Method and system for improving low-frequency oscillation suppression capability of power system stabilizer

Abstract

The invention relates to the field of power systems, in particular to a method and system for improving the low-frequency oscillation suppression capability of a power system stabilizer, and the method comprises the steps: building an excitation regulator hardware-in-loop simulation platform to form a closed-loop test environment; constructing a model containing prime mover forced oscillation and generator special operation conditions; configuring simulation parameters based on the actual constant value of the unit; injecting multi-modal disturbance and acquiring a dynamic signal; calculating a multi-dimensional index for evaluating the low-frequency oscillation suppression capability of the PSS; generating a full-working-condition adaptability judgment result through a dynamic weight distribution algorithm; and if not, optimizing the PSS parameters by taking the weighted comprehensive score as a target, and repeating the test until the requirements are met. According to the invention, the problem that the low-frequency oscillation suppression capability of the traditional PSS parameter setting is insufficient under a special working condition is solved.

Description

Technical Field

[0001] This invention relates to the field of power systems, and specifically to a method and system for improving the low-frequency oscillation suppression capability of power system stabilizers. Background Technology

[0002] Power system stabilizers (PSS), as an additional control component of excitation regulators, are widely recognized as the most economical and effective means of suppressing low-frequency oscillations. However, with the high proportion of renewable energy integration and the normalization of deep peak shaving by thermal power units, the adaptability defects of traditional PSS under special operating conditions are becoming increasingly apparent: when the unit speed or grid frequency fluctuates frequently near the primary frequency regulation dead zone, excessively low speed inequality will amplify the power response of small frequency difference disturbances (±0.05Hz), inducing forced oscillations in the prime mover. The turbine's integrated valve position remains stagnant in the inflection point region of the flow characteristics for a long time. Repeated adjustments of the valve position further exacerbate power oscillations. During deep peak shaving of the generator, the fixed reactance parameters lead to amplification of the sensorless speed calculation error, significantly reducing the PSS phase compensation capability. The mechanical power signal in the PSS model, after being filtered by fixed parameters, produces significant phase lag in the low-frequency band, resulting in excessive phase deviation between the output torque and the speed, with the damping component approaching zero or even becoming negative. The combination of these factors leads to poor low-frequency oscillation suppression performance of PSS under special operating conditions, posing significant safety hazards. Even more challenging is that traditional PSS performance testing relies on field tests or offline simulations, which cannot reproduce the dynamic coupling effects of valve inflection point nonlinearity, small frequency difference disturbances, and deep phase advance operation. This results in insufficient verification of the practicality and robustness of PSS parameters under all operating conditions, making it difficult to meet the stringent requirements of new power systems for the dynamic stability of units.

[0003] The existing technology CN103187732A proposes a method for optimizing power system stabilizer parameters. This method determines the PSS parameter sets corresponding to different power ranges through field tests and calls up the appropriate parameters based on the actual operating range. However, this method is only based on steady-state power partitioning and fails to consider the coupling effects of dynamic operating conditions such as valve inflection point nonlinearity, small frequency difference disturbances, and deep peak-shaving leading phase operation. This results in insufficient verification of parameter adaptability and fails to fundamentally solve the aforementioned technical problems. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for improving the low-frequency oscillation suppression capability of power system stabilizers, so as to solve the problem of insufficient low-frequency oscillation suppression capability of traditional PSS under special operating conditions.

[0005] To achieve the above objectives, the following technical solution is adopted.

[0006] A method for improving the low-frequency oscillation suppression capability of a power system stabilizer includes the following steps:

[0007] S1. Establish a hardware-in-the-loop simulation platform for the excitation regulator to form a closed-loop test environment. S2. Construct a model including the forced oscillation condition of the prime mover and the special operating condition of the generator. S3. Configure the parameters of the excitation regulator and the hardware-in-the-loop simulation platform based on the actual setpoint of the generator excitation regulator. S4. Inject multi-modal disturbances into the hardware-in-the-loop simulation platform and collect dynamic signals. S5. Calculate multi-dimensional indicators for evaluating the low-frequency oscillation suppression capability of the PSS based on the dynamic signals. S6. Perform weighted processing on the multi-dimensional indicators using a dynamic weight allocation algorithm to generate a full-condition adaptability judgment result for the PSS parameters. S7. If the judgment result shows that the full-condition adaptability requirement is not met, optimize and adjust the PSS parameters using the weighted comprehensive score of the multi-dimensional indicators as the objective function. S8. Configure the optimized PSS parameters into the excitation regulator and repeat the steps of injecting disturbances, collecting signals, calculating indicators, and judging adaptability until the full-condition adaptability requirement is met.

[0008] Optionally, the specific steps for building the excitation regulator hardware-in-the-loop simulation platform in S1 include:

[0009] The refined coupled dynamic model is run through a real-time simulator. The refined coupled dynamic model includes a steam turbine governor using the IEEE G1 model, a synchronous generator using a fifth-order practical model, and the grid impedance of the Thevenin equivalent circuit.

[0010] The generator stator voltage U output by the real-time simulator abc With rotor current I f Input excitation regulator;

[0011] The thyristor trigger pulse signal or UK control voltage generated by the excitation regulator is fed back to the real-time simulator to form a closed-loop test environment.

[0012] Optionally, the forced oscillation model of the prime mover in S2 includes:

[0013] Within the preset valve position inflection point range, a piecewise function is used to construct a flow nonlinear characteristic model, where the flow rate is linearly related to the valve position when it is below the inflection point, and a quadratic function is used for fitting when it is above the inflection point.

[0014] By setting a rotational speed unequal rate below a preset threshold, the power response of small frequency difference disturbances within the ±0.05Hz frequency band is amplified.

[0015] The generator special operating condition model is a deep peak-shaving leading phase operation model, which reproduces phase compensation error by adjusting the reactive power limit.

