A plasticizer production process optimization method and system based on quality fluctuation analysis

Abstract

This application discloses a method and system for optimizing plasticizer production processes based on quality fluctuation analysis, belonging to the field of chemical process optimization and control. The method includes: real-time acquisition of the temperatures at the top, sensitive plate, and bottom of a distillation column; calculation of the first temperature difference fluctuation rate between the top and the sensitive plate, and the second temperature difference fluctuation rate between the sensitive plate and the bottom; inputting the first and second temperature difference fluctuation rates and the feed load into a component distribution model constructed based on a three-dimensional lookup table and trilinear interpolation algorithm, outputting the enrichment layers and mass concentration percentages of light component impurities; calculating the target reflux ratio using a PID algorithm based on the identification results, and adjusting the reflux pump frequency using a ramp function control method; simultaneously, calculating the steam flow compensation amount according to the heat balance compensation formula, and synchronously adjusting the steam feed to the reboiler at the bottom of the column. This application has the effect of improving the stability of finished product quality and significantly reducing the energy consumption of the distillation process.

Description

Technical Field

[0001] This application belongs to the field of chemical process optimization and control, specifically relating to a method and system for optimizing plasticizer production process based on quality fluctuation analysis. Background Technology

[0002] With the continuous advancement of the fine chemical industry, plasticizers, as indispensable additives in plastic products, have seen their production processes refined and intelligently managed, becoming crucial for enhancing product competitiveness. In the industrial production process of plasticizers, the refining of crude esters after esterification is a core step determining the purity and performance of the final product. This typically requires effective removal of excess alcohol raw materials and light component impurities through distillation and dealcoholization processes. This process not only affects the chemical and physical stability of the plasticizer but also directly impacts the processing quality and application safety of subsequent downstream products.

[0003] The precision of the distillation and de-alcoholization process is a crucial indicator of plasticizer production quality. This process aims to achieve efficient separation of light components and precise purification of the target product by adjusting key parameters such as the reflux ratio, top pressure, and reboiler heat load of the distillation column. Under ideal conditions, stable operating parameters maintain gas-liquid balance within the column, ensuring the finished product flash point meets industry standards and minimizing heat consumption during production.

[0004] However, traditional distillation control methods often rely on fixed reflux ratios or manual adjustments based on periodic sampling and testing results, making it difficult to cope with component shocks caused by fluctuations in the conversion rate of the upstream esterification reaction. When the alcohol content in the crude ester entering the distillation column changes dynamically, the fixed reflux ratio mode lacks the ability to detect the distribution of impurities within the column, leading to frequent fluctuations in the flash point of the finished product due to excessive impurities, and even causing product quality defects. Simultaneously, experience-based hysteresis adjustments cannot provide real-time feedback on the nonlinear changes in the temperature difference between the sensitive plates and different sections within the column, causing the system to maintain high energy consumption even when component concentrations decrease, resulting in severe reboiler steam redundancy. Furthermore, control methods lacking a linkage compensation mechanism are prone to causing pressure imbalances within the column, making it difficult to maximize production efficiency while ensuring quality stability, thus becoming a bottleneck restricting the optimization of plasticizer processes. Therefore, a plasticizer production process optimization scheme based on quality fluctuation analysis is desired. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for optimizing the plasticizer production process based on quality fluctuation analysis, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for optimizing plasticizer production process based on quality fluctuation analysis includes the following specific steps: The temperatures of the top region, the sensitive plate region, and the bottom region of the distillation column are acquired in real time. Based on the temperature, the rate of change of the first temperature difference between the top of the column and the sensitive plate over time is calculated as the first temperature difference fluctuation rate, and the rate of change of the second temperature difference between the sensitive plate and the bottom of the column over time is calculated as the second temperature difference fluctuation rate. The first temperature difference fluctuation rate, the second temperature difference fluctuation rate, and the current feed load are input into the pre-built component distribution model. The component distribution model outputs the enrichment layer and mass concentration percentage of light component impurities based on a three-dimensional lookup table and a trilinear interpolation algorithm. Based on the enrichment layer and the percentage of mass concentration, the target reflux ratio is calculated using a proportional-integral-derivative control algorithm, and the operating frequency of the reflux pump is adjusted using a ramp function control method to make the actual reflux ratio approach the target reflux ratio. While adjusting the reflux ratio, the steam flow compensation amount is calculated based on the change in reflux ratio, the specific heat capacity of the reflux liquid, the saturation temperature inside the column, the temperature of the reflux liquid, and the latent heat of vaporization of the steam, according to the heat balance compensation formula. The steam feed rate to the reboiler at the bottom of the column is then adjusted according to the steam flow compensation amount.

[0007] Furthermore, the calculation of the first temperature difference volatility and the second temperature difference volatility adopts the least squares linear regression analysis within a preset time sliding window, and the calculated slope is used as the temperature difference volatility. The preset time sliding window contains multiple consecutive sampling points.

[0008] Furthermore, the methods for constructing component distribution models include: Collect historical operating data within a predetermined historical period. Each historical data point includes the measured value of the first temperature difference fluctuation rate, the measured value of the second temperature difference fluctuation rate, the measured value of the feed load, and the corresponding actual light component enrichment layer and actual mass concentration percentage. A three-dimensional lookup table is constructed using the first temperature difference fluctuation rate, the second temperature difference fluctuation rate, and the feed load as three dimensions. Each dimension is divided into multiple discrete intervals to form grid nodes. Each historical data is classified into the corresponding grid node, and the arithmetic mean of the actual light component enrichment layer and the actual mass concentration percentage corresponding to multiple historical data falling into the same grid node is stored in that node. For empty nodes with no historical data, the inverse distance weighted interpolation method in three-dimensional space is used to fill them with the stored values ​​of neighboring non-empty nodes.

[0009] Furthermore, the enrichment sites and mass concentration percentages of light component impurities are output using a trilinear interpolation algorithm, specifically including: Based on the current first temperature difference fluctuation rate, second temperature difference fluctuation rate, and feed load, locate the adjacent node intervals in each dimension in the three-dimensional lookup table; Calculate the normalized weights of the current input value within the intervals of adjacent nodes; Based on normalized weights, the output values ​​stored at the eight adjacent nodes are summed in a weighted manner, and the calculation result is used as the final output of the enrichment sites and mass concentration percentages of light component impurities.

[0010] Furthermore, the operating frequency of the reflux pump is adjusted using a ramp function control method, specifically including: receiving the frequency command corresponding to the target reflux ratio through the frequency converter, and linearly changing the operating frequency of the reflux pump motor from the current value to the target value according to the preset adjustment time constant.

