Mechanical pressing rice bran oil processing key technology

Abstract

The application discloses a key processing technology for mechanical pressing of rice bran oil, and through the construction of a two-dimensional dynamic phase trajectory of the fusion of micro driving information entropy and macro energy efficiency elasticity index, and the combination of multi-rate time alignment and nonlinear topological boundary discrimination mechanism, false alarms caused by complex material fluctuations can be removed from the root, and invalid oscillation of an executing mechanism can be effectively avoided; the early micro abnormality awareness capability before macro temperature rise can be provided, adaptive linkage interception can be implemented as soon as possible before destructive thermal degradation accumulates to a critical point; the control robustness of the pressing system in a complex environment is enhanced, the oil extraction efficiency and the final oil quality are effectively considered, and a novel and practical technical path is provided for the high-reliability continuous production of the rice bran oil pressing process.

Description

Technical Field

[0001] This invention relates to the field of mechanically pressed rice bran oil technology, specifically a key process for mechanically pressing rice bran oil processing. Background Technology

[0002] With the continuous evolution of edible oil processing technology, mechanical pressing has gradually become an important process for extracting high-quality rice bran oil due to its ability to preserve the natural active components of oil at lower temperatures. In continuous, full-load industrial rice bran pressing, the fluid dynamics and tribological state inside the pressing chamber are constantly undergoing highly nonlinear dynamic evolution due to the complexity of raw material sources and the multi-source heterogeneity of physicochemical characteristics such as moisture, particle size, and age. To ensure the stability of continuous production, industrial control systems have generally introduced automated monitoring methods to maintain a balance between the work done in pressing and pushing oil by adjusting equipment operating parameters. In recent years, the control concept of pressing processes has gradually evolved from a single macroscopic empirical setting to dynamic adaptive adjustment based on sensor feedback signals, aiming to further improve the oil extraction yield and minimize unintended mechanical or thermodynamic damage.

[0003] While existing monitoring solutions have promoted process automation to some extent, they still face the following technical limitations when dealing with complex and ever-changing real-world pressing conditions: While microscopic monitoring strategies based on high-frequency local signals offer agile responses, they are prone to misinterpreting instantaneous interference from normal large-particle materials as abnormal dry friction. This singularity in identification dimensions leads to frequent false alarms, resulting in high-frequency ineffective oscillations in the actuators and making it difficult to maintain a stable working state.

[0004] While macroscopic monitoring strategies based on global energy efficiency degradation can objectively reflect the actual load on equipment, their data representation suffers from time lag limitations. By the time macroscopic congestion is detected and adjustment commands are issued, the nutrient degradation caused by prolonged overheating in the pressing chamber is usually irreversible. This mismatch in spatiotemporal scale between microscopic false alarms and macroscopic lag makes it difficult to effectively balance quality preservation and congestion prevention.

[0005] Therefore, the present invention provides a key process for mechanically pressing rice bran oil. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a key process for mechanically pressing rice bran oil. By constructing a two-dimensional dynamic phase trajectory that integrates microscopic driving information entropy and macroscopic energy efficiency elasticity index, and combining multi-rate time alignment and nonlinear topological boundary discrimination mechanisms, it fundamentally eliminates false alarms caused by complex material fluctuations and effectively avoids ineffective oscillations in the actuator. It possesses the ability to detect microscopic anomalies at an extremely early stage, preceding macroscopic temperature rise, and can immediately implement adaptive linkage interception before destructive thermal degradation accumulates to a critical point. It enhances the control robustness of the pressing system in complex environments, effectively balancing oil extraction efficiency and final oil quality, thereby solving the technical problems described in the background art.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: A key process for mechanically pressing rice bran oil includes the following steps: S1: Obtain the remote sensing dataset of the pressing equipment after de-identification processing, extract the phased energy supply time-series waveform corresponding to the power input port, perform time-frequency domain transformation and dissipation calculation on the phased energy supply time-series waveform, and generate the first data to characterize the physical friction state of the pressing micro-interface. The first data is the transient information entropy of the driving power spectrum. S2: Extract global input energy load parameters and effective conversion work factor from the remote sensing dataset of the pressing equipment. Perform correlation evolution analysis on global input energy load parameters and effective conversion work factor within a dynamic time window. Calculate and generate second data to characterize the state of macroscopic extrusion work efficiency decay. The second data is the dynamic elasticity index of energy efficiency output. The sampling frequency for obtaining the second data is lower than the sampling frequency for obtaining the first data. S3: Call the pre-configured multi-rate smooth interpolation alignment algorithm to map the first data and the second data to the same time series anchor point. Use the first data as the horizontal axis feature quantity and the second data as the vertical axis feature quantity to construct a two-dimensional phase trajectory vector representing the nonlinear dynamic evolution path. S4: Obtain the preset nonlinear topological envelope surface function, substitute the two-dimensional phase trajectory vector into the nonlinear topological envelope surface function to run the topological space crossing degree discrimination, and generate a confirmation indication signal to characterize the cross-scale structural congestion risk when it is determined that the vector tip of the two-dimensional phase trajectory vector breaks through the spatial boundary defined by the nonlinear topological envelope surface function. S5: Based on the received confirmation indication signal, synchronously output the first control command and the second control command, which respectively expand the geometry of the end resistor cross section and lower the current working drive frequency operating baseline.

