NMP waste liquid recovery rectification intelligent control system

Abstract

The application discloses a kind of NMP waste liquid recovery rectification intelligent control system, it is related to NMP waste liquid recovery processing technical field, including data acquisition module, concentration prediction module, benefit evaluation module, optimization adjustment module, alarm module and data storage module, concentration prediction module includes real-time analysis unit and historical analysis unit, benefit evaluation module according to the NMP concentration of future time and according to real-time waste liquid recovery data predicts the expected effect of different parameter combinations, obtains comprehensive benefit score, passes through optimization adjustment module and is automatically adjusted system parameter to optimize system, with the effect that concentration prediction module fuses real-time and historical data, corrects parameter and improves prediction accuracy, reduces lag and purity fluctuation, through benefit evaluation module quantifies parameter benefit, and optimization adjustment module automatically selects optimal parameter, improves intelligentization and recovery efficiency, by setting interference compensation term dynamic weight, enhance anti-interference ability, guarantee system stable operation.

Description

Technical Field

[0001] This invention relates to the field of NMP waste liquid recovery technology, specifically to an intelligent control system for NMP waste liquid recovery distillation. Background Technology

[0002] Methylpyrrolidone, or NMP for short, is a colorless, transparent, oily liquid with a slight amine odor. It is highly hygroscopic and miscible with solvents such as water, alcohol, and ether. NMP is one of the most commonly used and important auxiliary materials for lithium-ion batteries. It is a widely used solvent in the front-end preparation process of lithium-ion batteries. As a PVDF solvent, it participates in the dispersion of conductive materials in the slurry, forming a slurry with a uniform medium that remains stable for a long time within a certain viscosity range.

[0003] However, NMP is only used as the main liquid carrier of slurry in the coating stage of lithium battery manufacturing and does not enter the lithium battery product. During the coating and baking stage, about 95% of the NMP will evaporate. The waste liquid needs to be recycled and purified through distillation to reduce costs and reduce environmental pollution.

[0004] Distillation technology is currently the most important and effective method for recovering NMP waste liquid. However, existing NMP waste liquid recovery distillation control systems still face several technical bottlenecks in practical applications, mainly in the following three aspects:

[0005] The system is highly dependent on human experience, resulting in control lag: Operating parameters of existing systems (such as reboiler temperature, top pressure, and reflux ratio) largely rely on the operator's experience for setting and adjustment. The nonlinear, large-lag, and multivariate coupling characteristics of the distillation process make it difficult for operators to respond promptly and accurately to fluctuations in feed concentration, composition, and other operating conditions. This leads to control actions often lagging behind process changes, causing the distillation process to remain in a suboptimal state for extended periods.

[0006] Unstable recovery efficiency and product purity: Existing technologies lack real-time and accurate prediction of key indicators (such as NMP concentration at the top of the column), making proactive adjustments impossible. When feed conditions change, the purity of the product collected at the top of the column can easily fluctuate or even become substandard, requiring rework. This increases energy consumption and reduces overall recovery efficiency and economic benefits.

[0007] Low overall energy efficiency: To ensure product purity, existing technologies often employ conservative operating procedures, such as excessively high reflux ratios and reboiler temperatures. While this ensures purity, it significantly increases steam and electricity consumption, resulting in high energy costs and failing to achieve the optimal balance between purity, recovery rate, and energy consumption.

[0008] Existing technologies lack an intelligent control system capable of deeply integrating real-time and historical data, accurately predicting key indicators, and automatically optimizing to maximize the overall benefits in terms of quality, yield, and energy consumption. Therefore, there is an urgent need in this field for a new system that can overcome these shortcomings and achieve intelligent, adaptive optimization control of the NMP waste liquid recovery distillation process, thereby improving control accuracy, stability, and overall economic efficiency.

[0009] Currently, the purity of NMP at the top of the column mainly relies on offline laboratory testing, which has a strong lag. During this period, if parameters such as feed composition and column temperature fluctuate, causing the purity to deviate from the target, the system cannot detect it in real time. At the same time, the core control parameters of the distillation system are mostly set manually within a fixed range, which lacks real-time performance and accuracy. This can easily lead to short-term product purity failure or excessive adjustment that wastes energy. Furthermore, it lacks the ability to respond to changes in the environment or equipment status, has poor anti-interference capabilities, and reduces system stability. Summary of the Invention

[0010] The purpose of this invention is to provide an intelligent control system for NMP waste liquid recovery and distillation, which solves the problems mentioned in the background art.

