De-slicing melt viscosity stability control method for polyester device

Abstract

The invention discloses a polyester device de-slicing melt viscosity stability control method, and relates to the technical field of production control, and the method comprises the following steps: S1, obtaining a plate repair operation plan and historical operation data, constructing a plate repair density map based on time distribution and frequency distribution of spinning assemblies, and obtaining a plate repair density map; combining the liquid level of the final polymerization reaction kettle, the melt retention time and the melt viscosity to establish a sensitivity matrix, dividing operation risk grades, and determining a high-fluctuation risk operation window. According to the method, a high-risk operation window is identified through a repair plate density map and a viscosity sensitivity matrix, disturbance trend pre-judgment is realized by combining continuous monitoring of key process parameters, and a whole-process closed-loop control mechanism is established by adopting feedforward control, vacuum adjustment and viscosity springback suppression means; the viscosity stability and the anti-disturbance capability of a polyester device in a de-slicing production mode are effectively improved, and the stable operation of a spinning end and the consistency of product quality are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of production control technology, specifically to a method for stabilizing the viscosity of melt during the deslicing process in polyester plants. Background Technology

[0002] The stable control of melt viscosity in polyester units without chip transfer refers to a dynamic control method established in continuous polyester production where the traditional chip transfer process is eliminated and the polyester melt is directly delivered to the spinning end. This method addresses the issue of fluctuations in melt flow rate and residence time caused by operations such as chip repair stops, metering pump start-ups and shutdowns, and melt reflux, which in turn lead to changes in final polymerization reaction conditions and fluctuations in melt viscosity. The core of this method is a dynamic control approach that includes measures such as staggered chip repair, stabilizing the load difference between prepolymerization and final polymerization pumps, rapid and slight adjustment of vacuum, linkage correction between filter pressure difference and stirring current, and selective filter switching. This method monitors key viscosity-influencing factors such as acid value, residence time, reaction temperature, vacuum, stirring speed, and catalyst dosage. When the online viscometer is unstable, alternative monitoring parameters such as stirring current, filter pressure difference, and pre-spinning component pressure are introduced to quickly identify and accurately correct viscosity disturbances. This ensures continuous and stable melt viscosity in an operating environment with full-supply spinning and frequent partial stops, guaranteeing consistent product quality and stable unit operation.

[0003] The existing technology has the following shortcomings:

[0004] In existing technologies, polyester plants, under continuous production conditions involving chip removal, typically rely on periodic maintenance of spinning components through plate trimming to maintain stable operation at the spinning end. However, when multiple short fiber lines undergo plate trimming simultaneously within the same time window, and the trimming duration exceeds the predetermined cycle, the liquid level in the final polymerization reactor experiences an abnormal rise. This passively extends the actual residence time of the melt in the final polymerization reactor beyond the process-set upper limit, causing excessive polycondensation and resulting in a rapid surge in melt viscosity within a short period. Subsequently, after the plate trimming is completed and the multi-metering pumps resume operation, the pump load rapidly recovers within a very short time, drastically shortening the melt residence time and causing a rapid decrease in viscosity, forming a bidirectional spike disturbance characterized by short-cycle, high-amplitude spikes. Existing technologies lack effective early identification mechanisms and dynamic buffering measures for this type of disturbance. Online viscosity control strategies often fail due to response lag, easily leading to a momentary imbalance in the melt rheological properties at the spinning end, causing a concentrated outbreak of filament breakage, and even forcing the entire production line to reduce load or shut down, severely impacting the stability of production capacity and product quality.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a method for stabilizing the viscosity of melt during the deslicing process in polyester plants, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for stabilizing the viscosity of melt in a polyester descaling unit, comprising the following steps:

[0008] S1. Obtain the plate repair operation plan and historical operation data. Construct a plate repair density map based on the time distribution and frequency distribution of the spinning components. Combine the liquid level of the final polymerization reactor, melt residence time and melt viscosity to establish a sensitivity matrix, classify the operation risk level, and determine the high fluctuation risk operation window.

[0009] S2, within the high-fluctuation risk operation window, activate the continuous monitoring mechanism to collect the slope of the change in the current of the stirrer motor of the final polymerization reactor, the rate of change in the differential pressure of the melt filter, and the pressure drift amplitude of the pre-spinning component. When any parameter exceeds the preset threshold and continues to exceed the set time, a dual threshold alarm is triggered and the trigger time is recorded.

[0010] S3, after the alarm is triggered, feedforward control is implemented to set the output difference between the metering pumps of the prepolymerization section and the final polymerization section to a fixed value, and the feed rate of the prepolymerization section is adjusted according to the plate density spectrum to maintain the stability of the liquid level in the final polymerization reactor.

[0011] S4. Within the preset time window after the plate repair is completed, the vacuum degree is adjusted in steps of ±1 to ±2 Pa. The vacuum degree reset point is determined by combining the melt residence time correction value and the slope of the final polymerization stirrer current change, so as to achieve rapid recovery of residence time.

[0012] S5. After the vacuum degree is reset, establish the mapping relationship between the output and the filter pressure difference. Calculate the proportional coefficient based on the output decrease and the pressure difference decrease. Adjust the target residence time of the final polymerization stage according to the proportional coefficient to reduce the viscosity rebound amplitude.

[0013] S6 performs shift-by-shift retrospective analysis on the time series data of three monitoring indicators within the high-risk operation window, the vacuum degree adjustment sequence, the production and pressure difference ratio coefficient, the final polymerization residence time correction value, and their corresponding viscosity response. Based on the analysis results, it dynamically updates the plate density map, feedforward control parameters, and alarm thresholds to form an adaptive closed-loop control system for viscosity fluctuations.

[0014] Preferably, step S1 includes:

[0015] Acquire and organize the plate repair operation plan and historical operation fluctuation data, extract plate repair records for no less than 30 consecutive days, and data fields include plate repair start time, plate repair end time, spinning position number, duration of each plate repair, number of digits involved in each plate repair, and start / stop status of the corresponding metering pump during each plate repair. At the same time, export four operational data items: final polymerization reactor liquid level, final polymerization melt pump speed, melt viscosity, and pelletizer start / stop status. Match all data with a unified timestamp at a time granularity of 5 minutes to form a complete historical operation time series.

[0016] The 24 hours of each day are divided into 48 fixed half-hour periods. The number of repairs, the average repair duration, and the average number of simultaneous repairs are statistically analyzed in each period to generate a repair density map and identify high-frequency repair segments and strong disturbance segments.

[0017] High-density time periods are extracted from the plate density spectrum. Combined with the liquid level fluctuation amplitude, pump speed change rate and viscosity change value of the corresponding time periods, a multi-parameter linkage viscosity sensitivity matrix is ​​constructed. The normalized sensitivity value is then used to quantify the viscosity disturbance intensity.

[0018] High-risk operating periods are defined as those with sensitivity values ​​higher than the preset quantile and corresponding time periods that are high-frequency or strongly disturbed in the plate density spectrum. A risk time distribution table containing risk level, liquid level fluctuation, pump speed change and viscosity change is output as the basis for triggering subsequent monitoring and control logic.

[0019] Preferably, step S2 includes:

[0020] Five minutes before the start of the high volatility risk operation window, the continuous monitoring process is initiated, and three process parameters to be monitored are determined: the motor current of the agitator in the final polymerization reactor, the differential pressure of the melt filter, and the pressure of the pre-spinning component. All three data are continuously recorded with a sampling period of 5 seconds and are aligned with a unified timestamp.

[0021] For each set of sampled data, add historical feature information of plate repair behavior corresponding to the current high volatility risk operation window. The added content includes: the average plate repair duration in the corresponding time period, the maximum number of plates repaired simultaneously in actual statistics, and the maximum fluctuation amplitude of the liquid level in the final polymerization reactor caused by the last plate repair operation. The above three parameters are used to characterize the disturbance background features in the current time period and assist in subsequent trend identification.

[0022] The judgment criteria are set as follows: the slope of the stirrer current change is ±0.25 amperes per minute, the rate of pressure difference change is ±15 kPa per minute, and the pressure drift threshold of the front spinning component is ±0.08 MPa. Combined with the duration requirement, a dual threshold judgment logic is formed.

[0023] If any parameter continuously exceeds the above threshold and continues to exceed the set time, an alarm event is triggered and the trigger time, direction of change, cumulative change value and alarm type are recorded, providing a trigger basis for subsequent vacuum adjustment and control response.

