be implemented on the hydropower unit speed governor.

[0097] The aforementioned loading for pre-rehearsal will be performed based on the identified trigger mode.

[0098] Specifically, we first construct a set of typical operating conditions for the entire operating cycle of hydropower units.

[0099] For example, operating conditions can be divided into steady-state operating conditions and transient operating conditions, with transient operating conditions further divided into transitional operating conditions and extreme operating conditions.

[0100] For steady-state operating conditions, it is further divided into five steady-state sub-operating conditions: no load, 25% rated load, 50% rated load, 75% rated load, and 100% rated load.

[0101] For the transitional operating conditions, it is further divided into four transient sub-conditions: ±10% step change in the active load of the unit's current stable operation before the step disturbance occurs, ±20% step change in the active load of the unit's current stable operation before the step disturbance occurs, 50% load shedding, and 100% load shedding. "100% load shedding" refers to the process where the unit, operating stably at 100% rated load under rated speed and rated head, suddenly sheds all load, and the unit transitions from rated load to no-load operation. The logic for 50% load shedding is the same as that for 100% load shedding.

[0102] For extreme operating conditions, they are further divided into three transient sub-conditions: high head and low load, sudden change in turbulence intensity, and power grid frequency fluctuation.

[0103] The typical operating condition set constructed above includes a total of 12 typical operating conditions in 3 major categories.

[0104] If the identification triggering mode is a steady-state condition triggering pre-tuning mode, it is loaded onto the hydropower unit digital twin platform for pre-simulation, including: loading the PID control parameter update amount onto the hydropower unit digital twin platform, and pre-simulating the currently operating steady-state sub-condition and its adjacent steady-state sub-conditions; for example, only simulating the currently operating steady-state sub-condition and the two adjacent currently operating steady-state sub-conditions, such as simulating 25% rated load, 50% rated load, and 75% rated load conditions when the current load is 50% of the rated load, to verify the condition adaptability of the parameters.

[0105] If the identified triggering mode is a transient operating condition-triggered emergency adjustment mode, it is loaded onto the hydropower unit's digital twin platform for pre-simulation, including: loading the PID control parameter update onto the hydropower unit's digital twin platform, and pre-simulating the current operating transient sub-conditions and load shedding conditions. For example, simulating the current operating transient sub-conditions and shedding 50% and 100% load to verify the emergency adjustment capability and stability of the parameters.

[0106] Furthermore, Based on the pre-rehearsal evaluation indicators, the pre-rehearsal results are obtained, including: comparing the pre-rehearsal evaluation indicators with the corresponding indicator qualification thresholds; if all pre-rehearsal evaluation indicators are qualified, the pre-rehearsal results are qualified; otherwise, the pre-rehearsal results are unqualified.

[0107] The pre-performance evaluation indicators include at least one of the following: adjustment time, overshoot, peak value of unit swing displacement, number of oscillations, and steady-state error.

[0108] The acceptable threshold values ​​are set according to GB / T 9652.1-2019 Technical Conditions for Speed ​​Regulation Systems of Hydropower Turbines. See the table below for an example:

[0109] "Steady-state value" refers to the stable value of the controlled output when the governor's closed-loop control system enters a stable state after a disturbance occurs. It is obtained under two operating conditions: Grid-connected operating conditions such as load step: The steady-state value is the setpoint of the active power load of the unit after the disturbance (or the measured stable value within the range of setpoint ± steady-state error). No-load conditions such as load shedding: The steady-state value is the rated speed of the unit (or the measured stable value within the range of rated speed ± steady-state error).

[0110] In steady-state error, "rated value" refers to the rated parameter of the unit corresponding to the controlled variable, and there is a one-to-one correspondence between them: Grid-connected operating condition: The controlled variable is the active power of the unit, and the rated value is the rated active power of the hydropower unit (rated output as indicated on the unit nameplate, in MW). No-load / load shedding condition: The controlled variable is the unit speed, and the rated value is the rated speed of the hydropower unit (the rated speed marked on the unit nameplate, in r / min).