[0011] To achieve the above objectives, the present invention provides the following technical solution: an intelligent control system for NMP waste liquid recovery and distillation, comprising a data acquisition module, a concentration prediction module, a benefit evaluation module, an optimization and adjustment module, an alarm module, and a data storage module;

[0012] The data acquisition module collects waste liquid recovery data in real time, including feed information, column bottom temperature, column top pressure, reflux information, energy consumption information, and column top NMP concentration. After dynamic calibration, the data is sent to the data storage module, which includes real-time waste liquid recovery data and historical waste liquid recovery data.

[0013] The concentration prediction module performs weighted analysis on real-time waste liquid recovery data and historical waste liquid recovery data to predict the NMP concentration in the future. The concentration prediction module includes a real-time analysis unit and a historical analysis unit. The real-time analysis unit is used to integrate and analyze various real-time parameters of waste liquid recovery, and the historical analysis unit is used to analyze the deviation between historical waste liquid recovery data and real-time waste liquid recovery data.

[0014] The benefit assessment module predicts the expected effects of different parameter combinations based on future NMP concentrations and real-time waste liquid recovery data, and obtains a comprehensive benefit score.

[0015] A target benefit score is set. When the difference between the target benefit score and the comprehensive benefit score is greater than the set value, the system parameters are automatically adjusted through the optimization and adjustment module to optimize the system.

[0016] Optionally, the concentration prediction module performs the following prediction process:

[0017]

[0018]

[0019] C pred The predicted NMP concentration at the top of the tower represents the predicted NMP concentration per unit time in the future.

[0020] C real These are real-time parameter values, representing the NMP concentration trend values ​​calculated based on the current process parameters.

[0021] C hist This is a historical deviation correction term, calculated based on the average deviation between historical predicted concentrations and actual concentrations.

[0022] D tot The total fluctuation of real-time parameters is obtained by calculating the sum of the standard deviations of the top temperature, top pressure, and feed rate over the past unit time.

[0023] D ref The standard fluctuation threshold is set to 0.5, which is used to determine whether the system is in a stable state.

[0024] The unit of time is 1 hour.

[0025] Optionally, the real-time analysis unit first obtains the real-time temperature, pressure, feed rate, and reflux ratio at the top of the column from the real-time waste liquid recovery data. It then compares these values ​​with their standard values, weights and fuses the ratios, and analyzes the changing trend of NMP concentration under the current process conditions to obtain the real-time parameter value C. real .

[0026] Optionally, the historical analysis unit calculates the difference between the actual average concentration and the predicted average concentration over a past unit of time, and introduces a correction coefficient to obtain the historical deviation correction term C. hist According to the historical deviation correction term C hist Reverse correction of real-time parameter value C real .

[0027] Optionally, the benefit evaluation module calculates the comprehensive benefit score according to the following steps:

[0028] (1) Based on the top extraction rate F3 and the predicted NMP concentration C at the top of the tower. pred Calculate the amount of NMP extracted from the top of the tower: NMP extracted from the top of the tower = F3 × C pred ;

[0029] (2) Based on the feed rate F1 and the feed NMP concentration C feed Calculate the NMP feed rate: NMP feed rate = F1 × C feed ;

[0030] (3) Calculate the NMP recovery rate YH, the process is as follows:

[0031]

[0032] (4) Obtain the tower reboiler heating consumption E from the energy consumption information. heat and return pump consumption E pump Calculate the total energy consumption: Total energy consumption = E heat +E pump ;

[0033] (5) Calculate the relative energy consumption ratio E all The process is as follows:

[0034]

[0035] Where E ref Baseline energy consumption;

[0036] (6) The predicted NMP concentration C at the top of the tower pred NMP recovery rate YH and relative energy consumption ratio E all The overall benefit score A is obtained by weighting and merging the results using weighted coefficients W1, W2, and W3. The specific process is as follows:

[0037]

[0038] Where W1 + W2 + W3 = 1, and W1, W2 and W3 are preset values.

