Wide load intelligent coordination optimization system for thermal power unit based on decoupling control algorithm

Abstract

The application discloses a wide-load intelligent coordination optimization system of a thermal power unit based on a decoupling control algorithm and belongs to the technical field of automatic control of thermal power generation. The system comprises a multi-source data acquisition module, which acquires in real time boiler-side furnace temperature field distribution data, steam turbine-side main steam pressure fluctuation sequences and power grid-side load instructions and change rate limit values; a time-frequency domain composite decoupling module, which performs unified time-frequency domain transformation processing on the temperature field data and the pressure fluctuation sequences, generates a decoupling characteristic matrix containing low-frequency common-mode components and high-frequency differential-mode components; a disturbance inverse compensation calculation module, which calculates a phase compensation coefficient according to the high-frequency differential-mode components and generates a feedforward compensation signal, which is superimposed on the conventional PID regulation loop on the boiler side and the steam turbine side; and an adaptive load response execution module, which performs rate consistency constraint on the regulation instruction based on the load change rate limit value and outputs a fuel actuating mechanism driving signal and a regulating valve actuating mechanism driving signal.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for thermal power generation, specifically to a wide-load intelligent coordination and optimization system for thermal power units based on a decoupled control algorithm. Background Technology

[0002] When thermal power units participate in deep peak shaving, they need to quickly track grid dispatch commands over a wide load range. Strong coupling and large inertia exist between furnace combustion dynamics and main steam pressure. Existing wide-load coordinated control technologies generally employ PID-based multivariate decoupling schemes, using preset decoupling matrices or model predictive control to coordinate the adjustment of boiler fuel quantity and turbine valve opening. In furnace temperature field measurements, signal transmission delays are inconsistent across numerous temperature measurement points, and main steam pressure fluctuations include both low-frequency trend components caused by load baseline changes and high-frequency dynamic differences caused by fuel disturbances and valve actions. Conventional decoupling techniques typically decouple variables by amplitude in the time domain, lacking methods for joint time-frequency domain characteristic analysis of temperature field and pressure fluctuations, making it difficult to effectively separate low-frequency common-mode trends from high-frequency differential-mode disturbances. Feedforward compensation often uses fixed-parameter models, with compensation signals acting on both trends and disturbances simultaneously, leading to over-adjustment between boiler-side fuel commands and turbine-side valve commands, resulting in significant fluctuations in main steam pressure. Furthermore, existing compensation strategies do not dynamically adjust the compensation amount based on the instantaneous polarity of the disturbance signal and the load change rate limit specified by the power grid. When dispatch instructions impose different limits on the load change rate, fixed feedforward compensation can easily exceed the rate tolerance of the actuator, causing load tracking deviation or triggering safety protection. How to extract time-frequency domain common-mode and differential-mode features from the temperature field and main steam pressure fluctuations at multiple measurement points and construct a decoupled description for disturbance compensation, and how to generate accurate phase compensation signals based on the instantaneous polarity of the differential-mode disturbance and the load change rate constraint, have become problems that need to be solved to improve the quality of wide-load coordinated control. Summary of the Invention

[0003] This invention provides a wide-load intelligent coordination optimization system for thermal power units based on a decoupled control algorithm, in order to solve the problems of incomplete decoupling between furnace combustion dynamics and main steam pressure fluctuations, and the disconnect between compensation strategies and load change rate constraints in existing wide-load coordinated control.

[0004] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a wide-load intelligent coordination and optimization system for thermal power units based on a decoupling control algorithm. The system includes a multi-source data acquisition module, a time-frequency domain composite decoupling module, a disturbance inverse compensation calculation module, and an adaptive load response execution module.

[0005] The multi-source data acquisition module is used to acquire in real time the boiler-side operating parameter set, turbine-side operating parameter set, and grid-side dispatch command signal set of the thermal power unit during wide load operation. The boiler-side operating parameter set includes furnace temperature field distribution data; the turbine-side operating parameter set includes main steam pressure fluctuation sequence; and the grid-side dispatch command signal set includes load target value and load change rate limit value.

[0006] The time-frequency domain composite decoupling module is used to perform unified time-frequency domain transformation processing on the furnace temperature field distribution data and the main steam pressure fluctuation sequence, generating a decoupling feature matrix containing low-frequency common-mode components and high-frequency differential-mode components. By separating the frequency components of the boiler heat storage trend and the turbine power response, the low-frequency common-mode components accurately reflect the macroscopic energy balance trend of the boiler-turbine coupling, while the high-frequency differential-mode components focus on the rapid fluctuations induced by combustion disturbances and differences in valve operation, providing a precise frequency partitioning basis for decoupling compensation.

[0007] As a preferred embodiment of the present invention, the time-frequency domain composite decoupling module includes a time delay alignment unit, a multi-scale frequency decomposition unit, and a feature fusion mapping unit. The time delay alignment unit receives the furnace temperature field distribution data and the main steam pressure fluctuation sequence, performs cross-correlation peak extraction processing on the time series of each temperature measurement point in the furnace temperature field distribution data and the main steam pressure fluctuation sequence to determine the transmission delay time of each temperature measurement point signal relative to the main steam pressure signal, and performs shift resampling processing on the furnace temperature field distribution data based on the transmission delay time to generate time-aligned temperature field data; thereby eliminating the phase shift caused by spatial differences in the temperature signals of multiple measurement points and ensuring the synchronization of subsequent frequency decomposition. Preferably, the time delay alignment unit uses a cross-correlation function peak detection method to determine the transmission delay time and uses cubic spline interpolation to perform shift resampling processing on the furnace temperature field distribution data to achieve delay compensation for non-integer sampling points while preserving signal details.

[0008] The input of the multi-scale frequency decomposition unit is connected to the output of the time delay alignment unit. It receives the time-aligned temperature field data and the main steam pressure fluctuation sequence, and performs bandpass pre-filtering and low-frequency trend term separation processing on the time-aligned temperature field data and the main steam pressure fluctuation sequence, respectively, to generate the low-frequency common-mode component and high-frequency differential-mode component of the time-aligned temperature field data and the main steam pressure fluctuation sequence. Further, the multi-scale frequency decomposition unit includes a Chebyshev bandpass filter bank and a low-frequency trend reconstruction subunit. The Chebyshev bandpass filter bank consists of a set of second-order Chebyshev bandpass filters with center frequencies distributed at logarithmic intervals. The passband bandwidth of each filter is set as a preset percentage value of the distance between adjacent center frequencies; specifically, this preset percentage value is 20%. This filter bank is used to perform multi-band bandpass filtering on the time-aligned temperature field data and the main steam pressure fluctuation sequence, generating a set of sub-band signals in multiple frequency bands. The logarithmic interval frequency band division method matches the multi-timescale characteristics of the thermal power unit's thermal process. The 20% passband overlap can avoid mode aliasing while retaining the dynamic information of the transition frequency band. The low-frequency trend reconstruction subunit is used to receive the set of sub-band signals, reconstruct and sum the sub-band signals above the preset cutoff frequency to generate the high-frequency differential mode component, and reconstruct and average the sub-band signals below the preset cutoff frequency to generate the low-frequency common mode component, thereby cleanly separating the slow trend caused by boiler heat storage from the rapid changes in pressure fluctuations.

[0009] The input of the feature fusion mapping unit is connected to the output of the multi-scale frequency decomposition unit. It is used to matrix-arrange the low-frequency common-mode component and the high-frequency differential-mode component according to their respective timestamps to generate the decoupled feature matrix, so that subsequent compensation calculations can directly index the corresponding components in a unified time-frequency feature space.

