Vacuum degree segmented gradient control method and system for vacuum distillation system

Abstract

The present application relates to the technical field of vacuum distillation process control and chemical separation automation, in particular to a vacuum degree segmented gradient control method and system for a vacuum distillation system, comprising: a loading gas release inflection point temperature band, a distillation stage parameter, a flow state control interval boundary, a control parameter table, a feedforward parameter table, a precursor threshold value, historical batch time delay data and a gas release prediction coefficient; obtaining pressure, temperature, distillate flow, bypass valve opening and roots pump frequency, calculating comprehensive pressure, pressure change rate and Knudsen number; judging gas sudden increase trend and predicting pressure peak value disturbance and peak arrival time, generating target pressure in sections; performing pressure feedback control according to flow state interval, and performing pre-pumping in a feedforward trigger window; determining segmented switching time and switching fraction collection channels according to gas phase temperature change rate, change acceleration, deviation and distillate flow; realizing pressure smooth transition and fraction high-purity separation.

Description

Technical Field

[0001] This invention relates to the field of vacuum distillation process control and chemical separation automation technology, specifically to a method and system for segmented gradient control of vacuum degree in a vacuum distillation system. Background Technology

[0002] In the current vacuum distillation and purification process of industrial base oil or waste engine oil, the material will periodically produce multi-component cross-distillation and concentrated gas release phenomena at different temperature ranges, and the gas dynamic characteristics of the system show significant nonlinear characteristics with pressure changes. To control the vacuum level of the system, existing solutions generally adopt a passive constant pressure control architecture, which uses a single vacuum gauge for feedback and fixed parameters for full-range adjustment. Although this solution has a certain control capability under normal steady-state operation, it is highly dependent on the range of a single sensor and lacks the ability to predict material phase changes. This results in a mismatch between the pumping capacity during the release period and the dynamic load, leading to pressure collapse and adjustment lag caused by feedback distortion. In addition, fixed parameter adjustment is difficult to cope with the complex flow regime changes when distilling near the boiling point, which can easily lead to misjudgment of overlapping boiling range fraction switching and deterioration of separation purity, making it difficult to support high-purity and stable separation of continuous batches. Therefore, overcoming the control instability caused by nonlinear physical characteristics and improving the timeliness and accuracy of pressure smooth transition and precise fraction separation in vacuum distillation systems have become urgent technical problems to be solved. Summary of the Invention

[0003] The purpose of this invention is to provide a vacuum degree segmented gradient control method and system for a vacuum distillation system, which avoids the pressure fluctuation amplitude exceeding the preset fluctuation threshold and the full-range control response lag caused by existing control methods relying on a single signal and fixed adjustment parameters. It also accurately determines the segmented switching nodes of overlapping boiling ranges, thereby achieving a smooth pressure transition and high-purity separation of distillate.

[0004] The objective of this invention can be achieved through the following technical solutions: A method for segmented gradient control of vacuum degree in a vacuum distillation system includes: loading the release inflection point temperature band, distillation stage parameters, flow control interval boundaries, control parameter table, feedforward parameter table, precursor threshold, historical batch time delay data, and release prediction coefficient; The pressure readings inside the vacuum distillation vessel by the capacitive thin-film vacuum gauge and the Pirani vacuum gauge, as well as the liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening degree and Roots pump frequency, are obtained. The combined pressure, pressure change rate and Knudsen number are calculated to form the sampling data. The gas surge trend is judged by a combination score of pressure change rate, liquid phase temperature and execution load determined by bypass valve opening and Roots pump frequency. The pressure peak disturbance and peak arrival time are predicted by combining the gas release prediction coefficient. The target pressure of each segment is generated segment by segment according to the basic pressure drop gradient and compensation pressure drop in the control parameter table. A stage control instruction table and a feedforward control state table including pre-extraction state are also generated. The flow control interval is divided according to the comprehensive pressure and Knudsen number. The corresponding proportional, integral and differential parameter groups are read to perform pressure feedback control. Pre-evacuation is performed when the peak arrival time falls into the feedforward trigger window determined by the Roots pump frequency rise time and the preset safety advance time. Calculate the rate of change and acceleration of the vapor phase temperature, as well as the deviation between the current state and the parameters of the distillation stage. Determine the timing of segment switching based on the rate of change, acceleration, deviation, and distillate flow rate, apply the target pressure for the next segment, and switch the fraction collection channel.

[0005] Furthermore, the overall pressure is determined based on a preset weighted relationship between the readings of the capacitive diaphragm vacuum gauge and the Pirani vacuum gauge; When the reading of the capacitive thin-film vacuum gauge is below the low-pressure switching threshold, the reading of the Pirani vacuum gauge is used as the overall pressure. When the reading of the capacitive thin-film vacuum gauge is between the low-pressure switching threshold and the high-pressure switching threshold, the two types of pressure readings are weighted and fused according to the weights that change with the capacitive thin-film vacuum gauge reading. When the reading of the capacitive thin-film vacuum gauge is above the high-pressure switching threshold, the reading of the capacitive thin-film vacuum gauge is used as the overall pressure. The pressure change rate is the ratio of the change in the filtered composite pressure over a predetermined differential time span to that time span; the Knudsen number is calculated based on the absolute temperature of the gas phase, the composite pressure, the hydraulic diameter, and the calibration coefficient.

[0006] Furthermore, the combined scoring includes: determining the pressure rise score based on the continuous increase in the pressure change rate; determining the temperature score based on whether the liquid phase temperature has entered the release inflection point temperature zone; and determining the execution load score based on the bypass valve opening and the Roots pump frequency. The gas surge index is obtained by weighting the pressure rise score, temperature score, and execution load score. When the gas surge index reaches the precursor threshold, the precursor-to-peak delay in the historical batch delay data is corrected by combining the current average pressure change rate to obtain the peak arrival time. The pressure peak disturbance is determined based on the average pressure change rate and liquid phase temperature deviation. When the gas surge index does not reach the precursor threshold, the current segmented target pressure is maintained.

[0007] Furthermore, the control parameter table includes the basic pressure drop gradient and the pressure drop amount for boiling range overlap compensation; The target pressure of adjacent segments is determined based on the basic pressure drop gradient and the pressure drop compensation amount of boiling range overlap. The target pressure of the next segment is determined by subtracting the basic pressure drop gradient from the target pressure of the previous segment, and is corrected by adding the compensation amount of overlap correction coefficient. The overlap correction factor is determined based on the deviation between the current segment's distillation platform duration and the distillation stage parameters, and increases with the increase of the distillation platform duration; The stage control instruction table includes the target pressure, entry conditions, minimum holding time, exit conditions, and fraction collection channel number for each segment.