[0011] Furthermore, the heat balance compensation formula is: ; in, This is the steam flow compensation amount. The preset thermal balance compensation constant, This represents the change in reflux ratio. The specific heat capacity of the reflux liquid. The saturation temperature inside the tower. The temperature of the reflux fluid. The latent heat of vaporization of steam; saturation temperature inside the tower The pressure at the top of the tower is obtained from the relationship between the saturation temperature and pressure of water vapor.

[0012] Furthermore, it also includes a steady-state detection step: When the first temperature difference fluctuation rate and the second temperature difference fluctuation rate are both continuously lower than the second preset fluctuation threshold for a predetermined duration, it is determined that the current component distribution in the tower has entered a stable high-purity zone. In response to entering the stable high-purity zone, the system automatically executes energy-saving control commands, gradually reducing the reflux ratio to the preset energy-saving reflux ratio mode, and simultaneously reducing the steam feed to the bottom reboiler.

[0013] Furthermore, the component distribution model has a self-learning mechanism, which includes: periodically comparing the enrichment degree of light components predicted by the model with the actual test results; when the deviation exceeds the preset allowable error range, using the gradient descent method to correct the stored values ​​of each node in the three-dimensional lookup table with the goal of minimizing the loss function between the predicted value and the actual test results.

[0014] Furthermore, the target reflux ratio is calculated using a feedforward-feedback composite control architecture through a proportional-integral-derivative control algorithm. The feedforward part directly calculates the coarse adjustment of the reflux ratio based on the enrichment degree of light components output by the component distribution model, while the feedback part performs fine adjustment based on the deviation between the temperature of the sensitive plate and the set value. The coarse adjustment and the fine adjustment are superimposed to form the final target reflux ratio.

[0015] A plasticizer production process optimization system based on quality fluctuation analysis includes: The temperature acquisition unit is used to acquire the temperature of the top area, the sensitive plate area, and the bottom area of ​​the distillation column in real time. The volatility calculation unit is used to calculate the rate of change of the first temperature difference between the top of the tower and the sensitive plate over time as the first temperature difference volatility, and the rate of change of the second temperature difference between the sensitive plate and the bottom of the tower over time as the second temperature difference volatility. The component identification unit is used to input the first temperature difference fluctuation rate, the second temperature difference fluctuation rate and the current feed load into the pre-built component distribution model. The component distribution model is based on a three-dimensional lookup table and outputs the enrichment layer and mass concentration percentage of light component impurities through a trilinear interpolation algorithm. The reflux ratio adjustment unit is used to calculate the target reflux ratio based on the enrichment layer and the percentage of mass concentration using a proportional-integral-derivative control algorithm, and to adjust the operating frequency of the reflux pump using a ramp function control method so that the actual reflux ratio approaches the target reflux ratio. The steam compensation unit is used to calculate the steam flow compensation amount based on the change in reflux ratio, the specific heat capacity of the reflux liquid, the saturation temperature inside the column, the temperature of the reflux liquid, and the latent heat of vaporization of the steam, according to the heat balance compensation formula, and to adjust the steam feed rate of the reboiler at the bottom of the column based on the steam flow compensation amount.

[0016] In summary, this application includes at least one of the following beneficial technical effects: 1. By introducing the first and second temperature difference fluctuation rates as feedforward control parameters, this invention enables early prediction of the risk of light component impurities shifting downstream within the distillation column, significantly improving response speed compared to traditional manual lag sampling and adjustment. Real-time closed-loop adjustment of the reflux ratio effectively eliminates the impact of front-end esterification reaction conversion rate fluctuations on finished product quality, resulting in a significantly higher flash point qualification rate for 2-octyl terephthalate compared to traditional processes, thus enhancing the product's market competitiveness.

[0017] 2. This invention accurately identifies the component state within the tower through a component distribution model. Under stable operating conditions with high crude ester purity and low volatility, it can automatically switch to an energy-saving operation mode, lowering the reflux ratio to a preset energy-saving value. Combined with the linkage compensation adjustment of steam feed, it avoids the heat energy waste of the traditional fixed reflux ratio mode, significantly reduces reboiler steam consumption, creates significant economic benefits for enterprises, and conforms to the development trend of green chemical industry.

[0018] 3. This invention constructs a fully automated system from signal acquisition and model recognition to closed-loop control, greatly reducing the frequency of operator intervention in the distillation process and lowering labor intensity. The system's self-learning and self-optimization capabilities enable it to continuously evolve control strategies based on historical data, effectively solving the problems of delayed manual intervention and difficulty in passing on production experience in plasticizer distillation, thus achieving standardization and intelligent transformation of the production process.

[0019] 4. By adjusting the steam feed rate and reflux ratio in a coordinated manner, this invention optimizes process parameters while maintaining the gas-liquid momentum balance and pressure stability inside the distillation column, effectively preventing drastic pressure fluctuations caused by sudden changes in reflux flow. Comprehensive safety interlock protection functions and a high-frequency sampling monitoring mechanism ensure the robustness of the unit in handling complex operating conditions and reduce the risk of unplanned shutdowns. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall method for optimizing plasticizer production processes based on quality fluctuation analysis; Figure 2 This is a schematic diagram illustrating the principle of component distribution identification and quality fluctuation analysis based on temperature difference fluctuation rate. Figure 3 It is a logic diagram of real-time calculation of the target reflux ratio and closed-loop adaptive adjustment of the liquid phase reflux flow rate; Figure 4 This is a schematic diagram of the multi-level interaction and data flow between the variable frequency drive of the reflux pump and the steam regulation and compensation of the reboiler. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the following description is provided in conjunction with the appendix. Figure 1-4 The present invention will be further described in detail with reference to specific embodiments. Example 1

[0022] In this embodiment, the plasticizer production process optimization method based on mass fluctuation analysis is applied to a continuous distillation line for 2-octyl terephthalate. The core equipment of this distillation line is a distillation column with multiple side-stream feeds and a tray structure, whose internal environment involves complex gas-liquid two-phase mass transfer processes. The plasticizer production process optimization method based on mass fluctuation analysis includes the following specific steps: The first step, S1, involves acquiring temperature signals at different heights within the distillation column in real time and calculating the key temperature difference fluctuation rate based on these signals. This is achieved through the following sub-steps: Step S101: Multi-point temperature sensor arrangement and signal acquisition; High-precision resistance temperature sensors are arranged in the top area, sensitive plate area, and bottom area of ​​the distillation column. The sensitive plate is determined as follows: A steady-state model of the distillation column is established using chemical process simulation software (such as Aspen Plus), and sensitivity analysis is performed within a given range of feed component fluctuations. The plate with the largest temperature change rate with respect to component is selected as the sensitive plate.