[0024] Preferably, step S3 includes:

[0025] After a viscosity disturbance alarm occurs, the output difference between the prepolymerization section metering pump and the final polymerization section metering pump is set to 3.2 tons per hour, which is used as the control benchmark value for stabilizing the liquid level in the final polymerization reactor. The fixed output difference adjustment strategy is then immediately implemented.

[0026] Based on the time period records corresponding to the high-fluctuation risk operation window in the plate repair density spectrum, the average number of plates repaired, the duration of a single plate repair operation, and the maximum fluctuation amplitude in the historical liquid level fluctuation curve are extracted within the corresponding time period. When the average number of plates repaired exceeds 10, the duration of a single plate repair exceeds 20 minutes, and the liquid level fluctuation amplitude of the final polymerization reactor is greater than ±35 mm, it is determined that there is a significant disturbance intensity. Subsequently, the speed of the metering pump in the prepolymerization section is increased by 4% on the basic set value to build an upstream feeding buffer before the output of the final polymerization section has fully recovered, enhance the stability of liquid level control, and lay a feedforward guarantee for the realization of continuous melt delivery.

[0027] The liquid level data of the final polymerization reactor is continuously collected and recorded every 10 seconds. During the 5-minute adjustment period, the liquid level fluctuation is observed to ensure that it remains within ±20 mm. The prepolymerization pump speed is then finely adjusted in real time according to the liquid level change trend to achieve dynamic optimization of the pump differential control status.

[0028] When the liquid level fluctuates within ±25 mm for 15 consecutive minutes without any new alarms, record the metering pump speed, instantaneous output difference, stable liquid level value, and plate repair behavior summary information between the pre-polymerization section and the final polymerization section. Use the current operating status as the initial working condition boundary for vacuum adjustment to ensure that subsequent control strategies have a stable material basis.

[0029] Preferably, step S4 includes:

[0030] After all spinning plate repair operations are completed, the metering pumps are restored to operation, and the melt delivery is restored to normal, this moment is set as the starting time point. A 5-minute preparation time window is started. During this time window, the liquid level of the final polymerization reactor, the pump speed of the final polymerization metering pump, and the motor current of the final polymerization agitator are continuously collected. It is determined whether the liquid level fluctuation is less than or equal to ±25 mm, the pump speed fluctuation is less than or equal to ±3%, and the current change rate is less than or equal to 0.05 amperes per minute to confirm whether the working condition has entered a stable state.

[0031] After the three process parameters of liquid level, pump speed and current are all within a stable range, perform a small step-by-step adjustment of vacuum degree. Each adjustment of vacuum degree is in units of ±1 Pa to ±2 Pa. After each adjustment, the vacuum degree is kept unchanged for no less than 2 minutes. At the same time, the trend of liquid level change and the slope of stirrer current change are collected. The cumulative adjustment range does not exceed ±6 Pa.

[0032] The instantaneous residence time of the melt in the final polymerization reactor is calculated based on the liquid level and pump speed values. It is compared with the set target value. Combined with the judgment that the agitator current slope is kept within ±0.03 amperes per minute for a continuous 5-minute period, it is determined whether the vacuum adjustment has met the reset condition. If the condition is met, the current vacuum value is locked.

[0033] After vacuum conditioning is completed, the fluctuations in liquid level, viscosity, stirrer current, and pump speed are monitored. When the liquid level is continuously maintained within ±20 mm for 10 minutes, the measured viscosity value is stable, the stirrer current fluctuation does not exceed ±0.02 amperes, and the pump speed is maintained within ±2% of a fixed value, the melt rheological state is considered to have been fully restored. The current operating data is recorded as the initial reference for subsequent control steps. At the same time, the records of each vacuum adjustment, response time, and recovery cycle are saved as optimization basis.

[0034] Preferably, step S5 includes:

[0035] The actual output data of the metering pump in the final polymerization stage and the pressure difference data of the inlet and outlet of the melt filter were collected simultaneously. The collection time period was set to 10 consecutive minutes after the vacuum degree was restored, and the sampling frequency was once every 5 seconds. The collected data were smoothed and filtered before being used as the analysis sample.

[0036] The stable output value and corresponding pressure difference value before plate repair are used as reference benchmarks and compared with the lowest output value and pressure difference value in the current operating stage. The output reduction and pressure difference reduction are calculated respectively, and the ratio coefficient between the two is calculated.

[0037] If the ratio coefficient increases by more than 10% compared to the benchmark value of the previous operating cycle, the target dwell time of the final polymerization stage will be reduced by 5% to 10% based on the standard value. If the ratio coefficient is lower than the benchmark value of the previous cycle, the target dwell time will be increased by 3% to 5%. The adjustment cycle is once every 2 minutes, and the adjustment range each time shall not exceed 3%.

[0038] The corrected target residence time was applied to the feed rate control of the metering pump in the final polymerization stage, and the stability of the agitator current and the viscosity change trend were continuously monitored to determine whether the correction was effective. The proportional coefficient and correction results were recorded for subsequent optimization analysis.

[0039] Preferably, step S6 includes:

[0040] Using one operating shift as a cycle, extract all control parameter data within the high-fluctuation risk operating window during continuous operation, and establish time series records at 5-second intervals. The parameters include the trend of agitator current change, the rate of change of melt filter differential pressure, the amplitude of melt pressure drift, vacuum regulation history, metering pump output difference, and viscosity change corresponding to the target residence time correction value.

[0041] The collected data were divided into three levels according to the maximum viscosity fluctuation range: mild, moderate and severe disturbance. The corresponding parameters were classified and modeled to determine their correlation with viscosity recovery time and viscosity stability value.

[0042] Based on the analysis results, the marked sections of the plate density map are dynamically updated, the feedforward production difference control range and the monitoring alarm trigger threshold are corrected, and the original control parameters are automatically replaced in the next shift operation cycle.

[0043] The updated control settings are used as the starting conditions for the viscosity disturbance response in the next operating cycle, and the acquisition and optimization operations continue to be performed to achieve closed-loop control.

[0044] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0045] This invention, by constructing a plate repair density map based on the timing pattern of plate repair operations and historical operational fluctuation data, combined with a sensitivity matrix of final polymerization reactor liquid level, melt residence time, and viscosity changes, achieves for the first time quantitative identification of high-risk operating windows. Furthermore, by continuously monitoring key process parameters (including agitator current slope, filter differential pressure rate, and pre-spinning component pressure drift), a disturbance trend perception model is built to trigger control responses in advance. Based on this, a closed-loop control system for viscosity disturbances is formed through feedforward setting of the pre-polymerization and final polymerization yield difference, step-by-step vacuum adjustment, viscosity rebound suppression mechanisms, and adaptive updates of control parameters. Compared with traditional fixed threshold hysteresis adjustment methods, this method significantly improves the adaptability and disturbance resistance of the equipment to complex fluctuating conditions, ensuring high stability of polyester melt viscosity even under frequent plate repairs and multi-line load variations. This guarantees continuous and stable rheological properties at the spinning end, reduces filament breakage, and improves product consistency and production line stability. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0047] Figure 1This is a flowchart of the method for controlling the viscosity stability of melt during the deslicing process in the polyester apparatus according to the present invention. Detailed Implementation

[0048] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0049] This invention provides, for example Figure 1 The method for stabilizing the melt viscosity of the polyester deslicing unit shown includes the following steps:

[0050] Acquire the plate repair operation plan and historical operation fluctuation data, and generate a plate repair density map based on the time distribution and frequency distribution of the spinning component plate repair; establish a sensitivity matrix by combining the liquid level of the final polymerization reactor, the residence time of the melt in the final polymerization reactor and the melt viscosity; classify the risk level of the operation status in different time periods and identify the operation windows with high fluctuation risk.

[0051] To identify high-risk operating periods during the continuous production of polyester units in the dechipping process, the following steps were used to construct a plate density map and establish a viscosity sensitivity matrix based on actual operating variables, thereby classifying the risk level.

[0052] Acquire and organize the plate repair operation plan and corresponding historical operational data. Specifically, extract plate repair operation plans for no less than 30 consecutive days from the shift logs of the polyester staple fiber spinning workshop, ensuring that the data covers three shift cycles: day shift, afternoon shift, and night shift. Data fields include: plate repair start time, plate repair end time, spinning station number involved in plate repair, number of plates involved in a single plate repair, and whether the metering pump corresponding to each plate repair is shut down. At the same time, export historical operational process data that is completely aligned with the plate repair plan time points from the DCS platform in the polyester main control room, including the liquid level curve of the final polymerization reactor (in mm), the speed curve of the final polymerization melt pump (in rpm), the online measurement value of polyester melt viscosity (in Pa·s), and the start / stop status of the pelletizer every hour. This data is organized into a time series in a uniform format with a 5-minute time granularity and synchronized using timestamps. Through this step, the correspondence between plate repair activities and key process parameters can be comprehensively obtained, providing a foundation for the next step of constructing a data map.

