Intelligent control response delay compensation method and system for industrial valves

Abstract

The application provides an intelligent control response delay compensation method and system for an industrial valve, and relates to the technical field of industrial valve control. The method comprises the following steps: obtaining a valve position feedback signal and an original control signal of a controller in a valve control loop; performing adaptive notch processing on the valve position feedback signal; performing Hilbert transform on the original control signal; extracting a fundamental component from the valve position feedback signal; calculating an instantaneous phase difference between an instantaneous phase of the original control signal and an instantaneous phase of the fundamental component; moving the instantaneous phase of the original control signal by the instantaneous phase difference while keeping the instantaneous amplitude unchanged to generate a phase reconstruction control signal, and performing valve control. The application solves the technical problem that, in the prior art, a fixed parameter compensation method cannot adapt to time-varying physical hysteresis due to drift of an industrial valve after long-term operation, thereby affecting the control precision of the industrial valve, and inhibits low-frequency oscillation of a control loop and improves the switching control precision of the industrial valve.

Description

Technical Field

[0001] This application relates to the field of industrial valve control technology, and in particular to a method and system for intelligent control response delay compensation for industrial valves. Background Technology

[0002] In actual operation, valves are exposed to complex media environments with high pressure differentials, high temperatures, strong corrosion, or containing solid particles for extended periods. This inevitably leads to wear on their mechanical friction pairs, a gradual increase in static friction between the valve stem and packing, a degradation of the elastic modulus of the sealing elements, and a significant nonlinear drift in the response characteristics of pneumatic or electric actuators over time. Simultaneously, the deformation of the valve core and seat under high-speed media scouring further alters the valve's flow characteristic curve, resulting in an expanded dead zone, changes in starting torque requirements, and a dynamic increase in pure time lag. These progressive physical degradations cause the valve's actual dynamic behavior to deviate from its initial commissioning state, and this deviation is unpredictable and time-varying, making it difficult to mitigate through offline tuning.

[0003] Currently, existing delay compensation methods typically employ fixed-parameter compensation strategies, where the compensation amount is not adjusted according to changes in operating conditions and valve status. When frictional torque increases or actuator stiffness drift causes a change in the hysteresis phase, a mismatch occurs between the fixed compensation parameters and the actual required compensation amount. This not only fails to effectively eliminate response delay but may also introduce additional phase shifts, exacerbating the periodic oscillations of the controlled flow and thus affecting the control accuracy of industrial valves.

[0004] In summary, existing technologies suffer from the technical problem that drift occurs in industrial valves after long-term operation, causing fixed parameter compensation methods to be unable to adapt to time-varying physical hysteresis. When the hysteresis accumulates to a certain extent, it will induce low-frequency oscillations in the control loop, further affecting the control accuracy of industrial valves. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for intelligent control response delay compensation for industrial valves, in order to solve the technical problem in the prior art where drift occurs after long-term operation of industrial valves, causing fixed parameter compensation methods to be unable to adapt to time-varying physical hysteresis. When the hysteresis accumulates to a certain extent, it will induce low-frequency oscillations in the control loop, further affecting the control accuracy of industrial valves.

[0006] In view of the above problems, this application provides a method and system for intelligent control response delay compensation for industrial valves.

[0007] In a first aspect, this application provides a method for intelligent control response delay compensation for industrial valves. This method is implemented through an intelligent control response delay compensation system for industrial valves. The method includes: acquiring a valve position feedback signal and a raw control signal output by a controller in the valve control loop; performing adaptive notch filtering on the valve position feedback signal to identify the frequency of the current low-frequency oscillation; performing a Hilbert transform on the raw control signal to extract the instantaneous amplitude and instantaneous phase; extracting a fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal and acquiring the instantaneous phase of the fundamental frequency component; calculating the instantaneous phase difference between the instantaneous phase of the raw control signal and the instantaneous phase of the fundamental frequency component; shifting the instantaneous phase of the raw control signal forward by the instantaneous phase difference while maintaining the instantaneous amplitude unchanged to generate a phase reconstruction control signal, and using the phase reconstruction control signal for valve control.

[0008] Optionally, the valve position feedback signal is input to a notch filter, wherein the center frequency of the notch filter is adjustable, and an error signal is obtained through the notch filter; with the goal of minimizing the energy of the error signal, the center frequency of the notch filter is iteratively updated online using a least mean square algorithm until the center frequency converges to a stable value, wherein the stable value is the frequency of low-frequency oscillation.

[0009] Optionally, in each control cycle, the instantaneous squared value of the error signal is calculated as an energy estimate; the sensitivity of the energy estimate to the center frequency is used as gradient information; the center frequency is adjusted in a fixed step size along the negative direction of the gradient information so that the energy of the error signal gradually decreases.

[0010] Optionally, the valve position feedback signal is simultaneously input into a sine wave generator with the same center frequency as the notch filter to obtain a reference sine signal; the reference sine signal is multiplied by the error signal and then integrated over time, and the result is used as gradient information.

[0011] Alternatively, the Hilbert transform can be implemented using a finite impulse response (FIR) filter, whose coefficients are updated online based on the current oscillation frequency.

[0012] Optionally, the valve position feedback signal is input to a bandpass filter, the passband center frequency of which is set to the current low-frequency oscillation frequency identified by the notch filter, to obtain a filtered signal containing only the fundamental frequency component of the oscillation; a Hilbert transform is performed on the filtered signal to generate an orthogonal signal that is 90 degrees out of phase with the filtered signal; the filtered signal is used as the real part and the orthogonal signal is used as the imaginary part to construct a complex analytic signal, and the instantaneous phase is extracted from the complex analytic signal.

[0013] Optionally, the ratio of the orthogonal signal to the filtered signal is calculated, and an arctangent operation is performed on the ratio; the result of the arctangent operation is subjected to phase unwinding processing to obtain the instantaneous phase.

[0014] Optionally, based on the amplitude of the low-frequency oscillation in the valve position feedback signal, a pre-trained gain mapping relationship is invoked to generate a variable gain; the phase reconstruction control signal is multiplied by the variable gain and then superimposed on the original control signal to generate a compensated control signal acting on the valve actuator.

[0015] Optionally, when the instantaneous phase difference continuously exceeds the phase compensation limit and the low-frequency oscillation amplitude attenuation is less than the preset significant attenuation, a gain attenuation strategy is triggered.

[0016] Secondly, this application also provides an intelligent control response delay compensation system for industrial valves, used to execute the intelligent control response delay compensation method for industrial valves as described in the first aspect. The intelligent control response delay compensation system for industrial valves includes: a signal acquisition module for acquiring the valve position feedback signal and the original control signal output by the controller in the valve control loop; an adaptive notch filtering module for performing adaptive notch filtering on the valve position feedback signal to identify the frequency of the current low-frequency oscillation; a Hilbert transform module for performing a Hilbert transform on the original control signal to extract the instantaneous amplitude and instantaneous phase; a fundamental frequency component extraction module for extracting the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal and acquiring the instantaneous phase of the fundamental frequency component; an instantaneous phase difference calculation module for calculating the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component; and a phase reconstruction module for shifting the instantaneous phase of the original control signal forward by the amount of the instantaneous phase difference while keeping the instantaneous amplitude unchanged, generating a phase reconstruction control signal, and using the phase reconstruction control signal for valve control.