[0023] In this embodiment, a platinum resistance thermometer is selected, whose measurement accuracy class meets the Class A standard, that is, the tolerance range at 0 degrees Celsius is controlled within ±0.15 degrees Celsius.

[0024] The sensor probe is mounted on the tower wall via a flange, and its temperature sensing element penetrates deep into the main gas or liquid phase zone inside the tower to ensure the representativeness of the collected data.

[0025] Temperature signal acquisition is performed by the analog input module of the distributed control system, with a sampling frequency set to 1 Hz, meaning a data snapshot is taken every 1 second.

[0026] Step S102, Analog signal conversion and digital filtering: The acquired original analog electrical signal, i.e., the standard current signal of 4 to 20 mA, is converted into a digital quantity by a 16-bit analog-to-digital converter.

[0027] To eliminate signal spikes caused by fluid turbulence or thermal noise inside the distillation column, the central controller performs median filtering on the acquired temperature sequence. The filtering window width is set to 5 sampling points, that is, the median of every 5 consecutive temperature data is taken as the output value of the window, and abnormal noise points that deviate from the mean within the window by more than 3 standard deviations are removed, thereby obtaining stable and reliable real-time temperature data.

[0028] Step S103: Define temperature difference parameters; After obtaining stable temperature data, the central controller calculates the difference between the top temperature of the tower and the temperature of the sensitive plate, and defines this difference as the first temperature difference; at the same time, it calculates the difference between the temperature of the sensitive plate and the temperature of the tower bottom, and defines this difference as the second temperature difference.

[0029] Step S104: Calculate the temperature difference fluctuation rate; For the first temperature difference and the second temperature difference, calculate their rate of change over time, i.e., the first temperature difference fluctuation rate and the second temperature difference fluctuation rate.

[0030] The calculation uses least squares linear regression analysis within a preset time window. The preset sliding window duration is 5 minutes, which means that 300 consecutive sampling points are collected to form a calculation window.

[0031] The specific formula is as follows: ; in, This is the total number of sampling points within the preset sliding window, with a value of 300. For the first Each sampling time point is represented in minutes. This represents the temperature difference at the corresponding time point, in degrees Celsius.

[0032] The slope calculated using this formula is the temperature fluctuation rate, and its unit is degrees Celsius per minute.

[0033] The first and second temperature difference volatility rates are calculated using the same formula described above, with each rate rate calculated by substituting its corresponding temperature difference data sequence.

[0034] Step S105: Assign physical meaning to volatility; The calculated first temperature difference volatility reflects the rate of downward penetration of light components at the top of the tower. When this value is positive and the absolute value increases, it indicates that the trend of light components migrating downwards intensifies.

[0035] The second temperature difference fluctuation rate characterizes the dynamic trend of contamination of the heavy component area in the bottom of the tower by light components. When this value is positive, it indicates that light components have penetrated into the bottom area of ​​the tower.

[0036] In summary, step S1 completes the acquisition, processing, and feature extraction of temperature signals, transforming the original temperature distribution data within the column into two key characteristic parameters: the first temperature difference fluctuation rate and the second temperature difference fluctuation rate. These two parameters characterize the migration dynamics of light component impurities from the column top to the sensitive plate and from the sensitive plate to the column bottom, respectively, providing a quantitative basis for the subsequent accurate identification of the enrichment degree and spatial distribution of light component impurities within the column. Based on this quantitative basis, the next step of column component state identification and process parameter control can be performed.

[0037] Next, for step S2, the enrichment level of light component impurities in the current column is identified, which includes the following implementation steps: Step S201: Construct input feature vector; The central controller encapsulates the first temperature difference fluctuation rate and the second temperature difference fluctuation rate calculated in real time in step S1 into a two-dimensional input feature vector, which is used as input data for subsequent model recognition.

[0038] Step S202: Import the component distribution model; the two-dimensional input feature vector is imported into the component distribution model that has been pre-built and stored in the non-volatile memory of the central controller.

[0039] Step S203: Clarify the structure and construction method of the component distribution model; the component distribution model is a mathematical model constructed based on multidimensional nonlinear mapping logic, which contains a three-dimensional lookup table.

[0040] The three dimensions of this three-dimensional lookup table correspond to the first temperature difference fluctuation rate, the second temperature difference fluctuation rate, and the current feed load, respectively. The first temperature difference fluctuation rate dimension is divided into 10 discrete intervals with a step size of 0.05℃ / min; the second temperature difference fluctuation rate dimension is divided into 10 discrete intervals with a step size of 0.05℃ / min; and the feed load dimension is divided into 20 discrete intervals with a step size of 5% of the rated load, thus forming a 10×10×20 grid node matrix. Each node stores a preset output value, which includes two fields: the light component enrichment layer identifier and the mass concentration percentage.

[0041] During the model building phase, the output values ​​of each node in the lookup table are determined by collecting production operation data within a predetermined historical period. The specific construction method is as follows: First, historical operating data from the past 365 days is collected. Each historical data record contains a complete set of input-output pairs. The inputs are the measured values ​​of the first temperature difference fluctuation rate, the second temperature difference fluctuation rate, and the feed load. The outputs are the actual light component enrichment layers and the actual mass concentration percentage obtained by manual analysis under this operating condition.

[0042] Then, all historical data are categorized according to the intervals of the input values ​​in each dimension of the three-dimensional lookup table. For multiple historical data falling into the same grid node, the arithmetic mean of their output values ​​is taken as the storage value of that node.

[0043] For empty nodes with no historical data, the inverse distance weighted interpolation method in three-dimensional space is used to fill the empty nodes by utilizing the output values ​​of neighboring non-empty nodes. The specific formula is as follows: ; in, The output value to be filled for empty nodes. The number of neighboring non-empty nodes participating in the interpolation is set to the 8 non-empty nodes closest to the empty node. For the first The output value is stored at each of the nearest non-empty nodes; For empty nodes and the first The Euclidean distance between three neighboring non-empty nodes in three-dimensional space; It is the power exponent, with a value of 2.

[0044] Through the above construction process, the three-dimensional lookup table of the component distribution model is fully established, realizing the mapping relationship from input parameters to output results.