[0053] A table of time and frequency distribution of patch repair behavior was constructed, and a patch repair density map was generated. In practice, each 24-hour day was divided into 48 half-hour periods, which were used as the horizontal axis to classify and statistically analyze all patch repair records. The vertical axis consisted of two dimensions: the number of patch repairs and the number of simultaneous patch repairs. For each time period, the total number of patch repairs, the average duration (in minutes), and the average number of simultaneous patch repairs were calculated for the entire data period, and these were marked using color depth or numerical value to form a two-dimensional heatmap-like patch repair density map. Specifically, if the frequency of patch repairs in a certain time period exceeded the 80th percentile of the total sample, it was marked as a high-frequency segment; if the average patch repair duration exceeded 25 minutes and the number of simultaneous patch repairs exceeded 6, it was marked as a strong disturbance segment. In this way, the map can clearly reflect the concentrated areas of patch repair behavior each day, laying the foundation for identifying potentially high-risk periods.

[0054] A multi-parameter linkage viscosity sensitivity matrix was established. Combining the high-density sections of the repair plate identified in step two, data on the liquid level changes in the final polymerization reactor, the rotational speed changes of the final polymerization melt pump, and the viscosity changes of the polyester melt were extracted from synchronized historical data corresponding to these sections, constructing a three-dimensional parameter set in units of "time periods." For each time period, the fluctuation amplitude of the liquid level in the final polymerization reactor (difference between the maximum and minimum values, in mm), the rate of change of the rotational speed of the final polymerization melt pump within that time period (rpm / min), and the maximum viscosity change within that time period (in Pa·s) were calculated. These three data points were normalized and summarized into a sensitivity value; a higher sensitivity value indicates a stronger perturbation of viscosity within that time period. For example, if the liquid level fluctuation is greater than 50 mm, the pump speed change exceeds 15 rpm, and the viscosity change is greater than 0.08 Pa·s within a certain time period, it is identified as a high-sensitivity section. This sensitivity matrix covers all 48 time periods and can be used for horizontal comparison of the viscosity fluctuation sensitivity across different time periods.

[0055] Based on the intersection of the viscosity sensitivity matrix and the plate density map, operational risk levels are classified, and high-fluctuation risk windows are identified. The specific classification principles are as follows: Time periods with sensitivity values ​​higher than a set percentile (e.g., the 85th percentile) and falling within a high-frequency or strongly disturbed segment of the plate density map are classified as high-risk operational periods; sensitivity values ​​between the 60th and 85th percentiles and plate density at a medium frequency are classified as medium-risk operational periods; the remaining periods are low-risk operational periods. The final output is an operational risk time distribution table arranged chronologically. This table marks each daily time period by risk level as red (high risk), yellow (medium risk), and green (low risk). Each risk interval is accompanied by corresponding actual reference data such as liquid level fluctuations, pump speed changes, and viscosity changes, which can be directly used for trigger logic judgments and control decisions in subsequent control steps. Unlike existing technologies that rely solely on human experience to assess risks or use fixed schedules to predict risks, this step enables dynamic risk identification based on operational data. It can update the risk distribution in real time as the plate repair behavior and production status change, thereby improving the pertinence and foresight of polyester melt viscosity control.

[0056] The purpose of this step is to lay a data-driven risk identification foundation for achieving stable melt viscosity control in polyester plants under continuous production conditions without chipping. Since the chip transfer link is eliminated, the polyester melt must be continuously and stably supplied to the spinning end. However, the trimming operation during spinning inevitably leads to partial shutdowns, metering pump starts and stops, melt backflow, and other operational behaviors, resulting in melt flow fluctuations, abnormal rises and falls in the final polymerization reactor level, and unstable residence time, ultimately causing sudden changes in viscosity. However, these disturbances exhibit clear temporal distribution characteristics and regularities, and are not entirely random. Therefore, this step collects and organizes trimming plans and historical operating data to construct a trimming density map reflecting the concentration of trimming behavior over time. Furthermore, by combining the final polymerization reactor level, melt residence time, and viscosity changes to establish a multidimensional sensitivity matrix, the impact of different operating conditions on viscosity fluctuations at different times can be quantified. Based on the matrix results, further operational risk level classification is performed, clearly defining time periods with potential severe disturbance risks as "high volatility risk windows." This provides a precise basis for determining the timing of subsequent monitoring mechanisms, the implementation of pre-control strategies, and vacuum adjustment operations. This step significantly differs from existing technologies that rely on experience-based judgment or static timetables, achieving a shift from passive response to proactive identification. It enhances the system's ability to perceive the sources of viscosity fluctuations and is a crucial preliminary step in the entire dynamic control strategy.

[0057] Within the high-fluctuation risk operating window, a continuous monitoring mechanism is established and activated to collect the slope of the change in the motor current of the agitator in the final polymerization reactor, the rate of change in the differential pressure of the melt filter, and the pressure drift amplitude of the pre-spinning component. When the rate of change of any parameter exceeds the preset threshold and the duration exceeds the corresponding set time, a dual threshold alarm is triggered and the alarm trigger time is recorded.

[0058] To enable rapid identification and trend assessment of melt viscosity disturbances in polyester units within high-risk operating windows, after identifying the high-risk operating window, continuous data collection and trend analysis are performed on key process parameters affecting viscosity stability. A dual-threshold alarm triggering mechanism based on numerical thresholds and duration-based determination is established. The specific operating procedure is as follows:

[0059] The process parameters for continuous monitoring were clearly defined, and the physical sampling paths and frequencies were set. Combining the modified plate density map and viscosity sensitivity matrix constructed in the previous steps, three physical process parameters closely related to viscosity changes were selected as monitoring targets within the time period identified as a high-risk operating window: the current value of the agitator motor in the final polymerization reactor (in amperes), the pressure difference between the front and rear ends of the melt filter (in kilopascals), and the melt pressure at the inlet of the pre-spinning assembly (in megapascals). All three data points were obtained through local measurement points in existing industrial process control equipment. The agitator current was collected through the feedback value from the frequency converter, the pressure difference was obtained through a differential pressure sensor, and the pre-spinning assembly pressure was collected through a pressure gauge located at the outlet of the metering pump. All parameters were continuously recorded with a sampling period of 5 seconds and time-aligned according to a unified timestamp format to ensure consistency across parameters in the time dimension. Unlike the traditional approach of monitoring only a single melt viscosity index, this implementation method uses three physical variables with fast dynamic response and high data stability as early warning indicators of viscosity anomalies.

[0060] Parameter acquisition is initiated 5 minutes before the start of a high-risk fluctuation period, and the corresponding plate-repairing behavior characteristics are read simultaneously. When the acquisition time approaches the start of a high-risk operating window (e.g., 08:00 to 10:00 daily), the monitoring logic is activated 5 minutes in advance, and historical plate-repairing behavior data for that period is extracted, including the average plate-repairing duration, the maximum number of plates repaired simultaneously, and the level fluctuation amplitude caused by the last plate-repair. This data will be added to each parameter acquisition record as background characteristics for that period, ensuring that each set of samples includes not only the current process status but also the corresponding historical disturbance behavior background, which helps in subsequent trend identification. After the monitoring logic is activated, the above three physical parameters are acquired every 5 seconds, while retaining the sampling history within a sliding window of at least 10 minutes to determine the trend of variable changes.

[0061] Set thresholds for the rate of change and duration of each parameter, and establish a dual-judgment mechanism. To ensure the accuracy of anomaly identification, it is necessary not only to determine whether a single sampled value exceeds the set range, but also to determine whether the change has a continuous trend. Therefore, the following judgment rules are set for the three parameters respectively:

[0062] The slope threshold for the change in the agitator motor current of the final polymerization reactor is set to ±0.25 amperes per minute, calculated as the difference between the current sampling point and the previous sampling point divided by the sampling interval. If three consecutive sampling points show the same direction of change and the cumulative change exceeds 0.1 amperes, it is marked as an upward or downward trend.

[0063] The threshold for the rate of change of differential pressure in the melt filter is set to ±15 kPa per minute. If the differential pressure changes by more than 10 kPa and does not recover within four consecutive sampling cycles, it is judged as amplified differential pressure fluctuation.

[0064] The drift threshold of the pressure of the pre-spinning component is set at ±0.08 MPa. If the pressure continuously increases or decreases non-periodically beyond this threshold and does not recover to within ±0.02 MPa of the original value within 60 seconds, it is considered that there is an unstable drift phenomenon.