[0008] Furthermore, the boundary of the flow control zone includes a first flow boundary and a second flow boundary; The flow regime control region includes the viscous flow region where the Knudsen number is less than the first flow regime boundary, the transition flow region where the Knudsen number is located between the first flow regime boundary and the second flow regime boundary, and the molecular flow region where the Knudsen number is not less than the second flow regime boundary. Each flow control interval corresponds to a set of proportional gain, integral gain, derivative gain, output limiting, and integral limiting. The field controller reads the corresponding parameter set according to the current flow control interval and limits the integral term and control output. A switching hysteresis band is set between adjacent flow control intervals.

[0009] Furthermore, the control parameter table also includes platform slope threshold, minimum effective distillation flow rate, acceleration threshold, predetermined count, and deviation threshold; Determining the timing of segment switching includes: smoothing the gas phase temperature and calculating the rate and acceleration of change; when the absolute value of the rate of change is not greater than the plateau slope threshold and the distillation flow rate is not less than the minimum effective distillation flow rate, it is determined that the distillation plateau is in place. Within the distillation platform, a weighted deviation assessment is performed between the current state, which consists of comprehensive pressure, vapor temperature, rate of change, acceleration of change, and distillation flow rate, and the parameters of each distillation stage to obtain the comprehensive deviation. When the changing acceleration reaches the acceleration threshold, the count of positive acceleration within the observation period consisting of a predetermined continuous sampling period reaches the predetermined count, and the minimum comprehensive deviation is not greater than the deviation threshold, the segment switching time is determined.

[0010] Furthermore, the feedforward control state table includes multiple states, including the pre-evacuation state. When the peak arrival time is greater than zero and falls within the feedforward trigger window, the Roots pump frequency increment and bypass valve pre-opening degree corresponding to the pressure peak disturbance are determined by looking up the feedforward parameter table, and the Roots pump frequency and bypass valve opening degree are increased after limiting. If the peak arrival time is not greater than zero or does not fall within the feedforward trigger window, pre-evacuation is not performed and the current control state is maintained. When both the overall pressure and the pressure change rate drop below the preset threshold, the Roots pump frequency and bypass valve opening are gradually restored.

[0011] Furthermore, it also includes: after the current batch ends, saving the overall pressure, liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening, Roots pump frequency, segment switching position, peak prediction error, pressure steady-state error, and pressure overshoot; The release inflection point temperature band and distillation stage parameters are updated based on the segment switching position using a historical batch weighted average method. The release prediction coefficient, feedforward parameter table, and parameter group corresponding to each flow control interval are updated based on the peak prediction error, pressure steady-state error, and pressure overshoot.

[0012] A vacuum degree segmented gradient control system for a vacuum distillation system includes: The initialization module is used to load the gas release inflection point temperature band, distillation stage parameters, flow control interval boundaries, control parameter table, feedforward parameter table, precursor threshold, historical batch delay data, and gas release prediction coefficient. The data acquisition and calculation module is used to acquire the pressure readings inside the vacuum distillation vessel from the capacitive thin-film vacuum gauge and the Pirani vacuum gauge, as well as the liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening degree and Roots pump frequency, and to calculate the comprehensive pressure, pressure change rate and Knudsen number to form the sampling data. The gradient generation module includes a scoring unit, a prediction unit, a target pressure generation unit, and a state table generation unit. It is used to judge the gas surge trend based on the combined scoring of pressure change rate, liquid phase temperature, and execution load, predict the pressure peak disturbance and peak arrival time by combining the gas release prediction coefficient, generate the target pressure of each segment according to the basic pressure reduction gradient and compensation pressure reduction in the control parameter table, and generate the stage control instruction table and the feedforward control state table including the pre-extraction state. The segmented execution module is used to divide the flow control interval according to the comprehensive pressure and Knudsen number, read the corresponding proportional, integral and differential parameter group for pressure feedback control, and perform pre-evacuation when the peak arrival time falls into the feedforward trigger window determined by the Roots pump frequency rise time and the preset safety advance time. The switching control module is used to calculate the rate of change and acceleration of the gas phase temperature, as well as the deviation between the current state and the parameters of the distillation stage. Based on the rate of change, acceleration, deviation, and distillation flow rate, it determines the timing of segment switching, applies the target pressure for the next segment, and controls the switching mechanism to switch the corresponding fraction collection channel.

[0013] Compared with the prior art, the present invention has the following advantages: 1. This invention loads the gas release inflection point temperature band, distillation stage parameters, flow control interval boundaries, control parameter table, feedforward parameter table, precursor threshold, historical batch time delay data, and gas release prediction coefficient. Combined with multi-source acquisition from capacitive thin-film vacuum gauge, Pirani vacuum gauge, liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening, and Roots pump frequency, it integrates to form comprehensive pressure, pressure change rate, and Knudsen number. With segmented weight switching and threshold hysteresis, it achieves stable measurement across the entire range, avoiding feedback distortion and switching jumps from a single vacuum gauge. Attached Figure Description

[0014] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart of a segmented gradient control method for vacuum degree in a vacuum distillation system provided in an embodiment of this application; Figure 2 This is a schematic diagram of a vacuum distillation system vacuum degree segmented gradient control system provided in an embodiment of this application. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0016] Please see Figure 1 A method for segmented gradient control of vacuum degree in a vacuum distillation system includes: loading the release inflection point temperature band, distillation stage parameters, flow control interval boundary, control parameter table, feedforward parameter table, precursor threshold, historical batch delay data, and release prediction coefficient. The pressure readings inside the vacuum distillation vessel by the capacitive thin-film vacuum gauge and the Pirani vacuum gauge, as well as the liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening degree and Roots pump frequency, are obtained. The combined pressure, pressure change rate and Knudsen number are calculated to form the sampling data. The gas surge trend is judged by a combination score of pressure change rate, liquid phase temperature and execution load determined by bypass valve opening and Roots pump frequency. The pressure peak disturbance and peak arrival time are predicted by combining the gas release prediction coefficient. The target pressure of each segment is generated segment by segment according to the basic pressure drop gradient and compensation pressure drop in the control parameter table. A stage control instruction table and a feedforward control state table including pre-extraction state are also generated. The flow control interval is divided according to the comprehensive pressure and Knudsen number. The proportional, integral and differential parameter group of the corresponding flow control interval is read to perform pressure feedback control. Pre-evacuation is performed when the peak arrival time falls into the feedforward trigger window determined by the Roots pump frequency rise time and the preset safety advance time. Calculate the rate of change and acceleration of the vapor phase temperature, as well as the deviation between the current state and the parameters of the distillation stage. Determine the timing of segment switching based on the rate of change, acceleration, deviation, and distillate flow rate, apply the target pressure for the next segment, and switch the fraction collection channel.