[0045] Step S204: Perform model recognition calculation; during operation, the model locates the corresponding data node in the three-dimensional lookup table based on the input two-dimensional feature vector and the current feed load.

[0046] Since the input value will not fall exactly on the node, the model uses a trilinear interpolation algorithm to calculate the output. That is, in three-dimensional space, the output values ​​stored at the eight adjacent nodes are weighted according to the relative position of the input value in each dimension of the lookup table.

[0047] The specific calculation formula is as follows: First, normalized weights are calculated for each of the three dimensions. Let the input value in the first dimension, temperature fluctuation rate, be located at node [node]. With nodes Between, the weight factor of that dimension. , ,in For input values, and These represent the values ​​of two adjacent nodes in this dimension; the second temperature difference fluctuation dimension uses the same weighting calculation method as the feed load dimension.

[0048] Then, the output values ​​of the eight adjacent nodes are weighted and summed, as shown in the following formula: ; in, , , These are the starting node indices of the intervals to which the input value belongs in the first temperature difference volatility dimension, the second temperature difference volatility dimension, and the feed load dimension, respectively. , , These are the weighting factors corresponding to each dimension; The output value stored at the corresponding node is a vector containing the identifier of the light component enrichment layer and the percentage of mass concentration. This is the final output result obtained from the interpolation calculation.

[0049] Through the above trilinear interpolation operation, the model can analyze the dynamic evolution of the temperature gradient in real time and output continuous recognition results.

[0050] Step S205: Output the identification results and their significance for the process; the model output includes the enrichment layers of light component impurities in the spatial dimension and the estimated mass concentration percentage.

[0051] The specific discrimination logic is as follows: When the first temperature difference fluctuation rate shows a positive growth trend and the absolute value exceeds the first preset fluctuation threshold of 0.5 degrees Celsius per minute, the model identifies that the concentration of light components in the top region of the current distillation column is increasing significantly. The light component is unreacted 2-ethylhexanol, and the model outputs the estimated value of the remaining time for the impurities to diffuse to the sensitive plate region based on the interpolation result.

[0052] When the second temperature difference fluctuation rate shows a positive change, the model further determines that the light components have begun to migrate towards the bottom of the tower, and provides corresponding enrichment layer information and concentration estimates based on the output results.

[0053] This identification process does not rely on delayed manual testing, but rather achieves rapid perception of component fluctuations through microscopic changes in temperature profiles.

[0054] In summary, step S2, by importing the real-time temperature difference fluctuation rate into the component distribution model, accurately identifies the enrichment degree and spatial distribution of light component impurities in the tower, providing a direct basis for the subsequent linkage adjustment of reflux ratio and steam feed rate.

[0055] Next, for step S3, the target reflux ratio is calculated and adjusted in real time, including the following specific implementation steps: Step S301: Determine the control objective and PID parameter tuning; Based on the enrichment degree of light components identified in step S2, the central controller calculates the target reflux ratio value that meets the finished product flash point qualification index through a preset proportional integral derivative control algorithm.

[0056] The proportional gain, integral time, and derivative time of the PID algorithm are pre-tuned based on the hysteresis characteristics of the distillation column. The tuning method adopts the critical proportional gain method commonly used in engineering, which involves making the system oscillate with constant amplitude under pure proportional action, recording the critical oscillation period and critical proportional gain, and then calculating the initial values ​​of the proportional gain, integral time, and derivative time according to the Ziegler-Nichols empirical formula. The values ​​are then fine-tuned based on the actual control effect during subsequent operation.

[0057] Step S302: Calculate the target reflux ratio. When an increase in the enrichment of light components is detected, the central controller calculates the required increase in liquid phase reflux flow rate based on the output of the PID algorithm, and then determines the target reflux ratio.

[0058] The adjustment range of the target reflux ratio is preset according to the process design conditions. The first preset reflux ratio is 1.5:1, which corresponds to the stable working condition with low alcohol content in the crude ester; the second preset reflux ratio is 2.0:1, which corresponds to the enhanced separation working condition when the enrichment degree of light components increases.

[0059] The PID controller outputs a continuous control quantity based on the deviation between the measured value and the target value of the current reflux ratio, combined with the calculation results of the proportional, integral, and derivative components.

[0060] Step S303: Conversion and output control command; The calculated target reflux ratio value is converted into a preset standard electrical signal and output to the reflux pump frequency converter. In this embodiment, a current signal of 4 to 20 mA is used as the analog signal transmission method, where 4 mA corresponds to the lower limit of the reflux ratio and 20 mA corresponds to the upper limit of the reflux ratio.

[0061] The analog output module of the central controller linearly maps the PID calculation results to the corresponding current values, which are then transmitted to the analog input terminals of the frequency converter via shielded cables.

[0062] Step S304, Inverter execution and ramp control: After receiving the command, the reflux pump inverter analyzes the target operating frequency based on the received current signal, and then changes the operating frequency of the reflux pump motor to adjust the liquid phase reflux flow rate.

[0063] To prevent sudden surges in reflux flow from causing abrupt changes in the liquid layer thickness on the trays and severe impacts on the gas-liquid balance within the tower, the frequency converter's adjustment process employs ramp function control. This means that a set adjustment time constant is established within the frequency converter, allowing the motor frequency to gradually change to the target value according to a preset slope, rather than through a step-like abrupt change.

[0064] In this embodiment, the adjustment time constant is set to 45 seconds, that is, the reflux pump frequency is linearly changed from the current value to the target value within 45 seconds. This time constant takes into account both response speed and control stability, so that it can respond in time when the composition fluctuates, while avoiding impact on the operating conditions inside the tower.

[0065] Step S305: Achieve closed-loop adaptive regulation; by changing the frequency of the reflux pump, the system achieves closed-loop adaptive regulation of the liquid phase reflux flow rate.

[0066] The added cold reflux liquid enters the top of the column and comes into contact with the rising gas phase. On the one hand, it cools down the gas phase and causes the light components in the gas phase to condense. On the other hand, the liquid phase washes the light components back to the top region of the column, thereby enhancing the separation effect of the rectification section on the light components in the rising gas phase. The light components are effectively suppressed in the top region of the column and prevented from migrating down the column and contaminating the finished product.

[0067] In summary, through step S3 above, a complete closed-loop control link is established from component state identification to reflux ratio adjustment. Based on the changes in the enrichment level of light components, the system accurately calculates the required reflux ratio using a PID algorithm and smoothly adjusts the variable frequency drive using a ramp function, ensuring both response speed and stability of the tower's operating conditions, thus creating conditions for subsequent linkage compensation adjustment of the steam feed.