[0065] If any of the above parameters meets the conditions of "change amplitude exceeding the threshold" and "duration exceeding the set duration" (e.g., no drop for 20 consecutive seconds), then the alarm trigger condition has been met.

[0066] Finally, an alarm event is triggered immediately upon meeting the alarm conditions, and the trigger time and related data are recorded. After an alarm event occurs, the timestamp, trigger parameter type, direction of change, cumulative change value, and complete parameter datasets for the preceding 5 minutes and the following 2 minutes are automatically recorded. The alarm source is also marked, such as whether it is triggered by stirrer current, differential pressure, or pressure drift. If two or more parameters simultaneously meet the dual threshold conditions, the alarm event is defined as a composite alarm. This alarm information will serve as the trigger point for vacuum adjustment in subsequent control steps and will also be used as the starting point for tracking viscosity change trends. All alarm events will be summarized at the end of the shift and correlated with the final viscosity change results to provide input for future plate repair behavior prediction and sensitivity matrix updates. Compared to the traditional hysteresis strategy of "responding only when the viscosity value exceeds a certain set upper limit," this method, by monitoring multiple precursor physical variables in real time and introducing dual determination of change rate and duration, can identify viscosity fluctuation trends earlier and more accurately, improving the forward control capability for melt rheological stability.

[0067] The main function of this step is to establish a proactive early warning mechanism for identifying melt viscosity fluctuation trends in real time, under the continuous operation conditions of the polyester unit with "de-chipping" of the process. This is achieved through continuous monitoring of key process parameters during periods of high volatility risk. Since the traditional chip buffering stage has been eliminated, the melt is directly supplied from the polyester system to the spinning process. The system is extremely sensitive to short-cycle disturbances (such as metering pump start-up and shutdown caused by plate repair, and liquid level fluctuations). Any undetected fluctuations can trigger drastic viscosity changes, leading to serious consequences such as filament breakage, spinning unit collapse, and abnormal spinning quality. This step uses three highly viscosity-related physical variables—the agitator motor current in the final polymerization reactor, the melt filter pressure difference, and the inlet pressure of the pre-spinning assembly—as monitoring targets. A "dual-threshold" triggering mechanism is constructed based on the rate and duration of continuous change, enabling early identification of viscosity fluctuation trends. Unlike existing technologies that rely on setting an upper limit for a single point value of the viscometer and then passively responding, this step uses both trend slope and anomaly persistence for judgment, independent of the stability and hysteresis of the viscometer itself, resulting in higher sensitivity and robustness. The triggered alarm event not only identifies the starting point of the potential risk's evolution but also provides a clear trigger point and basic data for subsequent control operations such as vacuum adjustment and feed regulation, greatly improving the initiative, real-time performance, and traceability of the entire viscosity control process. Therefore, this step plays a crucial role in the entire viscosity stability control scheme, serving as a core prerequisite for achieving refined, closed-loop viscosity disturbance control.

[0068] When an alarm occurs, a feedforward control strategy is implemented to set the production difference between the prepolymerization section metering pump and the final polymerization section metering pump to a fixed value, and adjust the feed rate of the prepolymerization section metering pump according to the time distribution pattern in the plate density spectrum to maintain the liquid level of the final polymerization reactor within the high-risk operating window, thus providing a stable initial operating condition for subsequent vacuum regulation.

[0069] To address existing viscosity disturbance alarms and ensure stable melt levels in the final polymerization reactor of the polyester unit within high-risk operating windows, preventing abnormal melt residence times caused by the combined effects of metering pump start-ups / shutdowns and plate repairs, and thus avoiding sudden viscosity changes, a process control method based on feedforward yield differential regulation is implemented. This method consists of the following steps:

[0070] A target value for the flow rate difference between the prepolymerization stage metering pump and the final polymerization stage metering pump is set as a control benchmark. In continuous polyester production, the prepolymerization stage metering pump is responsible for transporting the melt that has completed the initial polymerization reaction to the final polymerization reactor, while the final polymerization stage metering pump transports the final polymerized melt to the spinning metering device. The flow rate difference between the two, i.e., the flow rate difference, is a key control parameter for maintaining a stable liquid level in the final polymerization reactor. In this embodiment, through analysis of process operation data over 30 consecutive days, the average flow rate difference between the prepolymerization stage and the final polymerization stage metering pumps was 3.2 tons per hour when the liquid level in the final polymerization reactor was within a stable range of ±25 mm. Therefore, 3.2 tons per hour is set as the target value for the flow rate difference control. When a viscosity disturbance alarm occurs, this fixed difference control strategy is immediately activated to forcibly maintain the flow rate difference between the two pumps at no less than this value.

[0071] Based on the concentration of plate repairs in the plate repair density map during a given time period, the feed rate of the metering pump in the prepolymerization stage is dynamically fine-tuned. By accessing the time period records corresponding to the current high-risk operating window in the plate repair density map, the average number of plates repaired, the duration of a single repair, and historical liquid level fluctuation data for the same time period are retrieved. For example, during the morning shift (07:00 to 11:00), if the map shows an average number of plates repaired exceeding 10, a repair duration generally exceeding 20 minutes, and historical liquid level fluctuations exceeding ±35 mm, it indicates a severe impact on melt balance during this period. In this case, the speed of the metering pump in the prepolymerization stage is increased by 4% to achieve incremental forward feeding, thus creating an advance feeding buffer before the final polymerization stage production recovers. This adjustment does not rely on a passive response after liquid level fluctuations have already occurred, but rather proactively makes slight corrections to the upstream feed based on the time density and disturbance intensity of plate repair behavior, exhibiting clear feedforward control characteristics.

[0072] The system continuously monitors the liquid level change trend in the final polymerization reactor, judges the effectiveness of feedforward control in real time, and optimizes and corrects the pump differential control status. During the execution of pump differential control and feed fine-tuning, the liquid level value is collected every 10 seconds through the level gauge to generate a continuous liquid level change curve. If the liquid level fluctuates within ±20 mm within 5 minutes after adjustment and there is no continuous unidirectional upward or downward trend within 10 minutes, the current pump differential and fine-tuning strategy is deemed effective. If the liquid level continues to rise after 5 minutes of adjustment and exceeds ±40 mm within 10 minutes, it indicates insufficient prepolymerization feed, and the pump speed needs to be further increased by 2%. If the liquid level continues to drop, it indicates excessive feed, and the prepolymerization pump speed should also be appropriately reduced. Through this closed-loop judgment logic of "setting-response-judgment-optimization", the feedforward control has dynamic adjustment capability and improves its adaptability to liquid level anomalies.

[0073] After the liquid level in the final polymerization reactor stabilizes, the pump speed, actual yield difference, and liquid level trend curve are recorded, and the current stable operating condition is locked as the initial operating boundary for subsequent vacuum adjustment. When the liquid level remains stable within a ±25 mm fluctuation range for more than 15 minutes without any further alarms, the system determines that the feedforward control strategy is complete. At this point, the current pre-polymerization section pump speed, final polymerization section pump speed, stable liquid level center value, instantaneous yield difference between the two pumps, and summary information on plate repair behavior within this section are recorded as initial operating reference conditions for subsequent vacuum adjustment. Locking this state ensures that the vacuum adjustment process has a clear and stable material basis, avoiding vacuum regulation failure or viscosity control lag due to an unstable liquid level base.

[0074] The purpose of this step is to proactively stabilize the liquid level in the final polymerization reactor when a viscosity disturbance warning has occurred within a high-fluctuation risk operating window. This is achieved by implementing a feedforward control strategy based on pump differential settings, effectively suppressing sudden changes in melt residence time and providing a stable initial operating condition for subsequent vacuum adjustment and viscosity recovery control. In the de-chipping operation mode of the polyester unit, due to the elimination of chip storage buffers, all melt must be directly and continuously supplied to the spinning section. Therefore, under conditions of concentrated plate repair on multiple short fiber production lines and frequent start-stop of metering pumps, the residence time of the melt in the final polymerization reactor will change drastically with fluctuations in melt flow rate. If not adjusted in time, this can lead to a rapid increase or decrease in melt viscosity within a short period, resulting in unstable product quality, frequent spinning head breaks, or even complete shutdowns. This step uses a fixed output difference between the metering pumps in the pre-polymerization section and the final polymerization section as a control benchmark to maintain a relatively stable supply and discharge balance in the final polymerization reactor, avoiding liquid level instability caused by drastic pump speed changes. Simultaneously, by combining the intensity of time-period disturbances identified in the plate density map, the pump speed in the pre-polymerization section is fine-tuned to increase the feed rate in advance when a declining production trend is anticipated, forming an effective buffer and reducing drastic fluctuations in liquid level during the abrupt changes before and after plate repair. Real-time tracking and feedback of liquid level changes not only confirms the effectiveness of the control strategy but also allows for adjustments and optimizations based on liquid level trends, enhancing the adaptive capability of the entire control process. Ultimately, a controllable, continuous, and flow-balanced reaction environment is established based on a stable liquid level, creating well-defined and parameter-stable initial conditions for the next stage of vacuum adjustment, significantly improving the accuracy and efficiency of subsequent viscosity recovery control. Therefore, this step is a crucial bridging link in transforming viscosity disturbance identification into viscosity intervention operations and a core measure to ensure stable operation of the unit and the continuity of the viscosity control chain.

