An adaptive adjustment method for an oil refining process

Abstract

The application relates to the field of industrial automation control and discloses an adaptive adjustment method for a grease refining process, which comprises the following steps: synchronously acquiring raw material flow data and load current signals and storing the data into a historical state register sequence; extracting high-frequency power spectrum amplitudes in the load current signals to determine a dynamic correction coefficient reflecting a pipeline slip state; compensating a physical volume model according to the dynamic correction coefficient and calculating material dynamic time delay parameters; reversely searching time sequence associated process parameters in the sequence by using the dynamic time delay parameters; and correcting control gain according to the numerical deviation between the process parameters and the load current signals by a state observer, so as to adjust frying heating power. Through closed-loop compensation of nonlinear rheological characteristics of a transport process, the application ensures that the extracted historical parameters have physical and causal consistency, solves the time delay distortion problem caused by non-uniform materials, and improves system identification accuracy and operation stability.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, and in particular to an adaptive adjustment method for the oil refining process. Background Technology

[0002] Current large-scale continuous production lines typically employ distributed control systems to manage the process flow. These systems utilize sensors to collect operating parameters such as temperature, pressure, and flow rate, and adjust actuators based on preset thresholds. This conventional control method is based on a steady-state assumption, maintaining stable production through fixed feedback loops to meet standard process requirements. While hardware measures such as increasing sensor density can effectively eliminate causal misalignments caused by uneven material distribution in time and space, existing solutions neglect the dynamic evolution of materials during long-distance transport. For example, Chinese invention patent CN118853288B discloses an efficient refining control method for animal fats. The method and system switch stages based on oil quality parameters. The underlying logic is based on the quasi-steady-state assumption of detection-comparison-action. In large-scale production lines, the rheological properties of materials change with moisture content, and the slip ratio and lag time in pipelines fluctuate randomly. However, under continuous processing conditions with a daily processing capacity of 300 tons, due to the dynamic fluctuations in the initial moisture and properties of the raw rapeseed, the rheological properties of the material after roasting undergo high-frequency changes. These changes cause the viscous resistance and slip ratio of the material in the pipeline transportation process to deviate from the original design intention, making the physical movement time of the material from the roasting unit to the pressing unit exhibit random nonlinear variation characteristics.

[0003] Existing technologies typically set the transport delay as a static constant, causing a phase deviation between the feedback reference point of the control system and the actual material state in the causal dimension. Since the controller cannot accurately correlate the current pressing load feedback with the corresponding historical roasting state, the feedback loop is prone to overshoot or periodic oscillation when dealing with material disturbances, which restricts the system's control accuracy for acid value and coefficient of variation. Simply relying on increasing the density of sensor deployment or tightening alarm thresholds not only increases the hardware deployment load of the system, but also fails to solve the causal misalignment problem caused by time-space mismatch at the control logic level. How to construct an adaptive control strategy that can identify the rheological state of the system in real time and achieve alignment with the physical entity dimension has become an important challenge to improve the operational quality of large-scale industrial refining processes.

[0004] Therefore, how to align the control logic with the physical entity in a variable time-delay system affected by rheological characteristics is the technical problem to be solved by this invention. Summary of the Invention

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an adaptive adjustment method for an oil refining process, comprising the following steps:

[0006] Step S101: Obtain the raw material flow data of the controlled object and the load current signal of the pressing motor, and write the raw material flow data and load current signal into the historical state register sequence according to a unified time base.

[0007] Step S102: Extract the power spectrum amplitude in the frequency range of 10Hz to 50Hz from the load current signal and determine it as a characteristic quantity. Calculate the offset between the characteristic quantity and the preset damping reference value, and determine the offset as a dynamic correction coefficient reflecting the slip state of the pipeline material.

[0008] Step S103: Adjust the boundary parameters of the physical volume model of the conveying pipeline using dynamic correction coefficients, and perform volume integration calculation on the time axis of the raw material flow data under the adjusted physical volume model until the integration result reaches the equivalent volume of the conveying pipeline. Then, determine the corresponding integration time as the dynamic time delay parameter of the material passing through the conveying pipeline.

[0009] Step S104: Starting from the current sampling time, reverse addressing is performed in the historical state register sequence according to the dynamic delay parameter to extract specific frame historical process parameters that have a causal correspondence with the current load current signal.

[0010] Step S105: Calculate the numerical deviation between the historical process parameters of a specific frame and the load current signal, use the state observer to capture the rate of change of the numerical deviation, and iteratively correct the adjustment gain of the adaptive control law based on the rate of change.

[0011] Step S106: Adjust the heating power of the frying unit according to the modified adjustment gain to complete the closed-loop control of the oil refining process.

[0012] Preferably, step S102 further includes the following sub-steps: step S1021, performing a fast Fourier transform on the load current signal to obtain a power density distribution map; step S1022, identifying energy peaks in the power density distribution map that belong to the frequency range, and defining the average amplitude of the energy peak as a characteristic quantity to characterize the rheological resistance fluctuation of the material during the pressing process.