[0010] The disturbance inverse compensation calculation module is used to calculate the phase compensation coefficient between the boiler fuel quantity adjustment command and the turbine valve opening adjustment command based on the high-frequency differential mode component in the decoupling feature matrix, and to generate a feedforward compensation signal based on the phase compensation coefficient, which is then superimposed on the conventional PID control loops on the boiler side and the turbine side. By extracting the instantaneous polarity change of the high-frequency differential mode component, the dynamic mismatch direction between the boiler combustion response and the turbine valve action is identified in real time. The phase compensation coefficient calculated accordingly can effectively compensate for the phase lag caused by the delay in pulverized coal conveying and the steam volume inertia, thereby suppressing the mutual disturbance between the boiler and turbine during wide load regulation.

[0011] As a preferred embodiment of the present invention, the disturbance inverse compensation calculation module includes a feedforward gain calculation unit and a compensation signal generation unit. The feedforward gain calculation unit receives the high-frequency differential mode component in the decoupling characteristic matrix and the load change rate limit value, performs instantaneous extreme value capture processing on the high-frequency differential mode component, determines the positive and negative polarity pulse sequences of the high-frequency differential mode component, and calculates the phase compensation coefficient based on the ratio of the duration of the positive and negative polarity pulse sequences to the load change rate limit value. The compensation signal generation unit receives the phase compensation coefficient, performs phase compensation calculation on the initial fuel quantity adjustment command and initial valve opening adjustment command output by the conventional PID control loop, and generates the feedforward compensation signal; the feedforward compensation signal includes a fuel quantity advance compensation component and a valve lag compensation component.

[0012] More preferably, the feedforward gain calculation unit includes a zero-crossing detection subunit, a pulse generation subunit, and a compensation coefficient mapping subunit. The zero-crossing detection subunit is used to perform zero-crossing point identification processing on the high-frequency differential mode component, recording the sign change direction of the high-frequency differential mode component at each zero-crossing point, and generating a sign change sequence. Specifically, the zero-crossing detection subunit employs a hysteresis comparator, and the hysteresis width of the hysteresis comparator is set to 5% of the root mean square amplitude of the high-frequency differential mode component to suppress false zero-crossings caused by high-frequency noise and ensure the stability and reliability of the pulse sequence. The pulse generation subunit is used to receive the sign change sequence, perform peak-hold processing on the absolute value of the amplitude of the high-frequency differential mode component between two adjacent zero-crossing points, and generate the positive and negative polarity pulse sequences, wherein positive polarity pulses correspond to high-frequency differential mode component intervals with positive amplitudes, and negative polarity pulses correspond to high-frequency differential mode component intervals with negative amplitudes. The compensation coefficient mapping subunit is used to receive the positive and negative polarity pulse sequence and the load change rate limit value, perform timing processing on the duration of each polarity pulse, calculate the reciprocal of the ratio of the duration of each polarity pulse to the load change rate limit value, and use the reciprocal as the phase compensation coefficient; this mapping relationship enables the feedforward compensation gain to be automatically enhanced when the load change rate required by the power grid is relatively strict, thereby accelerating the dynamic balance process of the boiler and turbine.

[0013] The adaptive load response execution module is used to perform rate consistency constraint output processing on the boiler fuel quantity adjustment command and the turbine valve opening adjustment command according to the load change rate limit value, generating the final fuel actuator drive signal and valve actuator drive signal. By applying rate constraints that match the dispatch command, it ensures that the action rate of the boiler fuel quantity and the turbine valve opening is strictly synchronized and limited to the load change rate allowed by the power grid, avoiding large fluctuations in turbine inlet pressure or over-adjustment of load due to inconsistent actuator actions.

[0014] Preferably, the adaptive load response execution module includes a rate constraint limiting unit and an instruction conversion unit. The rate constraint limiting unit receives the boiler fuel quantity regulation instruction and turbine valve opening regulation instruction obtained by superimposing the feedforward compensation signal output by the disturbance inverse compensation calculation module onto the output of a conventional PID control loop, along with the load change rate limit value. Based on the load change rate limit value, it calculates the maximum change rate of boiler fuel quantity and the maximum change rate of turbine valve opening. It then performs first-order lag filtering and limiting processing on the corrected boiler fuel quantity regulation instruction and the corrected turbine valve opening regulation instruction, respectively, to generate a rate-constrained fuel quantity regulation sequence and a valve opening regulation sequence. Specifically, the rate constraint limiting unit employs a variable time constant first-order hysteresis filter. The time constant of the first-order hysteresis filter is adaptively adjusted according to the load change rate limit value, and the time constant is inversely proportional to the load change rate limit value. When the power grid requires rapid load change, the reduced filter time constant ensures the rapid transmission of the command signal, while when the load change rate is slow, the increased time constant can fully smooth the command and prevent sudden changes in fuel quantity and valve opening from causing system oscillations.

[0015] The instruction conversion unit is used to receive the fuel quantity adjustment sequence after rate constraint, and convert the fuel quantity adjustment sequence after rate constraint into the driving signal of the fuel actuator according to the preset linear mapping relationship table between pulverized coal flow and fuel actuator opening, so as to provide the actuators such as coal feeder and coal mill with directly executable driving quantity; the turbine valve opening adjustment sequence directly drives the valve actuator in the form of opening command.

[0016] The system described in this invention decouples the furnace temperature field and main steam pressure in the time and frequency domains, extracts the high-frequency differential mode component reflecting the dynamic differences between the turbine and boiler, and generates an adaptive phase compensation feedforward signal accordingly. Combined with the rate consistency constraint linked with the grid dispatch command, it achieves precise coordination between the turbine and boiler of the thermal power unit within a wide load range, effectively reduces the amplitude of main steam pressure fluctuations, improves the speed of load response and steady-state accuracy, and enhances the adaptability of the unit to grid dispatch.

[0017] The technical effects and advantages provided by the present invention in the above technical solution are as follows: The time-frequency domain composite decoupling module extracts the cross-correlation peak values ​​and performs cubic spline interpolation resampling on the signals from each measuring point in the furnace temperature field distribution data and the main steam pressure fluctuation sequence. This eliminates the transmission delay differences between multiple measuring points in the temperature field, obtaining time-aligned temperature field data. Based on this, a Chebyshev bandpass filter bank with center frequencies distributed at logarithmic intervals and a passband bandwidth equal to a preset percentage of the distance between adjacent center frequencies is used to perform multi-band bandpass filtering and low-frequency trend separation on the time-aligned temperature field data and the main steam pressure fluctuation sequence. Sub-band signals below the preset cutoff frequency are reconstructed into low-frequency common-mode components, while sub-band signals above the preset cutoff frequency are reconstructed into high-frequency differential-mode components. These are then arranged by timestamp to generate a decoupling feature matrix. This process clearly separates the slow changes reflecting the overall trend of boiler heat storage and load from the transient changes reflecting local combustion disturbances or rapid valve actions. This allows subsequent compensation stages to intervene only in the high-frequency differential-mode components, avoiding contamination of compensation by trend quantities. This suppresses over-adjustment between fuel quantity and valve opening, reduces the dynamic deviation of the main steam pressure, and improves parameter stability.