[0075] Within the preset time window after the plate repair is completed, based on the changing trend of the liquid level and viscosity in the final polymerization reactor, stepwise micro-vacuum regulation control is performed. The vacuum degree is adjusted in increments of ±1 Pa to ±2 Pa. The vacuum degree reset point is determined by combining the dynamic correction value of the melt residence time and the convergence speed of the slope of the final polymerization stirrer current change, so that the residence time of the melt in the final polymerization reactor can be quickly restored to the process target range.

[0076] To quickly restore the normal residence time of the polyester melt in the final polymerization reactor after the plate repair operation, and to avoid aggravated melt viscosity fluctuations due to delayed or over-adjusted vacuum levels, a step-by-step vacuum micro-adjustment method based on liquid level trends and viscosity stability is adopted. This method controls the operation rhythm through a preset time window and uses the characteristics of the final polymerization stirrer motor current change and the dynamic correction value of the residence time as judgment criteria to achieve precise control of the vacuum adjustment process. The specific process is as follows:

[0077] After all the plate repair work is completed and all spinning stations are back in operation, the vacuum conditioning preparation stage begins, with a time window set as the starting point for operation. Specifically, the time point when the last set of spinning plates is repaired, the corresponding metering pump is restarted, and melt delivery returns to normal is set as "T0". A fixed 5-minute preparation window is set after T0 to verify whether the pump differential control in the previous stage has been completed and whether the liquid level has stabilized. Within this window, the liquid level value (in millimeters), the pump speed of the final polycondensation metering pump (in revolutions per minute), and the current value of the final polycondensation agitator motor (in amperes) are continuously collected, and their fluctuation amplitude is judged to be stable. The judgment criteria are: liquid level fluctuation amplitude not exceeding ±25 mm, pump speed fluctuation not exceeding ±3%, and agitator current change rate less than 0.05 amperes per minute. If all three data meet the conditions, the vacuum conditioning start-up stage begins, ensuring that vacuum intervention is performed on the basis of material stability.

[0078] Perform stepwise micro-adjustments of vacuum level in increments of ±1 Pa to ±2 Pa, and observe the impact on liquid level and agitator current in real time. The initial vacuum level is based on the vacuum level before plate repair. When determining the direction, if the liquid level is still trending upward, it indicates that the melt residence time may be prolonged, and the vacuum level of the reactor needs to be increased; conversely, if the liquid level is too low, it indicates insufficient material residence, and the vacuum level needs to be decreased. Each vacuum adjustment should be strictly controlled within 1 Pa to 2 Pa, and the adjusted value should be kept unchanged for 2 minutes to ensure sufficient system reaction. The operation method is: "Adjust vacuum level—observe liquid level and current trends—determine whether to adjust again." After each adjustment, the adjustment range, adjustment time, direction of liquid level change, and slope of agitator current change should be recorded for subsequent analysis. During the entire adjustment process, the cumulative adjustment range should not exceed ±6 Pa to avoid sudden changes in the devolatilization rate inside the reactor due to excessive adjustment, which could cause viscosity overshoot or fluctuation rebound.

[0079] By combining the dynamic correction value of melt residence time with the slope of stirrer current change, it is determined whether the vacuum adjustment has reached the reset condition. The residence time of the melt in the final polymerization reactor is controlled by the combined effects of liquid level, melt viscosity, pump speed, and vacuum degree. To obtain the dynamic correction value, the following judgment is adopted in this embodiment: the instantaneous theoretical residence time is calculated based on the current liquid level and pump speed, and compared with the process set target value (e.g., 300 seconds ± 15 seconds) to obtain the current deviation degree; at the same time, the slope of stirrer current change is observed. If the slope remains within ±0.03 amperes per minute for 5 consecutive minutes, it indicates that the viscosity inside the reactor tends to stabilize and the reactant state has been adjusted. After both judgments are met, the vacuum adjustment is considered complete. At this time, the current vacuum value is locked, and the corresponding liquid level value, pump speed value, measured viscosity value, and stable stirring current section are recorded as the recovery node after this viscosity disturbance intervention.

[0080] After vacuum conditioning is completed, the current operating state is used as a stable operating baseline and as the starting reference for the next production cycle. The criteria for determining that the system has returned to a steady state are: the liquid level continuously fluctuates within ±20 mm for 10 minutes; the measured viscosity value returns to the target range without a continuous upward or downward trend; the stirring current remains stable with fluctuations not exceeding ±0.02 amperes; and the pump speed in the final polymerization stage remains within ±2% of a fixed value. Once these conditions are met, the melt reaction conditions and material state are considered fully restored, and the current data can be used as the initial baseline for the next conditioning cycle. Simultaneously, the adjustment values, trigger periods, response times, and total recovery time required during this vacuum conditioning process are summarized and recorded, and compared with the conditioning curve formed during the previous plate repair process to identify optimization space and provide a direction for correcting subsequent vacuum conditioning response strategies. This method differs from traditional manual, experience-based vacuum conditioning methods. It does not rely on subjective judgment but forms a highly structured conditioning process by fixing the step size, clarifying the adjustment sequence, and quantifying recovery standards, possessing good process adaptability and repeatability.

[0081] The purpose of this step is to achieve a rapid return to normal melt residence time in the final polymerization reactor by precisely controlling the vacuum level after the polyester unit's plate repair operation. This stabilizes melt viscosity and ensures the continuity of downstream spinning processes and the consistency of product quality. In continuous polyester production, when the staple fiber production line performs plate repair operations, several metering pumps often need to be shut down. This leads to an increased residence time of some melt in the final polymerization reactor, resulting in increased viscosity. After plate repair, as the metering pumps are restarted and melt production resumes, the residence time in the reactor quickly shortens again. If the vacuum level is not adjusted in time, a sudden drop in viscosity will occur. This rapid, bidirectional fluctuation in viscosity can cause problems such as filament breakage and unstable spinning positions, and in severe cases, may even lead to a complete line shutdown. To address this technological challenge, this step, after the pump speed and liquid level have initially stabilized, enters a 5-minute vacuum adjustment window, using ±1 Pa to ±2 Pa increments to gradually adjust the reactor vacuum level. By monitoring the slope of the final polymerization stirrer motor current change and the dynamic correction value of the melt residence time in real time, it is determined whether the vacuum adjustment has reached the stable target. When the viscosity control trend tends to be stable, the stirrer current has no sudden changes, and the residence time returns to the process set value range, the vacuum reset can be confirmed as successful. Compared with the existing process of making a large adjustment to the vacuum level at once or relying on manual judgment of the adjustment range, this method not only has a faster response speed and more precise adjustment range, but also is gentler on the reaction process, avoiding impact on the polymerization reaction and improving the stability and controllability of the entire melt viscosity control. Therefore, this step plays the role of a "critical relay point" in the entire viscosity dynamic control process, and is the core control node for transitioning from the production disturbance equilibrium stage to the viscosity return to stability stage.

[0082] After vacuum conditioning is completed, a mapping relationship between actual output change and filter pressure difference change is established. The proportional coefficient is calculated based on the real-time output decrease and the corresponding pressure difference decrease. The target residence time of the final polymerization stage is then corrected based on the proportional coefficient to reduce the viscosity rebound amplitude caused by output recovery from the source.