[0013] Preferably, step S103 further includes the following sub-steps: Step S1031, obtaining the calibration volume value of the physical volume model. Step S1032: calibrate the volume value. The equivalent pipeline volume after compensation is obtained by multiplying it with the dynamic correction factor. Step S1033: According to the current time... The order of recursion to historical moments, for raw material flow data Perform integration to determine if the condition is met. Duration This is a dynamic delay parameter.

[0014] Preferably, it further includes: real-time monitoring of the pressure difference signal at both ends of the fine filter and calculating the rising slope of the pressure difference signal; when the rising slope continues to exceed a preset threshold, a subtraction compensation weight is superimposed in the adjustment gain to reduce the heating power of the frying unit.

[0015] Preferably, in step S105, the state observer determines the correction weight of the adjustment gain according to the following formula: ,in, To adjust the gain correction weight, The preset convergence factor, This represents the absolute value of the numerical deviation. The standard deviation of a specific parameter in the historical state register sequence. This is the preset minimum value.

[0016] Preferably, after step S106, the method further includes: real-time acquisition of the steady-state error value of the controlled object after adjustment; calculation of the coefficient of variation of the steady-state error value within a preset sliding window; and locking the current adjustment gain as the steady-state control parameter for the production batch when the coefficient of variation is less than 5%.

[0017] Preferably, the method further includes: receiving a process correction matrix from a remote platform, wherein the process correction matrix contains thermo-mechanical coupling characteristic values ​​under different raw material moisture contents; and injecting the process correction matrix as an initial weight vector into the state observer to preset the nonlinear mapping relationship between numerical deviation and adjustment gain.

[0018] Preferably, in step S101, the sampling frequency of the raw material flow data is synchronized with the sampling frequency of the load current signal, and the sampling accuracy of the load current signal is 12 bits to 16 bits.

[0019] Preferably, step S104 includes: retrieving the data timestamp integrity in the historical state register sequence; when data is missing, performing cubic spline interpolation on the historical state register sequence to reconstruct the missing data nodes and ensure the physical continuity of historical process parameters for a specific frame.

[0020] Preferably, in step S105, the specific frame historical process parameters include frying temperature and dwell time. The state observer sets the step size for adjusting the gain by calculating the energy coupling sensitivity between the frying temperature and the load current signal.

[0021] The beneficial effects of this invention are:

[0022] 1. In the oil refining process, a time-varying integral traceability mechanism is established. The real physical transport delay between units is determined by the reverse accumulation of real-time volumetric flow rate. This allows the feedback benchmark of the control system to shift from simple time dimension alignment to physical entity causal alignment. This approach enables the current pressing load feedback to be accurately traced back to the historical roasting state with physical correlation, thereby eliminating the control law misalignment interference caused by material spatial transport lag in industrial control. It also avoids overshoot or periodic oscillation in the feedback loop when dealing with raw material property fluctuations, ensuring that the system exhibits high convergence characteristics when handling large-scale materials.

[0023] 2. By extracting the frequency component of the current fluctuation of the pressing motor as a rheological damping feature characterizing the resistance to material flow, and dynamically correcting the equivalent physical volume boundary of the pipeline accordingly, this invention achieves closed-loop compensation for the nonlinear rheological characteristics of the transportation process. This mechanism enables the flow integral addressing process to sense changes in material moisture and viscosity. Even when there are natural fluctuations in the moisture content of the raw material, it can ensure that the historical parameters extracted from the state register queue have the accuracy of physical causal dimension. This solves the problem of residual calculation distortion caused by assuming that the time delay is a static constant in the traditional control method, and improves the adaptive identification depth of the system for complex non-homogeneous materials.

[0024] 3. By adopting a coordinated adjustment mode of thermo-mechanical coupling state matrix and differential pressure gradient monitoring, the dynamic decoupling of multiple variables is achieved while ensuring the precision of fine filtration. This strategy drives the iterative update of characteristic heat transfer coefficient components by observing the effective dynamic residuals after spatiotemporal alignment. Combined with the real-time capture of the differential pressure gradient of the fine filter, the frying temperature control action and the fine filtration cycle adjustment form a logical complementarity. On a production line with a daily processing capacity of 300 tons, this effectively solves the technical contradiction between the stability of key product indicators and the service life of equipment, and suppresses acid value and coefficient of variation to a low level. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0026] Figure 1 This is a flowchart of the closed-loop control for adaptive adjustment in the oil refining process of the present invention;

[0027] Figure 2 This is a logic block diagram for the frequency domain feature extraction and dynamic correction of load current in this invention. Detailed Implementation

[0028] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail 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 should fall within the protection scope of the present invention.

[0029] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0030] Secondly, an embodiment or embodiment referred to herein refers to a specific feature, structure or characteristic that may be included in at least one implementation of the present invention. An embodiment appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0031] This invention is described in detail with reference to the schematic diagrams. When describing the embodiments of this invention, for ease of explanation, the cross-sectional views of the device structure will be partially enlarged without adhering to the general scale. Moreover, the schematic diagrams are only examples and should not limit the scope of protection of this invention. In addition, in actual manufacturing, the three-dimensional spatial dimensions of length, width and depth should be included.