[0008] (III) Beneficial Effects This invention provides a key process for mechanically pressing rice bran oil, which has the following beneficial effects: This invention constructs a two-dimensional mapping model between microscopic frictional state and macroscopic extrusion energy efficiency to achieve precise topological decoupling between non-destructive anomalies and systemic collapse risks. It obtains feature representations at different time scales. On the one hand, it extracts the transient information entropy of the driving power spectrum to quantitatively characterize the frictional disorder of the microscopic interface without delay. On the other hand, it extracts the dynamic elasticity index of global energy efficiency output to quantitatively characterize the proportion of macroscopic fatigue degradation caused by the conversion of additional consumed electrical energy into destructive thermal energy. This invention introduces a multi-rate smooth interpolation alignment algorithm to deeply fuse high-frequency microscopic data and low-frequency macroscopic data at a unified time series anchor point to construct a two-dimensional phase trajectory vector characterizing the nonlinear dynamic evolution path. This invention reconstructs the hyperbolic open topological boundary defined by the upper limit of microscopic dissipation and the lower limit of macroscopic energy efficiency online, and continuously discriminates the spatial crossing degree of the phase trajectory vector. The high-frequency pseudo-anomalies caused by normal turbulence of the filtered material are only confirmed by generating a confirmation signal when the microscopic high entropy and macroscopic low elasticity simultaneously exceed the safety boundary. This integrates the originally isolated mechanical resistance gap adjustment and electrical drive frequency adjustment into a closed-loop control, realizing the transformation from single information decision-making to cross-scale spatiotemporal collaborative decision-making and giving full play to the synergistic gain efficiency. This invention constructs a two-dimensional dynamic phase trajectory that integrates microscopic driving information entropy and macroscopic energy efficiency elasticity index, and combines multi-rate time alignment and nonlinear topological boundary discrimination mechanism. This enables the elimination of false alarms caused by complex material fluctuations at the source, effectively avoiding ineffective oscillations of the actuator. It possesses the ability to detect microscopic anomalies at an extremely early stage, preceding macroscopic temperature rise, and can immediately implement adaptive linkage interception before destructive thermal degradation accumulates to the critical point. It enhances the control robustness of the pressing system in complex environments, effectively balancing oil extraction efficiency and final oil quality, and provides a novel and practical technical path for the highly reliable continuous production of rice bran oil pressing. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the key process steps in the mechanical pressing of rice bran oil according to the present invention; Figure 2 This is a schematic diagram illustrating the verification indicator signal calculation steps of a key process for mechanically pressing rice bran oil processing according to the present invention. Figure 3 This is a schematic diagram illustrating the generation steps of the first and second control commands in a key process for mechanically pressing rice bran oil according to the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] Please see Figure 1-3 This invention provides a key process for mechanically pressing rice bran oil, comprising the following steps: S1: Obtain the remote sensing dataset of the pressing equipment after de-identification processing, extract the phased energy supply time-series waveform corresponding to the power input port, perform time-frequency domain transformation and dissipation calculation on the phased energy supply time-series waveform, and generate the first data to characterize the physical friction state of the pressing micro-interface. The first data is the transient information entropy of the driving power spectrum. S2: Extract global input energy load parameters and effective conversion work factor from the remote sensing dataset of the pressing equipment. Perform correlation evolution analysis on global input energy load parameters and effective conversion work factor within a dynamic time window. Calculate and generate second data to characterize the state of macroscopic extrusion work efficiency decay. The second data is the dynamic elasticity index of energy efficiency output. The sampling frequency for obtaining the second data is lower than the sampling frequency for obtaining the first data. S3: Call the pre-configured multi-rate smooth interpolation alignment algorithm to map the first data and the second data to the same time series anchor point. Use the first data as the horizontal axis feature quantity and the second data as the vertical axis feature quantity to construct a two-dimensional phase trajectory vector representing the nonlinear dynamic evolution path. S4: Obtain the preset nonlinear topological envelope surface function, substitute the two-dimensional phase trajectory vector into the nonlinear topological envelope surface function to run the topological space crossing degree discrimination, and generate a confirmation indication signal to characterize the cross-scale structural congestion risk when it is determined that the vector tip of the two-dimensional phase trajectory vector breaks through the spatial boundary defined by the nonlinear topological envelope surface function. S5: Based on the received confirmation indication signal, synchronously output the first control command and the second control command: send the first control command to the displacement execution port to expand the geometry of the end resistance section, and send the second control command to the frequency conversion drive port to lower the current working drive frequency operating baseline; during the execution of the first control command and the second control command, continuously acquire the first data and the second data of the next sampling period to form a feedback closed-loop control loop that drives the two-dimensional phase trajectory vector back to the spatial boundary.

[0012] The steps for generating the first data in S1 are as follows: Adaptive empirical mode decomposition is performed on the phase energy supply time-series waveform to remove the fundamental frequency steady-state interference and extract the residual high-frequency energy components; the residual high-frequency energy components are intercepted within the set sliding time window, a short-time Fourier transform is performed to construct the reconstructed phase space, and the disorder distribution value in the corresponding high-frequency band is calculated according to the Shannon information entropy formula. The obtained dimensionless scalar is used as the transient information entropy of the driving power spectrum. By using adaptive empirical mode decomposition, short-time Fourier transform, and Shannon information entropy formula, and correspondingly acquiring the stator current output of the frequency converter in hardware, multi-dimensional decomposition calculations are performed at a sampling frequency of 1kHz. This achieves the removal of the low-frequency envelope effect brought by the 50Hz / 60Hz fundamental frequency of the power grid, and quantitatively amplifies the extremely small instantaneous force disorder inside the pressing chamber during the transition from fluid lubrication to dry friction, providing the ability to detect microscopic anomalies hundreds of milliseconds earlier than the macroscopic temperature rise.

[0013] The steps for generating the second data in S2 are as follows: extract the first relative rate of change of the global input energy load parameters within the current dynamic time window, and the second relative rate of change of the effective conversion work factor obtained based on the internal fluid dynamics ratio; input the first relative rate of change and the second relative rate of change into a preset recursive least squares algorithm model with a forgetting factor, identify the covariance and variance ratio of the first relative rate of change and the second relative rate of change online, and use the identified ratio as the dynamic elasticity index of energy efficiency output; Data is acquired through the underlying energy meter bus (1Hz low-frequency sampling), and the covariance and variance ratio of the first relative change rate and the second relative change rate are calculated using the recursive least squares method with a forgetting factor. By introducing the idea of ​​diminishing marginal output algorithm, instantaneous material fluctuations are filtered out at the macro level, and the fatigue degradation ratio of the extra consumed electrical energy being converted into destructive frictional heat (rather than oil extraction work) is quantitatively characterized, thus establishing a strict underlying energy efficiency defense line.

[0014] The steps for constructing a two-dimensional phase trajectory vector in S3 are as follows: The first data is input into a preset Kalman smoother model as a local high-frequency observation update variable, and the second data is input into the Kalman smoother model as a prior constraint variable for macroscopic state transition; through the inverse interpolation operation of the Kalman smoother model, the low-frequency sampled second data is reconstructed on the time axis into a high-frequency synchronization sequence matrix with the same timestamp as the first data, and a two-dimensional phase trajectory vector is generated based on the synchronized horizontal and vertical feature quantities; By specifying a high-frequency entropy value of 1kHz as the observation input and a low-frequency elasticity index of 1Hz as the state prior, inverse interpolation is performed, and the abstract concept is concretized in the field of industrial-grade multi-rate digital signal processing. The technical problem of jagged and torn topological trajectories caused by inconsistent sampling rates is completely overcome, ensuring the mathematical continuity and judgment stability of subsequent spatial over-limit derivative calculations.