[0083] To address the issue of a sudden drop in melt viscosity in the final polymerization stage reactor caused by the rapid recovery of melt output due to the concentrated activation of metering pumps after spinning trimming during continuous operation of a polyester unit, a parametric calculation mechanism based on the mapping relationship between the actual output variation and the melt filter differential pressure variation was constructed. By extracting the proportional coefficient between the two, the target residence time in the final polymerization stage is dynamically corrected, thereby mitigating the viscosity rebound trend from the inlet flow rate end. The specific operation includes the following steps:

[0084] After the vacuum conditioning phase is completed, the real-time production and filter differential pressure data synchronous acquisition program is started, and a baseline time period for analysis is established. After successful vacuum conditioning reset and melt residence time restored to the process setting range, the continuous stable operation preparation phase begins. At this time, a continuous 10-minute data acquisition period is selected, recording the flow output data of the metering pump in the final polycondensation stage in tons per hour, and simultaneously acquiring the inlet and outlet differential pressure values ​​of the melt filter (e.g., model 18F01) in the corresponding time period, in kPa. The sampling frequency is set to once every 5 seconds to ensure high time resolution. To eliminate false data interference caused by instantaneous equipment fluctuations, each sample must undergo smoothing filtering, retaining only valid trend values. Before acquisition, it must be confirmed that the metering pump drive equipment is not experiencing frequency hopping, pauses, or other abnormal states, and that the differential pressure transmitter output signal is within the equipment calibration range (typically 4–20mA corresponding to 0–250 kPa) to avoid misjudgment of the proportional coefficient.

[0085] Based on historical benchmark output and pressure differential values, the real-time data during the current output recovery phase after plate repair is compared to calculate the output reduction and pressure differential reduction, and further, a proportionality coefficient is calculated. Using the stable output value before plate repair as a reference (e.g., 35 tons per hour) and the corresponding pressure differential (e.g., 180 kPa) as a benchmark, the output reduction (7 tons / hour) and pressure differential reduction (45 kPa) during the plate repair recovery process are compared to obtain a proportionality coefficient of 0.155 tons / kPa. This proportionality coefficient reflects the degree of output fluctuation corresponding to each unit change in pressure differential under the operating conditions of this period. Different filter structures, melt viscosity levels, and the operating status of spinning components will affect this proportionality coefficient; therefore, this value should be recalculated in real-time after each plate repair operation and should not be used statically for extended periods. This proportionality coefficient will directly participate in the adjustment range calculation in subsequent target residence time corrections and is a core data source for accurately controlling the viscosity rebound amplitude.

[0086] Based on the current proportional coefficient and real-time production recovery rate, determine whether the reaction residence time in the final polymerization reactor needs adjustment, and perform a fine-tuning of the target value. The judgment criteria are: when the production recovery rate is faster than the pressure difference recovery rate, and the proportional coefficient exceeds the previous operating condition baseline value by more than 10%, it is considered that the current filter structure has not fully recovered to a stable filtration state, the melt devolatilization efficiency has not reached a steady state, and the viscosity rebound trend is relatively large, requiring early intervention in the final polymerization reaction time. The specific implementation method is: reduce the target residence time of the final polymerization stage from the standard value (e.g., 300 seconds) by 5% to 10%, that is, adjust it to between 270 seconds and 285 seconds; the adjustment range is jointly determined by the proportional coefficient and the pressure difference growth rate, and a step-by-step reduction method is preferred to avoid a large fluctuation at once. If the pressure difference recovery rate is faster than the production growth rate, it indicates that the melt devolatilization recovery is smooth and the reactor does not retain excessive material, then the residence time can be appropriately extended by 3% to 5% to prevent the viscosity from dropping too quickly. All residence time correction operations are performed in 2-minute cycles, with each cycle allowing a maximum adjustment of no more than 3% to ensure the stable and continuous operation of the polymerization process.

[0087] The aforementioned target residence time correction value was applied to the adjustment of the metering pump feed rate in the final polymerization stage. The operating effects before and after the correction were tracked in real-time based on viscosity trends to confirm the effectiveness of the control strategy. After the correction was implemented, the stability of the final polymerization stirrer motor current, the melt filter differential pressure curve, and the viscosity test results of the test samples needed to be monitored in real time. Viscosity monitoring could be based on an online viscometer or interval test results. If the viscosity fluctuation range remained within ±0.03 Pa·s after adjustment, and the stirring current change rate was less than 0.05 amperes per minute, the correction was considered effective. If a viscosity overshoot trend occurred (viscosity decline rate exceeded 0.05 Pa·s / 5 minutes), the previous adjustment range needed to be reversed by 50%, and gradual fine-tuning was performed. Continuous observation was conducted in 10-minute cycles. If there was no significant rebound in viscosity within two consecutive cycles, the proportional correction strategy was considered effective. Finally, the proportional coefficient, production recovery curve, differential pressure recovery trend, target residence time correction value, and viscosity feedback results were compiled and archived to provide quantitative data support and optimization paths for viscosity disturbance control in subsequent plate repair operations.

[0088] The purpose of this step is to proactively identify the risk of "viscosity rebound" that may be induced during the rapid recovery of production after the polyester unit has completed vacuum conditioning and initially achieved melt viscosity stabilization. By constructing a mapping relationship between actual melt production changes and filter differential pressure changes, the coupling characteristics between flow fluctuations and material filtration resistance are quantified. This allows for forward correction control of the residence time in the final polymerization reactor, preventing a new round of viscosity disturbances caused by sudden changes in reaction conditions. In the continuous production process of polyester unit de-chipping, after the plate repair operation, the metering pumps are started up intensively, and each spinning station quickly resumes production, resulting in a sharp increase in melt production in a short period. However, because the filter has not yet fully recovered to a smooth state, the actual melt flow capacity lags behind the pump speed recovery rate, resulting in a "flow-pressure mismatch." If the statically set residence time parameters are still used in this stage, the residence time of the melt in the final polymerization reactor may drop too quickly, and the melt may be discharged before the reaction is fully completed, leading to a rapid decrease in melt viscosity and forming a typical "rebound-type downward surge." This step establishes a proportional relationship model between the decrease in production volume and the decrease in differential pressure, extracts the proportional coefficient, and judges in real time the current filter patency and pump speed matching. This allows for adjustment of the target residence time in the final polymerization stage, proactively buffering potential underreaction and reducing viscosity downsurge. Compared to existing control methods that rely on passive feedback from viscometer results, this invention employs a feedforward correction method based on the relationship between process physical parameters. This method offers advantages such as early response, high adjustment accuracy, and small disturbance amplitude, significantly improving the initiative of viscosity control and the stability of the polymerization reaction system. It is a key technical step in achieving "reducing disturbances at the source."

[0089] All operating parameters within the high-risk operating window, including time series data of the three monitoring indicators, vacuum adjustment process, production and pressure difference ratio coefficient, final polymerization residence time correction value and corresponding viscosity response, are analyzed retrospectively at the shift level. Based on the analysis results, the plate density map, feedforward production difference control range and alarm trigger threshold are dynamically updated to achieve adaptive optimization of viscosity disturbance response strategy and form a closed-loop control system.

[0090] To achieve the closed-loop control objective of the viscosity dynamic control method and prevent control parameters from becoming ineffective due to changes in equipment status or fluctuations in production load during long-term operation, a systematic and structured retrospective analysis of various control data within the high-fluctuation-risk operating window is conducted after each operating cycle. This dynamically updates the previously set control baseline and continuously optimizes the plate repair identification mechanism, control strategy thresholds, and adjustment parameters. The process specifically includes the following steps:

[0091] Key control parameters for each operating shift are categorized, archived, and cleaned to form a complete foundation for historical data analysis. In this invention, operating shifts are based on an 8-hour cycle. At the end of each shift, all core control data related to viscosity fluctuations are immediately extracted. This includes: the real-time trend of the agitator motor current in the final polymerization reactor and its first derivative (in A / min); the rate of change of the pressure difference between the inlet and outlet of the melt filter (in kPa / min); the fluctuation range of the melt pressure at the inlet of the pre-spinning component (in kPa); the adjustment time and change range of the vacuum setpoint during each vacuum adjustment stage (in Pa); the difference in production setting between the prepolymerization stage metering pump and the final polymerization stage metering pump (in tons / hour), and the actual fluctuation range of this difference during operation; furthermore, it includes the proportional coefficient between the filter pressure difference and production obtained through the aforementioned proportional modeling, and the target residence time value (in seconds) corrected according to the proportion. All the above parameters are indexed by timestamps and recorded at 5-second intervals. The data must undergo a one-time verification process to remove records with sudden signal changes, sensor offline, or verification values ​​exceeding the device's measurement range, ensuring the accuracy of subsequent analysis.