[0017] This embodiment addresses the problems of pressure collapse during the gas release period, full-range adjustment lag, and misjudgment of overlapping boiling range switching caused by single vacuum gauge feedback and fixed parameter adjustment in the existing vacuum distillation process. By fusion of multi-source sensor data and feature extraction, it predicts the trend of sudden gas release of materials and performs pre-gas extraction in advance. Meanwhile, the dynamic matching adjustment parameters are divided according to the gas flow pattern, and the segment switching nodes are accurately determined by combining temperature evolution and distillation flow characteristics; the overall scheme realizes the transformation from passive constant pressure control to active gradient application and adaptive adjustment, and finally achieves the core effect of smooth pressure transition and high-purity separation of distillate. In applications, this method can be deployed in a collaborative architecture of edge computing servers and field programmable logic controllers; temperature and pressure data collected in real time by sensors flow into the edge server, and the server continuously outputs stage control instruction tables and feedforward control status tables to the programmable logic controller based on data fusion and feature extraction logic; When the system detects that the liquid phase temperature has entered the gas release sensitive zone and the pressure change rate continues to rise, the frequency of the Roots pump can be increased in advance and the bypass valve can be opened in advance before the peak is predicted, so as to suppress the pressure disturbance caused by gas release within the preset fluctuation threshold. When entering the overlapping boiling range stage, the system automatically adds compensation pressure drop and precisely switches the fraction collection channel according to the acceleration of gas phase temperature change and distillation flow characteristics, thereby completing the stable iteration of the entire separation process.

[0018] As a preferred implementation, the overall pressure is determined based on a preset weighted average of the readings from the capacitive diaphragm vacuum gauge and the Pirani vacuum gauge, calculated according to the following formula: in, To address the combined pressures, For the reading of a capacitive thin-film vacuum gauge, For the Pirani vacuum gauge reading; when hour, ;when hour, ;when hour, ; This is the low-voltage switching threshold. The high-pressure switching threshold is used; the pressure change rate is the ratio of the difference between the filtered composite pressure and the predetermined differential time span set according to the system sampling period to that time span; the Knudsen number is calculated according to the following formula: in, For calibration coefficients, The absolute temperature of the gas phase. The hydraulic diameter of the vacuum distillation vessel is given.

[0019] After obtaining the readings of the capacitive thin-film vacuum gauge and the Pirani vacuum gauge from the vacuum distillation vessel, the readings of the two types of vacuum gauges are fused and calculated according to a preset weighting relationship to generate a comprehensive pressure. and Knudsen number ; Among them, the fusion weight This is used to smoothly transition the effective reliability ratio of different vacuum gauges under the current device, current gas system and current installation location conditions, rather than mechanically applying the nominal range of the instrument without considering the field calibration conditions; therefore, and This defines the engineering switching threshold at which the dominance of the two types of readings shifts after calibration in this system; the weight allocation relationship is based on the consistency of the overall system's measured values, as long as... to The transitions between them can be continuous, monotonous, and without abrupt changes, thus meeting the control requirements. For example, when the reading of the capacitive thin-film vacuum gauge is at the low-voltage switching threshold. High voltage switching threshold When the weights are between these values, linear interpolation can be used to determine the weights. and The lower and upper boundaries of the fusion interval for the two types of vacuum gauge readings are jointly defined; to avoid sudden changes in control variables caused by single-point jitter when the sampled value falls exactly near the threshold point, it is preferable to first optimize the control variable in the field implementation. Perform threshold hysteresis detection, and then update the weights according to the above segmentation relationship; When the sampled value is within the threshold point or hysteresis band, the weight state that has been effective in the previous sampling period is maintained, so that the range handover process remains stable on the control link. Furthermore, after entering the hysteresis zone, only one shift of dominance is allowed along the current switching direction. If the corresponding exit boundary is not crossed, the weight dominance state will not be rewritten in reverse to avoid back-and-forth jitter within the same pressure platform. This embodiment uses a combination of segmented weight switching and threshold hysteresis to smoothly transition the complementary response regions of the two types of sensors, transforming the disadvantages of a single range into the advantages of stable measurement across the entire range. This avoids control oscillations caused by sensor switching and solves the problem of adjustment lag caused by feedback distortion. Based on this, Knudsen number Characterizing the mean free path of gas molecules and the characteristic dimensions of the pipeline, i.e., the hydraulic diameter. The ratio of is used to approximately reflect the rarefaction and flow state of the gas under the current pressure. Calibration coefficient In engineering, the gas constant, molecular collision characteristics, and unit conversion relationships are pre-absorbed, so that... It is kept as a dimensionless quantity, which makes it easy to use directly for flow regime interval discrimination later.

[0020] As a preferred implementation, the combined scoring includes: determining a pressure rise score based on the continuous increase in the pressure change rate; determining a temperature score based on whether the liquid phase temperature has entered the release inflection point temperature zone; determining an execution load score based on the bypass valve opening and the Roots pump frequency; and obtaining a gas surge index by weighting the pressure rise score, temperature score, and execution load score. If the gas surge index reaches the precursor threshold, the historical batch delay data is corrected by combining the current average pressure change rate to extract the delay from the precursor to the peak, thereby obtaining the peak arrival time, and the pressure peak disturbance is determined based on the average pressure change rate and the liquid phase temperature deviation; if the gas surge index does not reach the precursor threshold, the current segmented target pressure is maintained.