[0017] One or more technical solutions provided in this application have at least the following beneficial effects:

[0018] By acquiring the valve position feedback signal and the original control signal output by the controller in the valve control loop; performing adaptive notch filtering on the valve position feedback signal to identify the frequency of the current low-frequency oscillation; performing Hilbert transform on the original control signal to extract the instantaneous amplitude and instantaneous phase; extracting the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal and obtaining the instantaneous phase of the fundamental frequency component; calculating the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component; shifting the instantaneous phase of the original control signal forward by the amount of the instantaneous phase difference while keeping the instantaneous amplitude unchanged to generate a phase reconstruction control signal, which is then used for valve control. In other words, by extracting the dynamic phase deviation between the valve position feedback signal and the original command signal in real time without adding any additional mechanical action or control impact, and performing time-axis-leading phase reconstruction on subsequent commands based on this deviation, low-frequency oscillations in the control loop are effectively suppressed, improving the control accuracy of industrial valve switching.

[0019] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

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

[0021] Figure 1 This is a flowchart illustrating the intelligent control response delay compensation method for industrial valves proposed in this application.

[0022] Figure 2 This is a schematic diagram of the intelligent control response delay compensation system for industrial valves according to this application.

[0023] Figure labeling: Signal acquisition module 11, Adaptive notch filtering module 12, Hilbert transform module 13, Fundamental frequency component extraction module 14, Instantaneous phase difference calculation module 15, Phase reconstruction module 16. Detailed Implementation

[0024] This application provides an intelligent control response delay compensation method and system for industrial valves, solving the technical problem in existing technologies where drift occurs after long-term operation of industrial valves, causing fixed parameter compensation methods to be unable to adapt to time-varying physical hysteresis. When this hysteresis accumulates to a certain level, it induces low-frequency oscillations in the control loop, further affecting the control accuracy of the industrial valve. By extracting the dynamic phase deviation between the valve position feedback signal and the original command signal in real time without adding any extra mechanical action or control shock, and reconstructing the leading phase of subsequent commands on the time axis based on this deviation, the low-frequency oscillations of the control loop are effectively suppressed, improving the control accuracy of industrial valve switching.

[0025] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0026] Example 1, please refer to the appendix. Figure 1 This application provides a method for compensating for intelligent control response delay in industrial valves. The method is executed by an intelligent control response delay compensation system for industrial valves, and specifically includes the following steps: Acquire the valve position feedback signal and the original control signal output by the controller in the valve control loop.

[0027] Specifically, two signal lines are connected in parallel at the control cabinet's wiring terminals. One line is led out from the feedback output terminal of the valve positioner and connected to a high-precision current / voltage isolation transmitter to convert the current signal into a voltage signal suitable for the analog-to-digital converter input. The other line is led out from the analog output terminal of the controller, and after being processed by an isolator, it is connected to another channel of the analog-to-digital converter.

[0028] Set the sampling frequency, such as 100Hz, and simultaneously perform analog-to-digital conversion on both channels to ensure that the two digital values ​​are latched without time difference. In the digital signal processor, read the original ADC register value and convert the digital value back to engineering units using a calibration formula. If the input range of the acquisition card is 0-10V, corresponding to a digital value of 0-65535, the current voltage value can be converted into the corresponding control output percentage and valve position feedback percentage based on a linear relationship.

[0029] Perform sliding window averaging filtering on the original digital sequence, such as taking a window length of 10 sampling points, but the timestamps corresponding to the two signals must be retained during the filtering process.

[0030] The valve position feedback signal is an electrical signal detected in real time by a position sensor installed on the valve stem or valve core and converted into a value that precisely corresponds to the actual mechanical opening of the valve, such as 0% fully closed and 100% fully open. The raw control signal output by the controller is a command electrical signal directly output by the upper-level controller after calculating the deviation between the set value and the feedback value. It is also a standard signal of 4-20mA or 0-10V, representing the target position that the controller expects the valve to reach.

[0031] The valve position feedback signal is subjected to adaptive notch filtering to identify the frequency of the current low-frequency oscillation.

[0032] Furthermore, this application also includes the following steps: inputting the valve position feedback signal into a notch filter, wherein the center frequency of the notch filter is adjustable, and obtaining an error signal through the notch filter; with the goal of minimizing the energy of the error signal, using a least mean square algorithm to iteratively update the center frequency of the notch filter online until the center frequency converges to a stable value, wherein the stable value is the frequency of low-frequency oscillation.

[0033] Furthermore, this application also includes the following steps: in each control cycle, calculating the instantaneous squared value of the error signal as an energy estimate; using the sensitivity of the energy estimate to the center frequency as gradient information; adjusting the center frequency in a fixed step size along the negative direction of the gradient information, so that the energy of the error signal gradually decreases.

[0034] Furthermore, this application also includes the following steps: simultaneously inputting the valve position feedback signal into a sine wave generator with the same center frequency as the notch filter to obtain a reference sine signal; multiplying the reference sine signal with the error signal and then performing time integration, the result of which is used as gradient information.

[0035] Specifically, the valve position feedback signal acquired in real time is sent to a digital notch filter, and a characteristic signal, namely the error signal, that reflects the current notch filtering effect is generated by calculating the difference between the original signal and the filtered output signal.

[0036] In the program memory of the digital signal processor, a transfer function structure for a second-order infinite impulse response digital notch filter is predefined, consisting of a set of fixed notch filter coefficient expressions and two adjustable parameters (center frequency and notch bandwidth). During power-on initialization, these two adjustable parameters are assigned initial values. The initial value of the center frequency can be set based on field experience, such as selecting an intermediate value within the typical frequency range of valve oscillation (0.5Hz to 5Hz), usually set to 1Hz. The notch filter internally contains several delay units, multipliers, and adders, used to perform weighted summation operations on the current and historical values ​​of the input signal and the historical values ​​of the output signal. After each center frequency update, based on the new center frequency value and a pre-set quality factor, usually a fixed value between 10 and 30, the coefficients in the notch filter's difference equation are recalculated, including the forward and feedback coefficients. This recalculates the frequency response shape of the notch filter, including the depth and width of the notch notch and the gain flatness of the passband region. After the coefficients are calculated, they are written into the internal register of the notch filter for use in subsequent filtering operations.

[0037] At the arrival of each sampling period, the instantaneous value x(n) of the current valve position feedback signal is read and sent to the configured notch filter module. The notch filter, based on its internal difference equation, uses the current input value, historical input values ​​from previous time steps, and historical output values ​​from previous time steps to perform multiplication and addition operations to calculate the current filtered output value y(n), representing the remaining frequency components of the original input signal that were not attenuated by the notch filter. Since the notch filter has a significant attenuation at the center frequency, if the current center frequency is exactly equal to the dominant oscillation frequency in the input signal, this oscillation component will almost disappear in the output y(n). Within the current sampling period, the original input valve position feedback signal value x(n) is subtracted from the notch filter output value y(n) to obtain the current error signal value, representing the difference between the original input signal and the filtered signal; essentially, it is the portion of the spectrum that the notch filter has extracted from the original signal. When the center frequency of the notch filter is precisely locked at the true oscillation frequency, the error signal will mainly contain the energy of that oscillation frequency, and its amplitude will be approximately equal to the oscillation amplitude in the original signal. When the center frequency deviates from the true frequency, the energy of the error signal will decrease significantly because the notch filter fails to effectively capture and filter out this oscillation component. The error signal specifically refers to the difference between the original valve position feedback signal and the residual signal output after filtering by the notch filter, that is, the portion of oscillation energy in the original signal that was captured and filtered out by the notch filter.