[0015] The steps for obtaining the preset nonlinear topological envelope surface function in S4 are as follows: read the environmental and material degradation boundary matrix coefficients from the external structured carrier; reconstruct the high-order polynomial online based on the boundary matrix coefficients to form a two-dimensional constraint mapping surface jointly defined by the upper limit of microscopic dissipation and the lower limit of macroscopic energy efficiency; and use the two-dimensional constraint mapping surface as the nonlinear topological envelope surface function. By reading externally configured CSV / Excel matrix parameters (a manifestation of hardware-software decoupling), a polynomial two-dimensional mapping surface is run and constructed in memory; this eliminates the extremely low robustness caused by hardcoding one-dimensional thresholds in the program code; the two-dimensional constrained surface gives it the ability to distinguish between high entropy and high elasticity (normal interference of coarse-particle materials) and high entropy and low elasticity (structural congestion outbreaks), thus eliminating the false alarm oscillations of traditional single-parameter high-frequency algorithms.

[0016] The key parameter definitions, logical meanings, and acquisition methods involved in this embodiment are as follows: The parameter symbol for the transient information entropy of the driving power spectrum is: The algorithm characterizes the degree of disorder in the microscopic physical friction state inside the pressing chamber, used to quantify the energy surge characteristics caused by the instantaneous dry friction at the metal boundary due to the rupture of the normal fluid lubricating oil film; it relies on Hall current sensor data to acquire the three-phase current waveform of the stator side of the output port of the drive spindle frequency converter; the preprocessing algorithm relies on Shannon information theory and classical time-frequency domain signal analysis techniques in information theory; it performs adaptive empirical mode decomposition on the acquired kilohertz-level AC waveform sequence, separating and discarding low-frequency mode components containing the grid fundamental frequency, and screening out the residual high-frequency energy component sequence containing excitation characteristics; it applies short-time Fourier transform to the residual high-frequency energy component sequence, mapping it from the time domain to a two-dimensional reconstructed phase space intertwined with energy and frequency, and performs entropy calculation within a set specific high-frequency observation window; the calculation logic is as follows: acquire the residual high-frequency energy component sequence after empirical mode decomposition; sort the residual high-frequency energy component sequence according to... The system is divided into multiple data frame sequences according to a set short-time sliding window length. Fourier spectrum analysis is performed on each data frame sequence to obtain the power spectral density values ​​at specific high-frequency points within the corresponding data frame. The power spectral density value at a given high-frequency point is divided by the sum of the power spectral density values ​​at all selected high-frequency points within the data frame to obtain a normalized relative energy density parameter characterizing the energy proportion at that specific frequency point. Each normalized relative energy density parameter is multiplied by its natural logarithm to obtain the dissipation component value at each frequency point. The dissipation component values ​​at all frequency points within the data frame are summed. The negative of the summation result, i.e., the positive value, is obtained. To ensure that the calculated result falls within the universally comparable closed interval between zero and one, a maximum theoretical entropy limit parameter of the system, pre-set by experiments, is selected. The positive value obtained is divided by the maximum theoretical entropy limit parameter; the ratio is the final output transient information entropy of the driving power spectrum. .

[0017] The parameter symbol of the system's maximum theoretical entropy limit parameter is: The term "entropy" refers to the highest spectral energy divergence limit that can be achieved inside the pressing chamber under extreme dry friction conditions, and is used as the full-scale denominator for normalization calculations. Under offline no-load conditions, the input of rice bran material and lubricating grease is cut off, and the main shaft inverter is driven to its rated speed, forcibly creating a dry friction physical condition at the metal-to-metal boundary. Under this condition, high-frequency current data on the stator side is continuously collected, and information entropy is calculated according to the aforementioned formula. The peak average of the information entropy calculated over ten consecutive seconds is extracted as the base value of the maximum theoretical entropy limit parameter. In this embodiment, based on the calibration of the dry grinding bench experiment, the preferred value of the maximum theoretical entropy limit parameter is set to 3.85. This ensures that under normal and abnormal oil pressing conditions, the entropy value calculated in real time can be safely mapped within the dimensionless range of zero to one, avoiding numerical overflow.

[0018] The parameter symbol for the dynamic elasticity index of energy efficiency output is:This is used to quantify the decreasing proportion of the actual work done by the effective oil pushing in the pressing chamber when each additional unit of driving electrical energy is consumed in the macro system; it relies on the active power feedback sequence pushed by the smart meter of the industrial system bus, and the theoretical volume displacement estimation interface configured by the main control node according to the screw geometry; it is calculated through the marginal elasticity assessment in macroeconomics and the least squares recursive analysis model in the discipline of system identification; since the conversion of mechanical energy into destructive frictional heat will lead to a decrease in the effective oil output rate, the fatigue level is locked by continuously capturing the correlation between the rate of change of input electrical energy and the rate of change of output converted mechanical work; the calculation logic is as follows: obtain the total active power increment parameters collected in the current monitoring cycle, and at the same time extract the adjacent previous time from the buffer stack. The total active power absolute reference value is collected at each moment; the incremental parameter of total active power is divided by the absolute reference value of total active power to obtain the quotient as the first relative rate of change parameter characterizing the input energy fluctuation characteristics; based on the internal conversion ratio of fluid dynamics, the incremental parameter of effective conversion extrusion propulsion work in the same period is obtained, and the absolute reference value of effective conversion extrusion propulsion work collected at the immediately preceding moment is extracted; the incremental parameter of extrusion propulsion work is divided by the absolute reference value of extrusion propulsion work to obtain the quotient as the second relative rate of change parameter representing the output characteristics; a preset forgetting factor parameter is retrieved, and the state of the covariance matrix containing historical correlation information and the state of the variance scalar are updated synchronously through a recursive formula; the updated second relative rate of change parameter of the current period is compared with the first... The initial identification ratio is obtained by dividing the covariance of the relative rate of change parameter by the variance of the first relative rate of change parameter. Since the initial identification ratio is affected by the heterogeneous physical conversion principle of the underlying sensors during its derivation, residual dimensions exist. Therefore, it is necessary to extract the dimension conversion coefficient parameter pre-fixed in the underlying control node. The extraction source of the dimension conversion coefficient parameter does not rely on real-time calculation, but is pre-calibrated through offline static multiplication based on the electromechanical conversion efficiency constant of the drive motor and the mechanical transmission efficiency constant of the gearbox as stated on the manufacturer's nameplate of the pressing equipment. In this embodiment, based on the actual hardware configuration, the preferred value of the dimension conversion coefficient parameter is rigidly set to 0.88. The initial identification ratio is then compared with the dimension conversion coefficient parameter of 0.88. The conversion coefficient parameter is multiplied; the dimensionless conversion coefficient parameter is used to offset the fixed physical losses in the process of converting electrical energy into mechanical energy, eliminate the differences in physical properties in the calculation of electromechanical power ratio, and thus force heterogeneous physical quantities to be mapped into a unified dimensionless scalar value for subsequent pure numerical calculations; the dimensionless pure value obtained after multiplication is used as the estimated intermediate value of the elasticity coefficient; the control system obtains a preset logistic smoothing mapping model to perform extreme value constraints; the calculation logic of the logistic smoothing mapping model is as follows: obtain the difference between the estimated intermediate value of the elasticity coefficient and the theoretical optimal balance constant 1, and use the difference as the basic input quantity; retrieve the preset nonlinear attenuation gain coefficient (in this embodiment, the nonlinear attenuation gain coefficient is set to -5).0 (aiming to accelerate the convergence slope of both sides of the boundary); multiply the basic input quantity with the nonlinear decay gain coefficient to obtain the exponential term result; use the natural constant as the base and the exponential term result as the exponent to perform a power operation to obtain the natural power result; add the value 1 to the natural power result to obtain the denominator term; divide the value 1 by the denominator term to complete the nonlinear inverse curvature convergence calculation; the calculation logic ensures that no matter how anomalously the basic input quantity changes, the final division output result is always within the range of zero to one in the open interval, and the final output result is assigned as the dynamic elasticity index of energy efficiency output. .