[0021] After obtaining the pressure change rate, liquid phase temperature and executed load, the operating status is weighted and evaluated according to the combined scoring mechanism to determine the gas surge index. The combined scoring system comprehensively assesses factors such as the continuous increase in pressure change rate, whether the liquid phase temperature has entered the release inflection point temperature zone, and the current load status of valves and pumps, in order to achieve multi-dimensional identification of pre-release precursors. The pressure rise score is calculated by the ratio of the current pressure change rate to the set safe change rate threshold. The safe change rate threshold is set based on the absolute value of the pressure drop rate corresponding to the background limit pumping speed measured by the maximum pumping power when the system is unloaded and without material. It is used to characterize the maximum load limit that the system can handle. When the ratio is greater than 1, the score is truncated to full; the temperature score is determined using structured mapping logic, specifically, the temperature difference is defined. It equals the lower boundary of the release inflection point temperature band minus the current liquid phase temperature, when When the temperature difference exceeds the preset warning temperature range, the temperature score is 0; when When the preset warning temperature difference is: The preset warning temperature difference is pre-set based on material characteristics and heating rate, with a preferred value of 5°C to 15°C; if the liquid phase temperature has entered or exceeded the inflection point zone, then... If the temperature score is full, the load score is extracted as the average of the bypass valve opening percentage and the Roots pump operating frequency. The above three scores are weighted and summed according to the static weight coefficients that are a priori set to characterize the sensitivity of gas release precursors, so as to obtain a quantitative gas surge index. To ensure the predictive power, the weighting is prioritized for the source factors that cause phase changes, namely temperature. For example, the weighting of pressure rise is set to 0.3, temperature is set to 0.4, and execution load is set to 0.3, with each score having a maximum score of 1. This embodiment extracts precursor features from multiple dimensions, transforming delayed response into advance prediction, thereby avoiding dynamic mismatch between pumping capacity and release load. When the system detects that there is no historical batch delay data for the corresponding material, the precursor-to-peak reference delay adopts a preset constant default delay, which is determined by the background response time of the system when heating to the temperature range without material; if historical data exists, the system extracts the precursor-to-peak reference delay of the same temperature range from the historical batch. If the gas surge index reaches the precursor threshold, the system extracts the precursor-to-peak reference time delay from the same temperature range in historical batches, and introduces the ratio of the current average pressure change rate to the historical average as a scaling factor to correct the time delay and obtain the peak arrival time.

[0022] Meanwhile, the liquid phase temperature deviation is defined as the effective exceedance of the current liquid phase temperature to the lower boundary of the release inflection point temperature band. That is, only the portion exceeding or entering the lower boundary is included. If the current liquid phase temperature is below this lower boundary, the deviation is treated as zero to avoid deriving a negative pressure peak disturbance before entering the effective release temperature range. The specific formula is as follows:

[0023] The gas release prediction coefficient must have the following physical dimensions. The unit is used to characterize the cumulative time scale of the rate of change of pressure under a unit temperature deviation, which is transformed into an absolute pressure increment, thus ensuring that the formula satisfies strict pressure requirements on both sides of the equation. Dimensional balance is ensured to prevent dimensionless constants from being incorrectly substituted into calculations. Furthermore, to avoid distortion of the time delay correction result when the current average pressure change rate is close to zero, the sign is reversed, or it is affected by instantaneous noise disturbance, in the field implementation, it is preferable to first take all positive change rate samples in the direction of pressure increase within the continuous sliding observation window to calculate the arithmetic mean of the current average pressure change used to correct the time delay. If a negative change rate appears, it is directly discarded and not included in the mean, and a minimum positive lower limit is set for the mean. If the arithmetic mean does not reach the minimum positive lower limit, the peak arrival time will not be updated in this period. Instead, the previous effective prediction result will be maintained or the state of no effective peak prediction will be directly determined. Correspondingly, if the corrected peak arrival time is less than or equal to zero, the prediction is considered to have lost its feedforward significance. In this cycle, the current segmented target pressure is maintained and the system waits for reassessment in the next sampling cycle. If the precursor threshold is not reached, the current target pressure is maintained to avoid invalid actions.

[0024] In one preferred embodiment, the control parameter table includes the base pressure drop gradient. and compensation for pressure drop The target pressure of adjacent segments is determined based on the base pressure reduction gradient and the compensation pressure reduction, and is calculated according to the following formula: in, For the first Segmented target pressure, For the first Segmented target pressure, Based on the pressure reduction gradient, To compensate for the pressure drop caused by boiling range overlap. This is an overlap correction factor determined based on the duration of the current distillation plateau and the deviation of distillation stage parameters. The value increases with the duration of the distillation platform; the stage control instruction table includes the target pressure, entry conditions, minimum holding time, exit conditions, and fraction collection channel number for each segment.

[0025] After obtaining the basic pressure drop gradient and boiling range overlap compensation pressure drop from the control parameter table, the overlap correction coefficient is used to... The target pressure for adjacent segments is generated segment by segment; Among them, the overlap correction coefficient It is a proportional factor for pressure reduction buffering compensation in distillation stages where the boiling range overlap exceeds a preset overlap threshold; in the formula, the positive compensation term... This is used to offset part of the basic pressure drop gradient, so that the downward trend of the target absolute pressure is slowed down according to the overlap correction coefficient; It can be dynamically determined based on the current distillation platform duration and the parameter deviation of the distillation stage, and increases with the increase of the distillation platform duration to adapt to the tail characteristics when the light components are almost completely distilled. Overlap correction factor It is equal to the sum of a dimensionless basic bias constant and a dynamic gain term, with the basic bias constant preferably taking a value of 0.1 to 0.5; To ensure the uniformity of physical dimensions in the calculation, the dynamic gain term is obtained by multiplying the ratio of the current distillation platform duration to the dimensionless value of the reference period, where the reference period is the standard distillation time set according to historical experience, with the ratio of the absolute deviation of the vapor temperature to the dimensionless value of the preset allowable temperature drift range. This avoids the underlying mathematical logic fallacy caused by directly adding the dimensionless constant and the value with temperature dimensions. To maintain consistent control semantics for segmented pressure reduction, an upper limit is set for compensation items in the engineering implementation, so that a single compensation is only used to alleviate pressure reduction and does not reversely increase the target pressure of the next segment. This design transforms the coarse control with uniform settings throughout the process into fine control that adapts to the characteristics of the components, effectively solving the problem of deterioration in separation purity caused by cross distillation of multiple components near their boiling points; the stage control instruction table further outputs the control parameters of each dimension in a structured manner, making it easy for the field controller to directly call and execute them.

[0026] In a preferred embodiment, the flow control region boundary includes a first flow boundary and a second flow boundary; the flow control region includes a viscous flow region where the Knudsen number is less than that of the first flow boundary, a transition flow region where the Knudsen number is located between the first flow boundary and the second flow boundary, and a molecular flow region where the Knudsen number is not less than that of the second flow boundary. Each flow control interval corresponds to a set of proportional gain, integral gain, derivative gain, output limiting, and integral limiting. The corresponding proportional, integral, and derivative parameter set is read according to the current flow control interval, and the integral term and control output are limited. A switching hysteresis band is set between adjacent flow control intervals.