[0038] Perform a multiplication operation on the error signal to obtain its instantaneous squared value, which is the square of the instantaneous value of the error signal, and use it as the energy estimate.

[0039] At the beginning of each sampling period, the current center frequency value is read from the parameter register of the notch filter. This value, updated in the previous iteration, represents the current frequency position targeted by the notch filter. This current center frequency value is used as the frequency control word for the sine wave generator, ensuring that the frequency of the reference signal is always strictly synchronized with the center frequency of the notch filter without any deviation. The frequency control word is a digitally encoded value input to the digital oscillator, determining the frequency of the oscillator's output signal; essentially, it is the angular step size that the phase accumulator needs to advance in each sampling period.

[0040] The current center frequency value f0 and the system sampling period T are compared. s Multiplying this by the constant 2π yields an angle value, called the phase increment Δ. , denoted as Δ =2π·f0·T s The physical meaning of phase increment is the angle that the rotation vector of the digital oscillator needs to rotate through within the duration of a single sampling period. The higher the frequency, the larger the angle rotated within one period, and the faster the output waveform changes. A dedicated register called the phase accumulator is maintained in the data memory to store the current phase angle value. At each new sampling point, the current phase increment is added to the existing value in the phase accumulator, i.e., ... (n)= (n-1)+Δ The new value is accumulated and overwrites the old value. Starting from an initial angle of zero, a fixed angle Δ is added every sampling period. As the number of sampling points increases, the accumulated phase monotonically increases from zero. When it accumulates to 2π radians, the rotation vector completes a full circle, and subsequent angles enter the next cycle. The updated phase accumulator value is used as the index address and sent to the sine lookup table module. The sine lookup table is a pre-calculated data table stored in the program memory, where each storage unit stores the sine function value corresponding to an angle. Since the phase accumulator value may exceed the range of 0 to 2π, phase truncation or folding processing is required before looking up the table. This involves taking the remainder of the accumulated phase divided by 2π to map the angle back to the range of a complete circle, and then using the lookup table address corresponding to the remainder for the lookup. The value output by the lookup table is the reference sine signal value r(n) = sin[ [n], whose amplitude is normalized to 1, is a dimensionless pure number.

[0041] Within the same sampling period, the reference sinusoidal signal value r(n) is multiplied by the error signal value e(n) output by the notch filter, resulting in a product p(n) = e(n)·r(n). This multiplication is performed once per sampling period, outputting an instantaneous product value. If the error signal contains a component with the same frequency as the reference signal, a DC offset proportional to the error amplitude will appear in the product; if the frequency of the error signal is different from that of the reference signal, the product result will oscillate rapidly between positive and negative, and the average value of long-term integration will approach zero.

[0042] A circular buffer of length N sampling points is allocated in the data storage, where N is 128, to ensure that the buffer covers an integer number of low-frequency oscillation cycles. For example, at sampling rates of 1.2Hz and 100Hz, 128 points cover 1.28s, containing approximately 1.5 cycles. When a new sampling point arrives, the current product value is stored in the buffer, while the oldest product value in the buffer is removed. All N product values ​​in the buffer are then summed to obtain the gradient estimate for the current time step. This refers to gradient information; the sign clearly indicates the direction of the deviation between the current center frequency and the true oscillation frequency. Based on the sign of the gradient value, the frequency update module determines the adjustment direction of the center frequency: if... If (n) > 0, it means the error energy increases with increasing frequency; the current notch filter frequency is lower than the true value, and the center frequency needs to be increased. If (n) < 0, it means the current notch frequency is higher than the true value, and the center frequency needs to be reduced; if If (n)≈0, it means that the center frequency is close to the true value and has entered the locked state.

[0043] Based on the fundamental principle of the steepest descent method, the center frequency is adjusted along the negative direction of the current gradient. This is done according to the formula f0(n+1)=f0(n). μ· (n) Calculate the new center frequency value. The negative sign ensures adjustment in the direction of decreasing error energy: when the sensitivity is positive, it means the center frequency is lower than the true value. Increasing the frequency will increase the error energy. The negative sign makes the correction term negative, but since the sensitivity is positive, the actual result is f0(n+1) = f0(n). A positive value indicates an increase in the center frequency, bringing it closer to the true frequency. When the sensitivity is negative, the negative sign makes the correction term positive, decreasing the center frequency. After each update, the new frequency value is written to the notch filter's parameter register, ensuring the notch filter operates at the new center frequency at the start of the next sampling period.

[0044] The center frequency of the notch filter is repeatedly updated in each sampling period. As the number of iterations increases, the center frequency gradually approaches the true oscillation frequency, the absolute value of the sensitivity continuously decreases, and the frequency correction becomes smaller and smaller until the center frequency converges to a stable value. That is, after multiple iterations, the center frequency of the notch filter no longer changes significantly and stabilizes around a certain value with slight fluctuations. The stable value is the true frequency of the dominant low-frequency oscillation in the valve position feedback signal. In engineering, the convergence condition is usually that the change in the center frequency is less than a preset threshold, such as 0.001Hz, within a certain number of consecutive control periods.

[0045] The center frequency of the notch filter has converged to the true oscillation frequency, marking the completion of all work on the adaptive frequency tracking link.

[0046] In industrial valve control circuits, valve position feedback signals may experience continuous periodic fluctuations due to excessive valve friction, improper controller parameter tuning, or fluctuations in medium pressure. The frequency of these fluctuations is typically between 0.5Hz and 5Hz, far below the power grid frequency, and is therefore termed low-frequency oscillation.

[0047] The original control signal is subjected to Hilbert transform to extract the instantaneous amplitude and instantaneous phase.

[0048] Furthermore, this application also includes the following steps: the Hilbert transform is implemented through a finite impulse response filter, and the coefficients of the finite impulse response filter are updated online according to the current oscillation frequency.

[0049] Specifically, the Hilbert transform is implemented using a finite impulse response filter, designed based on the ideal frequency response characteristics of the Hilbert transform: in the positive frequency range, the amplitude gain is constant at 1 and the phase shift is constant at -90°; in the negative frequency range, the amplitude gain is constant at 1 and the phase shift is constant at +90°; and at zero frequency (DC), the gain is zero. To achieve this characteristic, the filter's unit impulse response coefficient must satisfy specific odd symmetry conditions, and the filter order must be even.

[0050] The infinitely long unit impulse response sequence of the ideal Hilbert transformer is calculated, truncated and smoothed using a Hamming window to obtain a finite-length coefficient sequence. The choice of the filter order M requires a balance between phase shift accuracy and computational complexity; a higher order provides better phase shift accuracy within the passband, but also increases computational cost. Considering the processing power and accuracy requirements of the industrial controller, an even value between 32 and 64 is typically chosen for M. The filter's operating frequency range also needs to be determined during the design process to ensure that the phase shift error meets requirements within the expected oscillation frequency range (e.g., 0.5Hz to 5Hz).

