Adaptive algorithm and system for improving frequency measurement precision and response speed of speed regulator

Abstract

The invention provides a self-adaptive algorithm and system for improving frequency measurement precision and response speed of a speed regulator. The method comprises the following steps: Step 1, signal acquisition and preprocessing; step 2, dynamically adjusting filtering parameters, and suppressing power grid signal noise; 3, extracting frequency characteristics in real time, and improving the frequency measurement resolution; 4, a synchronous clock is adopted as a reference frequency source, a clock reference value is periodically and automatically corrected, and long-term drift is eliminated; 5, a nonlinear dynamic model of the hydroelectric generating set is constructed, the nonlinear relation between the valve opening degree and output of the speed regulator is compensated in real time, and the frequency modulation precision is improved; and step 6, outputting the instantaneous frequency. According to the scheme, through the self-adaptive frequency measurement algorithm, high-precision clock calibration, a hardware acceleration mechanism and a nonlinear compensation strategy, a multi-dimensional optimization system is constructed, high-precision frequency measurement is carried out, a rapid effect can be achieved, and a hydroelectric system has long-term stability and high adaptability.

Description

Technical Field

[0001] The present invention relates to the technical field of hydropower station modulation, and in particular to an adaptive algorithm and system for improving frequency measurement accuracy and response speed of a speed regulator. Background Art

[0002] The speed regulator is the core equipment for peak and frequency regulation in hydropower stations. It responds to grid frequency fluctuations by adjusting the output of hydro-generator units to ensure stable grid operation. However, existing speed regulator control systems have significant problems: 1. Nonlinear and time-varying characteristics: The dynamic characteristics of hydropower units are complex. Traditional speed regulator control methods often use linear processing. This control method is difficult to adapt to the nonlinear changes of the system, resulting in insufficient frequency regulation accuracy.

[0003] 2. Existing frequency measurement methods have poor anti-interference capabilities: Existing frequency measurement channels are easily affected by noise and cannot accurately capture small changes in grid frequency (such as ±0.01Hz).

[0004] 3. Response delay: Traditional algorithms have low computational efficiency and insufficient hardware processing capabilities, resulting in frequency measurement response times exceeding 200ms, making it difficult to meet fast frequency modulation requirements.

[0005] 4. Time reference value drift: During long-term operation, the clock source becomes unstable and the reference frequency drift error accumulates. Time reference value drift will affect the reliability of regulation. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an adaptive algorithm and system for improving the frequency measurement accuracy and response speed of the speed regulator, so as to solve the technical defects of the frequency regulation of the speed regulator of the hydropower station.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is: An adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator includes the following steps: Step 1, signal acquisition and preprocessing; Step 2: Use an adaptive frequency measurement algorithm; utilize an improved minimum mean square error (LMS) adaptive filter to dynamically adjust filter parameters and suppress grid signal noise; Step 3: Combine short-time Fourier transform (STFT) and sliding window technology to extract frequency features in real time and improve frequency measurement resolution; Step 4: Use a synchronous clock as the reference frequency source and embed a periodic calibration algorithm to periodically and automatically calibrate the clock reference value to eliminate long-term drift. Step 5: Build a nonlinear dynamic model of the hydropower unit and design an adaptive control law based on Lyapunov stability theory to compensate for the nonlinear relationship between the governor valve opening and output in real time, thereby improving frequency regulation accuracy. Step 6. Output instantaneous frequency.

[0008] The specific steps of signal acquisition and preprocessing in Step 1 above are: Collect the three-phase voltage signal of the power grid, the output current signal of the hydro-generator set and the feedback signal of the governor valve opening; Signal preprocessing: First, filter, then remove the jump data, then suppress the noise, and finally synchronize the data.

[0009] In the above-mentioned jump data elimination, a threshold value of the difference between adjacent sampling points is set. If the difference between the current sampling value and the previous value exceeds the threshold, it is determined to be jump interference, which is directly eliminated and interpolated.

[0010] The calculation formula of the improved minimum mean square error LMS adaptive filter in Step 2 above is: ; in, For the The weight coefficient of the filter at the iteration, dimensionless; is the step size factor, i.e., the learning rate, which controls the convergence speed and stability. Its value range is usually , is the maximum eigenvalue of the autocorrelation matrix of the input signal; Is the error signal, that is, the difference between the expected signal and the actual output, unit: consistent; The first The sampling value, that is, The grid voltage or current signal value at each sampling moment, unit: V or A; The above formula is used to update the weight of the adaptive filter and dynamically adjust the filter parameters to suppress noise and improve frequency measurement accuracy.