[0027] After obtaining the combined pressure and Knudsen number, the current state of the system is divided into flow control intervals according to the first and second flow boundaries to determine the current viscous flow region, transition flow region, or molecular flow region, and the corresponding proportional-integral-differential parameter set and limiting parameters are read. The first and second flow regime boundaries are set according to the physical distribution law of the ratio of the molecular mean free path to the hydraulic diameter of the system in gas dynamics. In engineering applications, the first flow regime boundary is preferably set to 0.01 and the second flow regime boundary is preferably set to 10.0. When the Knudsen number is less than 0.01, it is determined to be a viscous flow region dominated by collisions between gas molecules; when it is greater than or equal to 10.0, it is determined to be a molecular flow region dominated by collisions between molecules and the tube wall; the region in between is the transition flow region. The flow control region reflects the physical state partitioning of the gas dynamics characteristics in the vacuum system. This embodiment transforms the disadvantage of the controlled object's gain drift exceeding the preset allowable range into the advantage of segmented and precise targeted adjustment by identifying gas flow state transitions and matching independent proportional, integral, and derivative gains and output limiting. In the specific engineering implementation, the limit of the integral term is forcibly limited by an anti-integral saturation mechanism: In each control cycle, once the calculated current control output reaches the set upper or lower limit of the physical output limit, and the integral accumulation direction indicated by the deviation term still attempts to further deepen the saturation trend, the system will dynamically freeze the integral term accumulation process of the current proportional integral derivative parameter group and keep it unchanged in the state of the previous cycle. The integral term is updated normally only when the sign of the system deviation is reversed, causing the direction of integral accumulation to leave the saturation region. This achieves stable regulation across the entire range, avoids sluggishness in the high-voltage section and oscillation in the low-voltage section, and solves the problem of control instability caused by nonlinear physical characteristics. Setting a switching hysteresis band between adjacent flow control intervals can prevent frequent switching of proportional-integral-derivative parameter groups caused by fluctuations in the Knudsen number near the boundary, thereby further improving the stability of field control.

[0028] In a preferred embodiment, the control parameter table also includes a platform slope threshold, a minimum effective distillation flow rate, an acceleration threshold, a predetermined count, and a deviation threshold; determining the segment switching timing includes: The gas phase temperature is smoothed and the rate and acceleration of change are calculated. If the absolute value of the rate of change is not greater than the plateau slope threshold and the distillation flow rate is not less than the minimum effective distillation flow rate, then it is determined to be on the distillation plateau; if the above conditions are not met simultaneously, it is determined to be in the non-distillation plateau state. The preferred platform slope threshold is 0.01℃ / s to 0.05℃ / s, and the minimum effective distillation flow rate is set according to 5% to 10% of the system's rated collection capacity. Within the distillation platform, a weighted deviation assessment is performed between the current state, which consists of comprehensive pressure, vapor temperature, rate of change, acceleration of change, and distillation flow rate, and the parameters of each distillation stage to obtain the comprehensive deviation. If the changing acceleration reaches the acceleration threshold, the count of positive acceleration within the observation period consisting of the predetermined continuous sampling period reaches the predetermined count, and the minimum comprehensive deviation is not greater than the deviation threshold, then the segment switching time is determined; if the above conditions are not met, the current segment target pressure and fraction collection channel are maintained.

[0029] After obtaining the gas phase temperature sequence and distillation flow rate, for the temperature signal accompanied by high-frequency thermal noise in the industrial field, the smoothing process adopts a moving average filtering algorithm after removing the maximum and minimum values ​​within the sliding window, so as to eliminate the peak noise higher than the preset frequency threshold caused by transient electromagnetic interference and material splashing. Based on the temperature sequence after smoothing filtering, time intervals are extracted. The time interval is at least 50 times the basic sampling time step, and the corresponding temperature state quantities are extracted according to this time interval. The rate of temperature change and acceleration of temperature change are calculated using the following formula: in, The current rate of change of gas phase temperature. The current gas phase temperature. For phase difference The gas phase temperature before the time interval, The acceleration of the gas phase temperature change at the current moment. For phase difference The rate of change of gas phase temperature before the time interval; This avoids the amplification of second derivative noise caused by directly using the high-frequency sampling step size at the bottom layer as the denominator; The system performs a comprehensive deviation assessment of the current distillation platform characteristics to determine the timing of segment switching; among them, the platform slope threshold is used to determine the upper limit of the derivative when the temperature is in a pseudo steady state, and the predetermined count is used to limit the minimum number of sampling points that continuously satisfy the positive acceleration in order to filter out single noise spikes. The positive acceleration is the case where the second derivative of the gas phase temperature is positive and the direction of temperature rise is consistent with that of the next target fraction stage; In engineering applications, the acceleration threshold is set according to the on-site temperature change rate requirements, preferably from 0.01℃ / s² to 0.1℃ / s²; the predetermined count is used to filter out occasional thermal noise fluctuations, preferably set to 3 to 5 consecutive sampling periods; The observation period is preferably a continuous fixed number of sampling period windows, with the length remaining unchanged within the same batch, and this window is aligned with the pressure and flow sampling period to ensure that the counting judgment and the current state assessment use the same time reference; This embodiment distinguishes between pseudo-steady-state platforms and true switching precursors by combining temperature evolution patterns, actual distillation flow rates, and multi-dimensional features of the template library. This improves the accuracy of determining the true physical switching point under noisy conditions, thereby achieving precise linkage of the fraction collection channel, avoiding upstream losses or downstream contamination, and solving the problem of misjudgment of switching under overlapping boiling ranges of multiple components. In practical implementation, since the dimensions and orders of magnitude of physical quantities such as comprehensive pressure, gas phase temperature, rate of change, acceleration of change and distillation flow rate are more than the preset proportional threshold, the system calculates the deviation of each feature and defines it as the absolute difference between the current sampled state value and the target expected value of the corresponding distillation stage parameter. The minimax normalization algorithm is used to uniformly map the deviations of each parameter to... Dimensionless intervals prevent features with values ​​below a preset threshold from being ignored during addition; The historical experience lower and upper bounds of each feature are configured according to the current distillation stage. If the upper and lower bounds of a feature are equal in the current stage, it indicates that the feature has no effective fluctuation information in the current stage. In this case, the single deviation is directly treated as zero or the previous effective sample value is used to avoid the calculation distortion caused by the denominator being zero. Then, according to preset weights, such as a gas phase temperature deviation weight of 0.4 and a pressure deviation weight of 0.3, the remaining weights of other dynamic characteristics, namely the deviations of change rate, change acceleration and distillation flow rate, are equally distributed, each accounting for 0.1. The weighted sum of each normalized individual deviation is calculated to obtain the comprehensive deviation between the current state and the parameters of a specific candidate distillation stage. Since multiple candidate feature templates for the next fraction are usually preset in the process, the system will calculate the comprehensive deviation between the current state and each candidate template in turn to form a set of candidate deviations; the minimum comprehensive deviation is the minimum value taken from this set of deviations, which is used to identify which preset feature of the subsequent component best matches the current fluid state. The preset weights are allocated based on the sensitivity of each feature variable to the critical state of physical phase transition, and remain unchanged within the same batch to ensure comparability between different sampling times. If the minimum overall deviation is not greater than the deviation threshold, it means that a specific fraction stage has been clearly matched. After determining that the overall switching conditions are met, the system loads the target pressure of the next segment of the matched stage and drives the corresponding collection channel switching mechanism to complete the accurate switching of the fraction collection path.