[0111] During the rehearsal, if all evaluation indicators for all working conditions meet the qualification threshold, the rehearsal result is deemed qualified; otherwise, the rehearsal result is deemed unqualified.

[0112] If the pre-run results are satisfactory, the PID control parameters will be sent to the MGC-5000 physical speed controller for execution, and the PID control parameters and pre-run results will be stored in the parameter optimization database.

[0113] If the pre-run results are unsatisfactory, the deep deterministic strategy gradient algorithm is immediately activated to re-explore the optimal PID control parameters in the digital twin environment and perform another pre-run verification. If the results still fail to meet the standards after multiple optimizations, the safety protection mechanism is triggered, switching to the original backup PID control parameter set of the MGC-5000 system, and issuing a warning to the operators. The number of consecutive runs can be set as needed, such as 3 runs.

[0114] This application employs the Deep Deterministic Policy Gradient (DDPG) algorithm, adapted to the continuous action space of PID control parameter continuous adjustment, to achieve automatic exploration of optimal parameters in a digital twin environment. Specifically as follows: Algorithm framework: The Actor-Critic dual-network architecture is adopted, which includes an Actor policy network (outputting the optimal PID control parameters and actions) and a Critic evaluation network (evaluating the value of actions). Both networks are set up with online networks and target networks respectively to realize experience playback and stable training.

[0115] State space S: Defined as the real-time operating state of the digital twin platform, including current operating conditions and hydraulic system parameters. The system consists of a 12-dimensional state vector, including unit oscillation data, turbulence intensity, control deviation e, and deviation change rate ec. The deviation change rate ec refers to the difference between the control deviation at time k and the control deviation at time k-1.

[0116] Action space A: defined as the adjustment amount of the PID control parameters. , is a 3-dimensional continuous motion vector, and the motion range is consistent with the output range of the fuzzy controller.

[0117] Reward function R: To optimize performance, a reward function is constructed, as shown in the following formula:

[0118] In the formula: To control the absolute value of the deviation, To adjust the time, For overshoot, This represents the peak value of the unit's swing displacement. The weighting coefficients are set to 0.4, 0.3, 0.2, and 0.1 respectively, based on the adjustment priority. The larger the reward function value, the better the control performance. The algorithm explores the optimal combination of PID control parameters by maximizing the cumulative reward. Automatic exploration and optimization process: (1) Initialize the parameters of the Actor and Critic networks, take the update amount of the unqualified PID control parameters in the pre-run as the initial action, and take the current transient sub-condition as the initial state; (2) The Actor network outputs the PID control parameters to adjust the action according to the current state, and inputs it into the digital twin platform for simulation; (3) After the simulation is completed, calculate the reward function value, feed it back to the Critic network, and evaluate the value of the current action; (4) Update the network parameters using the gradient descent method, and repeat the training through the experience playback mechanism to continuously optimize the action strategy; (5) When the cumulative reward converges or the maximum number of training steps is reached, output the optimal PID control parameters to complete the optimization process.

[0119] To optimize PID control parameters using a deep deterministic gradient algorithm, it is necessary to simulate 12 typical operating conditions to verify the adaptability of the parameters across all operating conditions.

[0120] This application realizes the pre-demonstration optimization of PID control strategy through a digital twin simulation platform coupled with multi-physics fields, completely avoiding the risk of equipment damage from actual machine trial tuning. At the same time, combined with deep reinforcement learning algorithm, the PID parameter optimization cycle is shortened from the original 72 hours to less than 2 hours, and the optimization efficiency is improved by more than 95%.

[0121] This application achieves advance prediction of the speed regulation system status and pre-optimization of control performance through real-time parameter identification and digital twin mapping, reducing fatigue damage to equipment caused by unit oscillation, extending equipment service life by more than 20%, and reducing the risk of unplanned downtime and operation and maintenance costs, with an average annual reduction of more than 15% in operation and maintenance costs.