[0018] The disturbance inverse compensation calculation module obtains the high-frequency differential mode component from the decoupled feature matrix, identifies its zero-crossing point using zero-crossing detection with hysteresis characteristics, and records the direction of sign change. Peak value maintenance is performed on the absolute value of the amplitude between adjacent zero-crossing points to generate a positive and negative polarity pulse sequence. The compensation coefficient mapping stage times the duration of each polarity pulse, taking the reciprocal of the ratio of the duration to the load change rate limit as the phase compensation coefficient. The compensation signal generation unit then calculates a feedforward compensation signal containing fuel quantity advance compensation and valve lag compensation components, which is superimposed on the output of the conventional PID control loop. The compensation coefficient dynamically changes with the pulse duration and the load change rate limit. When the disturbance rapidly reverses polarity within a short time or the grid's allowable load change rate is low, the compensation amount automatically decreases to prevent command overshoot; when the disturbance is persistent and the allowable rate is high, the compensation amount increases accordingly to accelerate disturbance suppression. Fuel quantity advance compensation enables the boiler to respond to pressure change trends in advance, while valve lag compensation releases or recovers turbine energy in a timely manner, effectively compensating for the phase difference between the large delay on the boiler side and the rapid response on the turbine side, and reducing the imbalance between the boiler and turbine.

[0019] The adaptive load response execution module calculates the maximum rate of change of boiler fuel quantity and turbine valve opening based on the same load change rate limit. A variable time constant first-order lag filter is used to rate-limit the superimposed feedforward boiler fuel quantity regulation command and turbine valve opening regulation command. The filter's time constant is inversely proportional to the load change rate limit. The rate-constrained fuel quantity regulation sequence is then converted into an actuator drive signal through a linear mapping between pulverized coal flow and fuel actuator opening. This process ensures that the output command under any operating condition strictly meets the dispatch rate requirements, resulting in a smooth and shock-free adjustment process. Simultaneously, the load change rate is fully released during wide load fluctuations, balancing rapid response and equipment safety. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0021] Figure 1 This is a schematic diagram of the structure of a wide-load intelligent coordination and optimization system for thermal power units based on a decoupling control algorithm; Figure 2 This is a schematic diagram of the structure of a multi-scale frequency decomposition unit; Figure 3 This is a schematic diagram of the disturbance inverse compensation calculation module; Figure 4 This is a schematic diagram of the cross-correlation function between furnace temperature field distribution data and main steam pressure fluctuation sequence, as well as the peak propagation delay time. Figure 5 It is the rate constraint response curve of the boiler fuel quantity adjustment command under different load change rate limits. Detailed Implementation

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

[0023] See Figure 1This invention provides a wide-load intelligent coordination and optimization system for thermal power units based on a decoupling control algorithm, comprising a multi-source data acquisition module, a time-frequency domain composite decoupling module, a disturbance inverse compensation calculation module, and an adaptive load response execution module. The multi-source data acquisition module acquires in real time the boiler-side operating parameter set, the turbine-side operating parameter set, and the grid-side dispatch command signal set of the thermal power unit during wide-load operation. The boiler-side operating parameter set includes furnace temperature field distribution data, the turbine-side operating parameter set includes the main steam pressure fluctuation sequence, and the grid-side dispatch command signal set includes the load target value and the load change rate limit value. The time-frequency domain composite decoupling module receives the furnace temperature field distribution data and the main steam pressure fluctuation sequence, performs unified time-frequency domain transformation on both, and generates a decoupling feature matrix containing low-frequency common-mode components and high-frequency differential-mode components. The disturbance inverse compensation calculation module calculates the phase compensation coefficient between the boiler fuel quantity adjustment command and the turbine valve opening adjustment command based on the high-frequency differential-mode component in the decoupling feature matrix, and generates a feedforward compensation signal based on the phase compensation coefficient, which is then superimposed onto the conventional PID control loops on the boiler side and the turbine side. The adaptive load response execution module performs rate consistency constraint output processing on the boiler fuel quantity adjustment command and the turbine valve opening adjustment command based on the load change rate limit value, and generates the final fuel actuator drive signal and valve actuator drive signal.

[0024] Example 1:

[0025] In practical implementation, the time-frequency domain composite decoupling module includes a time delay alignment unit, a multi-scale frequency decomposition unit, and a feature fusion mapping unit. The time delay alignment unit receives furnace temperature field distribution data and main steam pressure fluctuation sequences from the multi-source data acquisition module. The furnace temperature field distribution data consists of time series collected from multiple temperature measuring points located at different positions inside the furnace, while the main steam pressure fluctuation sequence is a continuous time series signal output from the pressure transmitter on the main steam pipeline. The time delay alignment unit performs cross-correlation peak extraction processing on the time series of each temperature measuring point in the furnace temperature field distribution data and the main steam pressure fluctuation sequence to determine the transmission delay time of each temperature measuring point signal relative to the main steam pressure signal. The cross-correlation peak extraction processing is implemented as follows: for the time series of a single temperature measuring point, denoted as... And the main steam pressure fluctuation sequence, denoted as ,calculate and cross-correlation function The calculation formula is:

[0026] in, Indicates the time of a single temperature measurement point The time series of collected temperature values Indicates time The time series of collected main steam pressure values. Represents the delay time variable. Indicates the delay time The cross-correlation function values ​​of the temperature measurement point time series and the main steam pressure fluctuation series are calculated. The integration interval covers the overlapping time range of the temperature measurement point time series and the main steam pressure fluctuation series. Time delay alignment unit search is performed. When the maximum value is obtained, the corresponding The value, will The value is determined as the transmission delay time of the temperature measurement point signal relative to the main steam pressure signal. The above cross-correlation peak extraction process is performed on all temperature measurement points in the furnace temperature field distribution data to obtain the transmission delay time corresponding to each temperature measurement point.

[0027] In practical implementation, the time delay alignment unit performs shift resampling on the furnace temperature field distribution data based on the transmission delay time corresponding to each temperature measuring point, generating time-aligned temperature field data. The shift resampling process is as follows: for each temperature measuring point's time series, the transmission delay time corresponding to that temperature measuring point is subtracted from the time coordinate of the time series to obtain the time-shifted time series. The sampling times of the time-shifted time series become non-uniformly distributed. The time-shifted time series is then resampled using cubic spline interpolation, with the resampling times consistent with the sampling times of the main steam pressure fluctuation sequence. The cubic spline interpolation method uses a piecewise cubic polynomial to reconstruct the time series after shifting. The function value, first derivative, and second derivative of the piecewise cubic polynomial remain continuous at the resampling time. The interpolation nodes are the four original sampling points in the time-shifted time series closest to the resampling time. After shift resampling, the time series of all temperature measuring points are time-aligned with the main steam pressure fluctuation sequence, forming time-aligned temperature field data.

[0028] In practical implementation, the input of the multi-scale frequency decomposition unit is connected to the output of the time delay alignment unit. The multi-scale frequency decomposition unit receives the time-aligned temperature field data and the main steam pressure fluctuation sequence. The multi-scale frequency decomposition unit performs bandpass pre-filtering and low-frequency trend term separation processing on the time-aligned temperature field data and the main steam pressure fluctuation sequence, respectively. The bandpass pre-filtering is implemented using a set of bandpass filters with different passband center frequencies. The bandpass filter bank filters the time series of each temperature measurement point and the main steam pressure fluctuation sequence in the time-aligned temperature field data, decomposing each input time series into multiple sub-band signals of different frequency bands. The passband center frequencies of the bandpass filter bank are distributed at logarithmic intervals, and the passband bandwidth is set according to a preset percentage of the distance between adjacent center frequencies to cover the main dynamic frequency range during the wide-load operation of the thermal power unit. The low-frequency trend term separation processing divides all sub-band signals obtained after the bandpass pre-filtering into low-frequency and high-frequency parts according to a preset cutoff frequency. All sub-band signals with center frequencies below the preset cutoff frequency are reconstructed and averaged to generate low-frequency common-mode components. All sub-band signals with center frequencies above the preset cutoff frequency are reconstructed and summed to generate high-frequency differential-mode components. The preset cutoff frequency is determined based on the boiler-side thermal inertia time constant and the turbine-side response time constant. Its value ensures that the low-frequency common-mode components primarily reflect the overall trend of the load target value, while the high-frequency differential-mode components primarily reflect the dynamic energy exchange disturbances between the boiler and turbine sides. The above low-frequency trend term separation process is performed on the time series of each temperature measurement point and the main steam pressure fluctuation series in the time-aligned temperature field data to obtain the low-frequency common-mode and high-frequency differential-mode components corresponding to the time-aligned temperature field data, as well as the low-frequency common-mode and high-frequency differential-mode components corresponding to the main steam pressure fluctuation series.