[0039] Optionally, when the overall benefit score is less than the target benefit score, and the difference between the overall benefit score and the target benefit score is greater than a set value, the optimization and adjustment module is automatically activated to adjust the system parameters, as follows:

[0040] The reboiler temperature T1, reflux ratio R, and feed rate F1 are obtained from the real-time waste liquid recovery data and constrained as follows:

[0041]

[0042] At least 50 sets of candidate parameters are randomly generated. Each set of candidate parameters includes the reboiler temperature T1, reflux ratio R, and feed rate F1 that meet the constraints.

[0043] Each set of candidate parameters was substituted into the concentration prediction module and the benefit evaluation module, respectively, and a comprehensive benefit score was obtained.

[0044] All comprehensive benefit scores are compared, and the system parameters are adjusted according to the candidate parameters with the highest values.

[0045] When the difference between the comprehensive benefit score and the target benefit score corresponding to the candidate parameter with the highest value is greater than the set value, the management personnel will be notified through the alarm module.

[0046] Optionally, the optimization and adjustment module includes a control terminal for setting system parameters and visualizing data.

[0047] Optionally, the alarm module includes an audible and visual alarm unit and a pop-up alarm unit. The audible and visual alarm unit is installed on the system equipment and includes an alarm light and a buzzer. The pop-up alarm unit is located in the control terminal and is used for remote early warning.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] I. This invention enhances the integration of various real-time parameters of current waste liquid treatment through the real-time analysis unit in the concentration prediction module to obtain real-time parameter values. Then, the historical analysis unit analyzes the deviation between historical prediction data and actual concentration, and uses the historical deviation to correct the real-time parameter values ​​in reverse. This avoids large fluctuations in NMP prediction values ​​due to accidental fluctuations, improves the accuracy of predicted NMP concentration data, and allows the concentration prediction module to predict the NMP purity change trend in advance. This eliminates the need for traditional offline detection, reduces waiting time, reduces lag, and avoids purity fluctuations caused by passive adjustment.

[0050] Second, this invention uses a benefit evaluation module to predict the expected effects of different parameter combinations based on real-time waste liquid recovery data, and obtains a comprehensive benefit score. The comprehensive benefit score allows for the quantification of parameters from different dimensions, enabling unified comparison when adjusting different parameters. Then, the optimization adjustment module automatically selects the optimal system parameters based on the highest value of the comprehensive benefit score, eliminating the need for manual adjustment, improving the system's intelligence level, ensuring the real-time performance and accuracy of system parameter adjustments, and improving the purity and recovery efficiency of NMP waste liquid recovery.

[0051] Third, by setting an interference compensation item, the present invention gives a high weight to the real-time parameter value when the system parameters fluctuate greatly, and a high weight to the historical deviation correction item when the system parameters are stable. It can dynamically adjust the weight ratio according to the real-time parameter fluctuations, so that the system can respond in a timely manner when the environment or equipment status changes, improve the system's anti-interference ability, and ensure the system can operate stably. Attached Figure Description

[0052] Figure 1 This is a block diagram of the system modules of the present invention;

[0053] Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] For examples, please refer to Figure 1 and Figure 2 :

[0056] This implementation provides an intelligent control system for NMP waste liquid recovery and distillation, including a data acquisition module, a concentration prediction module, a benefit evaluation module, an optimization and adjustment module, an alarm module, and a data storage module;

[0057] The data acquisition module collects waste liquid recovery data in real time, including feed information, column bottom temperature, column top pressure, reflux information, energy consumption information, and column top NMP concentration. After dynamic calibration, the data is sent to the data storage module, which includes real-time waste liquid recovery data and historical waste liquid recovery data.

[0058] The concentration prediction module performs weighted analysis on real-time waste liquid recovery data and historical waste liquid recovery data to predict the NMP concentration in the future. The concentration prediction module includes a real-time analysis unit and a historical analysis unit. The real-time analysis unit is used to integrate and analyze various real-time parameters of waste liquid recovery, and the historical analysis unit is used to analyze the deviation between historical waste liquid recovery data and real-time waste liquid recovery data.

[0059] The benefit assessment module acquires data from the concentration prediction module and predicts the expected effects of different parameter combinations based on real-time waste liquid recovery data, thereby obtaining a comprehensive benefit score.

[0060] A target benefit score is set. When the difference between the target benefit score and the comprehensive benefit score is greater than the set value, the system parameters are automatically adjusted through the optimization and adjustment module to optimize the system.