[0068] Finally, regarding step S4, the process of synchronously compensating and adjusting the steam feed to the reboiler at the bottom of the column is as follows: Step S401: Initiate the heat balance compensation logic; while the central controller performs reflux ratio adjustment in step S3, it simultaneously initiates the linkage compensation control of the steam feed to the bottom reboiler.

[0069] The trigger for compensation control is the change in reflux ratio. ,when When the absolute value of the reflux ratio is greater than 0.05, that is, when the adjustment range of the reflux ratio exceeds the preset minimum change threshold, the central controller triggers the compensation logic to prevent frequent small fluctuations from causing frequent valve operation.

[0070] An increase in liquid phase reflux will bring more cooling into the column, resulting in a decrease in the momentum of the rising gas phase inside the column. If no compensation is made, it will cause a drop in pressure inside the column and a decrease in gas-liquid mass transfer efficiency. Therefore, heat balance compensation logic needs to be implemented to maintain the energy balance inside the column.

[0071] Step S402: Calculate the steam flow compensation amount; the central controller calculates the required increase in steam flow based on the heat balance compensation formula, the specific calculation formula is as follows: ; in, This is the steam flow compensation amount, expressed in kilograms per hour. The preset heat balance compensation constant is dimensionless. This constant reflects the heat exchange efficiency and compensation response characteristics of the distillation column, and its value is determined through on-site calibration tests.

[0072] The specific method is as follows: Under stable operating conditions of the distillation column, the reflux ratio is artificially changed by a known quantity. The actual increase in steam flow rate required to maintain a stable temperature distribution within the tower is measured. Then substitute into the formula to calculate backwards. Repeat the experiment multiple times and take the average value as the final result. The value is 1.05 in this embodiment; The change in reflux ratio is the difference between the current target reflux ratio and the reflux ratio before adjustment, and it is dimensionless. The specific heat capacity of the reflux liquid is expressed in kilojoules per kilogram of degree Celsius. This value represents the average specific heat capacity of the reflux liquid within the operating temperature range. In this embodiment, the reflux liquid is crude 2-octyl terephthalate, and its specific heat capacity is 1.82 kilojoules per kilogram of degree Celsius, obtained by consulting a chemical property handbook.

[0073] This is the saturation temperature inside the tower, expressed in degrees Celsius. This temperature is determined by the pressure value measured by the pressure transmitter at the top of the tower. It is obtained through calculation using the Antoine equation. The expression for the Antoine equation is: ,in The unit is millimeters of mercury. The unit is Celsius. For 2-octyl terephthalate, the physical property constants are... , , The values ​​are respectively , , (These constants were obtained through regression experimental data and are applicable to the operating temperature and pressure range of this process.) Tower top pressure in this embodiment. The saturation temperature was calculated using the Antoine equation, with the pressure controlled between 1.5 kPa and 2.5 kPa (corresponding to 11.25 mmHg to 18.75 mmHg). Approximately 160 to 180 degrees Celsius; The temperature of the reflux liquid is measured in degrees Celsius and is collected in real time by a platinum resistance temperature sensor arranged on the reflux pipeline. The sampling frequency is consistent with that in step S101.

[0074] The latent heat of vaporization of steam is expressed in kilojoules per kilogram. This value is obtained by consulting a chemical property handbook based on the pressure and temperature of the heating steam. In this embodiment, the heating steam is saturated steam with a pressure of 0.5 MPa and a latent heat of vaporization of 2106 kilojoules per kilogram.

[0075] Step S403: Output control commands to the actuator; the central controller will calculate the steam flow compensation amount. The command is converted into an opening adjustment command for the steam inlet regulating valve and output to the regulating valve actuator.

[0076] The conversion process is based on the valve's flow characteristic curve, and the specific calculation formula is as follows: ; in, The percentage represents the adjustment increment of the valve opening. This is the current measured steam flow rate, in kilograms per hour, collected in real time by the steam flow meter. The flow characteristic curve is an inverse function of the valve's flow characteristic curve. In this embodiment, the steam inlet regulating valve adopts an equal percentage flow characteristic, and its flow characteristic curve equation is: ,in This represents the actual traffic volume. This represents the maximum flow rate when the valve is fully open. The adjustable ratio is set to 30. The valve opening is denoted by 0, and its value ranges from 1. This inverse function can be used to calculate the opening increment required to achieve the desired flow compensation.

[0077] The central controller outputs a 4 to 20 mA current signal to the electric positioner of the regulating valve. This current signal has a linear relationship with the valve opening; that is, 4 mA corresponds to the fully closed position and 20 mA corresponds to the fully open position. The electric positioner converts the received current signal into a 3 to 15 psi pneumatic signal, which drives the valve stem to change the valve opening, thereby increasing the steam intake based on the original steam flow rate.

[0078] Step S404: Achieve linkage compensation and maintain gas-liquid balance. Through the above linkage adjustment, the increase in steam flow rate offsets the influence of changes in liquid phase reflux flow rate on the momentum balance of rising gas phase in the column, thus maintaining the gas-liquid two-phase balance state inside the distillation column.

[0079] The compensated steam flow rate matches the momentum of the rising gas phase in the column with the increased amount of reflux liquid phase, ensuring that the gas-liquid contact efficiency on the tray does not decrease due to the adjustment of the reflux ratio, thereby guaranteeing the separation effect of the rectification section on light components and the purification effect of the stripping section on the target product.

[0080] After the compensation control is completed, the central controller continues to monitor the changing trends of the pressure at the top of the tower and the temperature of the sensitive plate. If the pressure fluctuation exceeds the preset stability threshold, a second fine adjustment is made until the system re-enters a steady state.

[0081] In summary, through step S4, this method achieves synchronous and coordinated compensation and adjustment of the reflux ratio and steam feed rate. This compensation mechanism, when dealing with component fluctuations, enhances the separation of light components by increasing the reflux ratio and maintains the energy balance within the column by simultaneously increasing the steam feed rate. This avoids pressure drop and reduced mass transfer efficiency within the column due to increased reflux flow, providing dual protection for the stable operation of the distillation process.

[0082] Furthermore, the method also includes a steady-state detection step.