[0092] After data preprocessing, various parameters are categorized according to different disturbance intensity levels, and their correlation with actual viscosity response trends is modeled and analyzed. The disturbance level is assessed based on the maximum fluctuation range of melt viscosity (in Pa·s) within the shift, and is divided into three categories: mild disturbance (<0.05 Pa·s), moderate disturbance (0.05–0.08 Pa·s), and severe disturbance (>0.08 Pa·s). The trend curves of the corresponding three monitoring indicators, vacuum adjustment frequency, proportional correction amplitude, and actual production recovery rate are cross-compared with the viscosity response's decline time, recovery stabilization time, and deviation from the stable value to extract typical causal relationships. For example, if three events occur within the same shift where the filter differential pressure change rate exceeds 40 kPa / min, and each is accompanied by a production increase exceeding 5 tons / hour, resulting in three consecutive decreases in melt viscosity exceeding 0.1 Pa·s, it can be determined that there is a systemic control response lag risk in the current plate repair window. This type of data will be categorized into a pool of severely disturbed cases and used as a direct reference for subsequent parameter updates.

[0093] Based on the aforementioned retrospective analysis results, the control baseline is dynamically adjusted according to the relationship between disturbance level and parameter response. For the plate repair density spectrum, if the plate repair frequency exceeds 3 times / hour for three consecutive shifts within the same time period (e.g., the 2nd to 3rd hour), and the viscosity disturbance is at or above the moderate to severe level, then this time period is marked as a "high-density, high-risk operating window," and the three-parameter monitoring mechanism is activated in advance in the next operating cycle. For the feedforward production difference control range, if the analysis results indicate that the set difference is insufficient to maintain the stability of the final polymerization reactor level, then the original control value of ±1 ton / hour is adjusted to ±1.5 ton / hour. Regarding the alarm threshold, if the viscosity disturbance occurs before the filter differential pressure rate exceeds 30 kPa / min in most disturbance cases, then the original setting of 25 kPa / min should be increased to 28-30 kPa / min to enhance the system's adaptability and reduce the false trigger rate. All adjustments must be accompanied by parameter sources and analysis records to ensure that each optimization is based on evidence and traceable.

[0094] The updated plate repair charts, feedforward setting parameters, alarm thresholds, etc., are written into the control strategy of the next operating cycle, forming an automatically evolving closed-loop control system. At the beginning of the next operating cycle, the control settings updated in the latest shift will be automatically referenced for predictive markers in the plate repair window, pump speed setting benchmarks, alarm judgment references, etc. The real-time monitoring module will be reinitialized based on the new parameters to ensure that the warning behavior conforms to the latest control trends. All operating data continues to be collected and stored in the new shift, serving as the input source for the closed-loop system to achieve continuous learning and optimization. In this way, the unit's operating strategy continuously evolves with the production status, forming a self-learning control structure that combines "memory, responsiveness, and adaptability," significantly improving the overall management capability of viscosity disturbances.

[0095] The purpose of this step is to establish a dynamically optimized closed-loop control mechanism by systematically and shift-by-shiftly analyzing key operational data within high-fluctuation risk operating windows. This enables the polyester unit to adaptively suppress and continuously stabilize melt viscosity disturbances in a de-chipping continuous production mode. During polyester unit operation, the periodic and multi-line concurrent characteristics of spinning assembly plate repair operations can easily trigger problems such as metering pump start-up and shutdown, abnormal fluctuations in the final polymerization kettle level, and sudden changes in reaction residence time within local time windows, leading to bidirectional high-amplitude disturbances in melt viscosity. These disturbances often exhibit repetitive scenarios and similar parameter characteristics. Therefore, by systematically archiving and analyzing monitoring indicators (including agitator current slope, filter differential pressure rate, and pre-spinning pressure fluctuation amplitude), vacuum adjustment response processes, output and differential pressure change ratios, target residence time correction values, and their resulting actual viscosity responses during each operation, the true triggers of disturbances and the quality of response effects can be identified. Furthermore, by dynamically revising the calibration rules for high-risk time windows in the plate repair density spectrum, the setting range for feedforward yield difference, and the trigger thresholds for the three monitoring indicators through analysis results, the control strategy can be continuously optimized, updated, and adapted according to the evolution of actual operating conditions. Compared with traditional control methods based on fixed rules or human experience, this step has a cyclical mechanism of "data-driven—parameter update—control improvement—new data re-driven," thereby constructing a closed-loop control system with memory capability, adaptive adjustment capability, disturbance absorption capability, and trend prediction capability. This not only improves the response speed and accuracy to melt viscosity anomalies but also significantly enhances the continuous operation stability of the unit and the consistency of spinning product quality under high-frequency plate repair and multi-condition changing environments. Therefore, this step plays a core role in the entire viscosity stability control method, connecting the preceding and following steps, providing a strong closed loop, and promoting optimization. It is an important supporting link for realizing intelligent and efficient continuous polyester production.

[0096] The above-mentioned method for stabilizing the melt viscosity of polyester deslicing unit enables early identification, dynamic buffering, precise adjustment, and continuous optimization of polyester melt viscosity disturbances. This effectively overcomes the problem of short-cycle, high-amplitude bidirectional spike disturbances caused by concentrated plate repair, drastic pump load fluctuations, and lag in online viscosity control in existing technologies. This invention constructs a plate repair density map based on the timing of plate repair operations and historical operational fluctuation data. Combined with a sensitivity matrix of final polymerization reactor level, melt residence time, and viscosity change, it achieves quantitative identification of high-risk operating windows for the first time. Furthermore, by continuously monitoring key process parameters (including agitator current slope, filter differential pressure rate, and pre-spinning component pressure drift), a disturbance trend perception model is constructed to trigger control responses in advance. Based on this, a closed-loop control system for viscosity disturbances is formed through feedforward setting of the pre-polymerization and final polymerization yield difference, step-by-step vacuum adjustment, viscosity rebound suppression mechanisms, and adaptive updates of control parameters. Compared with the traditional fixed threshold hysteresis adjustment method, this method significantly improves the adaptability and anti-disturbance capability of the equipment operation to complex fluctuation conditions, so that the polyester melt viscosity remains highly stable under the conditions of frequent plate repair and multi-line load changes, thereby ensuring the continuous and stable rheological properties at the spinning end, reducing filament breakage, and improving product consistency and production line stability.

[0097] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for stabilizing the viscosity of melt in a polyester deslicing unit, characterized in that, Includes the following steps: S1. Obtain the plate repair operation plan and historical operation data. Construct a plate repair density map based on the time distribution and frequency distribution of the spinning components. Combine the liquid level of the final polymerization reactor, melt residence time and melt viscosity to establish a sensitivity matrix, classify the operation risk level, and determine the high fluctuation risk operation window. S2, within the high-fluctuation risk operation window, activate the continuous monitoring mechanism to collect the slope of the change in the current of the stirrer motor of the final polymerization reactor, the rate of change in the differential pressure of the melt filter, and the pressure drift amplitude of the pre-spinning component. When any parameter exceeds the preset threshold and continues to exceed the set time, a dual threshold alarm is triggered and the trigger time is recorded. S3, after the alarm is triggered, feedforward control is implemented to set the output difference between the metering pumps of the prepolymerization section and the final polymerization section to a fixed value, and the feed rate of the prepolymerization section is adjusted according to the plate density spectrum to maintain the stability of the liquid level in the final polymerization reactor. S4. Within the preset time window after the plate repair is completed, the vacuum degree is adjusted in steps of ±1 to ±2 Pa. The vacuum degree reset point is determined by combining the melt residence time correction value and the slope of the final polymerization stirrer current change, so as to achieve rapid recovery of residence time. S5. After the vacuum degree is reset, establish the mapping relationship between the output and the filter pressure difference. Calculate the proportional coefficient based on the output decrease and the pressure difference decrease. Adjust the target residence time of the final polymerization stage according to the proportional coefficient to reduce the viscosity rebound amplitude. S6 performs shift-by-shift retrospective analysis on the time series data of three monitoring indicators within the high-risk operation window, the vacuum degree adjustment sequence, the production and pressure difference ratio coefficient, the final polymerization residence time correction value, and their corresponding viscosity response. Based on the analysis results, it dynamically updates the plate density map, feedforward control parameters, and alarm thresholds to form an adaptive closed-loop control system for viscosity fluctuations.