[0051] The frequency response of a Hilbert finite impulse response (FIR) filter is a function of frequency, and its actual phase shift at a specific frequency point depends on the relative relationship between the filter coefficients and the signal frequency. When the signal frequency is exactly at the filter's design center frequency, the phase shift is precisely -90°; when the signal frequency deviates from the design center frequency, the phase shift will deviate. Therefore, to ensure an accurate 90° phase shift at the currently identified oscillation frequency, the filter's design parameters need to be adjusted in real time based on the currently identified oscillation frequency, and a new set of filter coefficients optimally matched to that frequency needs to be generated. Specifically, a filter coefficient calculation function is pre-defined in the digital signal processor. Whenever the adaptive notch filter updates its identified frequency, the filter coefficient calculation function is called, using the current oscillation frequency as the design center frequency, and the preset filter order M and window function type as fixed parameters, to recalculate all M filter coefficients and write the new coefficients into the filter's coefficient register.

[0052] At the arrival of each sampling period, the instantaneous value of the original control signal at the current moment is read and fed into a Hilbert finite impulse response filter with pre-configured coefficients for real-time convolution. The filter, based on its difference equation, uses the current input value and the historical input values ​​from the previous M-1 moments to perform multiplication and addition operations to calculate the filtered output value at the current moment, which is the orthogonal signal of the original control signal. The finite impulse response filter has a fixed group delay, meaning the output signal is delayed by M / 2 sampling periods relative to the input signal. The group delay of the filter is M / 2 multiplied by the system sampling period. To ensure strict alignment of the real and imaginary parts of the signal on the time axis, a pure delay compensation unit with the same delay amount must be connected in series in the real part channel. That is, the original control signal value used when calculating the instantaneous amplitude and instantaneous phase should be the signal after subtracting the delay.

[0053] After Hilbert transform filtering and delay compensation, the original control signal after delay compensation is taken as the real part, and the orthogonal signal output by the Hilbert transform is taken as the imaginary part; together, they form a complex sequence. The instantaneous amplitude is extracted by performing a modulo operation on the complex sequence, i.e., the sum of the squares of the real and imaginary parts, and taking the square root of the sum to obtain the instantaneous amplitude. This operation is performed once at each sampling point, outputting a non-negative amplitude sequence representing the envelope strength of the original control signal at that moment.

[0054] The instantaneous phase is extracted by performing a four-quadrant arctangent operation on the complex sequence, outputting the principal phase value, which is limited to the range of -180° to +180°. Subsequently, the principal phase value sequence is input into the phase unwinding module, which detects the phase jump between adjacent sampling points and performs appropriate addition or subtraction of 360° compensation, finally outputting an instantaneous phase curve that changes continuously and smoothly over time without abrupt jumps.

[0055] An online update trigger mechanism is implemented to continuously monitor the identification frequency of the pre-adaptive notch filter output. When the detected change in the identification frequency exceeds a preset update threshold (e.g., 0.02Hz), it is determined that the current filter coefficients are insufficient to maintain the required phase shift accuracy, and the filter coefficients are immediately recalculated. The recalculation process involves redesigning a set of finite impulse response filter coefficients based on the new frequency values ​​and employing a smooth switching strategy to complete the transition between the old and new coefficients over several sampling periods, thus avoiding phase disturbances caused by abrupt coefficient changes. After the update is complete, the filter's operating frequency is re-matched to the current signal frequency, and the quadrature signal generation accuracy is restored to its optimal state.

[0056] In digital signal processing, the Hilbert transform is a linear time-invariant system that produces a 90° phase shift for all positive frequency components of the input signal and a +90° phase shift for all negative frequency components, while maintaining the signal amplitude unchanged. By inputting the original control signal as its real part into the Hilbert transform, its output is a signal of the same frequency that is orthogonal to the original signal (i.e., out of phase by 90 degrees). By constructing a complex analytic signal (the real part being the original signal and the imaginary part being the orthogonal signal), the instantaneous amplitude and instantaneous phase of the signal can be extracted simultaneously. The instantaneous amplitude refers to the magnitude of the complex analytic signal at any sampling time, reflecting the instantaneous size of the signal envelope at that moment; the instantaneous phase refers to the phase angle of the complex analytic signal at any sampling time, reflecting the periodic position of the signal at that moment.

[0057] By constructing a complex analytic signal and performing modulo operation and four-quadrant arctangent operation, the instantaneous amplitude and instantaneous phase of the original control signal can be obtained simultaneously in one processing flow.

[0058] Extract the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal, and obtain the instantaneous phase of the fundamental frequency component.

[0059] Furthermore, this application also includes the following steps: inputting the valve position feedback signal into a bandpass filter, wherein the passband center frequency of the bandpass filter is set to the current low-frequency oscillation frequency identified by the notch filter, to obtain a filtered signal containing only the fundamental frequency component of the oscillation; performing a Hilbert transform on the filtered signal to generate an orthogonal signal that is 90 degrees out of phase with the filtered signal; using the filtered signal as the real part and the orthogonal signal as the imaginary part to construct a complex analytic signal, and extracting the instantaneous phase from the complex analytic signal.

[0060] Furthermore, this application also includes the following steps: calculating the ratio of the orthogonal signal to the filtered signal, performing an arctangent operation on the ratio; and performing phase unwinding processing on the result of the arctangent operation to obtain the instantaneous phase.

[0061] Specifically, the currently identified low-frequency oscillation frequency is read from the convergence result of the adaptive notch filter and refreshed by the notch filter's iterative update module after each sampling period. The low-frequency oscillation frequency is then written into the filter control register as the center frequency parameter of the bandpass digital filter. Based on a preset quality factor Q value, typically between 20 and 30, the upper and lower passband frequencies of the filter are calculated to ensure the passband width is narrow enough to filter out interference components from adjacent frequencies, while also being wide enough to tolerate minor residual deviations that may exist in the frequency tracking stage. A bandpass filter is a frequency-selective digital filter that allows signals within a specific frequency range to pass through while attenuating all frequency components outside that range. The passband center frequency of the bandpass filter is dynamically set to the current low-frequency oscillation frequency identified by the adaptive notch filter. The passband width (i.e., the quality factor Q value) is specially designed so that only signals within an extremely narrow frequency band centered on this oscillation frequency can pass through, while DC components, high-frequency noise, and other harmonic components are significantly suppressed, thereby extracting a pure fundamental frequency oscillation signal. The passband center frequency is the frequency point with the maximum gain in the frequency response of the bandpass filter, that is, the center position of the signal frequency that the filter allows to pass. It is directly taken from the low-frequency oscillation frequency identification value output by the pre-adaptive notch filter after convergence, thus realizing the dynamic alignment of the passband position of the bandpass filter with the actual frequency of the signal.

[0062] At the arrival of each sampling period, the instantaneous value x(n) of the valve position feedback signal at the current moment is read and input into a bandpass digital filter with a pre-configured center frequency for real-time convolution filtering. The filter typically employs a fourth- or sixth-order infinite impulse response structure to achieve a balance between the steepness of the transition band and the computational complexity. After filtering, the original DC bias component in the output signal s(n) has been removed, and high-frequency noise has been significantly attenuated, leaving only a pure fundamental frequency oscillation signal with an approximately sinusoidal shape dominated by the current oscillation frequency.