[0083] The system continuously monitors the first and second temperature difference fluctuation rates calculated in step S1. The central controller compares the current values ​​of the two fluctuation rates with a preset second fluctuation threshold in each sampling period. In this embodiment, the second preset fluctuation threshold is set to 0.1 degrees Celsius per minute. This threshold is determined statistically based on the temperature fluctuation amplitude of the distillation column during stable normal operation. A value below this threshold indicates that the temperature distribution within the column tends to be stable and component migration is not significant.

[0084] The central controller determines the duration of the first and second temperature difference fluctuation rates. When both fluctuation rate values ​​remain below a second preset fluctuation threshold for a predetermined duration, the component distribution within the tower is determined to have entered a stable high-purity zone. In this embodiment, the predetermined duration is set to 10 consecutive minutes. This duration is determined by statistically analyzing the shortest stable duration under historical stable operating conditions. This avoids misjudgments caused by short-term fluctuations and allows for timely response to changes in operating conditions, enabling the entry into energy-saving mode.

[0085] Once the system determines that it has entered the stable high-purity zone, the central controller automatically executes energy-saving control commands. These commands consist of two interconnected adjustment parts: the first part is reflux ratio adjustment. The central controller gradually lowers the output frequency of the reflux pump inverter, gradually reducing the reflux ratio from the current high-load state to the preset energy-saving reflux ratio mode. The inverter's adjustment process uses ramp function control, with an adjustment time constant set to 60 seconds to ensure a smooth transition of the reflux ratio. In this embodiment, the preset energy-saving reflux ratio mode is set to 1.2:1. This ratio is determined based on the minimum reflux ratio of the distillation column under design conditions and considering a safety margin, ensuring both product quality meets standards and energy consumption is minimized.

[0086] The second part involves steam flow regulation. The central controller synchronously reduces the opening of the steam inlet regulating valve of the reboiler at the bottom of the tower. The proportion of the steam flow reduction is determined based on the principle of heat balance and is matched with the reduction in cooling capacity caused by the decrease in the reflux ratio. The specific steam reduction is calculated according to the formula... Calculation, where To reduce the amount of reflux, The thermal balance compensation constant is set to 1.05. The specific heat capacity of the reflux liquid is taken as 1.82 kJ / kg Celsius. The saturation temperature inside the tower. The temperature of the reflux fluid. The latent heat of vaporization of steam is taken as 2106 kJ per kilogram.

[0087] The central controller converts the calculated steam reduction into a valve opening adjustment command, which is then output to the electric positioner of the steam inlet regulating valve via a 4 to 20 mA current signal to achieve precise adjustment of the valve opening.

[0088] Through the above adjustments, the system synchronously reduces the reflux ratio and steam feed rate to energy-saving operating levels under stable conditions, avoiding excessive heat energy consumption when the composition is stable. The reduction in the reflux ratio decreases the liquid phase reflux flow, thereby reducing the amount of liquid phase that needs to be vaporized in the reboiler. The synchronous reduction in steam feed rate directly reduces heat energy consumption, and the two work together to ensure that the gas-liquid balance within the tower is not disrupted.

[0089] During energy-saving operation, the system continues to monitor changes in the first and second temperature difference fluctuation rates. If either fluctuation rate exceeds the second preset fluctuation threshold again, the system automatically exits the energy-saving mode and reverts to the enhanced separation control mode described in steps S3 and S4, ensuring that product quality is not affected.

[0090] This dynamic retreat strategy based on fluctuations monitors the changing trend of temperature difference fluctuation rate in real time. Under the premise of identifying stable component distribution and meeting quality standards, it actively switches the operating parameters to a low-energy consumption mode, which not only ensures the stability of product quality, but also significantly reduces the energy consumption of the distillation process, achieving dual optimization of quality and energy consumption. Example 2

[0091] Building upon Example 1, this example delves deeper into the construction and self-learning mechanism of the component distribution model. The component distribution model is not merely a static lookup table, but a dynamic evolutionary model with online parameter correction capabilities.

[0092] The central controller's storage module is equipped with a multi-dimensional historical database to record all process characteristic parameters within a predetermined historical period. In this embodiment, the predetermined historical period is set to the past 365 days, which covers various operating condition changes of the distillation column within a one-year cycle, including the impact of factors such as raw material batch differences, ambient temperature fluctuations, and equipment performance degradation.

[0093] The recorded historical parameters include top pressure, bottom pressure, feed temperature, feed flow rate, real-time temperature of each tray, reflux pump current, steam main pressure, and laboratory analysis data of the finished product 2-octyl terephthalate. The laboratory analysis data includes quality indicators such as flash point, acid value, and color. All data are timestamped and organized in time series format for easy subsequent correlation analysis.

[0094] During the operation of the component distribution model, the central controller periodically compares the enrichment level of light components predicted by the model with the test results of the actual samples collected from the sideline. In this embodiment, the comparison period is set to every 24 hours, that is, the model prediction effect of the previous day is evaluated once a day.

[0095] When the deviation between the model's predicted value and the actual test results exceeds the preset allowable error range, the system initiates the model parameter correction procedure. The allowable error range is set according to the stringency of the finished product quality standards. In this embodiment, the allowable deviation for the mass concentration percentage is set to ±0.5 percentage points, and the deviation for the enrichment layer is set to no more than one tray.

[0096] The model parameter correction procedure uses gradient descent to adjust the nonlinear mapping coefficients within the model. The specific correction process is as follows: First, we construct the loss function, which takes the form of mean squared error. The calculation formula is as follows: ; in, The number of samples participating in this correction is taken as the most recent 30 sets of valid data in this embodiment; The model predicts the first Percentage of mass concentration of each sample; For the first The actual percentage of the test mass concentration of each sample; The model predicts the first The enrichment level of each sample; For the first The actual enrichment level of each sample; The weighting coefficient is used to balance the contributions of mass concentration error and layer error to the loss function. In this embodiment... The value is set to 0.1 because mass concentration has a more direct impact on product quality.

[0097] Then, the model parameters are iteratively updated using gradient descent. The set of model parameters is denoted as... This includes the output values ​​of each node in the 3D lookup table and the weight coefficients of the interpolation algorithm. The parameter update formula is: ; in, The parameter values ​​before the update. For the updated parameter values, The learning rate is set to 0.01 in this embodiment. This value was determined through experiments to ensure both fast convergence and avoid oscillations. For loss function For parameters The gradient, which is the first-order partial derivative of the loss function with respect to each parameter, is calculated using the backpropagation algorithm.

[0098] The iteration stopping condition for gradient descent is set to one of the following two: first, the loss function value is less than a preset accuracy threshold, which is set to 0.01 in this embodiment; second, the number of iterations reaches a preset maximum number of iterations, which is set to 100 in this embodiment. Iteration stops when either condition is met, completing the current round of model parameter correction.