[0030] In one preferred implementation, the feedforward control state table includes multiple states, including the pre-evacuation state. If the peak arrival time is greater than zero and falls within the feedforward trigger window, the Roots pump frequency increment and bypass valve pre-opening degree corresponding to the pressure peak disturbance are determined by looking up the feedforward parameter table, and the Roots pump frequency and bypass valve opening degree are increased after limiting. If the peak arrival time is not greater than zero or does not fall within the feedforward trigger window, pre-evacuation is not performed and the current control state is maintained. When both the overall pressure and the pressure change rate drop below the preset threshold, the Roots pump frequency and bypass valve opening are gradually restored.

[0031] If the predicted peak arrival time is greater than zero and falls within the feedforward trigger window, then the feedforward pre-evacuation state is triggered. The lower boundary of the feedforward trigger window is determined by the superposition of the theoretical frequency rise time of the Roots pump, the transmission delay of the field communication bus, and the dead time of the mechanical action of the actuator, and the upper boundary of the window is determined by combining the preset safety advance time. To avoid ambiguity in the timing definition, in engineering implementation, it is preferable to interpret the peak arrival time as the remaining time from the current sampling time to the predicted peak occurrence time; The pre-evacuation state is allowed only when the remaining time is greater than or equal to the total execution delay and not greater than the sum of the total execution delay and the preset safety advance time; if the remaining time is less than the total execution delay, it means that even if an instruction is issued in this cycle, it is impossible to physically establish a sufficient pumping speed margin before the peak. In this case, feedback control is maintained and no new feedforward increments are added. If the remaining time is greater than the upper boundary of the window, continue to wait for the subsequent sampling period for re-evaluation; this anti-hysteresis timing verification design avoids the conflict between the millisecond-level calculation and command issuance time granularity of the system and the second-to-minute physical inertial response of the field valves and pumps, ensuring that after the pre-evacuation command is issued, the establishment of the real pumping speed margin can strictly precede the gas phase sudden increase disturbance in the physical evolution link to reach the vacuum vessel. In practical implementation, the feedforward parameter table pre-divides the possible range of pressure peak disturbance into several discrete disturbance intervals, each interval uniquely mapping a pair of definite Roots pump frequency increment setpoints and bypass valve pre-opening percentages. The system inputs the predicted peak pressure disturbance into the feedforward parameter table for lookup matching. The initial mapping relationship in the feedforward parameter table is based on the formula: in, The frequency increment setpoint for the Roots pump. This represents the percentage of the bypass valve's pre-opening. This is the median of the disturbance interval. and Based on the frequency conversion coefficient and opening conversion coefficient calibrated by the system's no-load pumping test, the set values ​​of the corresponding range are extracted, and after physical limiting processing of the system's maximum frequency and maximum opening, the frequency of the Roots pump is increased and the opening of the bypass valve is increased, thereby offsetting the upcoming centralized gas release load from the source. To prevent adjacent sampling periods from repeatedly hitting the same disturbance interval and continuously accumulating execution volume, in the field implementation, it is preferable to freeze the current lookup result after entering the pre-extraction state, and only allow recalculation of feedforward settings when the exit condition is met or the predicted peak interval changes across ranges. When the system detects that the liquid phase temperature has completely exceeded the upper limit of the release inflection point temperature zone, and the overall pressure and pressure change rate have both dropped below the preset threshold, the system can gradually restore the Roots pump frequency and bypass valve opening. This exit logic requires that the physical phase change cause be eliminated and the control result be met simultaneously, in order to prevent the feedforward pump from being withdrawn too early during the phase change period, which could cause a secondary pressure rebound oscillation. Preferably, the retreat is performed gradually by adopting a step-by-step retreat or a slope retreat by a fixed time, and the retreat is only started after the comprehensive pressure and pressure change rate have been kept below the threshold for several consecutive sampling cycles, so as to prevent control jitter caused by canceling the pre-extraction in a single retreat.

[0032] As a preferred implementation, it also includes: after the current batch ends, storing the overall pressure, liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening, Roots pump frequency, segment switching position, peak prediction error, pressure steady-state error, and pressure overshoot. The release inflection point temperature band and distillation stage parameters are updated based on the segment switching position using a historical batch weighted average method. The release prediction coefficient, feedforward parameter table, and proportional-integral-derivative parameter set corresponding to each flow regime control interval are updated based on the peak prediction error, pressure steady-state error, and pressure overshoot.

[0033] After the current batch ends, the system records data such as comprehensive pressure, liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening, Roots pump frequency, segment switching position, peak prediction error, pressure steady-state error, and pressure overshoot. It also updates the release inflection point temperature band, distillation stage parameters, release prediction coefficient, feedforward parameter table, and proportional-integral-derivative parameter groups corresponding to each flow control interval using a historical batch weighted average method. Specifically, the historical batch weighted average method employs first-order exponential smoothing filtering logic, meaning the current batch update value equals the current batch measurement value or calculation deviation multiplied by the smoothing coefficient, plus the previous batch record value multiplied by... The smoothing coefficient is set based on historical experience constants, with a preferred value of 0.15. Regarding the update mechanism for feedback control parameters, when the recorded pressure overshoot exceeds the preset upper limit, the system reduces the proportional gain of the corresponding flow control interval according to the preset decay step size and dynamically expands the integral anti-saturation limit. If the recorded pressure overshoot is greater than the preset upper limit, the corresponding integral gain and derivative gain will remain unchanged in this batch and will not participate in the dynamic update decay; if the pressure overshoot is not greater than the preset upper limit, the integral gain and derivative gain will be updated by weighted average of historical batches according to the first-order exponential smoothing filter logic. To reduce the proportional effect while releasing more integral adjustment space for steady-state elimination; for the update of feedforward and prediction parameters, if the peak prediction error is positive, i.e. the predicted pressure peak is less than the peak value, the system adjusts the corresponding gas release prediction coefficient accordingly based on the error deviation ratio. The determined update mapping rule is as follows: The update learning rate is a preset positive real number representing the iteration sensitivity, typically taken as 0.1 to 0.3 to ensure convergence stability. Conversely, if the peak prediction error is negative, meaning the predicted pressure peak is greater than the peak value leading to excessive pre-vacuuming, a callback degradation branch is triggered, and the determined mapping rule is as follows: The attenuation multiplier is a damping coefficient to prevent excessive backlash, and is preferably set to 0.5. Furthermore, for the update of the feedforward parameter table, if the pressure peak still exceeds the upper limit of the allowable deviation, it indicates that the feedforward force is insufficient. The system will then multiply the Roots pump frequency increment and bypass valve pre-opening degree of the corresponding interval in the feedforward parameter table by a proportional coefficient greater than 1, such as 1.1, to increase the pre-pumping margin of the next batch. For updating the temperature band of the gas release inflection point and the parameters of the distillation stage, the system extracts the real gas phase and liquid phase temperatures recorded at the current batch segment switching time as measured values, and then weights and fuses them with historical parameters according to the first-order exponential smoothing filtering logic mentioned above, so as to realize the automatic convergence of the temperature reference of each control node to the real drift characteristics on site. The aforementioned update mechanism enables the system to continuously correct its threshold deviation as the number of running batches increases, ensuring batch consistency and control stability during long-term operation. By continuously accumulating and utilizing historical operating results to clearly define the data closed-loop update rules, the system can gradually improve the accuracy of gas release peak prediction, pressure steady-state performance, and the effect of feedforward and feedback coordinated control, thereby forming a self-optimizing closed loop for industrial production.