[0063] A pure fundamental frequency signal sequence is used as the real part signal and input into a Hilbert transformer. Similar to the process described above, the output signal is orthogonal to this signal, maintaining a 90-degree phase difference across all frequencies. Because the design center frequency of the Hilbert transformer always remains consistent with the current oscillation frequency, the phase shift accuracy at the current frequency can be maintained within ±0.5°. Simultaneously, since the finite impulse response Hilbert transformer has a fixed group delay, appropriate delay compensation is required for the real part channel during engineering implementation to ensure strict alignment of the real and imaginary signals on the time axis.

[0064] The filtered signal is used as the real part, and the orthogonal signal output from the Hilbert transform is used as the imaginary part. Together, they form a complex sequence, i.e., a complex analytic signal. The orthogonal signal and the filtered signal are strictly aligned on the time axis, forming the real and imaginary parts of a complex pair. The built-in four-quadrant arctangent function of the digital signal processor is called, with the orthogonal signal value as the imaginary part input and the filtered signal value as the real part input. The output is the principal phase value at that moment. Specifically, the ratio of the imaginary part to the real part is calculated, and an arctangent operation is performed on this ratio to obtain a base angle. Then, the quadrant of the current phase angle is determined based on the signs of the real and imaginary parts. The corresponding angle offset correction is performed on the base angle, and finally, an angle value in the range of -180° to +180° is output. Special handling is required for the boundary cases where the real part is 0: when the real part is positive zero and the imaginary part is positive, the output is +90°; when the real part is negative zero and the imaginary part is negative, the output is -90°; and when both the real and imaginary parts are 0, the output is 0. The use of the four-quadrant arctangent function avoids the inherent defect of the single-valued arctangent function in that it cannot correctly distinguish between the first and third quadrants, or between the second and fourth quadrants.

[0065] The principal phase value calculated at the current sampling point is stored in a dedicated phase history buffer. This buffer must store at least the unwound phase value of the previous sampling point and the principal phase value of the current sampling point. A transition detection operation is performed: the difference between the current principal phase value and the unwound phase value of the previous sampling point is calculated. If this difference is greater than +180°, it indicates a negative cross-boundary transition from the positive half-cycle to the negative half-cycle; if the difference is less than -180°, it indicates a positive cross-boundary transition from the negative half-cycle to the positive half-cycle; if the difference is between ±180°, it indicates no cross-boundary transition has occurred, and no compensation operation is required.

[0066] A cumulative compensation register is maintained in the data memory to record the cumulative compensation values ​​that have been applied up to the current moment, in multiples of 360°. Based on the transition detection results from the previous step, the following operations are performed: if a negative cross-boundary transition is detected (difference greater than +180°), the cumulative compensation is subtracted by 360°; if a positive cross-boundary transition is detected (difference less than -180°), the cumulative compensation is added by 360°; if no transition occurs, the cumulative compensation remains unchanged. Subsequently, the current phase principal value is algebraically added to the cumulative compensation to obtain the instantaneous phase value after unwinding at the current moment.

[0067] The unwound instantaneous phase value is stored in the history buffer and used as the reference value for the transition detection at the next sampling point. The above steps are repeated in each sampling period to continuously output the instantaneous phase curve.

[0068] Because the result of the arctangent operation is forcibly restricted to the principal value range of -180° to +180°, whenever the actual phase crosses the boundary of this range, the calculated value will jump abruptly from +180° to -180° (or vice versa). Phase unwrapping is a post-processing algorithm that detects phase jumps between adjacent sampling points. When the jump exceeds positive -180°, it automatically adds or subtracts an integer multiple of 360°, restoring the phase value, which was originally folded within a finite range, to a true phase curve that changes continuously over time without abrupt jumps.

[0069] By using a narrow-band pass filter with dynamic center frequency tracking, the DC bias in the valve position feedback signal is completely filtered out. The passband width is only 0.06Hz, and the signal-to-noise ratio is increased to over 40dB. This ensures that the signal components used for subsequent phase calculations are pure and are not affected by static deviations and broadband noise.

[0070] Calculate the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component.

[0071] Specifically, it is ensured that the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component are calculated based on data from the same sampling time, and both have undergone phase dewinding processing, and are each a continuous and smooth time series. The instantaneous phase difference at the current time is obtained by subtracting the instantaneous phase of the fundamental frequency component from the instantaneous phase value of the original control signal.

[0072] Since the two phase values ​​may be at different accumulation period levels (e.g., the control signal phase may have accumulated to several hundred degrees, while the feedback signal phase may have accumulated to different values), the difference obtained by direct subtraction may exceed the reasonable range of ±180°. To give the phase difference a clear physical meaning, namely the relative offset within one oscillation cycle, the difference result is normalized: the difference result is modulo 360° and mapped to the interval between -180° and +180°. Specifically, if the instantaneous phase difference is greater than 180°, 360° is subtracted; if the instantaneous phase difference is less than 180°, 360° is added. The phase difference after this processing directly reflects the phase offset between the control command and the valve position response within a single oscillation cycle; a positive value indicates that the command leads the response. Point-to-point difference calculation refers to the operation of subtracting the values ​​at the same sampling moment in two time series. Since the two signals have the same sampling period and the sampling moments are strictly aligned, the subtraction operation can be performed directly in each sampling period without interpolation or resampling.

[0073] The instantaneous phase of the original control signal is shifted forward by the amount of the instantaneous phase difference, while keeping the instantaneous amplitude unchanged, to generate a phase reconstruction control signal, which is then used to control the valve.

[0074] Specifically, at the arrival of each sampling period, the instantaneous amplitude, instantaneous phase, and instantaneous phase difference extracted from the original control signal are read. At the same time, the DC bias component in the original control signal, i.e., the signal mean or set reference value, is read directly from the long-term average value of the original control signal.

[0075] The instantaneous phase of the original control signal is added to its instantaneous phase difference to obtain the reconstructed new phase value. Under normal operating conditions, the reconstructed new phase value is positive, therefore it shifts forward on the time axis relative to the original phase, meaning the control command is issued earlier. If the reconstructed new phase value exceeds the range of a single cycle, the sine function calculation automatically utilizes its periodicity for folding, eliminating the need for additional normalization of the phase value itself.

[0076] Using the reconstructed new phase value, combined with the instantaneous amplitude and DC bias component, the new instantaneous value of the control signal is calculated according to the standard sinusoidal signal reconstruction formula: New instantaneous control signal value = Instantaneous amplitude × sin(reconstructed new phase value) + DC bias component. This process is performed once per sampling period, outputting a sinusoidal waveform with DC bias. Its amplitude is exactly the same as the original signal, but its phase is advanced by the instantaneous phase difference in angle. The instantaneous amplitude must be coordinated with the DC bias during the phase reconstruction process to ensure that the physical dimensions of the reconstructed signal are consistent with the original signal.

[0077] The instantaneous value of the new control signal is written to the digital-to-analog converter (DAC) register of the digital signal processor. The DAC converts this digital quantity into an analog voltage or current signal, which is then isolated, amplified, and output to the input of the valve positioner or electro-pneumatic converter. This signal directly drives the actuator, replacing the uncompensated control signal output by the original controller. The phase-reconstructed control signal is a new control command signal generated based on the original control signal by shifting the instantaneous phase forward while maintaining the instantaneous amplitude. It retains the force information of the original command but adjusts the timing of the command issuance, allowing the actuator to begin action with the appropriate lead time. The generated phase-reconstructed control signal is converted into a standard industrial analog signal by the DAC, typically a 4mA to 20mA current loop or a 0V to 10V voltage, and sent to the input terminal of the valve positioner or actuator to drive the valve stem to the target position. The phase-reconstructed control signal replaces the control signal output by the original controller, serving as the actual drive command for the actuator.