[0099] Through the above correction process, the model can compensate for performance drift caused by fouling of the packing inside the distillation column, decrease in the heat transfer coefficient of the heat exchanger, or aging of the sensor, so that the component distribution model can maintain a good match with the actual process conditions.

[0100] The process of identifying the enrichment level in step S2 is further refined into the following three sub-steps in this embodiment: A. Feature Extraction: The central controller extracts feature vectors from the real-time temperature sequence. In addition to the first and second temperature difference fluctuation rates used in Example 1, the first and second partial derivatives of the temperature difference are also introduced. The first partial derivative of the temperature difference reflects the rate of temperature change, and its calculation formula is as follows: The same linear regression method as in step S104 is used for calculation; the second partial derivative of the temperature difference reflects the acceleration of the temperature difference change, that is, the rate of change of the first partial derivative with respect to time, and the calculation formula is as follows: The values ​​were obtained by applying linear regression to the first-order partial derivative sequence. Introducing the second-order partial derivative helps to capture the acceleration characteristics of component fluctuations, that is, to identify whether the trend of component concentration changes is accelerating and deteriorating, thereby achieving earlier warning.

[0101] B. Pattern Matching: The extracted feature vectors are matched with typical operating conditions in the model library based on similarity. The model library is pre-built and contains feature templates for various typical operating conditions. For example, the excess raw material alcohol condition is characterized by a rapid increase in the first temperature fluctuation rate while the second temperature fluctuation rate lags behind; the catalyst activity decline condition is characterized by a simultaneous and slow increase in both temperature fluctuation rates; and the system pressure fluctuation condition is characterized by periodic oscillations in the temperature fluctuation rate. Similarity matching uses Euclidean distance as a metric, calculated using the following formula: ; in, This represents the distance between the current feature vector and the template feature vector. In this embodiment, the dimension of the feature vector is... The value is 4, corresponding to the two temperature difference fluctuation rates and their first and second partial derivatives; The th feature vector of the current feature vector One component; The first feature vector of the template Each component is selected. The template condition with the smallest distance is chosen as the matching result for the current condition.

[0102] C. Concentration Inversion: Based on the matched operating condition type and the current bottom heat load, the mole fraction distribution curves of light components at each key section of the column are inverted. The concentration inversion uses a simplified model based on the process mechanism, which divides the distillation column into several segments along its height, assuming a linear gas-liquid equilibrium within each segment. The inversion formula is as follows: ; in, For the first Mole fraction of light components at key cross sections; The mole fraction of the light component in the previous section; and The inversion coefficients, which are related to the type of operating condition, are obtained by fitting historical data. This represents the current heat load at the bottom of the tower. The heat load at the bottom of the tower under the baseline operating conditions; This represents the temperature deviation at this cross section. By calculating segment by segment, a complete distribution curve of the mole fraction of light components within the tower is finally obtained, providing more detailed component distribution information for precise control.

[0103] IV. Feedforward-Feedback Composite Control Architecture; In the reflux ratio adjustment in step S3, to ensure the stability of the system, the central controller introduces a feedforward-feedback composite control architecture.

[0104] The feedforward part calculates the coarse adjustment of the reflux ratio based on the enrichment sites and mass concentration percentages of light component impurities output from the component distribution model. This is specifically achieved through the following mapping function: ; in, To adjust the reflux ratio, The distance weighting of enriched layers relative to the sensitive plate (the closer the enriched layer is to the sensitive plate, the greater the weight). The percentage of the mass concentration of the light component impurities. The preset influence factor is obtained through process simulation or empirical tuning; in this embodiment, it is set to 0.5. The feedforward controller adds this coarse adjustment to the current reflux ratio to form the target reflux ratio calculated by the feedforward: ; in, The target reflux ratio is calculated for feedforward. This represents the current reflux ratio.

[0105] The feedback section performs fine-tuning based on the deviation between the temperature of the sensitive plate and the set value. The feedback controller uses a proportional-integral-derivative control algorithm, which is the same as steps S301 to S302 in Example 1, to correct the residual deviation of the feedforward control.

[0106] The feedforward output and the feedback output are superimposed to form the final target value of the reflux ratio. The calculation formula is as follows: ; in This is the output value of the feedback controller.

[0107] In addition, the internal control logic of the return pump inverter is equipped with an anti-surge limiter. When the inverter output frequency is lower than 20% of the motor's rated frequency or higher than 105% of the rated frequency, the controller automatically limits the frequency change to prevent drastic fluctuations in motor speed caused by abnormal control signals or equipment failure, thus protecting equipment safety.

[0108] In summary, this application first deploys temperature sensors at the top, sensitive plate, and bottom of the distillation column to calculate the first and second temperature difference fluctuation rates to characterize the migration trend of light components. A three-dimensional lookup table model is constructed based on historical data, and trilinear interpolation is used to identify the enrichment layers and concentrations of light components. Based on the identification results, a PID algorithm is used to adjust the reflux ratio, and a ramp function is used to smoothly control the variable frequency drive; simultaneously, the steam feed rate is compensated synchronously according to the heat balance formula to maintain gas-liquid balance within the column. When the temperature difference fluctuation rate remains below a threshold, the system automatically switches to energy-saving mode, reducing the reflux ratio and steam consumption, thus optimizing energy consumption while ensuring quality.

[0109] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0110] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for optimizing a plasticizer production process based on mass fluctuation analysis, characterized by, Includes the following steps: The temperatures of the top region, sensitive plate region, and bottom region of the distillation column are acquired in real time. Based on the temperatures, the rate of change of the first temperature difference between the top and the sensitive plate over time is calculated as the first temperature difference fluctuation rate, and the rate of change of the second temperature difference between the sensitive plate and the bottom region over time is calculated as the second temperature difference fluctuation rate. The sensitive plate is determined by establishing a steady-state model of the distillation column using chemical process simulation software, performing sensitivity analysis within a given range of feed component fluctuations, and selecting the plate with the largest rate of temperature change with component as the sensitive plate. The first temperature difference fluctuation rate, the second temperature difference fluctuation rate, and the current feed load are input into the pre-built component distribution model. The component distribution model outputs the enrichment layer and mass concentration percentage of light component impurities based on a three-dimensional lookup table and a trilinear interpolation algorithm. Based on the enrichment layer and the percentage of mass concentration, the target reflux ratio is calculated using a proportional-integral-derivative control algorithm, and the operating frequency of the reflux pump is adjusted using a ramp function control method to make the actual reflux ratio approach the target reflux ratio. While adjusting the reflux ratio, the steam flow compensation amount is calculated based on the change in reflux ratio, the specific heat capacity of the reflux liquid, the saturation temperature inside the column, the temperature of the reflux liquid, and the latent heat of vaporization of the steam, according to the heat balance compensation formula. The steam feed rate to the reboiler at the bottom of the column is then adjusted according to the steam flow compensation amount.