[0029] In practical implementation, the input of the feature fusion mapping unit is connected to the output of the multi-scale frequency decomposition unit. The feature fusion mapping unit receives the aforementioned multiple low-frequency common-mode components and high-frequency differential-mode components. The feature fusion mapping unit performs matrix-based arrangement of the low-frequency common-mode components and high-frequency differential-mode components according to their respective timestamps to generate a decoupled feature matrix. The specific process of matrix arrangement is as follows: using the timestamp of the main steam pressure fluctuation sequence as the reference time axis, the low-frequency common-mode components corresponding to all temperature measurement points are arranged according to the measurement point number as rows of the low-frequency common-mode component matrix. The low-frequency common-mode components corresponding to the main steam pressure fluctuation sequence are added as additional rows to the low-frequency common-mode component matrix to form the low-frequency common-mode component matrix. Arrange the high-frequency differential mode components corresponding to all temperature measurement points into rows of the high-frequency differential mode component matrix according to the measurement point number. Add the high-frequency differential mode components corresponding to the main steam pressure fluctuation sequence as additional rows to the high-frequency differential mode component matrix to form the high-frequency differential mode component matrix. Decoupling feature matrix From the low-frequency common-mode component matrix With high-frequency differential mode component matrix It is constructed by piecing together elements along the direction of travel. Each row corresponds to a temperature measurement point or a main steam pressure measurement point, and each column corresponds to a sampling time. The first part of the columns stores the low-frequency common-mode component values, and the second part stores the high-frequency differential-mode component values. The generated decoupling feature matrix is ​​transmitted to the disturbance inverse compensation calculation module.

[0030] See Figure 4 The horizontal axis in the graph represents the delay time. The unit is seconds, and the vertical axis represents the cross-correlation function value. The value ranges from -1 to 1. The curve is the cross-correlation function between the furnace temperature measurement point time series and the main steam pressure fluctuation series. With delay time The curve shows a periodic fluctuation, exhibiting a clear peak-valley alternation characteristic. The maximum peak value is approximately 1, and the minimum valley value is close to -1, reflecting a strong correlation and time delay between the two signals. The delay time corresponding to the maximum peak value marked with a red dot in the figure is shown. The peak transmission delay time is 25.2 seconds, and this peak transmission delay time is vertically marked with a red dashed line in the figure.

[0031] The peak transmission delay time =25.2 seconds is the transmission delay time of the temperature measurement point signal relative to the main steam pressure signal. By performing this cross-correlation peak extraction processing on all temperature measurement point signals, the transmission delay of each measurement point is obtained, realizing the time delay alignment of the furnace temperature field distribution data. The determination of this delay time provides a key parameter for the subsequent time delay alignment unit to perform shift resampling processing on the temperature field data, ensuring the accurate time alignment of the temperature field data with the main steam pressure fluctuation sequence, thereby providing an accurate time basis for multi-scale frequency decomposition and feature fusion mapping.

[0032] Example 2:

[0033] In specific implementation, please refer to Figure 2 The multi-scale frequency decomposition unit includes a Chebyshev bandpass filter bank and a low-frequency trend reconstruction sub-unit. The Chebyshev bandpass filter bank is used to perform multi-band bandpass filtering on the time-aligned temperature field data and the main steam pressure fluctuation sequence, generating a set of sub-band signals in multiple frequency bands. The Chebyshev bandpass filter bank consists of a set of second-order Chebyshev bandpass filters, the number of which is determined based on the dynamic frequency range to be analyzed during the wide-load operation of the thermal power unit. The center frequencies of the second-order Chebyshev bandpass filters follow a logarithmic interval distribution, with the first... Center frequency The method for determining it is as follows:

[0034] in, This represents the lowest center frequency of the Chebyshev bandpass filter bank. The value is 0.001Hz, which corresponds to the lower limit of the frequency of the slowest dynamic process determined by the thermal inertia of the boiler side of the thermal power unit. Indicates the center frequency spacing factor. The value is set to 1.41, ensuring that adjacent center frequencies are arranged at equal intervals on a logarithmic coordinate system. The value of 1.41 is based on the fact that the spacing between adjacent center frequencies is divided into 1 / 2 octaves to take into account the frequency resolution of the slow-changing process on the boiler side and the fast-changing process on the turbine side. This represents the index of a second-order Chebyshev bandpass filter. The range of values ​​is , This represents the total number of second-order Chebyshev bandpass filters. A value of 12 ensures the highest center frequency coverage reaches 2.83Hz, corresponding to the upper frequency limit of the turbine-side control valve opening's fast response. The passband bandwidth of each second-order Chebyshev bandpass filter is set to 20% of the distance between adjacent center frequencies, where the distance between adjacent center frequencies is defined as... passband bandwidth The calculation method is as follows The passband bandwidth is set to 20% of the distance between adjacent center frequencies to maximize the frequency coverage of each passband while ensuring that there is no significant spectral aliasing between adjacent frequency bands. The stopband attenuation rate of each second-order Chebyshev bandpass filter is determined by the order of the second-order filter, the passband ripple is set to 0.5dB, and the stopband attenuation is no less than 30dB at twice the passband bandwidth from the center frequency.

[0035] In practice, the Chebyshev bandpass filter bank performs multi-band bandpass filtering on the time series of each temperature measurement point in the time-aligned temperature field data. For each temperature measurement point's time series, the time series is input into... In a series of second-order Chebyshev bandpass filters, each second-order Chebyshev bandpass filter outputs a sub-band signal corresponding to the center frequency of the passband, generating a total of [number missing] signals. Each sub-band signal constitutes a sub-band signal set corresponding to the temperature measurement point. The same multi-band bandpass filtering process is performed on the main steam pressure fluctuation sequence to generate a corresponding sub-band signal set. After performing multi-band bandpass filtering on the time series of all temperature measurement points and the main steam pressure fluctuation sequence, the Chebyshev bandpass filter bank outputs multiple sets of sub-band signals, each set containing... The total number of sub-band signals is equal to the number of temperature measurement points plus one.