[0078] For example, a steam flow control system in a chemical plant uses a DN200 pneumatic diaphragm regulating valve. Due to the large air volume of the actuator, an 80ms response delay occurs, resulting in a phase difference of approximately 72° at a low-frequency oscillation of 2.5Hz, with a valve position feedback oscillation amplitude of 8%. Operating at a sampling frequency of 1000Hz, after compensation is activated at t=2s, the signal converges to the true frequency of 2.5Hz in approximately 1.5s using an LMS adaptive notch filter (initial guess 2.0Hz, step size μ=0.001). Subsequently, Hilbert transforms are performed on the original control signal and the bandpass-filtered fundamental frequency component to extract the instantaneous phases φ1(t) and φ2(t), respectively, and the instantaneous phase difference Δφ=φ1 is calculated. φ2 is used to shift the phase of the original control signal forward by Δφ while keeping the amplitude unchanged to generate a phase reconstruction signal. Then, based on the real-time oscillation amplitude, a variable gain is generated by pre-trained S-shaped gain mapping (8% amplitude corresponds to a gain of 0.55, and the upper limit is 0.80) and superimposed on the original control signal. Ultimately, the phase difference gradually decreases from 72° to 23.7°, a reduction of 67.1%, and the oscillation amplitude decreases synchronously from 8% to 2.8%, a reduction of 65.1%. Since the phase difference does not continuously exceed the 115° compensation upper limit and the oscillation attenuation is greater than the 0.3 threshold, the gain attenuation strategy is not triggered, and the system operates stably.

[0079] By strictly maintaining the instantaneous amplitude constant, response hysteresis is eliminated without increasing any stress on the actuator. Because the instantaneous amplitude is kept constant, this compensation method does not amplify high-frequency noise or introduce additional resonant peaks, thus improving response speed while ensuring robustness.

[0080] Furthermore, this application also includes the following steps: generating a variable gain by calling a pre-trained gain mapping relationship based on the amplitude of the low-frequency oscillation in the valve position feedback signal; multiplying the phase reconstruction control signal by the variable gain and then superimposing it on the original control signal to generate a compensated control signal acting on the valve actuator.

[0081] Specifically, within each sampling period, the filtered fundamental frequency signal value s(n) at the current moment is read from the output data buffer of the pre-bandpass filter. To obtain a stable amplitude estimate, a sliding window peak detection method is used: within a time window of a complete oscillation period, for example, for a 1.2Hz signal, the window length is 1s, i.e., 100 sampling points. The maximum and minimum values ​​within the window are recorded, and the peak-to-peak value is calculated as the estimate of the current oscillation amplitude, i.e., the difference between the maximum and minimum values ​​divided by 2. Here, the peak-to-peak value is the total span between the maximum and minimum values ​​of the signal within a complete period. This effectively resists the influence of noise at individual sampling points and provides a stable amplitude estimate.

[0082] The oscillation amplitude is used as an index to input into a pre-trained gain mapping table. This pre-trained gain mapping table is stored as an array in the digital signal processor's program memory, recording the optimal gain coefficients corresponding to several discrete amplitude points. When the input amplitude is exactly equal to a discrete point in the table, the corresponding gain coefficient is directly output; when the input amplitude lies between two discrete points, linear interpolation is performed to calculate the corresponding gain coefficient. The process of establishing the pre-trained gain mapping table is as follows: During the debugging phase before system commissioning, by applying disturbance signals of different amplitudes to the valves, the oscillation suppression effect and system stability under different gain coefficients are tested. The gain-amplitude correspondence that attenuates the oscillation amplitude the fastest while ensuring system stability is selected, and this correspondence is fixed in the program as a lookup table. For example, three representative opening points are selected within the valve stroke range, such as 20%, 50%, and 80%. At each opening point, sinusoidal disturbance signals of different amplitudes are superimposed onto the valve control loop using a signal generator. The percentage of the peak-to-peak value of the valve position feedback signal oscillation to the total valve stroke is taken as 1%, 3%, 5%, 8%, and 12%, respectively, to simulate low-frequency oscillation conditions of varying severity. For each amplitude level, under the stability constraints of ensuring a phase margin of not less than 30° and a gain margin of not less than 6dB in the control system, candidate gain coefficients are traversed in the range of 0.05 to 0.95 with a step size of 0.05. The settling time required for the oscillation amplitude to decay to 25% of the initial amplitude under the action of each candidate gain is recorded. The gain coefficient that satisfies the stability constraints and minimizes the settling time is selected as the optimal gain for that amplitude level. Each amplitude level and its corresponding optimal gain are stored as discrete nodes in the read-only memory of the digital signal processor. Linear interpolation is used between nodes to calculate the gain corresponding to the intermediate amplitude, thus forming an initial gain mapping table. A typical design example of the mapping table is that an amplitude of 8% corresponds to a gain of 0.6, an amplitude of 5% corresponds to a gain of 0.4, an amplitude of 3% corresponds to a gain of 0.2, and an amplitude of 1% corresponds to a gain of 0.05. The mapping relationship ensures that strong compensation is applied when the oscillation is severe, and the compensation is automatically weakened after the oscillation decays, avoiding over-adjustment.

[0083] A multiplication operation is performed, multiplying the variable gain coefficient by the AC component of the phase reconstruction control signal to obtain a scaled compensation component. The amplitude of the compensation component is controlled by the variable gain; when the oscillation amplitude is large, the compensation component is also large; when the oscillation amplitude is effectively suppressed, the compensation component automatically shrinks. An addition operation is performed, algebraically adding the scaled compensation component to the original control signal value at the current moment to obtain the final compensated control signal value. This ensures that the basic information of the original control command is completely preserved, while incorporating the dynamic compensation component. The superimposed signal undergoes amplitude limiting processing to ensure that its value does not exceed the safe operating range of the actuator, such as a limiting protection of 4mA to 20mA.

[0084] The final control signal value is written to the digital-to-analog converter register, and after digital-to-analog conversion and isolation amplification, it is output to the valve actuator. The above steps are repeated in each sampling period to form a dynamically adjusted closed-loop system. When the current oscillation amplitude is large, the gain is automatically increased to enhance the compensation effect; when the current oscillation amplitude decreases, the gain is automatically decreased to avoid overcompensation.

[0085] Variable gain is an amplification factor whose value is not fixed and is adjusted in real time according to the current oscillation amplitude. Its value typically ranges from 0 to 1. When the oscillation amplitude is large, the gain is large; when the oscillation amplitude has been effectively suppressed, the gain automatically decreases to near zero, determining how much force is applied to the phase reconstruction signal to help suppress oscillation. The essential difference between variable gain and traditional fixed gain is that traditional fixed gain compensation remains constant throughout the entire operation. When the valve's physical characteristics drift, causing oscillation to intensify, the fixed gain may be insufficient; conversely, when the valve returns to stability, the same gain may be excessive, triggering new high-frequency flutter. The variable gain mechanism, through amplitude feedback, achieves adaptive adjustment, effectively avoiding both of these problems.