2. The plasticizer production process optimization method based on mass fluctuation analysis according to claim 1, characterized by, The calculation of the first and second temperature difference volatility rates uses least squares linear regression analysis within a preset time sliding window. The calculated slope is used as the temperature difference volatility rate. The preset time sliding window contains multiple consecutive sampling points.

3. The plasticizer production process optimization method based on mass fluctuation analysis according to claim 1, characterized by, The methods for constructing component distribution models include: Collect historical operating data within a predetermined historical period. Each historical data point includes the measured value of the first temperature difference fluctuation rate, the measured value of the second temperature difference fluctuation rate, the measured value of the feed load, and the corresponding actual light component enrichment layer and actual mass concentration percentage. A three-dimensional lookup table is constructed using the first temperature difference fluctuation rate, the second temperature difference fluctuation rate, and the feed load as three dimensions. Each dimension is divided into multiple discrete intervals to form grid nodes. Each historical data is classified into the corresponding grid node, and the arithmetic mean of the actual light component enrichment layer and the actual mass concentration percentage corresponding to multiple historical data falling into the same grid node is stored in that node. For empty nodes with no historical data, the inverse distance weighted interpolation method in three-dimensional space is used to fill them with the stored values ​​of neighboring non-empty nodes.

4. The method for optimizing plasticizer production process based on quality fluctuation analysis according to claim 3, characterized in that, The enrichment sites and mass concentration percentages of light component impurities are output using a trilinear interpolation algorithm, specifically including: Based on the current first temperature difference fluctuation rate, second temperature difference fluctuation rate, and feed load, locate the adjacent node intervals in each dimension in the three-dimensional lookup table; Calculate the normalized weights of the current input value within the intervals of adjacent nodes; Based on normalized weights, the output values ​​stored at the eight adjacent nodes are summed in a weighted manner, and the calculation result is used as the final output of the enrichment sites and mass concentration percentages of light component impurities.

5. The method for optimizing plasticizer production process based on quality fluctuation analysis according to claim 1, characterized in that, The operating frequency of the reflux pump is adjusted using a ramp function control method, specifically by receiving the frequency command corresponding to the target reflux ratio through the frequency converter, and adjusting the operating frequency of the reflux pump motor linearly from the current value to the target value according to the preset adjustment time constant.

6. The method for optimizing plasticizer production process based on quality fluctuation analysis according to claim 1, characterized in that, The thermal balance compensation formula is: ； in, This is the steam flow compensation amount. This is the preset thermal balance compensation constant. This represents the change in the reflux ratio. The specific heat capacity of the reflux liquid. The saturation temperature inside the tower. The temperature of the reflux liquid. The latent heat of vaporization of steam; saturation temperature inside the tower The pressure at the top of the tower is calculated using the Antoine equation, i.e. ,in For the pressure at the top of the tower, , , is the physical property constant corresponding to 2-octyl terephthalate.

7. The method for optimizing plasticizer production process based on quality fluctuation analysis according to claim 1, characterized in that, It also includes a steady-state detection step: When the first temperature difference fluctuation rate and the second temperature difference fluctuation rate are both continuously lower than the second preset fluctuation threshold for a predetermined duration, it is determined that the current component distribution in the tower has entered a stable high-purity zone. In response to entering the stable high-purity zone, the system automatically executes energy-saving control commands, gradually reducing the reflux ratio to the preset energy-saving reflux ratio mode, and simultaneously reducing the steam feed to the bottom reboiler.

8. The method for optimizing plasticizer production process based on quality fluctuation analysis according to claim 1, characterized in that, The component distribution model has a self-learning mechanism, which includes: periodically comparing the enrichment degree of light components predicted by the model with the actual test results; when the deviation exceeds the preset allowable error range, using the gradient descent method to correct the stored values ​​of each node in the three-dimensional lookup table with the goal of minimizing the loss function between the predicted value and the actual test results.

9. The method for optimizing plasticizer production process based on quality fluctuation analysis according to claim 1, characterized in that, The target reflux ratio is calculated using a proportional-integral-derivative control algorithm and a feedforward-feedback composite control architecture. The feedforward part directly calculates the coarse adjustment amount of the reflux ratio based on the enrichment level and mass concentration percentage of the light component impurities output by the component distribution model. The feedback part performs fine adjustment based on the deviation between the temperature of the sensitive plate and the set value. The coarse adjustment amount and the fine adjustment amount are superimposed to form the final target reflux ratio.

10. A plasticizer production process optimization system based on quality fluctuation analysis, applied to the plasticizer production process optimization method based on quality fluctuation analysis described in claims 1-9, characterized in that... include: The temperature acquisition unit is used to acquire the temperature of the top area, the sensitive plate area, and the bottom area of ​​the distillation column in real time. The volatility calculation unit is used to calculate the rate of change of the first temperature difference between the top of the tower and the sensitive plate over time as the first temperature difference volatility, and the rate of change of the second temperature difference between the sensitive plate and the bottom of the tower over time as the second temperature difference volatility. The component identification unit is used to input the first temperature difference fluctuation rate, the second temperature difference fluctuation rate and the current feed load into the pre-built component distribution model. The component distribution model is based on a three-dimensional lookup table and outputs the enrichment layer and mass concentration percentage of light component impurities through a trilinear interpolation algorithm. The reflux ratio adjustment unit is used to calculate the target reflux ratio based on the enrichment layer and the percentage of mass concentration using a proportional-integral-derivative control algorithm, and to adjust the operating frequency of the reflux pump using a ramp function control method so that the actual reflux ratio approaches the target reflux ratio. The steam compensation unit is used to calculate the steam flow compensation amount based on the change in reflux ratio, the specific heat capacity of the reflux liquid, the saturation temperature inside the column, the temperature of the reflux liquid, and the latent heat of vaporization of the steam, according to the heat balance compensation formula, and to adjust the steam feed rate of the reboiler at the bottom of the column based on the steam flow compensation amount.

Images