[0016] Optionally, the multimodal disturbances injected in S4 include repeated valve position inflection point adjustment disturbances, small frequency difference disturbances, and deep peak-shaving phase-advancing disturbances;

[0017] The dynamic signals include generator speed deviation Δω, active power oscillation ΔP, and power system stabilizer output voltage U. pss .

[0018] Optionally, the calculation method for the multi-dimensional indicators in S5 is as follows:

[0019] The damping ratio is obtained by extracting the eigenvalues ​​of the dominant oscillation mode;

[0020] Oscillation decay time Ts is the time it takes for the statistical power oscillation peak to decay to within 5%;

[0021] The oscillation energy ratio Er is obtained through the formula We obtain, where ΔP ref ΔP represents the energy of the non-PSS oscillation and the active power oscillation.

[0022] Optionally, the execution steps of the dynamic weight allocation algorithm in S6 include:

[0023] The energy quantization value E of the multimodal perturbation i Sensitivity coefficient α to predefined operating conditions i Enter formula

[0024] Generate weighting factor w i ;

[0025] When the energy proportion of a specific perturbation mode increases, its weighting factor is automatically increased;

[0026] The weighting factor and the adaptive threshold are linked to generate the judgment result, and the adaptive threshold is obtained by clustering historical oscillation data.

[0027] Optionally, the adaptive threshold generation step includes:

[0028] A sample library was constructed by collecting historical oscillation data from thermal power, hydropower, and gas turbine units.

[0029] Clustering algorithms are used to classify working conditions into categories and the mean μi and standard deviation σi of each category are calculated.

[0030] Based on the characteristics of the current operating conditions, historical categories are matched to generate a dynamic threshold. i =μ i ±2σ i .

[0031] Optionally, the formula for calculating the weighted comprehensive score S in S7 is as follows:

[0032]

[0033] Where X iX represents the actual indicator value. ideal Threshold is the ideal target value for the i-th metric. i The dynamic threshold generated for S7, w i As a weighting factor;

[0034] The full-condition adaptability requirement is S≥0.7 and damping ratioζ≥0.1.

[0035] Optionally, the PSS parameter optimization adjustment in S7 adopts the simplex method, and the optimization variables include:

[0036] Power system stabilizer gain parameters;

[0037] The time constant of the DC blocking element;

[0038] Time constant of the lead-lag element;

[0039] The slope parameter of the slope tracking function.

[0040] A system for improving the low-frequency oscillation suppression capability of power system stabilizers, comprising:

[0041] The simulation environment building unit is used to build a hardware-in-the-loop simulation platform for the excitation regulator to form a closed-loop test environment.

[0042] The disturbance generation unit is used to construct models containing forced oscillation conditions of prime movers and special operating conditions of generators;

[0043] The parameter configuration unit is used to configure the parameters of the excitation regulator and the hardware-in-the-loop simulation platform based on the actual set value of the unit's excitation regulator.

[0044] The test execution unit is used to inject multimodal disturbances through the hardware-in-the-loop simulation platform and to collect generator speed deviation, active power oscillation and PSS output voltage signal.

[0045] The index calculation unit is used to calculate the damping ratio, oscillation decay time, and oscillation energy ratio based on the collected dynamic signals.

[0046] The judgment unit is used to weight multi-dimensional indicators through a dynamic weight allocation algorithm to generate the full-condition adaptability judgment result of PSS parameters.

[0047] The optimization unit is used to optimize the PSS parameters with the weighted comprehensive score as the objective function when the judgment result fails to meet the standard.

[0048] The control unit is used to configure the optimized parameters to the excitation regulator and control the test execution unit to re-inject disturbances for iterative verification until the full-condition adaptability requirements are met.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] This application proposes a method to improve the low-frequency oscillation suppression capability of power system stabilizers (PSS). By building a hardware-in-the-loop (HIL) simulation platform for the excitation regulator, a model is constructed including forced oscillation conditions of the prime mover and special operating conditions of the generator. Multimodal disturbances are injected and dynamic signals are collected. Multi-dimensional indicators are calculated, and a dynamic weight allocation algorithm is used to generate a full-condition adaptability judgment result for PSS parameters. If the results do not meet the standards, optimization and adjustments are made until the full-condition adaptability requirements are met. This method directly solves the problem of insufficient low-frequency oscillation suppression capability of traditional PSSs under special operating conditions. Through the hardware-in-the-loop simulation platform, accurate simulation and parameter optimization of complex operating conditions are achieved, significantly improving the adaptability and robustness of PSSs under all operating conditions. Specifically, this method ensures the synchronization of electromagnetic transients and mechanical transients through a high-precision closed-loop test environment of a real-time simulator and the excitation regulator, enabling key indicators of PSSs under extreme conditions, such as damping ratio, decay time, and oscillation energy ratio, to obtain quantifiable and traceable evaluation benchmarks. The dynamic weight allocation algorithm automatically adjusts the weights of each mode based on the real-time disturbance energy ratio, ensuring that the evaluation results always focus on the weakest link and avoiding the evaluation distortion caused by traditional static weights. In addition, using the weighted comprehensive score S≥0.7 and damping ratio ζ≥0.1 as the unified convergence conditions, the simplex method is used to perform targeted optimization of PSS gain, DC blocking time constant, lead-lag elements, and ramp tracking slope. After each parameter adjustment, the same disturbance combination is re-injected for closed-loop verification, realizing a "verification-correction" closed-loop optimization mechanism to ensure that the PSS parameters can reach the optimal state under all operating conditions.