[0019] The parameter symbol of the forgetting factor is: The weight of historical observation data in the current covariance matrix calculation is determined by the exponential decay, thereby giving the recursive least squares algorithm the ability to track time-varying parameters. In this embodiment, the forgetting factor parameter... The value is limited to the range of 0.95 to 0.99, with a preferred value of 0.98. If the value is too small, the algorithm will be overly sensitive to instantaneous material turbulence within a single Hertz sampling period, leading to misjudgment. If the value is too large, the algorithm will fall into data saturation, resulting in a sluggish response to the slow energy efficiency degradation caused by real congestion. The value of 0.98 represents the experimentally optimal balance between identification sensitivity and filtering robustness.

[0020] The parameter set notation for the coefficients of the environmental and material degradation boundary matrix is ​​as follows: The curvature and extreme boundary of the polynomial mapping surface at different two-dimensional coordinate points are defined to provide a physical constraint baseline for risk adjudication in phase space. Instead of relying on real-time sensing data, the coefficients are obtained by reading structured files (CSV format files in this embodiment) stored in the edge node storage medium. The original calibration of the coefficient values ​​comes from accelerated dry friction coking destruction experiments conducted on mixed rice bran with different moisture contents and storage periods on a material thermosensitivity test bench. The mathematical features of the extreme boundary fitting surface are extracted using a multivariate nonlinear regression algorithm. The calculation logic is as follows: The row vector records matching the moisture content identifier of the current batch of rice bran in the external configuration data table are read. From the row vector records, the first constant offset coefficient parameter, the linear term stretching coefficient parameter about the first data on the horizontal axis, the linear term compression coefficient parameter about the second data on the vertical axis, and the cross-coupling attenuation coefficient parameter relating the first and second data are extracted sequentially. All extracted coefficients are arranged and merged into a fixed-length one-dimensional numerical array in a specific order. The one-dimensional numerical array is the environmental and material degradation boundary matrix coefficient used to reconstruct the final judgment surface. .

[0021] The steps for calculating the confirmation indication signal in this embodiment of the invention are as follows: Receives three heterogeneous input elements; the first input is the transient entropy of the drive power spectrum based on the kilohertz-level high-frequency sampling of the frequency converter. The second input is the dynamic elasticity index of energy efficiency output based on single-hertz low-frequency sampling of the system bus. The third input is the environmental and material degradation boundary matrix coefficients read from the non-volatile storage area. ; Transient information entropy of the input driving power spectrum With the dynamic elasticity index of energy efficiency output The clock update rate mismatch exists; a Kalman smoother model is used for data alignment calculation; the Kalman smoother model uses the energy efficiency output dynamic elasticity index of the previous low-frequency cycle at the current moment. The Kalman smoother model continuously receives and buffers the transient information entropy of dozens of high-density driving power spectra arriving within the current high-frequency cycle, using prior historical states as a basis for recursion. As an observation correction quantity; perform inverse state backtracking calculation, that is, use the fluctuation variance disturbance contained in the continuous high-frequency observation data to correct the blank time period between two low-frequency sampling points; fill in the generation and driving power spectrum transient information entropy within the blank time period. Energy efficiency sequence points after interpolation with a consistent number of points; Vector merging is performed at each time alignment point on the millisecond-level clock line; the transient information entropy of the driving power spectrum is then processed. The value is assigned to the horizontal axis component representing microscopic dissipation, and the corresponding interpolated energy efficiency sequence point is assigned to the vertical axis component representing macroscopic energy efficiency; a two-dimensional phase trajectory vector with length and pointing characteristics is generated in the virtual Cartesian coordinate plane in memory for a single timestamp. The loaded environmental and material degradation boundary matrix coefficients Extract; extract the coefficients of the environmental and material degradation boundary matrix. The system includes a first constant offset coefficient parameter, a first linear term stretching coefficient parameter on the horizontal axis, a second linear term compression coefficient parameter on the vertical axis, and a cross-coupling attenuation coefficient parameter. The first linear term stretching coefficient parameter on the horizontal axis is bound to the horizontal axis coordinate variable, and the second linear term compression coefficient parameter on the vertical axis is bound to the vertical axis coordinate variable. The cross-coupling attenuation coefficient parameter is forcibly used as the multiplication gain coefficient of the product term between the horizontal and vertical axis coordinate variables. All parameters and bound terms are algebraically summed to instantiate a continuous bivariate quadratic polynomial hyperbolic open topological boundary in the memory coordinate plane. This hyperbolic open topological boundary divides the coordinate system into a steady-state control quadrant representing safe work and an outer region representing the risk of coking instability. The generated two-dimensional phase trajectory vector containing horizontal and vertical coordinate values ​​is projected onto this space. Obtain the absolute coordinate point value of the two-dimensional phase trajectory vector vector after projection into the coordinate system, and perform spatial crossing degree discrimination; perform judgment calculation, and substitute the microscopic lateral component value and macroscopic longitudinal component value at this time into the instantiated nonlinear bivariate quadratic polynomial to calculate the actual output value of the polynomial equation; The absolute coordinate point value of the two-dimensional phase trajectory vector vector after projection into the coordinate system is obtained, and the spatial crossing degree discrimination calculation is performed. Specifically, the discrimination calculation is performed by substituting the microscopic lateral component value and the macroscopic longitudinal component value into the instantiated nonlinear bivariate quadratic polynomial for algebraic operation, and calculating the actual output result value of the polynomial equation. The preset absolute safety topological potential energy reference constant (in this embodiment, the absolute safety topological potential energy reference constant is 0, representing the zero potential energy contour base of the polynomial surface) is retrieved from the underlying memory of the control node. The actual output value is compared with the absolute safety topological potential energy reference constant. If the actual output value is less than the absolute safety topological potential energy reference constant, it means that although the roughness of the natural material causes the microscopic first data to fluctuate, the two-dimensional phase trajectory vector has not yet broken through the hyperbolic open topological boundary, and the system elastic support of the macroscopic second data is still stable. This state is characterized as normal processing noise. The confirmation indication signal is forcibly not output, and the current judgment value is discarded to wait for the next cycle. If the actual output value is greater than or equal to the absolute safety topological potential energy reference constant; the phase trajectory breaks through the hyperbolic open topological boundary in the lower right corner region representing high entropy and low elasticity, and the microscopic dry friction has substantially weakened the overall propulsion efficiency of the equipment; it is regarded as a real precursor to coking; a confirmation indication signal of cross-scale structural congestion risk is immediately generated and the output stage begins. The final output confirmation indication signal is a trigger marker, which encapsulates both the microscopic deviation amplitude and the macroscopic attenuation rate at this time. Within the instantaneous control cycle of receiving the confirmation indication signal, the subsequent control link will be directly invoked according to the trigger marker, and the first control command will be issued simultaneously to physically expand the mechanical gap of the end cake dispensing mechanism to release the hydrostatic pressure, and the second control command will be issued to electrically reduce the speed and frequency of the drive motor to suppress heat generation, thereby generating an electromechanical cooperative protection gain effect.