[0036] In practical implementation, the input of the low-frequency trend reconstruction subunit is connected to the output of the Chebyshev bandpass filter bank, and the low-frequency trend reconstruction subunit receives the sub-band signal set. The low-frequency trend reconstruction subunit reconstructs and sums the sub-band signals above the preset cutoff frequency to generate high-frequency differential-mode components. The preset cutoff frequency is denoted as... , The value is 0.05Hz. The value of 0.05Hz is chosen because the corner frequency corresponding to the equivalent time constant of the boiler-side fuel quantity adjustment to the main steam pressure response is approximately 0.05Hz, which is lower than... The frequency band primarily characterizes the common-mode following characteristics of the boiler and turbine sides when the target load value changes, and is higher than... The frequency band primarily characterizes the differential-mode interaction between boiler-side combustion disturbances and turbine-side control valve disturbances. The low-frequency trend reconstruction subunit performs frequency band partitioning processing on each set of sub-band signals: for signals with a center frequency higher than... The sub-band signals output by the second-order Chebyshev bandpass filter are algebraically summed at each sampling time to obtain the high-frequency differential-mode components corresponding to the set of sub-band signals. For each temperature measurement point in the time-aligned temperature field data, a high-frequency differential-mode component is generated for the sub-band signal set corresponding to the time series, resulting in a total of high-frequency differential-mode components equal to the number of temperature measurement points. For the sub-band signal set corresponding to the main steam pressure fluctuation sequence, a high-frequency differential-mode component corresponding to the main steam pressure fluctuation sequence is generated.

[0037] In practical implementation, the low-frequency trend reconstruction subunit reconstructs and averages sub-band signals below the preset cutoff frequency to generate low-frequency common-mode components. For center frequencies below... The sub-band signals output by the second-order Chebyshev bandpass filter are algebraically summed at each sampling time, and then divided by the number of sub-band signals involved in the summation to obtain the low-frequency common-mode component corresponding to the set of sub-band signals. For each temperature measurement point's time series in the time-aligned temperature field data, a low-frequency common-mode component is generated for each corresponding sub-band signal set, resulting in a total of low-frequency common-mode components equal to the number of temperature measurement points. For the sub-band signal set corresponding to the main steam pressure fluctuation sequence, a low-frequency common-mode component corresponding to the main steam pressure fluctuation sequence is generated. The low-frequency trend reconstruction sub-unit outputs the generated multiple low-frequency common-mode components and multiple high-frequency differential-mode components in parallel to the feature fusion mapping unit.

[0038] Example 3:

[0039] In specific implementation, please refer to Figure 3The disturbance inverse compensation calculation module includes a feedforward gain calculation unit and a compensation signal generation unit. The feedforward gain calculation unit receives the high-frequency differential mode component from the decoupling feature matrix output by the feature fusion mapping unit in the time-frequency domain composite decoupling module, and simultaneously receives the load change rate limit value from the multi-source data acquisition module. The feedforward gain calculation unit performs instantaneous extremum capture processing on the high-frequency differential mode component to determine the positive and negative polarity pulse sequences of the high-frequency differential mode component. The instantaneous extremum capture process is as follows: the feedforward gain calculation unit scans the first high-frequency differential mode component corresponding to the main steam pressure fluctuation sequence point by point, identifying all local maxima and local minima sampling points in the time series of the first high-frequency differential mode component; the determination condition for a local maxima sampling point is that the amplitude of the sampling point is greater than the amplitude of the previous adjacent sampling point and greater than the amplitude of the next adjacent sampling point; the determination condition for a local minima sampling point is that the amplitude of the sampling point is less than the amplitude of the previous adjacent sampling point and less than the amplitude of the next adjacent sampling point. The feedforward gain calculation unit divides all sampling points between an adjacent local maximum and a local minimum sampling point into a unipolar interval. Local maximum and local minimum sampling points alternate, forming multiple unipolar intervals with alternating polarities. For each unipolar interval, the feedforward gain calculation unit extracts the maximum absolute value of the amplitude among all sampling points within that interval and uses this maximum value as the pulse amplitude corresponding to that unipolar interval. The feedforward gain calculation unit determines the pulse polarity based on the sign of the amplitude of the sampling points within the unipolar interval: if the amplitude is positive, a positive polarity pulse is generated; if the amplitude is negative, a negative polarity pulse is generated. The feedforward gain calculation unit arranges all positive and negative polarity pulses corresponding to the unipolar intervals in chronological order, forming a positive and negative polarity pulse sequence. Each pulse in the sequence contains three attributes: pulse polarity, pulse amplitude, and pulse duration. The pulse duration is defined as the difference between the start and end times of the corresponding unipolar interval.

[0040] In practical implementation, the feedforward gain calculation unit calculates the phase compensation coefficient based on the ratio of the duration of the positive and negative polarity pulse sequences to the load change rate limit. For each pulse in the positive and negative polarity pulse sequences, the feedforward gain calculation unit obtains the pulse duration of that pulse, denoted as... Simultaneously, the load change rate limit value is obtained, denoted as... , Data is sent from the power grid dispatch center to the multi-source data acquisition module. The unit is megawatts per minute. The value ranges from 1 MW / min to 50 MW / min. Phase compensation coefficient. The calculation formula is:

[0041] in, Represents the phase compensation coefficient. The physical unit is , Used to characterize the required phase compensation between boiler fuel quantity regulation command and turbine valve opening regulation command; This indicates the pulse duration of the currently processed pulse in the positive and negative polarity pulse sequence. The unit is seconds. The result is obtained by subtracting the end time and the start time of the unipolar interval using the feedforward gain calculation unit. This indicates the load change rate limit value. The unit is megawatts per minute (MW / min), and it is used in calculations. The values ​​need to be converted to megawatt-seconds. The feedforward gain calculation unit calculates a corresponding phase compensation coefficient for each pulse in the positive and negative polarity pulse sequence. This forms a phase compensation coefficient sequence that is time-synchronized with the positive and negative polarity pulse sequence. Each of the phase compensation coefficients in the sequence... The value varies with the pulse duration. The frequency response is dynamically adjusted according to changes in the frequency differential mode component. When the duration of the unipolar pulse in the high-frequency differential mode component is short, the frequency response is adjusted accordingly. Larger values ​​require stronger feedforward compensation; when the duration of the unipolar pulse in the high-frequency differential mode component is long... Smaller values ​​require weaker feedforward compensation.

[0042] In practical implementation, the compensation signal generation unit receives the phase compensation coefficient sequence from the feedforward gain calculation unit. The compensation signal generation unit simultaneously receives the initial fuel quantity adjustment command output from the boiler-side conventional PID control loop and the initial valve opening adjustment command output from the turbine-side conventional PID control loop. The initial fuel quantity adjustment command is the fuel quantity adjustment amount calculated by the boiler-side conventional PID control loop based on the deviation between the load target value and the actual main steam pressure. The initial valve opening adjustment command is the valve opening adjustment amount calculated by the turbine-side conventional PID control loop based on the deviation between the load target value and the actual unit power output. The compensation signal generation unit performs phase compensation calculations on the initial fuel quantity adjustment command and the initial valve opening adjustment command to generate a feedforward compensation signal. The feedforward compensation signal includes a fuel quantity lead compensation component and a valve lag compensation component. The fuel quantity lead compensation component is generated as follows: the compensation signal generation unit acquires the time sequence of the initial fuel quantity adjustment command, calculates the first-order forward difference value of the initial fuel quantity adjustment command at each sampling time, and compares the first-order forward difference value with the phase compensation coefficient corresponding to the current sampling time. Multiplying these components yields the fuel quantity advance compensation component. The valve lag compensation component is generated as follows: the compensation signal generation unit acquires the time series of the initial valve opening adjustment command, calculates the first-order forward difference value of the initial valve opening adjustment command at each sampling time, and then multiplies the first-order forward difference value with the phase compensation coefficient corresponding to the current sampling time. Multiplying the negative values ​​yields the valve lag compensation component. The compensation signal generation unit superimposes the fuel quantity advance compensation component with the initial fuel quantity adjustment command at the same timestamp, and simultaneously superimposes the valve lag compensation component with the initial valve opening adjustment command at the same timestamp. The superimposed fuel quantity adjustment command and valve opening adjustment command are then transmitted to the adaptive load response execution module. The fuel quantity advance compensation component acts on the output of the conventional PID control loop on the boiler side, causing the boiler fuel quantity adjustment command to lead the initial fuel quantity adjustment command in phase. The valve lag compensation component acts on the output of the conventional PID control loop on the turbine side, causing the turbine valve opening adjustment command to lag the initial valve opening adjustment command in phase. The combination of advance and lag compensation offsets the phase difference between the boiler-side combustion process and the turbine-side valve action process.