[0122] See Figure 2 As shown in the figure, this application embodiment also provides an adaptive control system for a hydropower unit speed governor, which includes: The feature extraction module is used to perform multi-scale wavelet decomposition on the comprehensive state signal after the fusion of the first operating parameters of the hydropower unit in order to extract feature vectors that characterize the dynamic characteristics of the hydropower unit. The identification module is used to identify the hydraulic system transfer function of the hydropower unit by combining the second operating parameters and feature vector of the hydropower unit and using the recursive least squares method with a forgetting factor, so as to obtain the hydraulic system parameters. The correction quantity acquisition module is used to construct a fuzzy controller by taking the hydraulic system parameters and feature vector as input quantities and the correction quantity of the PID control parameters as output quantities, and to perform defuzzification using the centroid method to obtain the correction quantity of the PID control parameters. The control module is used to obtain the guide vane opening based on the correction amount of the PID control parameters, the reference amount of the PID control parameters, and the control deviation of the third operating parameter, so as to perform adaptive control of the hydropower unit governor.

[0123] Furthermore, in one embodiment, the control system further includes a fusion module, which is used to sum the signals of each of the first operating parameters after weighting them by their respective weighting coefficients to obtain a comprehensive state signal.

[0124] Furthermore, in one embodiment, the fusion module is also used to calculate the reciprocal of the signal variance of each first operating parameter and the sum of the reciprocals of the signal variance of all first operating parameters, so as to obtain the weighting coefficient of the signal of each first operating parameter.

[0125] Furthermore, in one embodiment, the feature extraction module is used to perform multi-scale wavelet decomposition on the integrated state signal after the fusion of the first operating parameters of the hydropower unit, in order to extract feature vectors characterizing the dynamic characteristics of the hydropower unit, including: Multi-scale wavelet decomposition was performed on the integrated state signal after the fusion of the first operating parameters of the hydropower unit to obtain low-frequency approximation coefficients and high-frequency detail coefficients; The energy entropy of the high-frequency detail coefficients is obtained based on the number of sampling points and the total signal energy. A feature vector is constructed using the mean amplitude, peak value, and vibration frequency of the energy entropy and the low-frequency approximation coefficient.

[0126] Furthermore, in one embodiment, the identification module is used to combine the second operating parameters and feature vector of the hydropower unit, and employ a recursive least squares method with a forgetting factor to identify the hydraulic system transfer function of the hydropower unit, so as to obtain the hydraulic system parameters, including: Discretize the transfer function of the hydropower unit's hydraulic system into a difference equation; Using the guide vane opening and eigenvector as inputs, and the unit speed and active load as outputs, the difference equations are identified using a recursive least squares method with a forgetting factor to obtain hydraulic system parameters.

[0127] Further, in one embodiment, the correction quantity acquisition module is used to construct a fuzzy controller using the hydraulic system parameters and feature vectors as inputs and the correction quantity of the PID control parameters as outputs, and to perform defuzzification using the centroid method to obtain the correction quantity of the PID control parameters, including: Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller with two inputs and three outputs is constructed. The input and output quantities are divided into several fuzzy subsets, and a fuzzy rule set is constructed based on the experience of tuning the PID control parameters of the hydropower unit speed governor. By combining fuzzy subsets and fuzzy rule sets, the centroid method is used to defuzzify the fuzzy controller in order to obtain the correction amount of the PID control parameters.

[0128] Further, in one embodiment, the control module is used to obtain the guide vane opening based on the correction amount of the PID control parameters, the reference amount of the PID control parameters, and the control deviation of the third operating parameter, including: The update value of the PID control parameters is obtained based on the correction value of the PID control parameters and the reference value of the PID control parameters. The guide vane opening is obtained based on the PID control parameter update, the control deviation of the unit speed, the control deviation of the unit's active load, and the PID discrete control equation of the hydropower unit governor.