2. The method for stabilizing the viscosity of melt in the de-chipping process of a polyester plant according to claim 1, characterized in that, Step S1 includes: Acquire and organize the plate repair operation plan and historical operation fluctuation data, extract plate repair records for no less than 30 consecutive days, and data fields include plate repair start time, plate repair end time, spinning position number, duration of each plate repair, number of digits involved in each plate repair, and start / stop status of the corresponding metering pump during each plate repair. At the same time, export four operational data items: final polymerization reactor liquid level, final polymerization melt pump speed, melt viscosity, and pelletizer start / stop status. Match all data with a unified timestamp at a time granularity of 5 minutes to form a complete historical operation time series. The 24 hours of each day are divided into 48 fixed half-hour periods. The number of repairs, the average repair duration, and the average number of simultaneous repairs are statistically analyzed in each period to generate a repair density map and identify high-frequency repair segments and strong disturbance segments. High-density time periods are extracted from the plate density spectrum. Combined with the liquid level fluctuation amplitude, pump speed change rate and viscosity change value of the corresponding time periods, a multi-parameter linkage viscosity sensitivity matrix is ​​constructed. The normalized sensitivity value is then used to quantify the viscosity disturbance intensity. High-risk operating periods are defined as those with sensitivity values ​​higher than the preset quantile and corresponding time periods that are high-frequency or strongly disturbed in the plate density spectrum. A risk time distribution table containing risk level, liquid level fluctuation, pump speed change and viscosity change is output as the basis for triggering subsequent monitoring and control logic.

3. The method for stabilizing the viscosity of melt in the de-chipping process of a polyester plant according to claim 1, characterized in that, Step S2 includes: Five minutes before the start of the high volatility risk operation window, the continuous monitoring process is initiated, and three process parameters to be monitored are determined: the motor current of the agitator in the final polymerization reactor, the differential pressure of the melt filter, and the pressure of the pre-spinning component. All three data are continuously recorded with a sampling period of 5 seconds and are aligned with a unified timestamp. For each set of sampled data, add historical feature information of plate repair behavior corresponding to the current high volatility risk operation window. The added content includes: the average plate repair duration in the corresponding time period, the maximum number of plates repaired simultaneously in actual statistics, and the maximum fluctuation amplitude of the liquid level in the final polymerization reactor caused by the last plate repair operation. The above three parameters are used to characterize the disturbance background features in the current time period and assist in subsequent trend identification. The judgment criteria are set as follows: the slope of the stirrer current change is ±0.25 amperes per minute, the rate of pressure difference change is ±15 kPa per minute, and the pressure drift threshold of the front spinning component is ±0.08 MPa. Combined with the duration requirement, a dual threshold judgment logic is formed. If any parameter continuously exceeds the above threshold and continues to exceed the set time, an alarm event is triggered and the trigger time, direction of change, cumulative change value and alarm type are recorded, providing a trigger basis for subsequent vacuum adjustment and control response.

4. The method for stabilizing the viscosity of melt in the de-chipping process of a polyester plant according to claim 1, characterized in that, Step S3 includes: After a viscosity disturbance alarm occurs, the output difference between the prepolymerization section metering pump and the final polymerization section metering pump is set to 3.2 tons per hour, which is used as the control benchmark value for stabilizing the liquid level in the final polymerization reactor. The fixed output difference adjustment strategy is then immediately implemented. Based on the time period records corresponding to the high-fluctuation risk operation window in the plate repair density spectrum, the average number of plates repaired, the duration of a single plate repair operation, and the maximum fluctuation amplitude in the historical liquid level fluctuation curve are extracted within the corresponding time period. When the average number of plates repaired exceeds 10, the duration of a single plate repair exceeds 20 minutes, and the liquid level fluctuation amplitude of the final polymerization reactor is greater than ±35 mm, it is determined that there is a significant disturbance intensity. Subsequently, the speed of the metering pump in the prepolymerization section is increased by 4% on the basic set value to build an upstream feeding buffer before the output of the final polymerization section has fully recovered, enhance the stability of liquid level control, and lay a feedforward guarantee for the realization of continuous melt delivery. The liquid level data of the final polymerization reactor is continuously collected and recorded every 10 seconds. During the 5-minute adjustment period, the liquid level fluctuation is observed to ensure that it remains within ±20 mm. The prepolymerization pump speed is then finely adjusted in real time according to the liquid level change trend to achieve dynamic optimization of the pump differential control status. When the liquid level fluctuates within ±25 mm for 15 consecutive minutes without any new alarms, record the metering pump speed, instantaneous output difference, stable liquid level value, and plate repair behavior summary information between the pre-polymerization section and the final polymerization section. Use the current operating status as the initial working condition boundary for vacuum adjustment to ensure that subsequent control strategies have a stable material basis.

5. The method for stabilizing the viscosity of melt in the de-chipping process of a polyester plant according to claim 1, characterized in that, Step S4 includes: After all spinning plate repair operations are completed, the metering pumps are restored to operation, and the melt delivery is restored to normal, this moment is set as the starting time point. A 5-minute preparation time window is started. During this time window, the liquid level of the final polymerization reactor, the pump speed of the final polymerization metering pump, and the motor current of the final polymerization agitator are continuously collected. It is determined whether the liquid level fluctuation is less than or equal to ±25 mm, the pump speed fluctuation is less than or equal to ±3%, and the current change rate is less than or equal to 0.05 amperes per minute to confirm whether the working condition has entered a stable state. After the three process parameters of liquid level, pump speed and current are all within a stable range, perform a small step-by-step adjustment of vacuum degree. Each adjustment of vacuum degree is in units of ±1 Pa to ±2 Pa. After each adjustment, the vacuum degree is kept unchanged for no less than 2 minutes. At the same time, the trend of liquid level change and the slope of stirrer current change are collected. The cumulative adjustment range does not exceed ±6 Pa. The instantaneous residence time of the melt in the final polymerization reactor is calculated based on the liquid level and pump speed values. It is compared with the set target value. Combined with the judgment that the agitator current slope is kept within ±0.03 amperes per minute for a continuous 5-minute period, it is determined whether the vacuum adjustment has met the reset condition. If the condition is met, the current vacuum value is locked. After vacuum conditioning is completed, the fluctuations in liquid level, viscosity, stirrer current, and pump speed are monitored. When the liquid level is continuously maintained within ±20 mm for 10 minutes, the measured viscosity value is stable, the stirrer current fluctuation does not exceed ±0.02 amperes, and the pump speed is maintained within ±2% of a fixed value, the melt rheological state is considered to have been fully restored. The current operating data is recorded as the initial reference for subsequent control steps. At the same time, the records of each vacuum adjustment, response time, and recovery cycle are saved as optimization basis.

6. The method for stabilizing the viscosity of melt in the de-chipping process of a polyester plant according to claim 1, characterized in that, Step S5 includes: The actual output data of the metering pump in the final polymerization stage and the pressure difference data of the inlet and outlet of the melt filter were collected simultaneously. The collection time period was set to 10 consecutive minutes after the vacuum degree was restored, and the sampling frequency was once every 5 seconds. The collected data were smoothed and filtered before being used as the analysis sample. The stable output value and corresponding pressure difference value before plate repair are used as reference benchmarks and compared with the lowest output value and pressure difference value in the current operating stage. The output reduction and pressure difference reduction are calculated respectively, and the ratio coefficient between the two is calculated. If the ratio coefficient increases by more than 10% compared to the benchmark value of the previous operating cycle, the target dwell time of the final polymerization stage will be reduced by 5% to 10% based on the standard value. If the ratio coefficient is lower than the benchmark value of the previous cycle, the target dwell time will be increased by 3% to 5%. The adjustment cycle is once every 2 minutes, and the adjustment range each time shall not exceed 3%. The corrected target residence time was applied to the feed rate control of the metering pump in the final polymerization stage, and the stability of the agitator current and the viscosity change trend were continuously monitored to determine whether the correction was effective. The proportional coefficient and correction results were recorded for subsequent optimization analysis.

7. The method for stabilizing the viscosity of melt in the de-chipping process of a polyester plant according to claim 1, characterized in that, Step S6 includes: Using one operating shift as a cycle, extract all control parameter data within the high-fluctuation risk operating window during continuous operation, and establish time series records at 5-second intervals. The parameters include the trend of agitator current change, the rate of change of melt filter differential pressure, the amplitude of melt pressure drift, vacuum regulation history, metering pump output difference, and viscosity change corresponding to the target residence time correction value. The collected data were divided into three levels according to the maximum viscosity fluctuation range: mild, moderate and severe disturbance. The corresponding parameters were classified and modeled to determine their correlation with viscosity recovery time and viscosity stability value. Based on the analysis results, the marked sections of the plate density map are dynamically updated, the feedforward production difference control range and the monitoring alarm trigger threshold are corrected, and the original control parameters are automatically replaced in the next shift operation cycle. The updated control settings are used as the starting conditions for the viscosity disturbance response in the next operating cycle, and the acquisition and optimization operations continue to be performed to achieve closed-loop control.

Images