[0051] This paper further refines the specific steps and methods for simulation platform construction, model building, multimodal disturbance injection, dynamic signal acquisition, multi-dimensional index calculation, dynamic weight allocation algorithm execution, adaptive threshold generation, and PSS parameter optimization and adjustment. The implementation of these technical features makes PSS parameter optimization more accurate and efficient. For example, the paper describes the simulation platform construction process, ensuring a high degree of consistency between the simulation environment and the actual operating environment by running a refined coupled dynamic model. It further refines the construction of models for forced oscillation conditions of the prime mover and special operating conditions of the generator, as well as the injection of multimodal disturbances, enabling a comprehensive evaluation of PSS performance under different operating conditions. The dynamic weight allocation algorithm and adaptive threshold generation further enhance the flexibility and adaptability of the evaluation system. The paper clarifies the specific methods for PSS parameter optimization and adjustment, using the simplex method to optimize key parameters and dynamically adjusting the optimization direction according to failure conditions to ensure the dynamic stability and damping characteristics of the PSS under complex operating conditions. The implementation of these technical features not only enhances the adaptability and robustness of PSS under all operating conditions, but also significantly improves the efficiency and accuracy of parameter optimization, providing a strong guarantee for the stable operation of the power system. Attached Figure Description

[0052] Figure 1 This is a schematic flowchart of an embodiment of a method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to the present invention.

[0053] Figure 2 This is a flowchart of PSS parameter determination and optimization under multimodal disturbances according to an embodiment of a method for improving the low-frequency oscillation suppression capability of a power system stabilizer based on the present invention.

[0054] Figure 3 This is a hardware-in-the-loop simulation data acquisition diagram based on an embodiment of a method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to the present invention.

[0055] Figure 4 This is a schematic diagram of a module according to an embodiment of a power system stabilizer low-frequency oscillation suppression capability enhancement system based on the present invention. Detailed Implementation

[0056] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0057] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0058] Example 1

[0059] like Figure 1 and Figure 2 As shown, the practical implementation of a method to improve the low-frequency oscillation suppression capability of a power system stabilizer includes the following steps:

[0060] The first step is to build a hardware-in-the-loop simulation platform for the excitation regulator to form a closed-loop test environment. This step requires using a real-time simulator to run a refined coupled dynamic model. The construction of this refined coupled dynamic model is a crucial step. The turbine governor uses the IEEE G1 model, which can accurately simulate the operating characteristics of the turbine governor under different operating conditions, providing a foundation for subsequent simulations of the prime mover's related operating conditions. The synchronous generator uses a fifth-order practical model, which considers multiple key parameters and operating state variables of the synchronous generator, and can more realistically reflect the dynamic behavior of the synchronous generator in the power system. The grid impedance is represented by the Thevenin equivalent circuit. This equivalent method simplifies the complex power grid into a circuit model that is easy to analyze and calculate, greatly improving the efficiency and accuracy of the simulation. The real-time simulator outputs the generator stator voltage U... abc With rotor current I f The precise input of these two signals to the excitation regulator is crucial for its accurate determination of the generator's operating status and the generation of appropriate additional excitation control signals. The thyristor trigger pulse signal or UK control voltage generated by the excitation regulator is then fed back to the real-time simulator. This high-resolution feedback signal greatly ensures the accuracy and timeliness of signal transmission in the closed-loop test environment, enabling the real-time simulator to quickly adjust model parameters based on the feedback signal. This allows for the simulation of a power system dynamic process that more closely resembles actual operating conditions, ultimately forming a closed-loop test environment.

[0061] Next, models were constructed for the forced oscillation condition of the prime mover and special operating conditions of the generator. The construction of the model for the forced oscillation condition of the prime mover is highly targeted. Within the preset valve position inflection point range, the flow nonlinear characteristic model is constructed using a piecewise function. When the flow is below the inflection point, the flow rate and valve position have a linear relationship. This linear relationship conforms to the basic law of flow rate change with valve position under certain operating conditions and can accurately describe the flow characteristics under this condition. When the flow rate is above the inflection point, a quadratic function is used for fitting. This is because after the valve position is above the inflection point, the flow rate change exhibits more complex nonlinear characteristics, and the quadratic function can better fit this trend, thus more accurately simulating the operation of the prime mover under this condition. By setting a speed unequal rate below a preset threshold, the power response of small frequency difference disturbances within the ±0.05Hz frequency band can be effectively amplified. In actual power system operation, small frequency difference disturbances within this frequency band may cause low-frequency oscillation problems. This method allows for a more prominent study and simulation of the impact of such disturbances on the system, providing a basis for subsequent evaluation and improvement of the PSS's ability to suppress small frequency difference disturbances. The generator special operating condition model adopts a deep peak shaving leading phase operation model. In actual implementation, the phase compensation error is reproduced by precisely adjusting the reactive power limit. During deep peak shaving leading phase operation, the reactive power output and phase compensation of the generator are significantly different from those under normal operating conditions. Through this precise adjustment, the electrical characteristics of the generator under special operating conditions can be realistically simulated, providing the necessary conditions for a comprehensive evaluation of the PSS performance under various operating conditions.

[0062] Based on the actual setpoints of the generator excitation regulator, the parameters of the excitation regulator and the hardware-in-the-loop simulation platform are configured. In actual operation, a detailed investigation and analysis of the various actual setpoints of the generator excitation regulator is required. These setpoints include, but are not limited to, excitation current limits, voltage regulation ranges, and the setpoints of various control parameters. Based on these actual setpoints, the internal parameters of the excitation regulator are precisely configured to ensure that the excitation regulator can control according to actual operating requirements during simulated operation. Simultaneously, the parameters of the hardware-in-the-loop simulation platform are configured synchronously. For example, model parameters in the real-time simulator, such as the electrical and mechanical parameters of the generator and the line parameters of the power grid, need to be set according to the parameters of the actual power system. Furthermore, parameters related to data acquisition and transmission, such as the sampling frequency and data transmission rate of the simulation platform, also need to be reasonably configured according to actual needs and system performance requirements to ensure that the entire simulation system can operate accurately and efficiently.