[0086] By linking the variable gain to the current oscillation amplitude, the system automatically determines the required compensation level. It provides strong compensation in the initial stages to quickly suppress oscillations and slight auxiliary compensation in the later stages to maintain steady-state operation. The compensation component is superimposed on the original control signal, rather than completely replacing it. This means that even if the compensation module outputs zero due to some abnormal reason, the final control signal remains equal to the original control signal, and the control system can still operate normally; only the compensation effect disappears. This ensures that the system's basic control functions are not lost under extreme conditions, providing high safety redundancy.

[0087] Furthermore, this application also includes the following steps: when the instantaneous phase difference continuously exceeds the phase compensation upper limit and the low-frequency oscillation amplitude attenuation is less than the preset significant attenuation, a gain attenuation strategy is triggered.

[0088] Specifically, a duration counter is set to record the number of sampling periods in which the instantaneous phase difference continuously exceeds the phase compensation upper limit. The phase compensation upper limit is dynamically set to 85% of the total lag phase at the current oscillation frequency, meaning the maximum effective compensation phase must not exceed 85% of the product of the current oscillation frequency and the inherent pure lag time of the actuator, with an absolute value of 90° as the hard limit cap. In each sampling period, the current phase difference value is compared with the phase compensation upper limit. If the current phase difference value is greater than the upper limit, the counter is incremented by 1; if the current phase difference value is less than or equal to the upper limit, the counter is reset and the timing restarts. When the accumulated value of the counter exceeds a preset duration threshold, such as 3 seconds, the condition that the instantaneous phase difference continuously exceeds the phase compensation upper limit is considered met.

[0089] Read the current oscillation amplitude and the initial amplitude at the compensation activation time, and calculate the current amplitude attenuation: Amplitude attenuation = (Initial amplitude at compensation activation time - Current oscillation amplitude) / Initial amplitude at compensation activation time × 100%. Compare the amplitude attenuation with the preset significant attenuation. If the amplitude attenuation is less than the preset significant attenuation, it indicates that the oscillation amplitude has not decreased significantly after compensation activation, and the compensation effect is not ideal; this condition is met. If the amplitude attenuation is greater than or equal to the preset significant attenuation, it indicates that the compensation has shown a significant effect, and there is no need to trigger protection action. For example, the preset significant attenuation is (A0-At) / A0 The calculation is 100%, where A0 is the initial peak-to-peak value of the oscillation recorded at the compensation activation moment, and At is the current peak-to-peak value of the oscillation after five consecutive oscillation cycles. If the preset significant attenuation level can be set to 30%, and the oscillation amplitude attenuates less than 30% within a specified time after compensation activation, the compensation effect is considered unsatisfactory, posing a risk of continued oscillation, and a protection strategy needs to be triggered. The setting of the preset significant attenuation level needs to strike a balance between prematurely triggering protection leading to interruption of normal compensation and prematurely triggering protection leading to continued deterioration of the oscillation.

[0090] When the instantaneous phase difference continuously exceeds the phase compensation upper limit and the low-frequency oscillation amplitude attenuation is less than the preset significant attenuation, it is determined that the current phase reconstruction compensation method can no longer effectively eliminate the oscillation, triggering a gain attenuation strategy. A reduced new gain value is written to the gain parameter register of the main controller, usually using a gradual attenuation method. If the current gain is K... p The new gain value is 0.9K. p This means the amplitude is reduced by 10% each time. This process is repeated in each sampling cycle until the oscillation amplitude decays beyond the preset significant decay level, indicating that the oscillation has been effectively suppressed, or until the main controller gain drops to near the preset safety lower limit. If the gain has dropped to the lower limit but the oscillation is still not eliminated, an emergency alarm is triggered, notifying the operator to intervene.

[0091] After the gain attenuation strategy takes effect, continuously monitor the trend of oscillation amplitude changes. When the oscillation amplitude attenuation exceeds the preset significant attenuation level, it indicates that the gain reduction strategy has taken effect and the oscillation is being effectively suppressed. Gradually restore the main controller's gain to its normal value, using a gradual increase method opposite to the reduction, or maintain the attenuated gain level. The specific strategy depends on the on-site operating conditions and the operator's preset selection. If the oscillation amplitude continues to worsen or other system anomalies occur, trigger a high-level safety alarm and recommend manual intervention.

[0092] In summary, the intelligent control response delay compensation method for industrial valves provided in this application has the following beneficial effects: It acquires the valve position feedback signal and the original control signal output by the controller in the valve control loop; performs adaptive notch filtering on the valve position feedback signal to identify the frequency of the current low-frequency oscillation; performs Hilbert transform on the original control signal to extract the instantaneous amplitude and instantaneous phase; extracts the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal and obtains the instantaneous phase of the fundamental frequency component; calculates the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component; shifts the instantaneous phase of the original control signal forward by the amount of the instantaneous phase difference while keeping the instantaneous amplitude unchanged to generate a phase reconstruction control signal, which is then used for valve control. In other words, by extracting the dynamic phase deviation between the valve position feedback signal and the original command signal in real time without adding any additional mechanical action or control shock, and performing advance phase reconstruction on the time axis for subsequent commands based on this deviation, it effectively suppresses low-frequency oscillations in the control loop and improves the control accuracy of industrial valve switching.

[0093] Example 2: Based on the same inventive concept as the intelligent control response delay compensation method for industrial valves in Example 1, this application also provides an intelligent control response delay compensation system for industrial valves. Please refer to the appendix. Figure 2 The intelligent control response delay compensation system for industrial valves includes: The system comprises: a signal acquisition module 11 for acquiring the valve position feedback signal and the original control signal output by the controller in the valve control loop; an adaptive notch filtering module 12 for performing adaptive notch filtering on the valve position feedback signal to identify the frequency of the current low-frequency oscillation; a Hilbert transform module 13 for performing a Hilbert transform on the original control signal to extract the instantaneous amplitude and instantaneous phase; a fundamental frequency component extraction module 14 for extracting the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal and acquiring the instantaneous phase of the fundamental frequency component; an instantaneous phase difference calculation module 15 for calculating the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component; and a phase reconstruction module 16 for shifting the instantaneous phase of the original control signal forward by the amount of the instantaneous phase difference while keeping the instantaneous amplitude unchanged to generate a phase reconstruction control signal for valve control.

[0094] Furthermore, the adaptive notch processing module 12 in the intelligent control response delay compensation system for industrial valves is also used to: input the valve position feedback signal into a notch filter, wherein the center frequency of the notch filter is adjustable, and obtain an error signal through the notch filter; with the goal of minimizing the energy of the error signal, use the least mean square algorithm to iteratively update the center frequency of the notch filter online until the center frequency converges to a stable value, wherein the stable value is the frequency of low-frequency oscillation.

[0095] Furthermore, the adaptive notch processing module 12 in the intelligent control response delay compensation system for industrial valves is also used to: calculate the instantaneous square value of the error signal as an energy estimate in each control cycle; use the sensitivity of the energy estimate to the center frequency as gradient information; and adjust the center frequency in a fixed step size along the negative direction of the gradient information so that the energy of the error signal gradually decreases.

[0096] Furthermore, the adaptive notch processing module 12 in the intelligent control response delay compensation system for industrial valves is also used to: simultaneously input the valve position feedback signal into a sine wave generator with the same center frequency as the notch filter to obtain a reference sine signal; multiply the reference sine signal with the error signal and then perform time integration, and use the result as gradient information.