[0034] Please see Figure 2 A vacuum degree segmented gradient control system for a vacuum distillation system, the system comprising: The initialization module is used to load the gas release inflection point temperature band, distillation stage parameters, flow control interval boundaries, control parameter table, feedforward parameter table, precursor threshold, historical batch delay data, and gas release prediction coefficient. The data acquisition and calculation module is used to acquire the pressure readings inside the vacuum distillation vessel from the capacitive thin-film vacuum gauge and the Pirani vacuum gauge, as well as the liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening degree and Roots pump frequency, and to calculate the comprehensive pressure, pressure change rate and Knudsen number to form the sampling data. The gradient generation module includes a scoring unit, a prediction unit, a target pressure generation unit, and a state table generation unit. It is used to judge the gas surge trend based on the combined scoring of pressure change rate, liquid phase temperature, and execution load, predict the pressure peak disturbance and peak arrival time by combining the gas release prediction coefficient, generate the target pressure of each segment according to the basic pressure reduction gradient and compensation pressure reduction in the control parameter table, and generate the stage control instruction table and the feedforward control state table including the pre-extraction state. The segmented execution module is used to divide the flow control interval according to the comprehensive pressure and Knudsen number, read the corresponding proportional, integral and differential parameter group for pressure feedback control, and perform pre-evacuation when the peak arrival time falls into the feedforward trigger window determined by the Roots pump frequency rise time and the preset safety advance time. The switching control module is used to calculate the rate of change and acceleration of the gas phase temperature, as well as the deviation between the current state and the parameters of the distillation stage. Based on the rate of change, acceleration, deviation, and distillation flow rate, it determines the timing of segment switching, applies the target pressure for the next segment, and controls the switching mechanism to switch the corresponding fraction collection channel.

[0035] This embodiment provides a control system that serves as the hardware and logic carrier for the above-mentioned vacuum degree segmented gradient control method of vacuum distillation system. It achieves the coordinated operation of complex control decisions and real-time execution on site through a modular architecture. The system specifically includes an initialization module, a data acquisition and calculation module, a gradient generation module, a segmented execution module, and a switching control module; In industrial deployments, the initialization module and gradient generation module can be deployed on an edge computing server, responsible for loading historical batch data, calculating and evaluating multi-dimensional features, and generating target control forms; the acquisition and computing module connects to various sensors via a fieldbus to acquire real-time data from multiple sources. The scoring and prediction units within the gradient generation module predict the peak gas release in advance by evaluating the coupling relationship between the rate of pressure change and temperature. The target pressure generation unit and the state table generation unit transform continuous nonlinear decision results into structured stage control instruction tables and feedforward control state tables. This discretization and form-based processing of complex decision results enables the segmented execution module and switching control module to be directly deployed in the field programmable logic controller. The field programmable logic controller (FPGA) only needs to execute pressure domain adaptive parameter lookup calls, feedforward pre-evacuation actions, and physical switching of collection channels according to the issued control instruction table, thereby ensuring the real-time response capability and industrial operation reliability of the control system without increasing the real-time computing burden of the underlying controller.

[0036] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for segmented gradient control of vacuum degree in a vacuum distillation system, characterized in that, include: Load the release inflection point temperature band, distillation stage parameters, flow control interval boundaries, control parameter table, feedforward parameter table, precursor threshold, historical batch delay data, and release prediction coefficient; The pressure readings inside the vacuum distillation vessel by the capacitive thin-film vacuum gauge and the Pirani vacuum gauge, as well as the liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening degree and Roots pump frequency, are obtained. The combined pressure, pressure change rate and Knudsen number are calculated to form the sampling data. Based on the combined score of the pressure change rate, liquid phase temperature and the execution load determined by the bypass valve opening and the Roots pump frequency, the gas surge trend is judged. Combined with the gas release prediction coefficient, the pressure peak disturbance and peak arrival time are predicted. The target pressure of each segment is generated segment by segment according to the basic pressure drop gradient and compensation pressure drop in the control parameter table. A stage control instruction table and a feedforward control state table including the pre-extraction state are generated. The flow control interval is divided according to the comprehensive pressure and Knudsen number. The proportional, integral and differential parameter group of the corresponding flow control interval is read to perform pressure feedback control. Pre-evacuation is performed when the peak arrival time falls into the feedforward trigger window determined by the Roots pump frequency rise time and the preset safety advance time. Calculate the rate of change and acceleration of the gas phase temperature, as well as the deviation between the current state and the distillation stage parameters. Determine the segment switching timing based on the rate of change, acceleration, deviation, and distillation flow rate, apply the target pressure for the next segment, and switch the fraction collection channel.

2. The vacuum degree segmented gradient control method for a vacuum distillation system according to claim 1, characterized in that, The overall pressure is determined based on the readings of the capacitive thin-film vacuum gauge and the Pirani vacuum gauge according to a preset weighted relationship, and is calculated according to the following formula: in, To address the combined pressures, For the reading of a capacitive thin-film vacuum gauge, For the Pirani vacuum gauge reading; when hour, ;when hour, ; when hour, ; This is the low-voltage switching threshold. The high-pressure switching threshold is defined as follows: the pressure change rate is the ratio of the difference between the filtered composite pressure and the predetermined differential time span set according to the system sampling period to that time span; the Knudsen number is calculated according to the following formula: in, For calibration coefficients, The absolute temperature of the gas phase. The hydraulic diameter of the vacuum distillation vessel is denoted as .