[0063] Multimodal disturbances are injected into a hardware-in-the-loop (HIL) simulation platform to collect dynamic signals. During the injection of multimodal disturbances, the valve position inflection point is repeatedly adjusted to generate repeated inflection point adjustment disturbances. This disturbance simulates the sudden changes in flow and power fluctuations caused by valve adjustments during the actual operation of the prime mover, and is an important disturbance method for studying the ability of the power supply system (PSS) to suppress low-frequency oscillations under relevant prime mover operating conditions. Small frequency difference disturbances simulate small frequency differences in the power system caused by load changes and frequency fluctuations by precisely adjusting the frequency of the input signal within a specific frequency range, in order to study the PSS's suppression effect on low-frequency oscillations under small frequency difference disturbances. Deep peak shaving leading-phase disturbances further apply disturbance signals based on the simulation of deep peak shaving leading-phase operation of the generator to examine the PSS's ability to suppress low-frequency oscillations under special generator operating conditions. High-precision sensors and data acquisition equipment are used for collecting dynamic signals. The generator speed deviation Δω can be measured using a speed sensor mounted on the generator shaft. This sensor can acquire the generator speed information in real time and accurately, converting it into an electrical signal and transmitting it to the data acquisition system. The active power oscillation ΔP can be obtained by measuring the voltage and current signals on the generator output lines and using a power calculation algorithm. The power system stabilizer output voltage U... pss Data can be collected directly from the PSS output. These collected dynamic signals will be transmitted to subsequent data processing and analysis units in a high-speed and stable data transmission manner, providing a data foundation for calculating multi-dimensional indicators.

[0064] Multi-dimensional indicators for evaluating the low-frequency oscillation suppression capability of the PSS (Power Suppressor System) are calculated based on dynamic signals. The damping ratio is calculated by extracting the eigenvalues ​​of the dominant oscillation mode. In practice, professional signal analysis algorithms and software are used to process the acquired dynamic signals. First, the time-domain signal is converted to the frequency-domain signal using methods such as Fourier transform. Then, modal analysis is used to identify the dominant oscillation mode of the system. For the identified dominant oscillation mode, eigenvalue solving algorithms are used to extract its eigenvalues. The damping ratio is obtained based on the real and imaginary parts of the eigenvalues ​​using a specific formula. The oscillation decay time Ts is statistically analyzed by monitoring the changes in the peak power oscillation. During data processing, the peak value of the active power oscillation ΔP is tracked in real time. When the peak value decays to within 5% of the initial value, the time is recorded; this time is the oscillation decay time Ts. The oscillation energy ratio Er is calculated strictly according to the formula. The calculation is performed by integrating the collected active power oscillation quantity ΔP over a certain time period using an integral algorithm to obtain the numerator value. The denominator is the non-PSS oscillation energy ΔP. refIn this case, under the same simulation conditions, the PSS needs to be removed from the system first, the simulation needs to be run again and relevant data collected, and the oscillation energy at this time needs to be calculated as the denominator. In this way, the oscillation energy ratio E can be accurately calculated. r This provides an important quantitative indicator for comprehensively evaluating the low-frequency oscillation suppression capability of PSS.

[0065] A dynamic weighting algorithm is used to weight multi-dimensional indicators to generate the PSS parameter's adaptability assessment result under all operating conditions. When implementing the dynamic weighting algorithm, the first step is to quantize the multimodal disturbance energy value E. i Sensitivity coefficient α to predefined operating conditions i Input a specific formula to generate the weighting factor w i Multimodal perturbation energy quantization value E i To obtain the data, energy analysis of various injected disturbance signals is required. For example, for repeated valve position inflection point adjustment disturbances, small frequency difference disturbances, and deep peak-shaving phase-advancing disturbances, their energy values ​​are calculated throughout the simulation process and normalized to obtain the corresponding quantized values. The predefined operating condition sensitivity coefficient αi is a coefficient pre-set based on research and experience on the impact of low-frequency oscillations on the power system under different operating conditions. When the energy proportion of a specific disturbance mode increases, the algorithm automatically increases its weight factor, a process that is automated through programming. The weight factor and adaptive threshold are linked to generate the judgment result, and the generation of the adaptive threshold is relatively complex. First, historical oscillation data of thermal power, hydropower, and gas turbine units are collected extensively. These data cover the oscillation information of different types of units under various operating conditions. Then, a clustering algorithm is used to process these data. The clustering algorithm can group similar data into one category based on the characteristics of the data, thereby classifying different operating condition categories. For each operating condition category, the mean μi and standard deviation σi of its various indicators (such as damping ratio, oscillation decay time, oscillation energy ratio, etc.) are calculated. In actual operation, based on the characteristics of the current operating conditions, such as unit type, operating load, and grid topology, the system matches the operating condition categories in historical data to find the most similar category. Based on the matching results, a dynamic threshold (Threshold) is generated. i =μ i ±2σ i The calculated weighting factors are combined with dynamic thresholds, and specific judgment rules are used to finally generate the PSS parameter's adaptability judgment result under all operating conditions.

[0066] If the assessment result indicates that the full-condition adaptability requirements are not met, the PSS parameters will be optimized and adjusted using the weighted comprehensive score of multi-dimensional indicators as the objective function. The formula for calculating the weighted comprehensive score S has a specific form, where X... i Threshold is the actual indicator value. iw is the dynamic threshold generated earlier. i This is used as a weighting factor. In the actual calculation of the weighted comprehensive score S, the previously calculated actual index values ​​X, such as damping ratio, oscillation decay time, and oscillation energy ratio, are used as weighting factors. i and the corresponding dynamic threshold Threshold i and weighting factor w i Substitute the values ​​into the formula for calculation. The full-condition adaptability requirement is explicitly stated as S≥0.7 and damping ratioζ≥0.1. When the judgment result does not meet this requirement, the PSS parameters are optimized and adjusted using the simplex method. Optimization variables include power system stabilizer gain parameters, DC blocking time constant, lead-lag time constant, and slope parameter of the ramp tracking function. After each parameter adjustment, the same multimodal disturbance combination needs to be re-injected for verification. During the verification process, the previous steps of injecting disturbances, acquiring signals, calculating indicators, and determining adaptability are repeated to ensure that the adjusted PSS parameters can truly improve its ability to suppress low-frequency oscillations under full-condition conditions.