[0022] The steps for data alignment calculation in the Kalman smoother model are as follows: Extract the dynamic elasticity index of energy efficiency output from the previous low-frequency cycle at the current moment and inject it into the Kalman smoother model as a priori historical state quantity to start the recursion; continuously acquire the transient information entropy of dozens of driving power spectra arriving at high density within the current high-frequency cycle, perform stack buffering in memory, and use it as the input of the observation correction sequence to the Kalman smoother model; activate the reverse state backtracking operation calculation instruction inside the Kalman smoother model, that is, use the fluctuation variance perturbation features contained in the observation correction sequence to smooth and correct the blank time period between two low-frequency sampling points.

[0023] The process of outputting the first and second control commands in S5 is as follows: Extract two derivative features of the two-dimensional phase trajectory vector, namely the transient deviation scalar representing the deviation of the current state of the first data, and the time derivative scalar representing the decay rate of the second data; after converting the transient deviation scalar through pulse width modulation, it is used as the first control command output; in parallel, the time derivative scalar is input into the gradient optimization algorithm to calculate the bias constraint, and is used as the second control command output.

[0024] Before calling the Kalman smoother model to perform the formal data alignment calculation, the matrix parameters are initialized: the state transition matrix is ​​constructed. Since the macroscopic energy efficiency state is assumed to conform to inertial stationary change in a very short time of milliseconds in this embodiment, the state transition matrix is ​​initialized as a first-order identity matrix; the process noise covariance parameter and the observation noise covariance parameter are configured; the process noise covariance parameter characterizes the solid-nature uncertainty of the macroscopic state transition, and the preferred value is set to 0.01; the observation noise covariance parameter characterizes the variance of microscopic measurement electromagnetic noise brought by the high-frequency current sensor, and the preferred value is set to 0.5; when fusing heterogeneous data, the smoothness of low-frequency but stable macroscopic power output data is trusted more, while the high-frequency fluctuating microscopic entropy value is regarded as a high-frequency disturbance around the stable baseline, thereby providing a mathematical anchor for subsequent reverse state backtracking.

[0025] For hydraulic / mechanical actuators (i.e., for gap adjustment), PWM pulse output (i.e., based on deviation) and gradient descent bias calculation (i.e., based on decay rate) for frequency converters (i.e., for spindle speed regulation); through decomposition and dimensionality reduction, a rigorous causal nesting in the control dimension is achieved—the transient stress field is dismantled by gap fine-tuning, and thermodynamic catastrophe is suppressed by speed degradation, integrating the originally isolated mechanical adjustment and electrical regulation under the trigger of a single signal.

[0026] This embodiment involves the following key parameters, the physical / logical meanings, acquisition methods, and calculation logic of which are as follows: The parameter sign of the transient deviation scalar is This is used to quantify the severity of the current microscopic physical friction state (i.e., the first data) exceeding the safety threshold boundary; it reflects the transient overflow of the fluid incompressible hard resistance that surges instantaneously inside the pressing chamber relative to the maximum theoretical bearing capacity; it relies on the nonlinear topological envelope surface function already constructed in the control node memory and the real-time updated two-dimensional phase trajectory vector coordinates; it originates from the point-to-surface distance measurement principle in analytical geometry; the calculation logic is as follows: obtain the two-dimensional phase trajectory vector at the moment of triggering the risk confirmation indication signal, extract the current macroscopic energy efficiency longitudinal absolute coordinate value; substitute the longitudinal absolute coordinate value into the preset nonlinear topological envelope surface function, and solve the equation root in reverse to calculate the corresponding horizontal axis safety extreme value; the calculation logic is as follows: multiply the extracted cross-coupling attenuation coefficient parameter with the longitudinal absolute coordinate value, and add the product result to the horizontal axis first data linear term stretching coefficient parameter to obtain the value used as the division benchmark. The first intermediate denominator; multiply the compression coefficient parameter of the second linear term of the vertical axis with the vertical absolute coordinate value, and add the product result to the first term constant offset coefficient parameter to obtain the second intermediate molecular weight; take the negative of the second intermediate molecular weight and divide it by the first intermediate denominator; the division quotient is the strictly permissible micro-dissipation horizontal axis safety extreme value under a specific macro-efficiency level; obtain the micro-dissipation horizontal absolute coordinate value of the current two-dimensional phase trajectory vector; subtract the horizontal axis safety extreme value from the micro-dissipation horizontal absolute coordinate value to obtain the difference value representing the absolute space limit difference; in order to make the control parameters have scale uniformity and meet the requirement that the output value range is limited to the interval between zero and one, the hyperbolic tangent function is called as a nonlinear compression mapping tool; specifically, the difference value is input into the hyperbolic tangent function for mapping calculation. Since the limit difference must be positive, the calculation output result of the hyperbolic tangent function is between the closed interval zero and one, and the mapping output result is the transient deviation scalar. .