[0032] Furthermore, in the description of this invention, it should be noted that the terms such as "upper," "lower," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or component referred to has a specific orientation, or is constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0033] Unless otherwise explicitly specified and limited, the terms installation, connection, and linking in this invention should be interpreted broadly. For example, they can refer to fixed connection, detachable connection, or integrated connection; similarly, they can refer to mechanical connection, electrical connection, or direct connection, or indirect connection through an intermediate medium, or internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0034] An adaptive adjustment method for an oil refining process includes the following steps:

[0035] Step S101: Obtain the raw material flow data of the controlled object and the load current signal of the pressing motor, and write the raw material flow data and load current signal into the historical state register sequence according to a unified time base.

[0036] Step S102: Extract the power spectrum amplitude in the frequency range of 10Hz to 50Hz from the load current signal and determine it as a characteristic quantity. Calculate the offset between the characteristic quantity and the preset damping reference value, and determine the offset as a dynamic correction coefficient reflecting the slip state of the pipeline material.

[0037] Step S103: Adjust the boundary parameters of the physical volume model of the conveying pipeline using dynamic correction coefficients, and perform volume integration calculation on the time axis of the raw material flow data under the adjusted physical volume model until the integration result reaches the equivalent volume of the conveying pipeline. Then, determine the corresponding integration time as the dynamic time delay parameter of the material passing through the conveying pipeline.

[0038] Step S104: Starting from the current sampling time, reverse addressing is performed in the historical state register sequence according to the dynamic delay parameter to extract specific frame historical process parameters that have a causal correspondence with the current load current signal.

[0039] Step S105: Calculate the numerical deviation between the historical process parameters of a specific frame and the load current signal, use the state observer to capture the rate of change of the numerical deviation, and iteratively correct the adjustment gain of the adaptive control law based on the rate of change.

[0040] Step S106: Adjust the heating power of the frying unit according to the modified adjustment gain to complete the closed-loop control of the oil refining process.

[0041] Preferably, step S102 further includes the following sub-steps: step S1021, performing a fast Fourier transform on the load current signal to obtain a power density distribution map; step S1022, identifying energy peaks in the power density distribution map that belong to the frequency range, and defining the average amplitude of the energy peak as a characteristic quantity to characterize the rheological resistance fluctuation of the material during the pressing process.

[0042] Preferably, step S103 further includes the following sub-steps: Step S1031, obtaining the calibration volume value of the physical volume model. Step S1032: calibrate the volume value. The equivalent pipeline volume after compensation is obtained by multiplying it with the dynamic correction factor. Step S1033: According to the current time... The order of recursion to historical moments, for raw material flow data Perform integration to determine if the condition is met. Duration This is a dynamic delay parameter.

[0043] Preferably, it further includes: real-time monitoring of the pressure difference signal at both ends of the fine filter and calculating the rising slope of the pressure difference signal; when the rising slope continues to exceed a preset threshold, a subtraction compensation weight is superimposed in the adjustment gain to reduce the heating power of the frying unit.

[0044] Preferably, in step S105, the state observer determines the correction weight of the adjustment gain according to the following formula: ,in, To adjust the gain correction weight, The preset convergence factor, This represents the absolute value of the numerical deviation. The standard deviation of a specific parameter in the historical state register sequence. This is the preset minimum value.

[0045] Preferably, after step S106, the method further includes: real-time acquisition of the steady-state error value of the controlled object after adjustment; calculation of the coefficient of variation of the steady-state error value within a preset sliding window; and locking the current adjustment gain as the steady-state control parameter for the production batch when the coefficient of variation is less than 5%.

[0046] Preferably, the method further includes: receiving a process correction matrix from a remote platform, wherein the process correction matrix contains thermo-mechanical coupling characteristic values ​​under different raw material moisture contents; and injecting the process correction matrix as an initial weight vector into the state observer to preset the nonlinear mapping relationship between numerical deviation and adjustment gain.

[0047] Preferably, in step S101, the sampling frequency of the raw material flow data is synchronized with the sampling frequency of the load current signal, and the sampling accuracy of the load current signal is 12 bits to 16 bits.

[0048] Preferably, step S104 includes: retrieving the data timestamp integrity in the historical state register sequence; when data is missing, performing cubic spline interpolation on the historical state register sequence to reconstruct the missing data nodes and ensure the physical continuity of historical process parameters for a specific frame.

[0049] Preferably, in step S105, the specific frame historical process parameters include frying temperature and dwell time. The state observer sets the step size for adjusting the gain by calculating the energy coupling sensitivity between the frying temperature and the load current signal.