[0067] The optimized PSS parameters are configured to the excitation regulator, and the steps of injecting disturbances, acquiring signals, calculating indicators, and determining adaptability are repeated until the full-condition adaptability requirements are met. In actual implementation, after obtaining new PSS parameters through optimization, these parameters are accurately input into the parameter setting module of the excitation regulator. Then, the hardware-in-the-loop simulation platform is restarted, and the same multimodal disturbances are injected in the same way as before, acquiring signals such as generator speed deviation, active power oscillation, and PSS output voltage. Using these signals, indicators such as damping ratio, oscillation decay time, and oscillation energy ratio are recalculated, and a new PSS parameter full-condition adaptability determination result is generated through a dynamic weight allocation algorithm. If the requirements are still not met, parameter optimization and adjustment continue, and this process is repeated until the full-condition adaptability requirements of S≥0.7 and damping ratioζ≥0.1 are finally met, thereby effectively improving the low-frequency oscillation suppression capability of the power system stabilizer.

[0068] Example 2

[0069] like Figures 1-3 As shown, a method for improving the low-frequency oscillation suppression capability of a power system stabilizer includes the following steps:

[0070] First, the hardware-in-the-loop (PSS) simulation system architecture is constructed, and a hardware-in-the-loop simulation platform based on a real-time simulator and excitation regulator is built. The simulator runs a refined model. In one possible implementation, the refined model includes the turbine and governor (using the IEEE G1 model), the synchronous generator (a fifth-order practical model), the transformer (T-type equivalent circuit), and the power grid (Thevenin equivalent impedance). The simulator outputs the generator stator voltage U. abc Rotor current I fThe excitation regulator outputs a control signal to the simulator to ensure the synchronization of the strong electromagnetic transient process.

[0071] Second, the key prime mover output model is constructed and parameters are configured. In the construction of the nonlinear model for the valve flow, a multi-segment fine fitting function is used to describe the inflection point of the flow characteristics. In one possible implementation, a piecewise function is used, where the flow rate and valve position at the inflection point are the intersection of the nonlinear characteristic model and the quadratic function. When the comprehensive valve position command V ≤ 75%, the flow rate Qv = k1·V (linear region); when V > 75%, Qv = k2·V 2 +b (nonlinear region), where k1, k2, and b are fitted using the measured flow curves of the steam turbine. In the construction of the overexcitation primary frequency regulation model, the speed unequal rate R is set to be 4%-5% lower than the national standard. Taking R = 1.67% as an example, under small frequency difference disturbances (Δf = ±0.033Hz~±0.05Hz), the frequency regulation power ΔP FR =Δf·P rated / R, so that the power response reaches 3 times the standard value, simulating an excessive frequency modulation parameter scenario.

[0072] Third, the excitation regulator system parameters are configured. Based on the actual set value of the unit's excitation regulator, relevant parameters of the excitation regulator are set, including but not limited to various limiting links such as voltage control main loop, PSS, low excitation limit, and V / Hz limit, to ensure consistency with the actual operating parameters.

[0073] IV. Conduct dynamic data acquisition. Real-time acquisition of Δω, ΔP, and U. abc U pss Dynamic index calculations are performed on signals such as Figure 4 As shown. In one possible implementation, the dominant oscillation mode is extracted, and the damping ratio is calculated. The peak-to-peak power of the statistical power oscillation decays from its maximum value to within 5% of the power loss Ts. Calculate U pss The phase difference with Δω and the oscillation energy ratio (Er), ΔP ref This refers to the energy of the PSS oscillation.

[0074] V. Develop a dynamic weight allocation algorithm.

[0075]

[0076] E i : Energy of the i-th perturbation mode; α i Modal sensitivity coefficient.

[0077] VI. Adaptive Threshold Generation for Multimodal Disturbances. Historical oscillation data of typical generating units (thermal power, hydropower, and gas turbine) under different operating conditions are collected to construct a threshold sample library. Clustering algorithms are used to classify operating conditions, and the mean value μ of the index for each category is calculated. i With standard deviation σ i Perform dynamic threshold calculation, match the current working condition category in real time, and generate dynamic thresholds:

[0078] Threshold i =μ i ±2σ i

[0079] For the current time window data, extract the same time domain, frequency domain, and operating condition features as the historical database, and calculate the distance between the real-time feature vector and the center point of each category.

[0080] VII. Establish a multimodal perturbation energy quantification model.

[0081] 1. Quantification of disturbance energy at the valve inflection point:

[0082]

[0083] Qv: Flow characteristic function of the control valve; ΔVk: Jump variable of valve position command in the inflection point region.

[0084] 2. Energy of small frequency difference disturbance (E) FReq ):

[0085]

[0086] RF R Rotational speed inequality

[0087] 3. Construct other perturbation energies.

[0088] 8. Perform parameter determination, optimization, and iterative verification. In one possible implementation, a comprehensive score S is used to determine the PSS parameters:

[0089]

[0090] Among them, X i X represents the actual indicator value. ideal Threshold is the ideal target value for the i-th metric. i For the generated dynamic threshold, w i This is the weighting factor.

[0091] Output the judgment result: S≥0.7 is qualified, S<0.7 requires optimization and iterative verification.

[0092] Parameter optimization uses the weighted comprehensive score S as the objective function, and in one possible implementation, it employs NeldE.r The Mead simplex method automatically optimizes the core parameters of the PSS (gain parameters, DC blocking parameters, lead / lag parameters, and ramp tracking function parameters), as shown in the flowchart below. Figure 2 As shown, the specific process is as follows:

[0093] Initialization optimization: Based on the actual setpoint of the excitation regulator, initial parameters are set, multi-modal mixed disturbances are injected, and signals are acquired through a real-time simulation platform to calculate the damping ratio ζ, decay time Ts, and oscillation energy ratio E under the current operating conditions. r ;

[0094] Dynamic weight allocation: Match the current working condition category, call the mean and standard deviation of damping ratio of the same working condition in the historical sample library, generate a dynamic threshold, and trigger parameter optimization if the measured damping ratio < 0.1 or S < 0.7.