[0027] The parameter sign of the time derivative scalar is Used to quantify the instantaneous acceleration of the macroscopic system's squeezed functional efficiency state (i.e., the second data) sliding towards irreversible decline; reflects the rate of loss of elastic support due to continuous heating; the data source depends on the dynamic elasticity index sequence of energy efficiency output for two consecutive cycles obtained from the low-frequency bus; the basic Euler forward difference model uses the difference quotient to replace the continuous derivative; the calculation logic is as follows: actively shield the interference of the high-frequency interpolation sequence, extract the latest real low-frequency physical sampling clock stamp to which the current moment belongs from the bottom layer; extract the vertical absolute coordinate value of macroscopic energy efficiency corresponding to the latest low-frequency physical sampling clock stamp as the energy efficiency index of the current cycle; forcibly cross all high-frequency interpolation data points generated by the Kalman smoother in the memory buffer stack, backtrack to extract the historical measured points that are exactly one Hz away from the current clock stamp, and set the historical measured points... The macroscopic energy efficiency vertical absolute coordinate value is used as the energy efficiency index for the historical period; the obtained absolute time step parameter is rigidly equivalent to the low-frequency physical sampling period (i.e., 1 second) of the underlying energy meter bus; the energy efficiency index of the historical period is subtracted from the energy efficiency index of the current period to obtain the positive drop reflecting the actual physical efficiency degradation; the positive drop is divided by the absolute time step parameter of 1 second to obtain the unstandardized absolute degradation slope; the pre-set empirically calibrated maximum allowable system collapse slope constant is retrieved, and the obtained absolute degradation slope is divided by the maximum allowable system collapse slope constant; if the division result is greater than 1, it is forcibly truncated and assigned the value 1; if the result is between zero and one, the original value is retained; this truncation operation constitutes the piecewise linear normalization logic, and the final output controlled value is the time derivative scalar that is completely immune to high-frequency computational noise interference. .

[0028] The symbol for the duty cycle parameter of pulse width modulation is: As the core execution quantity of the first control command, it defines the proportion of the flow time of the external hydraulic servo proportional valve in a single control cycle, directly determining the instantaneous expansion displacement of the resistance section at the end of the press; it relies on the digital pulse generator interface at the hardware level; it is a direct linear gain mapping mechanism; the calculation logic is as follows: read the generated transient deviation scalar between zero and one; obtain the preset maximum permissible single pressure relief coefficient parameter; multiply the transient deviation scalar directly by the maximum permissible single pressure relief coefficient parameter, and the product result is the pulse width modulation duty cycle finally output to the hardware interface. .

[0029] The maximum permissible system collapse slope constant represents the critical rate limit of energy efficiency loss per unit time before irreversible screw seizure occurs. Its value is determined through a destructive material thermal degradation bench experiment, using a step-feed of high-friction bran with extremely low moisture content to measure the average energy efficiency drop rate in the three seconds before the spindle torque rapidly diverges. Based on experimental calibration, the preferred value for the maximum permissible system collapse slope constant in this embodiment is set to 0.15 per second. The maximum permissible single pressure relief coefficient parameter limits the instantaneous opening stroke of the resistance section in a single closed-loop control to prevent excessive pressure relief leading to pressure build-up failure inside the pressing chamber. Its determination relies on hydraulic servo-end step response testing and fluid back pressure maintenance experiments. In this embodiment, the preferred value for the maximum permissible single pressure relief coefficient parameter is set to 12% (corresponding to the value 0.12). This value represents a rigid constraint balance boundary achieved between ensuring sufficient disintegration of micro-stress peaks and maintaining the macro-static pressure required for oil extraction.

[0030] The symbol for the speed offset adjustment ratio is: As the core output of the second control command, it represents the percentage reduction required to lower the current operating reference frequency of the spindle drive inverter. It is calculated based on the steepest descent method model in classical optimization theory, treating the energy efficiency degradation rate as a minimized cost function. The calculation logic is as follows: The generated time derivative scalar is read; the time derivative scalar acts as the local negative gradient replacement value of the cost function at time t; the pre-calibrated offline optimization learning step size coefficient parameter is retrieved; the time derivative scalar is multiplied by the optimization learning step size coefficient parameter to obtain the frequency reduction calculation for the current cycle; the frequency reduction calculation is compared with the preset inverter anti-stepping lower limit protection extreme value parameter and the smaller value is used to prevent excessive single speed reduction from causing the motor slip rate to exceed the limit. The percentage value obtained after the smaller value operation is the speed offset adjustment ratio normalized to the zero-to-one interval. The optimization learning step size coefficient parameter is set to 0.05.

[0031] The parameter symbols for the extreme values ​​of the inverter's anti-step-out lower limit protection are as follows: The purpose is to set the maximum safe boundary for the operating frequency of the drive motor within a single control cycle, and to prevent the slip rate from collapsing due to the motor speed not keeping up with the deceleration of the synchronous rotating magnetic field caused by a cliff-like speed reduction; it is based on the factory calibration of the motor's electrical system; specifically, it is estimated by looking up a table together the rated rotor inertia of the equipment's main shaft motor and the factory-nominated maximum critical torque slip rate; in this embodiment, based on the actual measurement verification of the large inertia rice bran extrusion load condition, the preferred value of the extreme value parameter of the anti-step loss protection is rigidly set to eight percent (i.e., the corresponding percentage value of 0.08); to ensure that when the gradient optimization algorithm seeks a large speed reduction under extreme conditions, the hardware execution layer always maintains the physical bottom line support of magnetic field coupling.

[0032] The steps for generating the first control command and the second control command are as follows: Acquire the confirmation indication signal, which carries both horizontal and vertical coordinate data at the trigger time; a determined two-dimensional phase trajectory vector coordinate data stack (including the coordinate points of the current point and the historical cycle); and various safety extreme value constants and optimization learning step size coefficient parameters that are solidified and read at the bottom layer. The control logic is decoupled and split; Parallel processing branch one: calls the current coordinate point data and the nonlinear topological envelope surface function, and calculates the output transient deviation scalar according to the hyperbolic tangent function mapping logic; Parallel processing branch two: calls the current coordinate point data and historical periodic point data, calculates the decay slope according to Euler forward difference logic, and calculates the output time derivative scalar according to piecewise linear normalization logic; Logic transformation and constraint judgment; Transformation branch 1: Perform multiplication and scaling on the extracted transient deviation scalar to convert it into a pulse width modulation duty cycle parameter; Transformation branch 2: Feed the extracted time derivative scalar into the gradient descent iterator and multiply it with the optimization learning step size coefficient parameter; Introduce key anti-collapse logic judgment: If the frequency drop calculation calculated by the product is greater than the preset anti-step loss lower limit protection extreme value parameter, the oversaturation blocking path is triggered, and the output is forcibly truncated and assigned to the preset anti-step loss lower limit protection extreme value parameter; If the product is less than the anti-step loss lower limit protection extreme value parameter, the normal optimization path is followed to maintain the original value; Finally, the output is locked as the speed offset adjustment ratio parameter; Within the same millisecond-level communication cycle, the preset hydraulic servo reference carrier frequency parameter is extracted; the calculated pulse width modulation duty cycle parameter is precisely modulated into the periodic time line of the hydraulic servo reference carrier frequency parameter; through the underlying high-speed digital IO interface, the modulated first control command pulse train is sent to the hydraulic servo proportional valve to drive the valve core to overcome static friction and perform instantaneous expansion of the resistance section; and through the industrial fieldbus protocol (in this embodiment, Modbus TCP is used), the second control command data frame encoded by the speed offset adjustment ratio parameter is sent to the main frequency converter to realize concurrent control of the mechanical execution domain and the electrical power domain.