[0050] Example 1: In a large-scale rapeseed oil refining production line with a daily processing capacity of 300t, the initial moisture and physical properties of the raw rapeseed fluctuate dynamically, and the rheological properties of the material undergo high-frequency changes after roasting. This causes the viscous resistance and slip ratio of the material during pipeline transportation to deviate from the predetermined state. When the physical movement time of the material from the roasting unit to the pressing unit exhibits random nonlinear variation characteristics, the error in the transportation time delay parameter causes a phase deviation between the feedback reference point of the control system and the material state in the physical causal dimension. If a static constant is used to set the time delay, the controller cannot accurately correlate the current pressing load feedback with the corresponding historical roasting state. The feedback loop will overshoot or periodic oscillation when dealing with material disturbances. For the above operating conditions, the control system collects the raw material flow data of the controlled object. and the load current signal of the press motor and raw material flow data With load current signal Stored in the historical state register sequence according to a unified time base; where For raw material flow data, The load current signal is extracted. The power spectrum amplitude within the 10Hz to 50Hz frequency range was identified as a characteristic quantity. Characteristic quantity extraction was based on the fluid elastohydrodynamic lubrication and torque transmission mechanism. Fluctuations in the material's sliding resistance at the pipe wall boundary layer generate a reverse pulsating torque through the press's screw shaft, which is ultimately superimposed on the electromagnetic torque demand of the drive motor. Background interference from the press's own mechanical meshing and bearing operation was eliminated. The system data acquisition environment was set to the press shaft operating at its steady-state rated speed. The control station pre-collected the no-load reference current spectrum corresponding to the rated speed under the initial state of pipeline emptying and press no-load operation. In the actual operating sequence, the processor extracted the current load current signal. After calculating the real-time power spectrum amplitude in the 10Hz to 50Hz range, the amplitude component of the no-load reference current spectrum at the corresponding frequency node is simultaneously subtracted to obtain the absolute high-frequency residual that only characterizes the fluctuation of pure material rheological resistance, which is then determined as a characteristic quantity. The offset between the characteristic quantity and the preset damping reference value is calculated, and the offset is determined as a dynamic correction coefficient reflecting the material slip state in the pipeline. The boundary parameters of the physical volume model of the conveying pipeline are adjusted using the dynamic correction coefficient, and the raw material flow data are analyzed under the adjusted physical volume model. The volume integral is performed over time until the integral result reaches the equivalent volume of the delivery pipeline. The volume integral operation follows the law of conservation of mass in fluid dynamics; the total volume of material flowing into the closed pipeline within a specific time period must equal the sum of the infinitesimal volumes of the fluid elements containing the material inside the pipeline. The processor uses the current sampling time... The upper limit of the integration calculation time is set using the historical data sequence with the sliding back time node as the lower limit for raw material flow data. Continue performing integral calculations until the cumulative integral value equals the compensated equivalent pipeline volume. To eliminate the drift in actual physical volume caused by coking and wear on the inner wall of the delivery pipeline after long-term service, the system has a built-in 72-hour adaptive baseline update procedure. During the maintenance and emptying cycle, the system commands the supply pump to inject a calibrated water-based fluid into the pipeline at a constant flow rate. It records the actual filling time from the opening of the inlet valve to the detection of a liquid level jump by the pressure sensor at the outlet. The processor recalculates and overwrites the original calibrated volume value based on the product of the actual filling time and the constant flow rate. To compensate for long-term aging errors at physical boundaries, the corresponding integral time is determined as the dynamic time delay parameter of the material passing through the conveying pipeline.