[0095] Parameter adjustment and iteration to improve the S value: Parameter adjustment is performed in the direction of parameter adjustment, for example, when E is under deep peaking and phase initiation conditions. r When E = 0.12 (exceeding the limit) and S = 0.65, the optimization algorithm increases the gain K1 from 8.0 to 9.6 and adjusts T1 from 0.1s to 0.15s (phase compensation shift), re-injects the same perturbation, and tests are performed. If E r If the value drops to 0.08, S rises to 0.88, and the damping ratio ζ ≥ 0.1, then the current operating condition is considered to meet the standard; otherwise, continue iterating until all preset operating conditions meet the condition that S ≥ 0.7 and the damping ratio ζ ≥ 0.1.

[0096] Example 3

[0097] like Figure 4 As shown, based on the above-mentioned method for improving the low-frequency oscillation suppression capability of a power system stabilizer, the constructed method for improving the low-frequency oscillation suppression capability of a power system stabilizer has specific implementation methods in actual operation.

[0098] The simulation environment construction unit is responsible for building a hardware-in-the-loop simulation platform for the excitation regulator to form a closed-loop test environment. This unit integrates a high-performance real-time simulator and related control and interface circuits. During the actual construction process, technicians will accurately construct a refined coupled dynamic model in the real-time simulator based on the actual parameters and operating requirements of the power system. For the turbine governor, the parameter settings using the IEEE G1 model will be set with reference to the actual turbine model, rated power, speed regulation characteristics, and other parameters. Parameters of the fifth-order practical model of the synchronous generator, such as stator resistance, reactance, rotor resistance, reactance, and inertia time constant, will be accurately input based on the generator's design drawings and actual operating test data. Thevenin equivalent circuit parameters of the grid impedance are obtained through measurement and calculation of the grid line parameters. The data transmission interface between the real-time simulator and the excitation regulator will use a high-speed, reliable communication protocol, such as fiber optic communication or high-speed Ethernet communication, to ensure the generator stator voltage U... abc With rotor current I f It can transmit data quickly and accurately to the excitation regulator, and the excitation additional control signal generated by the excitation regulator can also be fed back to the real-time simulator at high resolution, thus successfully building a closed-loop test environment.

[0099] The disturbance generation unit is used to construct models including forced oscillation conditions of the prime mover and special operating conditions of the generator. This unit is equipped with specialized modeling software and data processing modules. When constructing the forced oscillation model of the prime mover, technicians accurately set the preset valve position inflection point range in the modeling software based on the operating characteristics of the prime mover in the actual power system and historical fault data. For the flow nonlinear characteristic model, a piecewise function is constructed by writing a program. When the flow is below the inflection point, the linear relationship coefficient between the flow and the valve position is set; when it is above the inflection point, the coefficients of the quadratic function are determined to accurately fit the flow characteristics. When setting the speed unequal rate, the value of the speed unequal rate is precisely adjusted based on the analysis of its impact on power system stability to amplify the power response of small frequency difference disturbances within the ±0.05Hz frequency band. When constructing the special operating condition model of the generator, based on the technical requirements of deep peak shaving and leading phase operation of the generator and actual operating experience, the reactive power limit parameters are adjusted in the modeling software to accurately reproduce the phase compensation error, thus completing the construction of the special operating condition model of the generator.

[0100] The parameter configuration unit configures the parameters of the excitation regulator and the hardware-in-the-loop (HIL) simulation platform based on the actual setpoints of the unit's excitation regulator. This unit has a bidirectional communication interface with both the excitation regulator and the HIL simulation platform, as well as a parameter storage and management database. In practice, technicians first obtain the actual setpoints of the excitation regulator from its parameter setting panel or communication interface and store them in the database. Then, based on these actual setpoints, they send parameter configuration commands to the excitation regulator via the communication interface to accurately set its control and protection parameters. Simultaneously, parameters related to the HIL simulation platform, such as model parameters, data acquisition and transmission parameters from the real-time simulator, are read from the database and configured into the HIL simulation platform via the communication interface, ensuring that the parameter configuration of the entire system matches the actual unit and operational requirements.

[0101] The test execution unit injects multimodal disturbances through a hardware-in-the-loop simulation platform and acquires generator speed deviation, active power oscillation, and PSS output voltage signals. This unit includes a disturbance signal generator, a data acquisition card, and signal conditioning circuitry. When injecting multimodal disturbances, the disturbance signal generator generates signals for repeated valve position inflection point adjustment, small frequency difference disturbances, and deep peak-shaving leading phase disturbances, based on preset disturbance types and parameters. These disturbance signals are amplified and filtered by the signal conditioning circuitry before being injected into the system through the input interface of the hardware-in-the-loop simulation platform. For signal acquisition, the data acquisition card connects to the speed sensor mounted on the generator shaft, the voltage and current sensors on the generator output lines, and the PSS output terminal via sensor interfaces. The speed sensor converts the generator speed deviation Δω into an electrical signal, which is processed by the signal conditioning circuitry and then acquired by the data acquisition card. The signals acquired by the voltage and current sensors are processed by the signal conditioning circuitry and the power calculation module to calculate the active power oscillation ΔP, which is then acquired by the data acquisition card. The PSS output voltage U... pss The data is then directly acquired from the PSS output by the data acquisition card. These acquired signals are transmitted to the subsequent indicator calculation unit via the data acquisition card's communication interface using a high-speed, stable data transmission method.