[0033] The rigid physical time scale of the first control command pulse train is represented by the hydraulic servo reference carrier frequency parameter, which is the basic period for the digital pulse generator output level to flip. The hydraulic servo reference carrier frequency parameter is used to generate a small alternating chatter current in the electromagnetic coil of the hydraulic servo proportional valve, thereby effectively eliminating the mechanical static friction dead zone of the valve core and improving the dynamic response sensitivity when fine-tuning the duty cycle. In this embodiment, considering the coil inductance characteristics of the large-diameter hydraulic pressure relief proportional valve, the preferred value of the hydraulic servo reference carrier frequency parameter is set to 120 Hz.

[0034] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. 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 the invention. Therefore, the invention 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.

Claims

1. A key process for mechanically pressing rice bran oil, characterized in that, Includes the following steps: S1: Obtain the remote sensing dataset of the pressing equipment after de-identification processing, extract the phased energy supply time-series waveform corresponding to the power input port, perform time-frequency domain transformation and dissipation calculation on the phased energy supply time-series waveform, and generate the first data to characterize the physical friction state of the pressing micro-interface. The first data is the transient information entropy of the driving power spectrum. S2: Extract global input energy load parameters and effective conversion work factor from the remote sensing dataset of the pressing equipment. Perform correlation evolution analysis on global input energy load parameters and effective conversion work factor within a dynamic time window. Calculate and generate second data to characterize the state of macroscopic extrusion work efficiency decay. The second data is the dynamic elasticity index of energy efficiency output. The sampling frequency for obtaining the second data is lower than the sampling frequency for obtaining the first data. S3: Call the pre-configured multi-rate smooth interpolation alignment algorithm to map the first data and the second data to the same time series anchor point. Use the first data as the horizontal axis feature quantity and the second data as the vertical axis feature quantity to construct a two-dimensional phase trajectory vector representing the nonlinear dynamic evolution path. S4: Obtain the preset nonlinear topological envelope surface function, substitute the two-dimensional phase trajectory vector into the nonlinear topological envelope surface function to run the topological space crossing degree discrimination, and generate a confirmation indication signal to characterize the cross-scale structural congestion risk when it is determined that the vector tip of the two-dimensional phase trajectory vector breaks through the spatial boundary defined by the nonlinear topological envelope surface function. S5: Based on the received confirmation indication signal, synchronously output the first control command and the second control command, which respectively expand the geometry of the end resistor cross section and lower the current working drive frequency operating baseline.

2. The key process for mechanically pressing rice bran oil according to claim 1, characterized in that: The steps for generating the first data in S1 are as follows: Adaptive empirical mode decomposition is performed on the phase energy supply time-series waveform to remove the fundamental frequency steady-state interference and extract the residual high-frequency energy components; the residual high-frequency energy components are intercepted within the set sliding time window, a short-time Fourier transform is performed to construct the reconstructed phase space, and the disorder distribution value in the corresponding high-frequency band is calculated according to the Shannon information entropy formula. The obtained dimensionless scalar is used as the transient information entropy of the driving power spectrum.

3. The key process for mechanically pressing rice bran oil processing according to claim 1, characterized in that: The steps for generating the second data in S2 are as follows: extract the first relative rate of change of the global input energy load parameters within the current dynamic time window, and the second relative rate of change of the effective conversion work factor obtained based on the internal fluid dynamics ratio; input the first relative rate of change and the second relative rate of change into a preset recursive least squares algorithm model with a forgetting factor, identify the covariance and variance ratio of the first relative rate of change and the second relative rate of change online, and use the identified ratio as the dynamic elasticity index of energy efficiency output.

4. The key process for mechanically pressing rice bran oil processing according to claim 1, characterized in that: The steps for constructing a two-dimensional phase trajectory vector in S3 are as follows: The first data is input into a preset Kalman smoother model as a local high-frequency observation update variable, and the second data is input into the Kalman smoother model as a prior constraint variable for macroscopic state transition; through the inverse interpolation operation of the Kalman smoother model, the low-frequency sampled second data is reconstructed on the time axis into a high-frequency synchronization sequence matrix with the same timestamp as the first data, and a two-dimensional phase trajectory vector is generated based on the synchronized horizontal and vertical feature quantities.

5. The key process for mechanically pressing rice bran oil processing according to claim 1, characterized in that: The steps for obtaining the preset nonlinear topological envelope surface function in S4 are as follows: read the environmental and material degradation boundary matrix coefficients from the external structured carrier; reconstruct the high-order polynomial online based on the boundary matrix coefficients to form a two-dimensional constraint mapping surface jointly defined by the upper limit of microscopic dissipation and the lower limit of macroscopic energy efficiency, and use the two-dimensional constraint mapping surface as the nonlinear topological envelope surface function.