[0051] Starting from the current sampling time, based on the dynamic delay parameter Reverse addressing in the historical state register sequence to extract the current load current signal. Specific frame historical process parameters that exhibit causal correspondence, including the frying temperature at a specific historical moment. Calculate the historical process parameters and load current signal for a specific frame. The numerical deviation between them is calculated based on the principle of multiphysics energy equivalent mapping. The processor pre-reads the upper limit of the calibrated heating temperature of the roasting unit and the upper limit of the calibrated load current of the pressing motor as the mapping reference, and uses the historical roasting temperature. Divide by the calibrated upper limit of heating temperature to generate a dimensionless relative temperature parameter; simultaneously use the current load current signal. Dividing by the upper limit of the calibrated load current to generate a dimensionless relative current parameter, the processor subtracts the relative temperature parameter from the relative current parameter to generate a numerical deviation. The physical mechanism of this numerical deviation lies in the fact that during continuous oil processing, the material absorbs heat energy, causing the macromolecular chain segments to expand. This results in a monotonically negative correlation between the apparent viscosity and the rheological resistance of the final pressing in the overall transport process and the heat absorbed in the earlier stages. By performing dimensionless division on both, the system essentially maps the overall thermodynamic input intensity and the surface kinetic output load to a normalized energy conversion barrier domain. When the system is in an ideal thermomechanical equilibrium state without external disturbances, an increase in dimensionless relative temperature will inevitably be accompanied by a proportional and equal decrease in dimensionless relative current. The linear difference between the two in this dimensional domain always tends to approach the static zero point. When a sudden change in water content triggers nonlinear additional absorption of latent heat, the heat conversion path to viscosity reduction is blocked. At this time, the transient numerical peak generated by subtracting relative temperature from relative current is the net disturbance energy leakage that is not statically absorbed by the physical volume model. This physical difference, as an error signal, directly characterizes the thermomechanical mismatch of the physical system. The processor extracts the preceding information within the processing path. The ratio of the relative current parameter change gradient to the relative temperature parameter change gradient within a sliding window of data from each sampling period is calculated. This ratio is determined as the energy coupling sensitivity, which characterizes the direct driving strength of a unit dimensionless thermal energy input on the change in end rheological drag. The state observer uses the energy coupling sensitivity as a weighting coefficient to multiply into the basic control step size matrix to generate the dynamic adjustment step size for the current adjustment period. The state observer captures the rate of change of numerical deviation, and iteratively corrects the adjustment gain of the adaptive control law based on the rate of change. According to the adjusted gain The heating power of the frying unit is adjusted to complete the closed-loop control of the oil refining process. During this process, the dynamic correction coefficient, as a rheological damping characteristic, compensates for the equivalent physical volume boundary of the pipeline, enabling the flow integral addressing process to adapt to changes in material moisture and viscosity. This ensures the accuracy of historical parameters extracted from the historical state register sequence in terms of physical causality, eliminating control law misalignment interference caused by material spatial transport lag. In production operation, the system monitors the pressure difference signal across the fine filter and calculates the rising slope of the pressure difference signal. When the rising slope continuously exceeds a preset threshold, the gain is adjusted... The heating power of the stir-frying unit is reduced by superimposing a subtraction compensation weight. At the same time, the steady-state error value of the controlled object after adjustment is collected, and the coefficient of variation of the steady-state error value within a preset sliding window is calculated. When the coefficient of variation is less than 5%, the current adjustment gain is locked. As a steady-state control parameter for this production batch, this adjustment method ensures that the acid value of the product remains stable below 0.24 between batches, and the oleic acid retention rate is consistently maintained at 78.6%, while the fluctuation of the press motor current is controlled within a certain range. Within 3%.

[0052] Example 2: On a rapeseed oil refining production line verification platform with a daily processing capacity of 300t, the control system connects to the raw material flow sensor and the pressing motor current transformer via real-time industrial Ethernet. This platform is used to verify the control stability of the method of the present invention when dealing with step disturbances in raw material moisture content. The flow sensor has a 1% measurement accuracy, and the sampling frequency of the current transformer is set to 100Hz. The sampling period is set to 100ms. The technical consideration for determining this sampling period is to balance the wavefront capture accuracy when the material in the conveying pipeline moves at 1.5m / s with the processing load of the control station. Since the effective frequency bandwidth of the monitored signal is limited to 50Hz, to meet the Nyquist sampling criterion and reserve anti-aliasing filtering space, the sampling period is selected as one-tenth of the signal period, i.e., 100ms. To simulate electromagnetic interference in the industrial environment and verify the environmental current signal to the load, [further details are needed]. The experiment uses actively superimposed Gaussian white noise with a signal-to-noise ratio of 20dB. The experimental design includes the sample group of this invention, control group A, and control group B. Control group A uses a fixed time delay setting, while control group B uses only raw material flow data. The volume integral determines the time delay without introducing a dynamic correction coefficient; under the fluctuating condition where the raw material moisture content jumps from 8.5% to 12.4%, the increase in moisture content reduces the material viscosity and increases the pipe wall slip ratio, causing a nonlinear change in the flow resistance within the conveying pipeline. Control group A, using a fixed time delay parameter of 320s, addresses the historical roasting temperature... With current load current signal A 45-second misalignment on the physical time axis caused the feedback loop to generate a periodic oscillation with an amplitude of 15.6%. Although control group B dynamically adjusted the time delay through volume integration, the calculated time delay deviation still reached 12.8 seconds due to the lack of rheological compensation, and the standard deviation of the control deviation fluctuation was 4.12. The sample group of this invention used step S102 to extract the load current signal. The power spectrum amplitude within the 10Hz to 50Hz range is determined as a characteristic quantity. The offset of this characteristic quantity from the preset damping reference value is calculated to obtain a dynamic correction coefficient. This coefficient is reduced from 1.0 to 0.88, thus adjusting the dynamic time delay parameter. The time was automatically corrected from 320s to 282.4s.