[0129] Furthermore, in one embodiment, the control system further includes a pre-performance module, which is used to load the updated PID control parameters obtained from the correction amount of the PID control parameters and the reference amount of the PID control parameters onto the digital twin platform of the hydropower unit for pre-performance in order to obtain pre-performance evaluation indicators. Based on the aforementioned pre-simulation evaluation indicators, the pre-simulation results are obtained; If the pre-simulation results are satisfactory, adaptive control will be implemented on the hydropower unit speed governor.

[0130] Furthermore, in one embodiment, the control system further includes a trigger mode acquisition module, which is used to acquire an identification trigger mode based on the fourth operating parameter of the hydropower unit and the identification trigger threshold. The identification trigger mode includes a steady-state condition trigger pre-adjustment mode and a transient condition trigger emergency adjustment mode.

[0131] Furthermore, in one embodiment, if the identification triggering mode is a steady-state condition triggering pre-tuning mode, the pre-performance module is used to load the pre-performance onto the hydropower unit digital twin platform, including: loading the PID control parameter update onto the hydropower unit digital twin platform, and pre-performance the currently operating steady-state sub-condition and its adjacent steady-state sub-conditions. If the identification trigger mode is a transient operating condition trigger emergency adjustment mode, the pre-drill module is used to load it onto the hydropower unit digital twin platform for pre-drilling, including: loading the PID control parameter update amount onto the hydropower unit digital twin platform, and pre-drilling the current transient sub-operating condition and load shedding condition.

[0132] Furthermore, in one embodiment, the trigger mode acquisition module is used to acquire the identified trigger mode based on the fourth operating parameter of the hydropower unit and the identified trigger threshold, including: If the peak value of the unit's swing displacement is less than or equal to the peak displacement threshold, and the unit's active load fluctuation is less than or equal to the load fluctuation threshold, then the identification triggering mode is the steady-state condition triggering pre-adjustment mode. Otherwise, the identification trigger mode is a transient condition trigger emergency adjustment mode.

[0133] Furthermore, in one embodiment, the pre-exercise module is used to obtain pre-exercise results based on the pre-exercise evaluation indicators, including: The pre-performance evaluation indicators are compared with the corresponding qualification thresholds. If all pre-performance evaluation indicators are qualified, the pre-performance result is qualified; otherwise, the pre-performance result is unqualified.

[0134] Furthermore, in one embodiment, when the pre-simulation result is unqualified, the pre-simulation module is further configured to use the update amount of the PID control parameters with unqualified pre-simulation results as the initial action, use the currently running transient sub-condition as the initial state, and employ a deep deterministic strategy gradient algorithm to optimize the PID control parameters to obtain the optimized PID control parameter update amount, and then load it again onto the hydropower unit digital twin platform for pre-simulation. If the pre-simulation result is unqualified after multiple consecutive optimizations of the PID control parameters, a safety protection mechanism is triggered.

[0135] The functions of each module in the control system correspond to the steps in the control method embodiments described above, and their functions and implementation processes will not be elaborated here.

[0136] In the description of this application, it should be noted that the terms "upper," "lower," etc., indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Unless otherwise expressly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication between two elements. For those skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0137] It should be noted that in this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0138] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An adaptive control method for a hydropower unit speed governor, characterized in that, It includes: Multi-scale wavelet decomposition is performed on the integrated state signal after the fusion of the first operating parameters of the hydropower unit to extract feature vectors characterizing the dynamic characteristics of the hydropower unit. The first operating parameters include the intensity of water turbulence and the unit's oscillation displacement; Combining the second operating parameters and feature vector of the hydropower unit, the recursive least squares method with a forgetting factor is used to identify the hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters; the second operating parameters include guide vane opening, unit speed and unit active load; Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller is constructed. The centroid method is used for defuzzification to obtain the correction values ​​of the PID control parameters. The guide vane opening is obtained based on the correction value of the PID control parameters, the reference value of the PID control parameters, and the control deviation of the third operating parameter, so as to perform adaptive control of the hydropower unit governor; the third operating parameter includes the unit speed and the unit active load. The overall status signal is obtained by summing the signals of each of the first operating parameters after weighting them by their respective weighting coefficients.