[0011] In the above Step 2, the filter weights are continuously adjusted to make the output signal approach the expected signal, minimize the error, and thus filter out the noise interference in the grid signal.

[0012] The frequency estimation formula of the short-time Fourier transform STFT in Step 3 above is: ; in, For the moment Estimated grid frequency, in Hz; is the number of sampling points in the sliding window, which determines the frequency resolution; For the The grid voltage or current signal amplitude of each sampling point, unit: V or A; is the fundamental frequency; is the sampling period, unit: second; The above formula is used to perform time-frequency analysis on the power grid signal, extract the instantaneous frequency value at the current moment, and improve the frequency measurement resolution.

[0013] In Step 3 above, the frequency estimation formula of the short-time Fourier transform (STFT) intercepts the signal fragments through the sliding window, performs Fourier transform on the signal in each window, and calculates the fundamental frequency in the spectrum. The corresponding amplitudes are finally weighted averaged to obtain the instantaneous frequency estimate.

[0014] The periodic calibration formula in Step 4 above is: ; in, is the reference frequency after calibration, unit: Hz); The original reference frequency of the synchronous clock output, unit: Hz; The frequency offset is in ppm, part per million, which is caused by temperature or aging.

[0015] The control equation for the compensation of the nonlinear dynamic model of the hydropower unit in Step 5 above is: ; in, is the control output of the speed regulator; 、 : Proportional, integral, and differential gain coefficients, used to adjust control response speed and stability, are dimensionless; is the frequency deviation, that is, the difference between the target frequency and the actual frequency, unit: Hz; It is the nonlinear compensation amount, calculated in real time according to the unit dynamic model, unit: %.

[0016] A system using the above-mentioned adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator includes an adaptive frequency measurement algorithm module, a clock calibration module, a nonlinear dynamic compensation module, and a sampling sensor group.

[0017] The adaptive algorithm and system for improving the frequency measurement accuracy and response speed of a speed regulator provided by the present invention have the following beneficial effects: 1. High-precision frequency measurement: The adaptive algorithm increases the frequency measurement resolution to 0.01Hz, significantly enhancing the ability to capture small frequency fluctuations in the power grid; 2. Fast response: Hardware acceleration reduces the frequency measurement response time to 80ms, meeting the grid's second-level frequency regulation requirements; 3. Long-term stability: GPS clock and calibration algorithm combined, the reference drift rate is reduced by 90%; 4. Strong adaptability: The nonlinear compensation strategy effectively responds to the dynamic characteristics of hydropower units, and the frequency regulation accuracy is improved by 13%. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 It is a flow chart of the adaptive frequency measurement algorithm of the present invention. DETAILED DESCRIPTION

[0019] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0020] An adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator includes the following steps: Step 1, signal acquisition and preprocessing; Step 2: Use an adaptive frequency measurement algorithm; utilize an improved minimum mean square error (LMS) adaptive filter to dynamically adjust filter parameters and suppress grid signal noise; Step 3: Combine short-time Fourier transform (STFT) and sliding window technology to extract frequency features in real time and improve frequency measurement resolution; Step 4: Use the GPS synchronized clock as the reference frequency source and embed a periodic calibration algorithm to periodically and automatically calibrate the clock reference value to eliminate long-term drift. Step 5: Build a nonlinear dynamic model of the hydropower unit and design an adaptive control law based on Lyapunov stability theory to compensate for the nonlinear relationship between the governor valve opening and output in real time, thereby improving frequency regulation accuracy. Step 6. Output instantaneous frequency.

[0021] The specific steps of signal acquisition and preprocessing in Step 1 above are: This step completes the acquisition and preliminary processing of power grid signals through high-precision sensors and preprocessing modules. The main process is: first acquire the signal, then preprocess the signal. The preprocessing process is divided into filtering, noise suppression, and data synchronization. The specific process is as follows: 1. Acquisition signal type and acquisition equipment: Acquisition signal: Grid three-phase voltage signal (unit: V); Output current signal of the hydro-generator set (unit: A); Governor valve opening feedback signal (unit: %).

[0022] Hardware configuration: Voltage transformer (PT): transformation ratio 100V / 5V, collecting three-phase voltage signals of the power grid; Current transformer (CT): transformation ratio 1000A / 5A, collecting generator current signal; 16-bit high-precision AD converter, to collect valve signals.

[0023] 2. Preprocessing steps: 2.1. Filtering: Perform filtering with a cutoff frequency of 500Hz to eliminate high-frequency noise interference.

[0024] 2.2. Transition data removal: Set the difference threshold between adjacent sampling points (such as ±10% of the range). If the difference between the current sampling value and the previous value exceeds the threshold, it is determined to be jump interference, which is directly eliminated and interpolated (such as linear interpolation).