[0043] Example 4:

[0044] In specific implementation, the feedforward gain calculation unit includes a zero-crossing detection subunit, a pulse generation subunit, and a compensation coefficient mapping subunit. The zero-crossing detection subunit receives the high-frequency differential mode component from the decoupling feature matrix output by the feature fusion mapping unit in the time-frequency domain composite decoupling module. The zero-crossing detection subunit performs zero-crossing point identification processing on the first high-frequency differential mode component corresponding to the main steam pressure fluctuation sequence, recording the sign change direction of the first high-frequency differential mode component at each zero-crossing point, and generating a sign change sequence. The zero-crossing detection subunit uses a hysteresis comparator for zero-crossing point identification. The input of the hysteresis comparator is connected to the continuous time sequence signal of the first high-frequency differential mode component. The positive threshold level and the negative threshold level of the hysteresis comparator are not equal; the positive threshold level is set above zero, and the negative threshold level is set below zero. The hysteresis width of the hysteresis comparator is defined as the absolute value of the difference between the positive and negative threshold levels, and the hysteresis width is set to 5% of the root mean square amplitude of the first high-frequency differential mode component. The calculation method for the root mean square (RMS) amplitude of the first high-frequency differential mode component is as follows: Sum the squares of the amplitudes of all sampling points of the first high-frequency differential mode component within a preset time window, divide by the total number of sampling points within the preset time window, and then take the square root. The length of the preset time window is 10 times the period corresponding to the lowest frequency component in the first high-frequency differential mode component, to cover a sufficient number of signal fluctuation periods. The hysteresis width is set to 5% of the RMS amplitude of the first high-frequency differential mode component because: a 5% ratio filters out small-amplitude noise fluctuations in the high-frequency differential mode component while retaining effective zero-crossing events representing the dynamic energy exchange between the boiler side and the turbine side. The hysteresis comparator outputs a high level when the amplitude of the first high-frequency differential mode component rises from below the positive threshold level and crosses the positive threshold level, and outputs a low level when the amplitude of the first high-frequency differential mode component falls from above the negative threshold level and crosses the negative threshold level. Each time the output level flips, the zero-crossing detection subunit records the flip time as the zero-crossing point and simultaneously records the flip direction, which includes a positive zero-crossing (flipping from low to high) and a negative zero-crossing (flipping from high to low). The zero-crossing detection subunit arranges all zero-crossing points in chronological order and converts the flip direction of each zero-crossing point into a sign change direction. A positive zero-crossing corresponds to a sign change direction from negative to positive, and a negative zero-crossing corresponds to a sign change direction from positive to negative, forming a sign change sequence. The time interval between two adjacent elements in the sign change sequence is not fixed and depends on the actual fluctuation frequency of the first high-frequency differential mode component.

[0045] In specific implementation, the pulse generation subunit receives the symbol change sequence from the zero-crossing detection subunit. The pulse generation subunit performs peak hold processing on the absolute value of the amplitude of the first high-frequency differential mode component between two adjacent zero-crossing points in the symbol change sequence, generating a positive and negative polarity pulse sequence. The peak hold processing process is as follows: the pulse generation subunit, based on the first zero-crossing point in the symbol change sequence... The moment past midnight and the first A pulse generation interval is determined by each zero-crossing moment. This represents the index of the zero-crossing point in the sign change sequence. The value of is a positive integer and , No. The zero-crossing time is the starting time of the pulse generation interval, and the first zero-crossing time is the starting time of the pulse generation interval. The zero-crossing point marks the end of the pulse generation interval. The pulse generation subunit reads the amplitude of all sampling points of the first high-frequency differential mode component within the pulse generation interval and takes the absolute value of the amplitude at each sampling point. The pulse generation subunit compares all the absolute values ​​of the amplitude within the pulse generation interval and extracts the maximum value as the peak hold value corresponding to the pulse generation interval. The pulse generation subunit determines the polarity of the pulse based on the original amplitude sign of the sampling points of the first high-frequency differential mode component within the pulse generation interval: if the amplitude of all sampling points of the first high-frequency differential mode component within the pulse generation interval is positive, then the pulse generation interval corresponds to generating a positive polarity pulse; if the amplitude of all sampling points of the first high-frequency differential mode component within the pulse generation interval is negative, then the pulse generation interval corresponds to generating a negative polarity pulse. The pulse generation subunit sets the amplitude of positive pulses to the peak-hold value and assigns a positive sign, and sets the amplitude of negative pulses to the peak-hold value and assigns a negative sign. In the generated positive and negative pulse sequences, the positive pulses correspond to the first high-frequency differential mode component interval with positive amplitudes, and the negative pulses correspond to the first high-frequency differential mode component interval with negative amplitudes. The time position of each pulse in the positive and negative pulse sequences corresponds one-to-one with the time interval defined by two adjacent zero-crossing points in the sign change sequence.

[0046] In practical implementation, the compensation coefficient mapping subunit receives positive and negative polarity pulse sequences from the pulse generation subunit, and simultaneously receives load change rate limit values ​​from the multi-source data acquisition module. The compensation coefficient mapping subunit performs timing processing on the duration of each polarity pulse in the positive and negative polarity pulse sequence. The timing processing method is as follows: for the [number]th [pulse] in the positive and negative polarity pulse sequence... A polar pulse, This indicates the index of the polarity pulse within the positive and negative polarity pulse sequence. The value of is a positive integer, and the compensation coefficient mapping subunit obtains the first . The start and end times of the polarity pulse, where the start time is the nth polarity pulse in the symbol change sequence. The zero-crossing time is the termination time of the zero-crossing time in the symbol change sequence. At the zero-crossing moment, calculate the time difference between the ending moment and the starting moment; this time difference is the zero-crossing moment. The duration of the polarity pulse. The compensation coefficient mapping subunit calculates the duration of the first polarity pulse. The ratio of the duration of each polarity pulse to the load change rate limit value, denoted as [missing value]. , The unit is megawatts per minute, and the phase compensation coefficient is... The calculation formula is:

[0047] in, Indicates the first Phase compensation coefficients corresponding to each polarity pulse. The physical unit is ; Indicating the positive and negative polarity of the pulse sequence, the first... The duration of each polar pulse, The unit is seconds; This indicates the load change rate limit value. The unit is megawatts per minute (MW / min), and it is used in calculations. The values ​​are directly taken from the raw values ​​in megawatts per minute. The compensation coefficient mapping sub-unit calculates a corresponding phase compensation coefficient for each polarity pulse in the positive and negative polarity pulse sequence according to the above formula. The calculated phase compensation coefficients are then arranged according to the time order of the polarity pulses to form a phase compensation coefficient sequence. The first... Phase compensation coefficient The first and second polarity pulse sequences Each polarity pulse has a one-to-one correspondence in time. The compensation coefficient mapping subunit takes the reciprocal of the phase compensation coefficient calculated for each polarity pulse as the corresponding phase compensation coefficient, and the phase compensation coefficient sequence is output to the compensation signal generation unit.