2. The adaptive control method for the governor of a hydropower unit as described in claim 1, characterized in that: The weighting coefficient is obtained by dividing the reciprocal of the signal variance of the first operating parameter by the sum of the reciprocals of the signal variances of all the first operating parameters.

3. The adaptive control method for the governor of a hydropower unit as described in claim 1, characterized in that, Multi-scale wavelet decomposition is performed on the integrated state signal after fusing the first operating parameters of the hydropower unit to extract feature vectors characterizing the dynamic characteristics of the hydropower unit, including: Multi-scale wavelet decomposition was performed on the integrated state signal after the fusion of the first operating parameters of the hydropower unit to obtain low-frequency approximation coefficients and high-frequency detail coefficients; The energy entropy of the high-frequency detail coefficients is obtained based on the number of sampling points and the total signal energy. A feature vector is constructed using the mean amplitude, peak value, and vibration frequency of the energy entropy and the low-frequency approximation coefficient.

4. The adaptive control method for the governor of a hydropower unit as described in claim 1, characterized in that, Combining the second operating parameters and feature vector of the hydropower unit, a recursive least squares method with a forgetting factor is used to identify the hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters, including: Discretize the transfer function of the hydropower unit's hydraulic system; Using the guide vane opening and eigenvector as inputs, and the unit speed and active load as outputs, the recursive least squares method with a forgetting factor is used to identify the discretized hydraulic system transfer function of the hydropower unit in order to obtain the hydraulic system parameters.

5. The adaptive control method for the governor of a hydropower unit as described in claim 1, characterized in that, Using the hydraulic system parameters and eigenvectors as inputs, and the correction values ​​of the PID control parameters as outputs, a fuzzy controller is constructed. The centroid method is used for defuzzification to obtain the correction values ​​of the PID control parameters, including: Using the hydraulic system parameters and feature vectors as inputs and the correction values ​​of the PID control parameters as outputs, a fuzzy controller with two inputs and three outputs is constructed. The input and output quantities are divided into several fuzzy subsets, and a fuzzy rule set is constructed based on the experience of tuning the PID control parameters of the hydropower unit speed governor. By combining fuzzy subsets and fuzzy rule sets, the centroid method is used to defuzzify the fuzzy controller in order to obtain the correction amount of the PID control parameters.

6. The adaptive control method for a hydropower unit speed governor as described in claim 1, characterized in that, The guide vane opening is obtained based on the correction value of the PID control parameters, the reference value of the PID control parameters, and the control deviation of the third operating parameter, including: The update value of the PID control parameters is obtained based on the correction value of the PID control parameters and the reference value of the PID control parameters. The guide vane opening is obtained based on the PID control parameter update, the control deviation of the unit speed, the control deviation of the unit's active load, and the PID discrete control equation of the hydropower unit governor.

7. The adaptive control method for the governor of a hydropower unit as described in claim 1, characterized in that, Before adaptive control of the hydropower unit speed governor, the method further includes: The updated PID control parameters, obtained from the correction and baseline values ​​of the PID control parameters, are loaded onto the digital twin platform of the hydropower unit for pre-simulation to obtain pre-simulation evaluation indicators. Based on the aforementioned pre-simulation evaluation indicators, the pre-simulation results are obtained; If the pre-simulation results are satisfactory, adaptive control will be implemented on the hydropower unit speed governor.