[0025] 3. Noise suppression: Set the digital filter threshold (such as ±0.5V), and set the signal segment below the threshold to zero; 4. Data synchronization: Synchronize the AD sampling clock to the GPS reference clock to ensure sampling interval accuracy.

[0026] The calculation formula of the improved minimum mean square error LMS adaptive filter in Step 2 above is: ; in, For the The weight coefficient of the filter at the iteration, dimensionless; is the step size factor, i.e., the learning rate, which controls the convergence speed and stability. Its value range is usually , is the maximum eigenvalue of the autocorrelation matrix of the input signal; Is the error signal, that is, the difference between the expected signal and the actual output, unit: consistent; The first The sampling value, that is, The grid voltage or current signal value at each sampling moment, unit: V or A; The above formula is used to update the weight of the adaptive filter and dynamically adjust the filter parameters to suppress noise and improve frequency measurement accuracy.

[0027] In the above Step 2, the filter weights are continuously adjusted to make the output signal approach the expected signal, minimize the error, and thus filter out the noise interference in the grid signal.

[0028] The frequency estimation formula of the short-time Fourier transform STFT in Step 3 above is: ; in, For the moment Estimated grid frequency, in Hz; is the number of sampling points in the sliding window, which determines the frequency resolution (e.g. =1024); For the The grid voltage or current signal amplitude of each sampling point, unit: V or A; is the fundamental frequency (nominal frequency of the grid, such as 50Hz or 60Hz); is the sampling period, unit: second, is the sampling frequency, such as 1000Hz; The above formula is used to perform time-frequency analysis on the power grid signal, extract the instantaneous frequency value at the current moment, and improve the frequency measurement resolution.

[0029] In Step 3 above, the frequency estimation formula of the short-time Fourier transform (STFT) intercepts the signal fragments through the sliding window, performs Fourier transform on the signal in each window, and calculates the fundamental frequency in the spectrum. The corresponding amplitudes are finally weighted averaged to obtain the instantaneous frequency estimate.

[0030] The periodic calibration formula in Step 4 above is: ; in, is the reference frequency after calibration, unit: Hz); The original reference frequency of the synchronous clock output, unit: Hz; The frequency offset is in ppm, part per million, which is the deviation caused by temperature or aging (for example, ±0.1ppm).

[0031] The control equation for the compensation of the nonlinear dynamic model of the hydropower unit in Step 5 above is: ; in, is the control output of the speed regulator (such as valve opening instruction, unit: %); 、 : Proportional, integral, and differential gain coefficients, used to adjust control response speed and stability, are dimensionless; is the frequency deviation, that is, the difference between the target frequency and the actual frequency, unit: Hz; It is the nonlinear compensation amount, calculated in real time according to the unit dynamic model, unit: %.

[0032] A system using the above-mentioned adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator includes an adaptive frequency measurement algorithm module, a clock calibration module, a nonlinear dynamic compensation module, and a sampling sensor group.

[0033] Example: 1. System hardware configuration and initialization 1. FPGA hardware acceleration platform construction Hardware selection: A domestically produced independently controllable chip is used, equipped with 8 parallel DSP modules, each module supports 18×25-bit multipliers, and the total number of logic units is 326,080.

[0034] Architecture design: Adaptive filtering module: occupies 15% of FPGA resources, adopts pipeline structure, and completes weight update in a single cycle ( ).

[0035] STFT calculation unit: Configured with 4 parallel FFT cores and window length N=1024.

[0036] Nonlinear compensation module: Real-time calculation of Lyapunov compensation based on lookup table (LUT), with a storage depth of 1024×32 bits.

[0037] Clock configuration: The FPGA main frequency is 200MHz, the AD sampling clock is synchronized to the GPS reference clock, and the jitter is ≤1ps.

[0038] 2. High-precision clock source integration Installation and calibration: Deploy the GPS module to output 1PPS (pulse per second) signal, which is synchronized to the FPGA clock management unit.

[0039] Initialize the reference frequency =50.000000Hz, calibration period is 24 hours, and GPS frequency deviation is read after the calibration procedure is triggered (accuracy ±0.1ppm).

[0040] Hardware interface: The GPS module communicates with the FPGA through the RS-422 interface and performs even parity check.