[0048] Example 5:

[0049] In practical implementation, the adaptive load response execution module includes a rate constraint limiting unit and a command conversion unit. The rate constraint limiting unit receives boiler fuel quantity adjustment commands superimposed with feedforward compensation signals from the compensation signal generation unit of the disturbance inverse compensation calculation module, and turbine valve opening adjustment commands superimposed with feedforward compensation signals. Simultaneously, the rate constraint limiting unit receives load change rate limits from the multi-source data acquisition module. Based on the load change rate limits, the rate constraint limiting unit calculates the maximum rate of change of boiler fuel quantity and the maximum rate of change of turbine valve opening. The maximum rate of change of boiler fuel quantity is calculated by multiplying the load change rate limit by the boiler fuel quantity-load gain coefficient. The boiler fuel quantity-load gain coefficient is defined as the ratio of the boiler fuel quantity change corresponding to a unit load change under steady-state conditions. The boiler fuel quantity-load gain coefficient is obtained by fitting unit thermodynamic characteristic test data, and its typical range is 0.8 tons / hour / MW to 1.5 tons / hour / MW. The maximum rate of change of the turbine-side control valve opening is calculated by multiplying the load change rate limit by the turbine-side control valve opening-load gain coefficient. The turbine-side control valve opening-load gain coefficient is defined as the ratio of the change in turbine control valve opening to the change in unit load under steady-state conditions. The turbine-side control valve opening-load gain coefficient is obtained by fitting the characteristic curve data of the turbine regulating stage. The typical range of the turbine-side control valve opening-load gain coefficient is 0.5% / MW to 1.2% / MW.

[0050] In practical implementation, the rate constraint limiting unit performs first-order lag filtering and limiting processing on the corrected boiler fuel quantity adjustment command and the corrected turbine valve opening adjustment command, respectively, to generate the rate-constrained fuel quantity adjustment sequence and the rate-constrained valve opening adjustment sequence. The rate constraint limiting unit uses a variable time constant first-order lag filter, and the transfer function of the variable time constant first-order lag filter is... ,in, Represents a Laplace complex variable. This represents the time constant of a variable time constant first-order hysteresis filter. The time constant of this filter is adaptively adjusted based on the load change rate limit, and is inversely proportional to the load change rate limit. The calculation formula is:

[0051] in, This represents the time constant of a first-order lag filter with variable time constant. The unit is seconds; Indicates the rate limiting time constant factor. The data was obtained through test data of the closed-loop step response of the thermal power unit's coordinated control system. The identification method is as follows: Under coordinated control of the thermal power unit, a step disturbance with an amplitude of 5% of the rated load is applied to the target load value. The actual response curve of the boiler fuel quantity is recorded. A first-order inertial element is fitted to the actual response curve of the boiler fuel quantity. The fitted time constant is multiplied by the load change corresponding to the step disturbance amplitude to obtain the result. The experimental values, The typical value range is from 50 seconds·megawatts / minute to 150 seconds·megawatts / minute; This indicates the load change rate limit value. The unit is megawatts per minute (MW / min). When the load change rate limit increases, the time constant of the variable time constant first-order lag filter decreases, weakening the filtering effect and allowing for a faster rate of change in the adjustment command. Conversely, when the load change rate limit decreases, the time constant of the variable time constant first-order lag filter increases, strengthening the filtering effect and limiting the rate of change in the adjustment command. The rate constraint limiting unit inputs the corrected boiler fuel quantity adjustment command into the variable time constant first-order lag filter. The output signal of the variable time constant first-order lag filter is then subjected to limiting processing. The limiting value is set to the product of the maximum rate of change in boiler fuel quantity and the sampling period. The sampling period is the operation period of the rate constraint limiting unit, with a value of 100 milliseconds. The output signal after limiting processing constitutes the rate-constrained fuel quantity adjustment sequence. The rate constraint limiting unit inputs the corrected turbine valve opening adjustment command into another variable time constant first-order lag filter with the same structure. The output signal of the other variable time constant first-order lag filter is subjected to limiting processing. The limiting value of the limiting processing is set as the product of the maximum change rate of the turbine side valve opening and the sampling period. The output signal after limiting processing constitutes the valve opening adjustment sequence after rate constraint.

[0052] In practical implementation, the command conversion unit receives the fuel quantity adjustment sequence after rate constraints. The command conversion unit internally stores a preset linear mapping table between pulverized coal flow rate and the opening degree of the fuel actuator. This linear mapping table is established based on calibration data of the actual flow characteristics of the fuel actuators in the thermal power unit's fuel supply system. The fuel actuator is either a coal feeder speed regulating device or a coal mill output regulating baffle. The opening degree of the fuel actuator is defined as a percentage of the coal feeder speed or the opening degree of the coal mill output regulating baffle. The linear mapping table contains a one-to-one correspondence between multiple discrete pulverized coal flow rate values ​​and fuel actuator opening degree values. The unit of pulverized coal flow rate is tons per hour, and the unit of fuel actuator opening degree is a percentage. The pulverized coal flow rate interval between adjacent discrete points in the linear mapping table is 0.5 tons per hour. The instruction conversion unit uses the target fuel quantity value at each sampling moment in the rate-constrained fuel quantity adjustment sequence as the pulverized coal flow rate input value. It then searches a preset linear mapping table between pulverized coal flow rate and fuel actuator opening value for the two discrete pulverized coal flow rate values ​​closest to the input value. Linear interpolation is performed on the corresponding fuel actuator opening values ​​to obtain the fuel actuator drive signal corresponding to the pulverized coal flow rate input value. The linear interpolation method is as follows: Let the pulverized coal flow rate input value be... The linear mapping table with The closest smaller discrete pulverized coal flow rate is The larger discrete pulverized coal flow rate value is , The corresponding fuel actuator opening value is , The corresponding fuel actuator opening value is Then the fuel actuator drive signal The calculation formula is: The command conversion unit outputs the calculated fuel actuator drive signal to the fuel actuator drive interface circuit in the form of a 4mA to 20mA current signal or a 0V to 10V voltage signal.

[0053] See Figure 5 The figure illustrates the rate constraint limiting unit's application of boiler fuel quantity adjustment commands to different load change rate limits within the adaptive load response execution module based on Embodiment 5. and strain time constant The response curve is shown below. The horizontal axis represents time. (Unit: seconds), the vertical axis represents the boiler fuel quantity adjustment command (unit: tons / hour). The graph contains four curves: 1. The black dashed line represents a step change in the initial fuel quantity adjustment command, with a step amplitude of 10 tons / hour. The step occurs at... At the second mark, it reflects the ideal abrupt change in the load target value; 2. The red curve corresponds to Variable time constant The limiting response curve shows a slow, linear upward trend, reaching only about 4.2 tons / hour within 60 seconds, indicating a very strong rate limitation, significant filtering effect, and slow response speed to adjustment commands. 3. The blue curve corresponds to Variable time constant The curve rises exponentially and approaches the target of 10 tons / hour in about 40 seconds. The response speed is significantly faster than the red curve, showing the regulation characteristics under medium rate limit conditions. 4. The light green curve corresponds to Variable time constant The curve quickly approaches the target value, reaching 10 tons / hour in about 10 seconds, exhibiting a relatively weak rate limitation, rapid changes in adjustment commands, and a weak filtering effect.