[0097] Furthermore, the Hilbert transform module 13 in the intelligent control response delay compensation system for industrial valves is also used to: implement the Hilbert transform through a finite impulse response filter, wherein the coefficients of the finite impulse response filter are updated online according to the current oscillation frequency.

[0098] Furthermore, the fundamental frequency component extraction module 14 in the intelligent control response delay compensation system for industrial valves is also used to: input the valve position feedback signal into a bandpass filter, wherein the passband center frequency of the bandpass filter is set to the current low-frequency oscillation frequency identified by the notch filter, to obtain a filtered signal containing only the oscillation fundamental frequency component; perform a Hilbert transform on the filtered signal to generate an orthogonal signal with a phase difference of 90 degrees from the filtered signal; construct a complex analytic signal by using the filtered signal as the real part and the orthogonal signal as the imaginary part, and extract the instantaneous phase from the complex analytic signal.

[0099] Furthermore, the fundamental frequency component extraction module 14 in the intelligent control response delay compensation system for industrial valves is also used to: calculate the ratio of the orthogonal signal to the filtered signal, perform an arctangent operation on the ratio, and perform phase unwinding processing on the result of the arctangent operation to obtain the instantaneous phase.

[0100] Furthermore, the phase reconstruction module 16 in the intelligent control response delay compensation system for industrial valves is also used to: generate a variable gain by calling a pre-trained gain mapping relationship based on the amplitude of the low-frequency oscillation in the valve position feedback signal; multiply the phase reconstruction control signal by the variable gain and then superimpose it onto the original control signal to generate a compensated control signal acting on the valve actuator.

[0101] Furthermore, the intelligent control response delay compensation system for industrial valves is also used to trigger a gain attenuation strategy when the instantaneous phase difference continuously exceeds the phase compensation upper limit and the low-frequency oscillation amplitude attenuation is less than a preset significant attenuation.

[0102] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The intelligent control response delay compensation method and specific examples for industrial valves in the aforementioned Embodiment 1 are also applicable to the intelligent control response delay compensation system for industrial valves in this embodiment. Through the foregoing detailed description of the intelligent control response delay compensation method for industrial valves, those skilled in the art can clearly understand the intelligent control response delay compensation system for industrial valves in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0103] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0104] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for intelligent control response delay compensation for industrial valves, characterized in that, include: Acquire the valve position feedback signal and the original control signal output by the controller in the valve control loop; The valve position feedback signal is subjected to adaptive notch filtering to identify the frequency of the current low-frequency oscillation. Perform a Hilbert transform on the original control signal to extract the instantaneous amplitude and instantaneous phase; Extract the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal, and obtain the instantaneous phase of the fundamental frequency component; Calculate the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component; The instantaneous phase of the original control signal is shifted forward by the amount of the instantaneous phase difference, while keeping the instantaneous amplitude unchanged, to generate a phase reconstruction control signal, which is then used to control the valve.

2. The intelligent control response delay compensation method for industrial valves as described in claim 1, characterized in that, Adaptive notch filtering is performed on the valve position feedback signal to identify the frequency of the current low-frequency oscillation, including: The valve position feedback signal is input into a notch filter, wherein the center frequency of the notch filter is adjustable, and an error signal is obtained through the notch filter. With the goal of minimizing the energy of the error signal, the center frequency of the notch filter is updated online using the least mean square algorithm until the center frequency converges to a stable value, which is the frequency of low-frequency oscillation.

3. The intelligent control response delay compensation method for industrial valves as described in claim 2, characterized in that, With the goal of minimizing the energy of the error signal, the center frequency of the notch filter is updated online iteratively using the least mean square algorithm, including: In each control cycle, the instantaneous squared value of the error signal is calculated as an energy estimate; The sensitivity of energy estimation to the center frequency is used as gradient information; The center frequency is adjusted with a fixed step size along the negative direction of the gradient information, so that the energy of the error signal is gradually reduced.

4. The intelligent control response delay compensation method for industrial valves as described in claim 3, characterized in that, The sensitivity of the energy estimate to the center frequency is used as gradient information, including: The valve position feedback signal is simultaneously input into a sine wave generator with the same center frequency as the notch filter to obtain a reference sine signal. The reference sinusoidal signal is multiplied by the error signal and then integrated over time; the result is used as gradient information.

5. The intelligent control response delay compensation method for industrial valves as described in claim 2, characterized in that, Extracting the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal, and obtaining the instantaneous phase of the fundamental frequency component, includes: The valve position feedback signal is input into a bandpass filter, and the passband center frequency of the bandpass filter is set to the current low-frequency oscillation frequency identified by the notch filter to obtain a filtered signal containing only the oscillation fundamental frequency component. Perform a Hilbert transform on the filtered signal to generate an orthogonal signal that is 90 degrees out of phase with the filtered signal; The filtered signal is used as the real part and the orthogonal signal is used as the imaginary part to form a complex analytic signal, and the instantaneous phase is extracted from the complex analytic signal.

6. The intelligent control response delay compensation method for industrial valves as described in claim 5, characterized in that, Extracting the instantaneous phase from the complex analytic signal includes: Calculate the ratio of the orthogonal signal to the filtered signal, and then perform an arctangent operation on the ratio; The result of the arctangent operation is subjected to phase unwinding to obtain the instantaneous phase.

7. The intelligent control response delay compensation method for industrial valves as described in claim 5, characterized in that, The Hilbert transform is achieved using a finite impulse response (FIR) filter, whose coefficients are updated online based on the current oscillation frequency.

8. The intelligent control response delay compensation method for industrial valves as described in claim 1, characterized in that, After generating the phase reconstruction control signal, the process also includes: Based on the amplitude of the low-frequency oscillation in the valve position feedback signal, a variable gain is generated by calling the pre-trained gain mapping relationship; The phase reconstruction control signal is multiplied by the variable gain and then superimposed on the original control signal to generate a compensated control signal that acts on the valve actuator.

9. The intelligent control response delay compensation method for industrial valves as described in claim 1, characterized in that, When the instantaneous phase difference continuously exceeds the phase compensation limit and the low-frequency oscillation amplitude attenuation is less than the preset significant attenuation, the gain attenuation strategy is triggered.

10. A smart control response delay compensation system for industrial valves, characterized in that, The steps for implementing the intelligent control response delay compensation method for industrial valves according to any one of claims 1 to 9, wherein the intelligent control response delay compensation system for industrial valves comprises: The signal acquisition module is used to acquire the valve position feedback signal and the original control signal output by the controller in the valve control loop; An adaptive notch filtering module is used to perform adaptive notch filtering on the valve position feedback signal to identify the frequency of the current low-frequency oscillation. The Hilbert transform module is used to perform a Hilbert transform on the original control signal to extract the instantaneous amplitude and instantaneous phase. The fundamental frequency component extraction module is used to extract the fundamental frequency component with the same frequency as the low-frequency oscillation from the valve position feedback signal, and to obtain the instantaneous phase of the fundamental frequency component; The instantaneous phase difference calculation module is used to calculate the instantaneous phase difference between the instantaneous phase of the original control signal and the instantaneous phase of the fundamental frequency component; The phase reconstruction module is used to shift the instantaneous phase of the original control signal forward by the amount of the instantaneous phase difference, while keeping the instantaneous amplitude unchanged, to generate a phase reconstruction control signal, and to control the valve using the phase reconstruction control signal.

Images