[0053] By comparing the above data, the steady-state error of the sample group in this invention remained within 2.37% during the moisture fluctuation process, and the rate of change of numerical deviation captured by the state observer remained within 0.05. This proves that the frequency domain characteristics from 10Hz to 50Hz and the dynamic time delay parameters are consistent. There is a definite physical correspondence between them, and the adaptive correction mechanism ensures causal consistency between the extracted historical process parameters and the current production status. To ensure the deterministic convergence of the algorithm's underlying layers, the core boundary parameters of the system's internal operation are all based on the prior calibration of the production line's physical characteristics. The calibration boundary of the preset damping benchmark value originates from the average frequency domain value of the motor power noise floor recorded for 30 consecutive minutes when the entire pipeline was flushed with deionized water at a design flow rate of 1.0 m / s during the initial stage of production. The convergence factor in the adjustment formula is physically constrained to a range of 0.05 to 0.15, which is based on the maximum value of the steam valve used for roasting. The upper limit of the anti-saturation gain is derived by reverse derivation of the large controllable opening temperature rise rate; the standard deviation of the specific parameter in the formula is actually obtained by the distributed control system through periodic extraction of the temperature deviation sequence from the past 24 hours of normal production history database and dynamic sliding calculation of its statistical variance, so as to improve the stability of the algorithm against long-term mechanical wear drift; while the preset minimum value is fixed in the microprocessor memory as a constant of the order of ten to the power of negative four, eliminating the risk of interruption and crash caused by the denominator approaching zero when the underlying floating-point number executes the division assembly instruction; when the raw material moisture continues to increase and exceeds the critical point of 13.5%, the load current signal When the trend of change approaches saturation, the gain should be adjusted. The correction amount converged to the limiting range. The system stabilized the acid value of the product between batches below 0.24 and the oleic acid retention rate at 78.6% by adjusting the heating power of the roasting unit. The fluctuation of the press motor current was controlled within ±3%. After the noisy original signal without processing by the method of this invention was corrected by the adaptive control law, the standard deviation of the power output curve fluctuation decreased from 8.42 to 1.15, which confirms the technical effect of the method in eliminating transport hysteresis interference by compensating for the equivalent volume boundary of the pipeline under rheological characteristic changes. For the load current signal The 10Hz to 50Hz frequency range is used as the feature extraction logic. This range is determined based on the superposition characteristics of the mechanical resonance frequency of the press drive system and the pulsating frequency of the material flow. Power spectral density analysis identifies that the components below 10Hz are mainly composed of power frequency interference caused by power grid fluctuations, while the components above 50Hz belong to the higher-order harmonics of the motor and the noise floor of the sensor, and do not directly carry physical characteristics reflecting the rheological damping of the material. By setting this specific bandpass filter window, the injection of noise at non-target frequencies can be suppressed, ensuring that the dynamic correction coefficient is only driven by the pipeline slip state, and improving the physical volume model's assessment of the dynamic time delay parameters of the material passing through the conveying pipeline. The accuracy of the settlement.

[0054] Example 3: When the oil refining production line is transitioning from rapeseed to sunflower seeds as raw materials, the apparent viscosity and pipe wall slip coefficient of the roasted sunflower seeds differ from those of rapeseed. This causes a jump in the rheological damping characteristics within the conveying pipeline, leading to a mismatch error at the physical volume model boundary. If the system cannot calculate the load current signal... The mapping logic between the characteristic offset and the dynamic correction coefficient causes the state observer to diverge during switching transients because the rate of change of the captured numerical deviation exceeds the convergence region, thus affecting the adjustment gain. The correction direction deviated, inducing temperature oscillations in the frying unit; to determine the characteristic quantity conversion logic, this implementation uses a linear proportional operator. The offset of the feature quantity is converted into a dynamic correction coefficient, where the linear scaling operator... Based on the regression slope determined by pre-conducted rheological calibration experiments on the pipeline, the control system extracts the load current signal during material switching transients. The characteristic values ​​in the 10Hz to 50Hz range are calculated, and the offset from the reference value is determined. The linear scaling operator is then invoked. The dynamic correction coefficient is calculated, and this coefficient is used to reconstruct the equivalent volume boundary of the physical volume model, thus adjusting the calculated dynamic delay parameters. It can reflect the physical movement time of sunflower seed material in the pipeline.

[0055] Regarding the internal implementation path of the state observer, the system adopts a second-order discretized state-space model that includes a feedback gain matrix to calculate the historical process parameters and the current load current signal for a specific frame. The rate of change of numerical deviation, wherein the feedback gain matrix is ​​predetermined by the pole placement method, so that the observation residual converges to a steady-state error tolerance of 1% within 3 sampling periods. When the observation residual enters the preset steady-state range, the system will adjust the gain according to the physicochemical indicators after material switching. Adjust to 0.38; simultaneously acquire the pressure signal at the inlet of the fine filter. According to the pressure signal The ratio of the upward slope to the preset pressure warning value is used to calculate the subtraction compensation weight, and then added to the adjustment gain. In the correction loop, response hysteresis caused by abrupt changes in filter resistance is suppressed, where the pressure signal... The measurement resolution is 0.01 MPa. Through the parameter mapping logic and observer model reconstruction for the rheological property change conditions, the system can achieve physical causal tracing of the transport delay within 120 seconds after material switching. The prediction deviation of the state observer is maintained within 0.02, and the standard deviation of temperature fluctuation at the outlet of the frying unit is reduced to within 0.8℃. This realizes the control of the oil refining process under complex physical property change environment and confirms the synergistic adjustment effect of dynamic correction coefficient and state observer under deterministic mapping constraints.