[0041] 2. Implementation of Adaptive Frequency Measurement Algorithm 1. Signal acquisition and preprocessing 1. Signal Acquisition and Preprocessing ①Sensor configuration and signal type: Acquisition signal: The three-phase voltage signal of the power grid (unit: V) is collected through a voltage transformer (PT) with a transformation ratio of 100V / 5V and an accuracy of 0.2. The output current signal of the hydro-generator set (unit: A) is collected through a current transformer (CT) with a transformation ratio of 1000A / 5A and a bandwidth of 0~1kHz; The speed regulator valve opening feedback signal (unit: %) is collected by a 16-bit high-precision AD converter with a range of ±10V and a resolution of 16 bits.

[0042] Sampling parameters: The sampling frequency is 1000Hz, which is an integer multiple of the grid base frequency (50Hz); Anti-aliasing filter with a cutoff frequency of 500Hz.

[0043] ② Preprocessing steps: Transition data removal: Set the difference threshold of adjacent sampling points to ±10% of the range (for example, if the voltage signal range is ±10V, the threshold is ±1V); If the difference between the current sampling value and the previous value exceeds the threshold, it is determined to be a jump interference, which is directly eliminated and the data is supplemented by linear interpolation.

[0044] Noise suppression: Pre-analog filter (cut-off frequency 60Hz) eliminates high-frequency noise; The digital filter threshold is set to ±0.5V, and the signal segment below the threshold is reset to zero; A sliding average filter (window length 5 points) was used to further smooth the signal.

[0045] Data synchronization: The AD sampling clock is synchronized to the GPS reference clock (1PPS signal) to ensure sampling interval accuracy (jitter ≤ 1ps); The clock synchronization check between GPS module and FPGA is realized through RS-422 interface.

[0046] ②Noise suppression: The pre-analog filter (cut-off frequency 60Hz) eliminates high-frequency noise, and the digital filter threshold is set to ±0.5V. Signal segments below the threshold are directly set to zero.

[0047] 2. Improved LMS adaptive filtering Parameter settings: Filter order L=64, Step factor (learning rate) = 0.02, initial weight .

[0048] Expected signal For an ideal 50Hz sine wave, the error signal = ,in

[0049] Performance Verification: When the input SNR is 30dB, the output SNR is increased to 60dB (the noise power is reduced to 1% of the original value).

[0050] 3. Short-time Fourier transform (STFT) frequency estimation Window design: Hanning Window, window length N = 1024, overlap rate 50% (sliding step size 512 points).

[0051] Frequency resolution , increased to 0.01Hz through interpolation algorithm.

[0052] Real-time calculation: Output instantaneous frequency every 10ms , calculation formula: ; fundamental frequency =50Hz, sampling period

[0053] The measured data are shown in the following table:

[0054] 3. High-precision clock calibration 1. Reference frequency correction process Calibration trigger: Automatically start the calibration process at 00:00 every day to read the frequency deviation output by the GPS module .

[0055] Calculation formula: ; Example: If ,but:

[0056] Write to FPGA: Update the internal reference frequency register of FPGA through SPI interface, with writing accuracy of 0.001Hz.

[0057] 2. Long-term stability test The comparative experiments are shown in the following table:

[0058] 4. Nonlinear Dynamic Compensation Control 1. Construction of hydropower unit model Measured data collection: Guide vane opening-output curve: dead zone ±2%, saturation zone > 90% opening when output linearly decreases by 30%.

[0059] Dynamic response delay: The delay from valve action to output response is 80ms (measured average).

[0060] Mathematical model: ; in =80ms, is the nonlinear compensation term.

[0061] 2. Adaptive compensation strategy Control law design: Traditional PID parameters: .

[0062] Compensation Based on real-time calculation of Lyapunov function,

[0063] Note: The Lyapunov stability condition formula is: ; Formula meaning: Through Lyapunov function The derivative of To judge the stability of the system, if , the system is stable.

[0064] Parameter Description: : The rate of change of frequency deviation (unit: Hz / s).

[0065] Physical meaning: frequency deviation required Its rate of change The signs are opposite to ensure the error convergence.

[0066] Example: like and =−0.05Hz / s, then: x(-0.05)

[0067] Real-time adjustments: The compensation value is updated every 20ms, and the FPGA lookup table method achieves microsecond-level response.

[0068] The performance comparison is shown in the following table:

[0069] 5. System integration and full-condition testing 1. Full working condition verification Working condition 1 (steady-state frequency modulation): Simulation: When the frequency fluctuation is ±0.01Hz, the speed regulator output error is ≤0.005Hz; Actual measurement: Under the same conditions, the error is ≤0.007Hz.

[0070] Working condition 2 (noise environment): Simulation: When SNR=25dB, the frequency fluctuation is ≤0.012Hz; Actual measurement: Fluctuation under the same conditions ≤ 0.015Hz.

[0071] Working condition 3 (long-term operation): Simulation: 72-hour reference drift ≤ 0.08ppm; Measured: drift ≤ 0.12ppm.