3. The vacuum degree segmented gradient control method for a vacuum distillation system according to claim 1, characterized in that, The combined score includes: The pressure rise score is determined based on the continuous increase in the pressure change rate; the temperature score is determined based on whether the liquid phase temperature enters the gas release inflection point temperature zone; and the execution load score is determined based on the bypass valve opening and the Roots pump frequency. The gas surge index is obtained by weighting the pressure rise score, temperature score, and execution load score. If the gas surge index reaches the precursor threshold, the historical batch delay data is corrected in conjunction with the current average pressure change rate to extract the precursor-to-peak delay, thereby obtaining the peak arrival time, and the pressure peak disturbance is determined based on the average pressure change rate and liquid phase temperature deviation; if the gas surge index does not reach the precursor threshold, the current segmented target pressure is maintained.

4. The vacuum degree segmented gradient control method for a vacuum distillation system according to claim 1, characterized in that, The control parameter table includes the basic bucking gradient. and compensation for pressure drop The target pressure of adjacent segments is determined based on the base pressure reduction gradient and the compensation pressure reduction, and is calculated according to the following formula: in, For the first Segmented target pressure, This is an overlap correction factor determined based on the duration of the current distillation platform and the deviation of the distillation stage parameters. The value increases with the duration of the distillation platform; the stage control instruction table includes the target pressure, entry conditions, minimum holding time, exit conditions, and fraction collection channel number for each segment.

5. The vacuum degree segmented gradient control method for a vacuum distillation system according to claim 1, characterized in that, The boundary of the flow control zone includes a first flow boundary and a second flow boundary; The flow control region includes a viscous flow region where the Knudsen number is less than the first flow boundary, a transition flow region where the Knudsen number is located between the first flow boundary and the second flow boundary, and a molecular flow region where the Knudsen number is not less than the second flow boundary. Each flow control interval corresponds to a set of proportional gain, integral gain, derivative gain, output limiting, and integral limiting. The corresponding proportional, integral, and derivative parameter set is read according to the current flow control interval, and the integral term and control output are limited. A switching hysteresis band is set between adjacent flow control intervals.

6. The method for segmented gradient control of vacuum degree in a vacuum distillation system according to claim 1, characterized in that, The control parameter table also includes a platform slope threshold, minimum effective distillate flow rate, acceleration threshold, predetermined count, and deviation threshold; determining the segment switching timing includes: The gas phase temperature is smoothed, and the rate of change and the acceleration of change are calculated. If the absolute value of the rate of change is not greater than the plateau slope threshold and the distillation flow rate is not less than the minimum effective distillation flow rate, then it is determined that the distillation platform is in operation; if the above conditions are not met simultaneously, then it is determined that the distillation platform is not in operation. Within the distillation platform, a weighted deviation assessment is performed between the current state, which consists of the combined pressure, vapor temperature, the rate of change, the acceleration of change, and the distillation flow rate, and the parameters of each distillation stage to obtain the combined deviation. If the change in acceleration reaches the acceleration threshold, the count of positive acceleration within the observation period consisting of a predetermined continuous sampling period reaches the predetermined count, and the minimum comprehensive deviation is not greater than the deviation threshold, then it is determined that the segment switching time has been reached; if the above conditions are not met, the current segment target pressure and fraction collection channel are maintained.

7. The method for segmented gradient control of vacuum degree in a vacuum distillation system according to claim 1, characterized in that, The feedforward control status table includes multiple states, including the pre-evacuation status. If the peak arrival time is greater than zero and falls within the feedforward trigger window, the Roots pump frequency increment and bypass valve pre-opening degree corresponding to the pressure peak disturbance are determined by looking up the feedforward parameter table, and the Roots pump frequency and bypass valve opening degree are increased after limiting; if the peak arrival time is not greater than zero or does not fall within the feedforward trigger window, pre-evacuation is not performed and the current control state is maintained. When both the overall pressure and the pressure change rate drop below the preset threshold, the Roots pump frequency and bypass valve opening are gradually restored.

8. The method for segmented gradient control of vacuum degree in a vacuum distillation system according to claim 1, characterized in that, Also includes: After the current batch ends, save the overall pressure, liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening, Roots pump frequency, segment switching position, peak prediction error, pressure steady-state error, and pressure overshoot. The gas release inflection point temperature band and distillation stage parameters are updated according to the segment switching position using a historical batch weighted average method. The gas release prediction coefficient, feedforward parameter table, and proportional-integral-derivative parameter group corresponding to each flow control interval are updated according to the peak prediction error, pressure steady-state error, and pressure overshoot.

9. A segmented gradient control system for vacuum distillation system, applied to the segmented gradient control method for vacuum distillation system according to any one of claims 1-8, characterized in that, The system includes: The initialization module is used to load the gas release inflection point temperature band, distillation stage parameters, flow control interval boundaries, control parameter table, feedforward parameter table, precursor threshold, historical batch delay data, and gas release prediction coefficient. The data acquisition and calculation module is used to acquire the pressure readings inside the vacuum distillation vessel from the capacitive thin-film vacuum gauge and the Pirani vacuum gauge, as well as the liquid phase temperature, gas phase temperature, distillation flow rate, bypass valve opening degree and Roots pump frequency, and to calculate the comprehensive pressure, pressure change rate and Knudsen number to form the sampling data. The gradient generation module includes a scoring unit, a prediction unit, a target pressure generation unit, and a state table generation unit. It is used to determine the gas surge trend based on the combined scoring of the pressure change rate, liquid phase temperature, and execution load, predict the pressure peak disturbance and peak arrival time in combination with the gas release prediction coefficient, generate the target pressure of each segment segment according to the basic pressure reduction gradient and compensation pressure reduction in the control parameter table, and generate a stage control instruction table and a feedforward control state table including the pre-extraction state. The segmented execution module is used to divide the flow control interval according to the comprehensive pressure and Knudsen number, read the corresponding proportional, integral and differential parameter group for pressure feedback control, and perform pre-evacuation when the peak arrival time falls into the feedforward trigger window determined by the Roots pump frequency rise time and the preset safety advance time. The switching control module is used to calculate the rate of change, acceleration of change, and deviation of the current state from the distillation stage parameters of the gas phase temperature. Based on the rate of change, acceleration of change, deviation, and distillation flow rate, it determines the timing of segment switching, applies the target pressure of the next segment, and controls the switching mechanism to switch the corresponding fraction collection channel.

Images