[0102] The index calculation unit calculates the damping ratio, oscillation decay time, and oscillation energy ratio based on the acquired dynamic signals. This unit uses professional signal analysis software and calculation modules. When calculating the damping ratio, the signal analysis software first performs a Fourier transform on the acquired dynamic signals, converting the time-domain signal to a frequency-domain signal. Then, it uses a modal analysis algorithm to identify the dominant oscillation mode of the system. The eigenvalue solving algorithm extracts the eigenvalues ​​of the dominant oscillation mode, and the damping ratio is calculated based on the eigenvalues ​​and the damping ratio calculation formula. The oscillation decay time Ts is calculated by setting a power oscillation peak monitoring program in the signal analysis software to track the peak value of the active power oscillation ΔP in real time. When the peak value decays to within 5% of the initial value, the software records this time, which is the oscillation decay time Ts. The oscillation energy ratio E... r The software uses an integral algorithm to integrate the collected active power oscillation quantity ΔP over a certain time period to obtain the numerator value. Simultaneously, the software calls upon previously collected data from simulations without a power oscillation system (PSS) to calculate the PSS-free oscillation energy ΔP. ref As the denominator, the final oscillation energy ratio E is calculated. r .

[0103] The judgment unit uses a dynamic weight allocation algorithm to weight multi-dimensional indicators and generates a full-condition adaptability judgment result for the PSS parameters. This unit integrates the dynamic weight allocation algorithm program and the data storage and processing module. When implementing the dynamic weight allocation algorithm, the algorithm program first reads the multimodal perturbation energy quantization value E from the data storage and processing module. i and the predefined operating condition sensitivity coefficient α i Multimodal perturbation energy quantization value E i The disturbance signal is collected by the test execution unit, calculated using an energy analysis algorithm, and stored in the data storage and processing module. A predefined sensitivity coefficient α is used. i These values ​​are pre-set based on power system operation experience and expert knowledge and stored in this module. The algorithm program substitutes these values ​​into a specific formula to generate the weighting factor w. i When the energy proportion of a specific disturbance mode increases, the algorithm automatically increases its weight factor through logical judgment and calculation. Regarding the generation of adaptive thresholds, the data storage and processing module first reads historical oscillation data from the historical oscillation data database for thermal power, hydropower, and gas turbine units. This data is accumulated through long-term power system operation monitoring and data collection. Then, a clustering algorithm is used to process this data, dividing it into different operating condition categories and calculating the mean μi and standard deviation σi for each category. In actual operation, based on the characteristic information of the current operating condition, the judgment unit obtains this information from the system's operating condition monitoring module and matches it with the operating condition categories in the historical data. The most similar category is found, and a dynamic threshold (Threshold) is generated based on the matching result.i =μi±2σi. Finally, the algorithm combines the weighting factor with the dynamic threshold and generates the full-condition adaptability judgment result of the PSS parameters through preset judgment rules.

[0104] When the judgment result fails to meet the standard, the optimization unit optimizes the PSS parameters using a weighted comprehensive score as the objective function. This unit possesses a parameter optimization algorithm library and communication interfaces with other units. During the optimization process, when the unit receives a signal from the judgment unit indicating that the PSS parameter's full-condition adaptability judgment result fails to meet the standard (i.e., S < 0.7 or damping ratio ζ < 0.1), the optimization unit initiates the parameter optimization process. First, it obtains the actual values ​​X of the multi-dimensional indicators from the indicator calculation unit. i Dynamic threshold i And obtain the weight factor w from the decision unit i The weighted comprehensive score S is calculated by substituting the values ​​into the weighted comprehensive scoring formula. Using this score as the objective function, the simplex method from the parameter optimization algorithm library is called to optimize the PSS parameters. Optimization variables include power system stabilizer gain parameters, DC blocking time constants, lead-lag time constants, and slope parameters of the ramp tracking function. During optimization, the algorithm dynamically adjusts the parameter optimization direction based on failure conditions. For example, when insufficient suppression of small frequency difference disturbances is detected, the algorithm automatically increases the gain parameter and shifts the phase compensation time constant backward; when slow decay of valve position inflection point oscillations is detected, the slope of the ramp tracking function is increased. After each parameter adjustment, the optimization unit sends the adjusted parameters to the parameter configuration unit via the communication interface for subsequent verification.

[0105] The control unit configures the optimized parameters to the excitation regulator and controls the test execution unit to re-inject disturbances for iterative verification until the full-condition adaptability requirements are met. This unit is the core control part of the entire system, possessing powerful logic control and timing management capabilities. Upon receiving the optimized PSS parameters from the optimization unit, the control unit accurately configures these parameters into the excitation regulator via the communication interface with the excitation regulator. Then, it sends control commands to the test execution unit, instructing it to re-inject multimodal disturbances according to the previous disturbance type and parameters. Simultaneously, the control unit coordinates the index calculation unit and the judgment unit to perform signal analysis, index calculation, and adaptability judgment according to a predetermined process. Throughout the iterative verification process, the control unit monitors the judgment results in real time. If the judgment results meet the full-condition adaptability requirements of S≥0.7 and damping ratioζ≥0.1, the control unit stops the iteration, completing the optimization and improvement of the PSS parameters. If not, it continues to control the optimization unit to adjust parameters and repeats the above configuration, disturbance injection, index calculation, and judgment steps until the full-condition adaptability requirements are met.

[0106] During the actual operation of the entire system, the various units exchange data and transmit commands through a high-speed data bus and a unified communication protocol to ensure the coordinated operation of the system. For example, after the simulation environment construction unit sets up the closed-loop test environment, it sends a ready signal to the control unit. Upon receiving this signal, the control unit sends a start command to the disturbance generation unit and the parameter configuration unit, instructing them to begin building the model and configuring parameters. Once the model construction and parameter configuration are complete, each unit sends a completion signal back to the control unit, which then sends a start test command to the test execution unit, initiating the disturbance injection and signal acquisition process. This coordinated working mechanism ensures that the entire system can operate efficiently and accurately according to the predetermined process, thereby effectively improving the low-frequency oscillation suppression capability of the power system stabilizer and meeting the stringent requirements of high-proportion renewable energy access and deep peak shaving for the dynamic stability of the units.

[0107] As is known from common technical knowledge, this invention can be implemented through other embodiments that do not depart from its spirit or essential characteristics. Therefore, the disclosed embodiments described above are merely illustrative in all respects and are not the only ones. All modifications within the scope of this invention or its equivalents are included in this invention.