8. The adaptive control method for the governor of a hydropower unit as described in claim 7, characterized in that: Before performing multi-scale wavelet decomposition, the method further includes: obtaining an identification triggering mode based on the fourth operating parameter of the hydropower unit and the identification triggering threshold, wherein the identification triggering mode includes a steady-state condition triggering pre-adjustment mode and a transient condition triggering emergency adjustment mode. Among them, the steady-state operating condition is divided into several steady-state sub-operating conditions, the transient operating condition is divided into several transient sub-operating conditions, and at least one transient sub-operating condition is a load shedding condition; If the identification triggering mode is a steady-state condition triggering pre-tuning mode, it is loaded onto the hydropower unit digital twin platform for pre-performance, including: loading the PID control parameter update amount onto the hydropower unit digital twin platform, and pre-performance of the currently operating steady-state sub-condition and its adjacent steady-state sub-conditions. If the identification trigger mode is a transient operating condition trigger emergency adjustment mode, it is loaded onto the hydropower unit digital twin platform for pre-simulation, including: loading the PID control parameter update quantity onto the hydropower unit digital twin platform, and pre-simulating the current transient sub-operating condition and load shedding condition.

9. The adaptive control method for a hydropower unit speed governor as described in claim 8, characterized in that: The fourth operating parameter includes the unit's oscillation displacement and the unit's active load; The trigger thresholds include the peak displacement threshold and the load fluctuation threshold; Based on the fourth operating parameter of the hydropower unit and the identification trigger threshold, the identification trigger mode is obtained as follows: If the peak value of the unit's swing displacement is less than or equal to the peak displacement threshold, and the unit's active load fluctuation is less than or equal to the load fluctuation threshold, then the identification triggering mode is the steady-state condition triggering pre-adjustment mode. Otherwise, the identification trigger mode is a transient condition trigger emergency adjustment mode.

10. The adaptive control method for the governor of a hydropower unit as described in claim 7, characterized in that: The pre-performance evaluation indicators include at least one of the following: adjustment time, overshoot, peak value of unit swing displacement, number of oscillations, and steady-state error.

11. The adaptive control method for the governor of a hydropower unit as described in claim 7, characterized in that: Based on the aforementioned pre-rehearsal evaluation indicators, the pre-rehearsal results are obtained, including: The pre-performance evaluation indicators are compared with the corresponding qualification thresholds. If all pre-performance evaluation indicators are qualified, the pre-performance result is qualified; otherwise, the pre-performance result is unqualified.

12. The adaptive control method for the governor of a hydropower unit as described in claim 7, characterized in that: If the pre-simulation result is unsatisfactory, the method further includes: The update values ​​of the PID control parameters that fail the pre-simulation results are used as the initial actions, and the current transient sub-condition is used as the initial state. The PID control parameters are then optimized using a deep deterministic strategy gradient algorithm to obtain the optimized PID control parameter update values, which are then loaded back onto the hydropower unit digital twin platform for pre-simulation.

13. The adaptive control method for a hydropower unit speed governor as described in claim 12, characterized in that: If the pre-test results fail after multiple consecutive optimizations of the PID control parameters, the safety protection mechanism will be triggered.

14. An adaptive control system for a hydropower unit speed governor, characterized in that, It includes: The feature extraction module is used to perform multi-scale wavelet decomposition on the comprehensive state signal after the fusion of the first operating parameters of the hydropower unit, so as to extract feature vectors that characterize the dynamic characteristics of the hydropower unit. The first operating parameters include the water flow turbulence intensity and the unit swing displacement. The comprehensive state signal is obtained by summing the signals of each of the first operating parameters after weighting calculation with their respective weighting coefficients. The identification module is used to identify the hydraulic system transfer function of the hydropower unit by combining the second operating parameters and feature vector of the hydropower unit and using the recursive least squares method with a forgetting factor, so as to obtain the hydraulic system parameters; the second operating parameters include guide vane opening, unit speed and unit active load; The correction quantity acquisition module is used to construct a fuzzy controller by taking the hydraulic system parameters and feature vector as input quantities and the correction quantity of the PID control parameters as output quantities, and to perform defuzzification using the centroid method to obtain the correction quantity of the PID control parameters. The control module is used to obtain the guide vane opening based on the correction amount of the PID control parameters, the reference amount of the PID control parameters, and the control deviation of the third operating parameters, so as to perform adaptive control of the hydropower unit governor; the third operating parameters include the unit speed and the unit active load.

Images