6. The key process for mechanically pressing rice bran oil processing according to claim 1, characterized in that: The steps for calculating the confirmation signal are as follows: It receives three heterogeneous input elements; the first input is the transient information entropy of the drive power spectrum based on the high-frequency sampling of the inverter at the kilohertz level; the second input is the dynamic elasticity index of energy efficiency output based on the low-frequency sampling of the system bus at the single hertz level; the third input is the environmental and material degradation boundary matrix coefficients read from the non-volatile memory area. To address the clock update rate mismatch between the input driving power spectrum transient information entropy and the energy efficiency output dynamic elasticity index, a Kalman smoother model is invoked for data alignment calculation. The Kalman smoother model uses the energy efficiency output dynamic elasticity index of the previous low-frequency cycle as a priori historical state for recursion. The Kalman smoother model continuously receives and buffers dozens of high-density driving power spectrum transient information entropies arriving within the current high-frequency cycle as observation correction quantities. A reverse state backtracking operation is performed, utilizing the fluctuation variance disturbance contained in the continuous high-frequency observation data to correct the blank time interval between two low-frequency sampling points. Within the blank time interval, interpolated energy efficiency sequence points with the same number of driving power spectrum transient information entropies are generated. Vector merging is performed for each time alignment point on the millisecond clock line; the transient information entropy of the driving power spectrum is assigned as the horizontal axis component representing micro-dissipation, and the corresponding interpolated energy efficiency sequence points are assigned as the vertical axis component representing macro-energy efficiency; a two-dimensional phase trajectory vector with length and pointing characteristics is generated in the virtual Cartesian coordinate plane in memory for a single timestamp. The loaded environmental and material degradation boundary matrix coefficients are retrieved; the first constant offset coefficient parameter, the first linear term stretching coefficient parameter of the horizontal axis, the second linear term compression coefficient parameter of the vertical axis, and the cross-coupling attenuation coefficient parameter contained in the environmental and material degradation boundary matrix coefficient set are extracted; the first linear term stretching coefficient parameter of the horizontal axis is bound to the horizontal axis coordinate variable, the second linear term compression coefficient parameter of the vertical axis is bound to the vertical axis coordinate variable, and the cross-coupling attenuation coefficient parameter is forced as the multiplication gain coefficient of the product term of the horizontal axis coordinate variable and the vertical axis coordinate variable; all parameters and bound terms are algebraically summed to instantiate a continuous bivariate quadratic polynomial hyperbolic open topological boundary in the memory coordinate plane. The hyperbolic open topological boundary divides the coordinate system into a steady-state control quadrant region representing safe work and an outer region representing the risk of coking instability; the generated two-dimensional phase trajectory vector containing horizontal and vertical coordinate values ​​is projected onto this space. Obtain the absolute coordinate point value of the two-dimensional phase trajectory vector vector after projection into the coordinate system, and perform spatial crossing degree discrimination; perform judgment calculation, and substitute the microscopic lateral component value and macroscopic longitudinal component value at this time into the instantiated nonlinear bivariate quadratic polynomial to calculate the actual output value of the polynomial equation; Obtain the absolute coordinate point value of the two-dimensional phase trajectory vector vector after projection into the coordinate system, and perform spatial crossing degree discrimination calculation; Compare the actual output value with the absolute safety topological potential energy reference constant: if the actual output value is less than the absolute safety topological potential energy reference constant; This indicates that although the roughness of natural materials causes the microscopic first data to fluctuate, the two-dimensional phase trajectory vector has not yet broken through the hyperbolic open topological boundary, and the system elastic support of the macroscopic second data remains solid. This state is classified as normal processing noise; the confirmation indicator signal is forcibly not output, and the current judgment value is discarded to wait for the next cycle; If the actual output value is greater than or equal to the absolute safety topological potential energy reference constant; The phase trajectory breaks through the hyperbolic open topological boundary in the lower right region, which represents high entropy and low elasticity. Microscopic dry friction has substantially weakened the overall propulsion efficiency of the equipment; this is regarded as a real precursor to coking; a confirmatory indication signal of cross-scale structural congestion risk is immediately generated and the output stage begins. The final output confirmation signal is a trigger flag.

7. The key process for mechanically pressing rice bran oil processing according to claim 6, characterized in that: The trigger marker encapsulates both the microscopic deviation magnitude and the macroscopic attenuation rate at this moment. Within the instantaneous control cycle of receiving the confirmation indication signal, the subsequent control link will be directly invoked according to the trigger marker, and the first control command will be issued simultaneously to physically expand the mechanical gap of the end cake discharge to release the hydrostatic pressure, and the second control command will be issued to electrically reduce the speed and frequency of the drive motor to suppress heat generation, thereby generating an electromechanical cooperative protection gain effect.

8. The key process for mechanically pressing rice bran oil processing according to claim 1, characterized in that: The process of outputting the first and second control commands in S5 is as follows: Extract two derivative features of the two-dimensional phase trajectory vector, namely the transient deviation scalar representing the deviation of the current state of the first data, and the time derivative scalar representing the decay rate of the second data; after converting the transient deviation scalar through pulse width modulation, it is used as the first control command output; in parallel, the time derivative scalar is input into the gradient optimization algorithm to calculate the bias constraint, and is used as the second control command output.

9. The key process for mechanically pressing rice bran oil processing according to claim 1, characterized in that: The first control command is to send a command to the displacement execution port to expand the geometry of the end-stage resistance section, and the second control command is to send a command to the frequency converter drive port to lower the current working drive frequency operating baseline. During the execution of the first and second control commands, the first and second data of the next sampling period are continuously acquired to form a feedback closed-loop control loop that drives the two-dimensional phase trajectory vector back to the spatial boundary.

10. The key process for mechanically pressing rice bran oil according to claim 9, characterized in that: The steps for generating the first control command and the second control command are as follows: Acquire the confirmation indication signal, which carries both horizontal and vertical coordinate data at the trigger time; the determined two-dimensional phase trajectory vector coordinate data stack; and various safety extreme value constants and optimization learning step size coefficient parameters that are read and solidified at the bottom layer. The control logic is decoupled and split; Parallel processing branch one: calls the current coordinate point data and the nonlinear topological envelope surface function, and calculates the output transient deviation scalar according to the hyperbolic tangent function mapping logic; Parallel processing branch two: calls the current coordinate point data and historical periodic point data, calculates the decay slope according to Euler forward difference logic, and calculates the output time derivative scalar according to piecewise linear normalization logic; Logic transformation and constraint judgment; Transformation branch 1: Perform multiplication and scaling on the extracted transient deviation scalar to convert it into a pulse width modulation duty cycle parameter; Transformation branch 2: Feed the extracted time derivative scalar into the gradient descent iterator and multiply it with the optimization learning step size coefficient parameter; Introduce key anti-collapse logic judgment: If the frequency drop calculation calculated by the product is greater than the preset anti-step loss lower limit protection extreme value parameter, the oversaturation blocking path is triggered, and the output is forcibly truncated and assigned to the preset anti-step loss lower limit protection extreme value parameter; If the product is less than the anti-step loss lower limit protection extreme value parameter, the normal optimization path is followed to maintain the original value; Finally, the output is locked as the speed offset adjustment ratio parameter; Within the same millisecond-level communication cycle, the preset hydraulic servo reference carrier frequency parameter is extracted; the calculated pulse width modulation duty cycle parameter is precisely modulated into the periodic time line of the hydraulic servo reference carrier frequency parameter; through the underlying high-speed digital IO interface, the modulated first control command pulse train is sent to the hydraulic servo proportional valve to drive the valve core to overcome static friction and perform instantaneous expansion of the resistance section. It also sends a second control command data frame encoded by the speed offset adjustment ratio parameter to the main frequency converter via the industrial fieldbus protocol, thereby realizing concurrent control of the mechanical execution domain and the electrical power domain.

Images