[0056] Example 4: When the oil refining control system starts the production preparation process, due to the physical influence of the physical topology of the conveying pipeline on the material flow resistance, the system starts the calibration program of the preset damping reference value. By inputting the calibration material with a gradient viscosity in the pipeline, the load current signal of the pressing motor is extracted. The power spectral density distribution is analyzed and energy components in the 10Hz to 50Hz range are identified. The measured value of the material arrival time at the pipeline outlet is compared with the integral value of the flow rate. The phase deviation at different flow rates is calculated and the phase deviation is determined as the correction vector of the equivalent physical volume. The correspondence between the material rheological properties and the preset damping reference value is established in the register sequence of the distributed control system.

[0057] When the oil refining production line switches raw material types, the system calls the corresponding preset damping reference value in the register sequence as the initial boundary of the control law, synchronously collects the sampling sequence at the beginning of production, and calculates the load current signal. The average power spectrum amplitude is compared with the deviation between the current sampled value and the preset damping benchmark value. When the fluctuation rate of this deviation within the preset time window is less than 1%, the algorithm calls the linear proportional operator. The equivalent volume parameters of the physical volume model are refreshed, enabling the control model to sense changes in the physical state of the conveying pipeline. The heating power adjustment command of the roasting unit completes phase compensation before the material enters the pressing unit.

[0058] Example 5: In the deployment scenario of a newly commissioned oil refining production line, deviations in the physical dimensions of the conveying pipeline cause residuals in the physical volume model. The system initiates a standardized calibration procedure to determine the linear proportional operator. The feedback gain matrix of the state observer is used to introduce a calibration material with known viscosity into the pipeline while maintaining the raw material flow rate data. As a constant value, the physical time of material passing through the pipeline is recorded and the phase difference between this time and the integral time based on the pipeline design volume is calculated. Load current signals are obtained under different temperature gradients. For multiple sets of energy spectrum characteristic quantities in the frequency range of 10Hz to 50Hz, a first-order linear regression operation is performed between the offset values ​​of the characteristic quantities and the corresponding phase differences. The resulting regression slope is determined as the corresponding linear proportional operator for this production line environment. In the specific implementation of the first-order linear regression calculation, the calibration program drives the control station to lock the cooking unit at three characteristic temperature nodes of 60℃, 80℃, and 100℃ respectively. After the system reaches thermal steady state, the average energy spectrum characteristic quantity of the current working condition is collected and the reference amplitude of the no-load motor current under the pipeline emptying state is subtracted to obtain the pure characteristic offset value set. Simultaneously, the physical time phase difference value captured by the actual measurement under the corresponding working condition is divided by the standard integral time calculated by the pipeline theoretical design volume to obtain the dimensionless volume drift ratio set. The control station uses the least squares algorithm module, with the characteristic offset value set as the independent variable vector and the volume drift ratio set as the dependent variable vector, and inputs it into the standard linear fitting equation for an analytical solution. The unique coefficient obtained is the engineering calibration value of the linear proportional operator, thereby establishing the direct algebraic mapping between the motor electromagnetic frequency band information and the pipeline physical space volume shrinkage rate.

[0059] After the calibration procedure completes the system switch to online operation mode, the control station retrieves the preset damping reference value from the register sequence based on the material type of the production batch. Using the pole placement method, the poles of the feedback gain matrix of the state observer are set in the left half of the complex plane to ensure that the historical process parameters and load current signals for a specific frame are consistent. The rate of change of numerical deviation entered the steady-state range within 3 sampling periods. During the processing of sunflower seed oil, the system extracts energy characteristics from 10Hz to 50Hz in real time and processes them using a linear proportional operator. The conversion yields dynamic correction coefficients, which are then used to perform dynamic compensation on the physical volume model boundaries, thus adjusting the calculated dynamic delay parameters. The adjustment gain generated under this constraint matches the physical flow rate of the material in the pipeline. The heating power of the driving roasting unit is adjusted to ensure that the product acid value is kept below 0.24 and the oleic acid retention rate is kept at 78.6% during the production line operation.

[0060] In the closed-loop steady-state determination procedure of the oil refining process, the coefficient of variation within the preset sliding window is set at 5%. The technical consideration for determining this value lies in balancing the convergence speed of the control process with the stability of the product's acid value. By monitoring the steady-state error under different coefficient of variation thresholds online, 5% is determined as the critical operating point for trigger parameter locking. Within this boundary, it can be ensured that the system quickly returns to the steady-state range after responding to material disturbances. Simultaneously, in the setting process of the feedback gain matrix of the state observer, the current step response curves of the press under both no-load and full-load extreme conditions are collected offline. The time constant and damping ratio reflecting the system's inertial components are extracted, and these physical properties are mapped to the system matrix in the state space. The diagonalization mapping method is used to decouple the components of the feedback gain matrix from the eigenvalues ​​of the system matrix, allowing the observer's poles to be equivalently shifted in the left half of the complex plane according to the response speed requirements. This ensures convergence of the observation residuals while avoiding high-frequency characteristic amplification caused by excessive gain, ensuring that the rate of change of numerical deviation can drive the adjustment gain of the adaptive control law. Perform parameter correction.