Claims

1. A method for improving the low-frequency oscillation suppression capability of a power system stabilizer, characterized in that, Includes the following steps: S1. Build a hardware-in-the-loop simulation platform for the excitation regulator to form a closed-loop test environment. S2. Construct a model containing the forced oscillation condition of the prime mover and the special operating condition of the generator. S3. Configure the parameters of the excitation regulator and the hardware-in-the-loop simulation platform based on the actual set value of the generator excitation regulator. S4. Inject multimodal disturbances into the hardware-in-the-loop simulation platform and collect dynamic signals. S5. Calculate multi-dimensional indicators for evaluating the low-frequency oscillation suppression capability of PSS based on the dynamic signal; S6. Perform weighted processing on the multi-dimensional indicators through a dynamic weight allocation algorithm to generate the PSS parameter full-condition adaptability judgment result. S7. If the judgment result shows that the full working condition adaptability requirement is not met, the PSS parameter is optimized and adjusted using the weighted comprehensive score of the multi-dimensional indicators as the objective function. S8. Configure the optimized PSS parameters to the excitation regulator, and repeat the steps of injecting disturbance, acquiring signals, calculating indicators and determining adaptability until the full-condition adaptability requirements are met.

2. The method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The specific steps for building the excitation regulator hardware-in-the-loop simulation platform in S1 include: The refined coupled dynamic model is run through a real-time simulator. The refined coupled dynamic model includes a steam turbine governor using the IEEE G1 model, a synchronous generator using a fifth-order practical model, and the grid impedance of the Thevenin equivalent circuit. The generator stator voltage U output by the real-time simulator abc With rotor current I f Input excitation regulator; The thyristor trigger pulse signal or UK control voltage generated by the excitation regulator is fed back to the real-time simulator to form a closed-loop test environment.

3. The method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The forced oscillation model of the prime mover in S2 includes: Within the preset valve position inflection point range, a piecewise function is used to construct a flow nonlinear characteristic model. When the flow rate is below the inflection point, the flow rate and valve position are linearly related. When the flow rate is above the inflection point, a quadratic function is used for fitting. At the inflection point, the flow rate and valve position are the intersection of the nonlinear characteristic model and the quadratic function. By setting a rotational speed unequal rate below a preset threshold, the power response of small frequency difference disturbances within the ±0.05Hz frequency band is amplified. The generator special operating condition model is a deep peak-shaving leading phase operation model, which reproduces phase compensation error by adjusting the reactive power limit.

4. The method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The multimodal disturbances injected in S4 include repeated valve position inflection point adjustment disturbances, small frequency difference disturbances, and deep peak modulation phase advance disturbances. The dynamic signals include generator speed deviation Δω, active power oscillation ΔP, and power system stabilizer output voltage U. pss .

5. The method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The calculation method for the multi-dimensional indicators in S5 is as follows: The damping ratio is obtained by extracting the eigenvalues ​​of the dominant oscillation mode; Oscillation decay time Ts is the time it takes for the statistical power oscillation peak to decay to within 5%; Oscillation energy ratio E r Through formula Where ΔP ref ΔP represents the energy of the non-PSS oscillation and the active power oscillation.

6. The method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The execution steps of the dynamic weight allocation algorithm in S6 include: The energy quantization value E of the multimodal perturbation i Sensitivity coefficient α to predefined operating conditions i Enter formula Generate weighting factor w i ; When the energy proportion of a specific perturbation mode increases, its weighting factor is automatically increased; The weighting factor and the adaptive threshold are linked to generate the judgment result, and the adaptive threshold is obtained by clustering historical oscillation data.

7. A method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 6, characterized in that, The steps for generating the adaptive threshold include: A sample library was constructed by collecting historical oscillation data from thermal power, hydropower, and gas turbine units. Clustering algorithms are used to classify operating conditions and the mean μ of each index is calculated. i and standard deviation σ i ; Based on the characteristics of the current operating conditions, historical categories are matched to generate a dynamic threshold. i =μ i ±2σ i .

8. The method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The formula for calculating the weighted comprehensive score S in S7 is as follows: Where X i X represents the actual indicator value. ideal Threshold is the ideal target value for the i-th metric. i The dynamic threshold generated for S7, w i As a weighting factor; The full-condition adaptability requirement is S≥0.7 and damping ratioζ≥0.

1.

9. A method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to claim 1, characterized in that, The PSS parameter optimization adjustment in S7 adopts the simplex method, and the optimization variables include: Power system stabilizer gain parameters; The time constant of the DC blocking element; Time constant of the leading and lagging components; The slope parameter of the slope tracking function.

10. A system for improving the low-frequency oscillation suppression capability of a power system stabilizer, based on the method for improving the low-frequency oscillation suppression capability of a power system stabilizer according to any one of claims 1-9, characterized in that, include, The simulation environment building unit is used to build a hardware-in-the-loop simulation platform for the excitation regulator to form a closed-loop test environment. The disturbance generation unit is used to construct models containing forced oscillation conditions of prime movers and special operating conditions of generators; The parameter configuration unit is used to configure the parameters of the excitation regulator and the hardware-in-the-loop simulation platform based on the actual set value of the unit's excitation regulator. The test execution unit is used to inject multimodal disturbances through the hardware-in-the-loop simulation platform and to collect generator speed deviation, active power oscillation and PSS output voltage signal. The index calculation unit is used to calculate the damping ratio, oscillation decay time, and oscillation energy ratio based on the collected dynamic signals. The judgment unit is used to weight multi-dimensional indicators through a dynamic weight allocation algorithm to generate the full-condition adaptability judgment result of PSS parameters. The optimization unit is used to optimize the PSS parameters with the weighted comprehensive score as the objective function when the judgment result fails to meet the standard. The control unit is used to configure the optimized parameters to the excitation regulator and control the test execution unit to re-inject disturbances for iterative verification until the full-condition adaptability requirements are met.

Images