[0061] In this embodiment: Real-time waste liquid recovery data is collected by the data acquisition module. The concentration prediction module uses key parameters collected in real time, combined with historical waste liquid recovery data and interference compensation items. First, the real-time analysis unit in the concentration prediction module combines the current real-time parameters of waste liquid treatment. Then, the historical analysis unit analyzes the deviation between historical prediction data and actual concentration, and compensates for the results of the real-time analysis unit. This allows for the prediction of NMP purity change trends in advance, eliminating the need for traditional offline detection, reducing waiting time. After predicting the NMP purity per unit time in the future, subsequent system parameter optimization and adjustment can be carried out to avoid purity fluctuations caused by passive adjustment. Furthermore, by setting interference compensation items, the weight ratio is dynamically adjusted according to real-time parameter fluctuations to improve the system's anti-interference ability and ensure stable system operation.

[0062] After predicting the NMP concentration using the concentration prediction module, the benefit evaluation module combines real-time waste liquid recovery data to predict the expected effects of different parameter combinations, thus obtaining a comprehensive benefit score.

[0063] By combining and quantifying parameters from different dimensions through comprehensive benefit scoring, subsequent adjustments to different parameters can be compared in a unified manner. The optimization and adjustment module automatically selects the optimal system parameters for waste liquid recovery and distillation without manual adjustment, reducing human intervention, ensuring the real-time performance and accuracy of waste liquid recovery and treatment, and improving the purity and recovery efficiency of NMP waste liquid recovery.

[0064] Furthermore, the concentration prediction module combines real-time and historical waste liquid recovery data, and incorporates top temperature and pressure to predict the top NMP concentration at future times. The concentration prediction module includes a real-time analysis unit and a historical analysis unit, and the specific process is as follows:

[0065]

[0066] C pred The predicted NMP concentration at the top of the tower (%) represents the predicted NMP concentration per unit time in the future.

[0067] C real These are real-time parameter values, representing the NMP concentration trend values ​​calculated based on the current process parameters.

[0068] C hist This is a historical deviation correction term, which adjusts the real-time parameter value C based on the deviation between historical predictions and actual concentrations. real Provide compensation;

[0069] β is a dynamic weighting coefficient, used as a disturbance compensation term to reflect system stability. When system parameters fluctuate greatly, the real-time parameter value C... real When the weight is high and the system parameters are stable, the historical deviation correction term C is used. histWith high weighting, the weighting ratio can be dynamically adjusted according to real-time parameter fluctuations, improving the system's anti-interference capability;

[0070] β×C real This represents the impact of real-time system parameters on predicted concentration under the current system fluctuation state. When the system is disturbed, causing significant parameter fluctuations, the dynamic weighting coefficient β increases, and the real-time parameter value C... real The predicted NMP concentration C at the top of the tower pred The proportion increases, β×C real Increase the capacity to ensure that NMP concentration prediction can respond quickly to real-time changes and avoid lag;

[0071] (1-β)×C hist This represents the impact of historical parameter prediction deviations on the current predicted concentration under the current system fluctuation state, expressed as β×C. real Complementary, by using the prediction deviation of historical system parameters to correct the current prediction in reverse, thereby improving the accuracy of NMP concentration prediction;

[0072] The calculation process for the dynamic weighting coefficient β is as follows:

[0073]

[0074] D tot The total fluctuation of real-time parameters is obtained by analyzing the standard deviation of each parameter over a past unit of time, which comprehensively reflects the degree of fluctuation of the top temperature, top pressure and feed rate of the column.

[0075] D ref The standard fluctuation threshold is set to 0.5, and is determined by comparing it with the total fluctuation D of the real-time parameter. tot Comparison is used to determine system stability;

[0076] Used to reflect the degree to which the current system operating parameters deviate from the stable state, the total real-time parameter fluctuation D tot Compared with the standard fluctuation threshold D ref The larger the ratio, the worse the system stability, and the closer the dynamic weighting coefficient β is to 1, thus making the real-time parameter value C... real The predicted NMP concentration C at the top of the tower pred The proportion increased, and the historical deviation correction term C hist The predicted NMP concentration C at the top of the tower pred The proportion decreased;

[0077] Specifically, the concentration prediction module uses real-time collected process parameters of NMP waste liquid recovery as input and incorporates a historical deviation correction term C. histThe dynamic weighting coefficient β is used to predict the NMP concentration in the future, so as to quickly analyze the impact of interference factors such as feed component fluctuations and equipment status changes on the top concentration of the column, avoiding the lag of traditional offline detection, and enabling subsequent optimization and adjustment.