[0061] Specifically, the state vector of the above-mentioned second-order discretized state-space model is defined as a column vector containing the dimensionless temperature deviation component of the historical process parameters of a specific frame and the dimensionless current deviation component of the current load current signal. The state transition matrix A in its system state equation is essentially a diagonal matrix, representing the natural coupling attenuation rate of the thermal inertia of the frying unit and the pipeline transport delay over time. The input matrix B represents the mapping gain of the heating power increment on the temperature state variable. The output matrix C maps the state vector to the comprehensive physical damping characteristic quantity that can be collected at the pressing motor end. During the system initialization phase, the processor extracts the steady-state step transfer function under the reference operating condition offline and performs bilinear transformation discretization to obtain physical matrices A, B, and C with fixed dimensions and constant elements. Then, it solves the discrete algebraic Riccati equation within the preset complex plane convergence region and derives and calculates the corresponding constant state feedback gain matrix. Thus, the originally functional pole placement logic is implemented as a matrix multiplication and addition algebraic iteration process that can be directly executed by the processor.

[0062] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. An adaptive adjustment method for an oil refining process, characterized in that, Includes the following steps: Step S101: Obtain the raw material flow data of the controlled object and the load current signal of the pressing motor, and write the raw material flow data and load current signal into the historical state register sequence according to a unified time base. Step S102: Extract the power spectrum amplitude in the frequency range of 10Hz to 50Hz from the load current signal and determine it as a characteristic quantity. Calculate the offset between the characteristic quantity and the preset damping reference value, and determine the offset as a dynamic correction coefficient reflecting the slip state of the pipeline material. Step S103: Adjust the boundary parameters of the physical volume model of the conveying pipeline using dynamic correction coefficients, and perform volume integration calculation on the time axis of the raw material flow data under the adjusted physical volume model until the integration result reaches the equivalent volume of the conveying pipeline. Then, determine the corresponding integration time as the dynamic time delay parameter of the material passing through the conveying pipeline. Step S104: Starting from the current sampling time, reverse addressing is performed in the historical state register sequence according to the dynamic delay parameter to extract specific frame historical process parameters that have a causal correspondence with the current load current signal. Step S105: Calculate the numerical deviation between the historical process parameters of a specific frame and the load current signal, use the state observer to capture the rate of change of the numerical deviation, and iteratively correct the adjustment gain of the adaptive control law based on the rate of change. Step S106: Adjust the heating power of the frying unit according to the modified adjustment gain to complete the closed-loop control of the oil refining process.

2. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, Step S102 further includes the following sub-steps: Step S1021, perform a fast Fourier transform on the load current signal to obtain a power density distribution map; Step S1022, identify the energy peaks in the power density distribution map that belong to the frequency range, and define the average amplitude of the energy peaks as a characteristic quantity to characterize the rheological resistance fluctuations of the material during the pressing process.

3. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, Step S103 further includes the following sub-steps: Step S1031, obtaining the calibration volume value of the physical volume model. Step S1032: calibrate the volume value. The equivalent pipeline volume after compensation is obtained by multiplying it with the dynamic correction factor. Step S1033: According to the current time... The order of recursion to historical moments, for raw material flow data Perform integration to determine if the condition is met. Duration This is a dynamic delay parameter.

4. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, Also includes: Real-time monitoring of the pressure difference signal across the fine filter and calculation of the rising slope of the pressure difference signal; When the upward slope continues to exceed the preset threshold, a subtraction compensation weight is superimposed on the adjustment gain to reduce the heating power of the frying unit.

5. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, In step S105, the state observer determines the correction weights for the adjustment gain according to the following formula: ,in, To adjust the correction weights for the gain, The preset convergence factor, This represents the absolute value of the numerical deviation. The standard deviation of a specific parameter in the historical state register sequence. This is the preset minimum value.

6. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, The process after step S106 further includes: real-time acquisition of the steady-state error value of the controlled object after adjustment; calculation of the coefficient of variation of the steady-state error value within a preset sliding window; and locking the current adjustment gain as the steady-state control parameter for the production batch when the coefficient of variation is less than 5%.

7. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, Also includes: Receive a process correction matrix from a remote platform, wherein the process correction matrix contains thermo-mechanical coupling characteristic values ​​under different raw material moisture contents; The process correction matrix is ​​injected into the state observer as the initial weight vector to pre-determine the nonlinear mapping relationship between the numerical deviation and the adjustment gain.

8. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, In step S101, the sampling frequency of the raw material flow data is synchronized with the sampling frequency of the load current signal, and the sampling accuracy of the load current signal is 12 bits to 16 bits.

9. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, Step S104 includes: retrieving the integrity of data timestamps in the historical state register sequence; when data is missing, performing cubic spline interpolation on the historical state register sequence to reconstruct the missing data nodes and ensure the physical continuity of historical process parameters for a specific frame.

10. The adaptive adjustment method for an oil refining process according to claim 1, characterized in that, In step S105, the specific frame history process parameters include the frying temperature and dwell time. The state observer sets the step size for adjusting the gain by calculating the energy coupling sensitivity between the frying temperature and the load current signal.

Images