[0072] 2. Comparative Experiment Summary Table

[0073] Through the collaborative design of FPGA hardware acceleration, GPS clock calibration, adaptive filtering and nonlinear compensation, this solution has demonstrated significant advantages in both simulation and actual machine testing.

Claims

1. An adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator, characterized in that: The following steps are involved: Step 1, signal acquisition and preprocessing; Step 2: Use an adaptive frequency measurement algorithm; utilize an improved minimum mean square error (LMS) adaptive filter to dynamically adjust filter parameters and suppress grid signal noise; Step 3: Combine short-time Fourier transform (STFT) and sliding window technology to extract frequency features in real time and improve frequency measurement resolution; Step 4: Use a synchronous clock as the reference frequency source and embed a periodic calibration algorithm to periodically and automatically calibrate the clock reference value to eliminate long-term drift. Step 5: Build a nonlinear dynamic model of the hydropower unit and design an adaptive control law based on Lyapunov stability theory to compensate for the nonlinear relationship between the governor valve opening and output in real time, thereby improving frequency regulation accuracy. Step 6. Output instantaneous frequency.

2. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 1 is characterized in that: The specific steps of signal acquisition and preprocessing in Step 1 are as follows: Collect the three-phase voltage signal of the power grid, the output current signal of the hydro-generator set and the feedback signal of the governor valve opening; Signal preprocessing: First, filter, then remove the jump data, then suppress the noise, and finally synchronize the data.

3. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 2, characterized in that: In the jump data removal, a threshold value of the difference between adjacent sampling points is set. If the difference between the current sampling value and the previous value exceeds the threshold, it is determined to be jump interference, which is directly removed and interpolated.

4. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 3 is characterized in that: The calculation formula of the improved minimum mean square error LMS adaptive filter in Step 2 is: ； in, For the The weight coefficient of the filter at the iteration, dimensionless; is the step size factor, i.e., the learning rate, which controls the convergence speed and stability. Its value range is usually , is the maximum eigenvalue of the autocorrelation matrix of the input signal; Is the error signal, that is, the difference between the expected signal and the actual output, unit: consistent; The first The sampling value, that is, The grid voltage or current signal value at each sampling moment, unit: V or A; The above formula is used to update the weight of the adaptive filter and dynamically adjust the filter parameters to suppress noise and improve frequency measurement accuracy.

5. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 4 is characterized in that: In the aforementioned Step 2, the filter weights are continuously adjusted so that the output signal approaches the desired signal, the error is minimized, and thus the noise interference in the grid signal is filtered out.

6. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 5, characterized in that: The frequency estimation formula of the short-time Fourier transform STFT in Step 3 is: ； in, For the moment Estimated grid frequency, in Hz; is the number of sampling points in the sliding window, which determines the frequency resolution; For the The grid voltage or current signal amplitude of each sampling point, unit: V or A; is the fundamental frequency; is the sampling period, unit: second; The above formula is used to perform time-frequency analysis on the power grid signal, extract the instantaneous frequency value at the current moment, and improve the frequency measurement resolution.

7. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 6, characterized in that: In the above Step 3, the frequency estimation formula of the short-time Fourier transform (STFT) is to intercept the signal fragments through the sliding window, perform Fourier transform on the signal in each window, and calculate the fundamental frequency in the spectrum. The corresponding amplitudes are finally weighted averaged to obtain the instantaneous frequency estimate.

8. The adaptive algorithm for improving the frequency measurement accuracy and response speed of a speed regulator according to claim 7, characterized in that: The periodic calibration formula in Step 4 is: ； in, is the reference frequency after calibration, unit: Hz); The original reference frequency of the synchronous clock output, unit: Hz; The frequency offset is in ppm, part per million, which is caused by temperature or aging.

9. The adaptive algorithm for improving frequency measurement accuracy and response speed of a speed regulator according to claim 8, characterized in that: The control equation for the compensation of the nonlinear dynamic model of the hydropower unit in Step 5 is: ； in, is the control output of the speed regulator; 、 : Proportional, integral, and differential gain coefficients, used to adjust control response speed and stability, are dimensionless; is the frequency deviation, that is, the difference between the target frequency and the actual frequency, unit: Hz; It is the nonlinear compensation amount, calculated in real time according to the unit dynamic model, unit: %.

10. A system using the adaptive algorithm for improving frequency measurement accuracy and response speed of a speed regulator according to claim 9, characterized in that: The system includes an adaptive frequency measurement algorithm module, a clock calibration module, a nonlinear dynamic compensation module and a sampling sensor group.

Images