[0054] The curves in the figure reflect how the rate-constrained limiting unit adaptively adjusts the time constant of the first-order hysteresis filter according to the load change rate limit. The effect of the load change rate limit. The increase of the filter time constant Decreasing the filter weakens its effect, resulting in a faster response to the adjustment command, making it closer to the initial step command. Conversely, decreasing... This will increase the filtering time constant, limit the rate of change of the adjustment command, and make the response smoother.

[0055] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A wide-load intelligent coordination and optimization system for thermal power units based on decoupling control algorithm, characterized in that, include: The multi-source data acquisition module is used to acquire in real time the boiler-side operating parameter set, turbine-side operating parameter set, and grid-side dispatch command signal set of the thermal power unit during wide-load operation. The boiler-side operating parameter set includes furnace temperature field distribution data; the turbine-side operating parameter set includes main steam pressure fluctuation sequence; and the grid-side dispatch command signal set includes load target value and load change rate limit value. The time-frequency domain composite decoupling module is used to perform unified time-frequency domain transformation processing on the furnace temperature field distribution data and the main steam pressure fluctuation sequence to generate a decoupling feature matrix containing low-frequency common-mode components and high-frequency differential-mode components. The disturbance inverse compensation calculation module is used to calculate the phase compensation coefficient between the boiler fuel quantity adjustment command and the turbine valve opening adjustment command based on the high-frequency differential mode component in the decoupling feature matrix, and generate a feedforward compensation signal based on the phase compensation coefficient and superimpose it into the conventional PID control loops on the boiler side and the turbine side. The adaptive load response execution module is used to perform rate consistency constraint output processing on the boiler fuel quantity adjustment command and the turbine valve opening adjustment command according to the load change rate limit value, and generate the final fuel actuator drive signal and valve actuator drive signal.

2. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 1, characterized in that, The time-frequency domain composite decoupling module includes: The time delay alignment unit is used to receive the furnace temperature field distribution data and the main steam pressure fluctuation sequence, perform cross-correlation peak extraction processing on the time series of each temperature measuring point in the furnace temperature field distribution data and the main steam pressure fluctuation sequence, determine the transmission delay time of each temperature measuring point signal relative to the main steam pressure signal, and perform shift resampling processing on the furnace temperature field distribution data according to the transmission delay time to generate time-aligned temperature field data. A multi-scale frequency decomposition unit, whose input is connected to the output of the time delay alignment unit, is used to receive the time-aligned temperature field data and the main steam pressure fluctuation sequence, and to perform bandpass pre-filtering and low-frequency trend term separation processing on the time-aligned temperature field data and the main steam pressure fluctuation sequence, respectively, to generate the low-frequency common-mode component and the high-frequency differential-mode component of the time-aligned temperature field data and the main steam pressure fluctuation sequence. The feature fusion mapping unit, whose input is connected to the output of the multi-scale frequency decomposition unit, is used to matrix-arrange the low-frequency common-mode component and the high-frequency differential-mode component according to their respective timestamps to generate the decoupled feature matrix.

3. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 2, characterized in that, The multi-scale frequency decomposition unit includes: Chebyshev bandpass filter bank is used to perform multi-band bandpass filtering on the time-aligned temperature field data and the main steam pressure fluctuation sequence, respectively, to generate a set of sub-band signals in multiple frequency bands; The Chebyshev bandpass filter bank consists of a group of second-order Chebyshev bandpass filters with center frequencies distributed at logarithmic intervals. The passband bandwidth of each filter is set as a preset percentage of the distance between adjacent center frequencies. The low-frequency trend reconstruction subunit has its input end connected to the output end of the Chebyshev bandpass filter bank. It is used to receive the sub-band signal set, reconstruct and sum the sub-band signals above the preset cutoff frequency to generate the high-frequency differential mode component, and reconstruct and average the sub-band signals below the preset cutoff frequency to generate the low-frequency common mode component.

4. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 1, characterized in that, The disturbance inverse compensation calculation module includes: The feedforward gain calculation unit is used to receive the high-frequency differential mode component in the decoupling feature matrix and the load change rate limit value, perform instantaneous extreme value capture processing on the high-frequency differential mode component, determine the positive and negative polarity pulse sequence of the high-frequency differential mode component, and calculate the phase compensation coefficient based on the ratio of the duration of the positive and negative polarity pulse sequence to the load change rate limit value. The compensation signal generation unit is used to receive the phase compensation coefficient, perform phase compensation calculation on the initial fuel quantity adjustment command and initial valve opening adjustment command output by the conventional PID control loop, and generate the feedforward compensation signal; the feedforward compensation signal includes a fuel quantity advance compensation component and a valve lag compensation component.

5. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 4, characterized in that, The feedforward gain calculation unit includes: The zero-crossing detection subunit is used to perform zero-crossing point identification processing on the high-frequency differential mode component, record the sign change direction of the high-frequency differential mode component at each zero-crossing point time, and generate a sign change sequence. The pulse generation subunit is used to receive the sign change sequence, perform peak hold processing on the absolute value of the amplitude of the high-frequency differential mode component between two adjacent zero crossings, and generate the positive and negative polarity pulse sequences, wherein the positive polarity pulse corresponds to the high-frequency differential mode component interval with positive amplitude and the negative polarity pulse corresponds to the high-frequency differential mode component interval with negative amplitude. The compensation coefficient mapping subunit is used to receive the positive and negative polarity pulse sequence and the load change rate limit value, perform timing processing on the duration of each polarity pulse, calculate the reciprocal of the ratio of the duration of each polarity pulse to the load change rate limit value, and use the reciprocal as the phase compensation coefficient.

6. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 1, characterized in that, The adaptive load response execution module includes: The rate constraint limiting unit is used to receive the boiler fuel quantity regulation command and turbine valve opening regulation command obtained by superimposing the feedforward compensation signal output by the disturbance inverse compensation calculation module onto the output of the conventional PID control loop, as well as the load change rate limit value. Based on the load change rate limit value, it calculates the maximum change rate of boiler fuel quantity and the maximum change rate of turbine valve opening. It performs first-order lag filtering and limiting processing on the corrected boiler fuel quantity regulation command and the corrected turbine valve opening regulation command respectively to generate the rate-constrained fuel quantity regulation sequence and valve opening regulation sequence. The instruction conversion unit is used to receive the fuel quantity adjustment sequence after rate constraint, and convert the fuel quantity adjustment sequence after rate constraint into the driving signal of the fuel actuator according to the preset linear mapping relationship table between pulverized coal flow rate and fuel actuator opening degree.

7. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 2, characterized in that, The delay alignment unit uses the cross-correlation function peak detection method to determine the transmission delay time, and uses cubic spline interpolation to perform shift resampling processing on the furnace temperature field distribution data.

8. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 3, characterized in that, The center frequencies of the Chebyshev bandpass filter bank are distributed at logarithmic intervals, and the passband bandwidth of each filter is set to 20% of the distance between adjacent center frequencies.

9. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 5, characterized in that, The zero-crossing detection subunit uses a hysteresis comparator to identify the zero-crossing point of the high-frequency differential mode component, and the hysteresis width of the hysteresis comparator is set to 5% of the root mean square amplitude of the high-frequency differential mode component.

10. The intelligent coordinated optimization system for thermal power units with wide load range based on decoupling control algorithm according to claim 6, characterized in that, The rate constraint limiting unit employs a variable time constant first-order hysteresis filter. The time constant of the first-order hysteresis filter is adaptively adjusted according to the load change rate limit value, and the time constant is inversely proportional to the load change rate limit value.

Images