[0078] Among them, the real-time parameter value C real The real-time analysis unit is used to analyze the changing trend of NMP concentration under the current process conditions. During the NMP waste liquid distillation process, the column top temperature, pressure, reflux ratio, and feed rate directly determine the gas-liquid equilibrium state. For example, an increase in column top temperature leads to increased volatilization of light components, causing an increase in NMP concentration. The real-time parameter value C... real The process of obtaining the result is as follows:

[0079]

[0080] In the above formula, C real These are real-time parameter values, representing the NMP concentration trend of the current system operating parameters;

[0081] T2 is the real-time temperature at the top of the column, which is collected by a thermocouple sensor at the top of the column and directly affects the gas-liquid separation effect of NMP and impurities.

[0082] T 2,ref The standard temperature at the top of the column represents the reference temperature at the top of the column when the NMP purity meets the standard.

[0083] k1 is the influence coefficient of the top temperature of the column, with a value of 40. It is a weight value determined through system debugging, reflecting the degree of influence of each top temperature parameter on the concentration.

[0084] The top temperature of the column is a key parameter affecting NMP concentration. This is determined by comparing the real-time top temperature T2 with the standard top temperature T. 2,ref Compare the results and multiply them by the influence coefficient of the tower top temperature, k1, that is... The influence value of the top temperature can be used to quantify the impact of the deviation of the top temperature from the standard temperature on the NMP concentration trend;

[0085] P1 is the real-time pressure at the top of the column, which is collected by the pressure transmitter at the top of the column. Pressure fluctuations will affect the volatility of the components.

[0086] P 1,ref K is the standard tower top pressure, and k2 is the tower top pressure influence coefficient, with a value of 20.

[0087] R is the real-time reflux ratio, which is calculated by the ratio of the reflux flow rate to the output flow rate at the top of the tower.

[0088] k3 is the influence coefficient of the tower top pressure, with a value of 15;

[0089] F1 is the real-time feed rate, which is collected by the electromagnetic flow meter at the feed inlet and represents the speed at which NMP waste liquid enters the distillation column;

[0090] F 1,ref The standard feed rate is given by k4, which is the influence coefficient of the top pressure of the column, and its value is 25.

[0091] k1, k2, k3, and k4 are all weight values ​​determined during system debugging, reflecting the degree of influence of each parameter on the concentration.

[0092] Furthermore, the historical bias correction term C hist The historical analysis unit yields the historical deviation correction term C, used to compensate for the systematic error in prediction. hist The calculation process is as follows:

[0093]

[0094] C hist This is a historical deviation correction term, which adjusts the real-time parameter value C based on the deviation between historical predictions and actual concentrations. real Provide compensation;

[0095] This represents the average predicted NMP concentration at the top of the tower over past unit time periods, reflecting the overall trend of historical predictions. The average NMP concentration at the top of the column over a past unit of time was collected using an online gas chromatograph. Compared with the predicted average NMP concentration at the top of the tower per unit time in the past Subtracting them yields the historical deviation value;

[0096] γ is a correction factor, initially set to 0.3. The correction factor γ is then compared with... Multiplication is used to control the strength of compensation for historical biases in the current forecast. > This indicates that historical forecasts are lower than expected. If the result is positive, then add them together. This will cause the historical bias correction term C to... hist Increase, when < ， This indicates that historical forecasts are too high. If the value is negative, then it is further compared with... Adding them together will reduce the historical bias correction term C. hist ;

[0097] Specifically, if past forecasts have consistently been low, compensation is provided using a correction coefficient γ to improve forecast stability and accuracy. The real-time parameter value C... real It may be prone to deviation due to sensor drift and measurement noise.

[0098] By calculating the actual average NMP concentration at the top of the tower over past unit time periods. Compared with the predicted average NMP concentration at the top of the tower per unit time in the past The historical deviation value is obtained, and the historical deviation value is used to correct the current prediction, thereby improving the accuracy of the predicted NMP concentration data.

[0099] in:

[0100] The following table provides an example of parameter calculations for the concentration prediction module:

[0101]

[0102] Furthermore, the benefit assessment module acquires data from the concentration prediction module and predicts the expected effects of different parameter combinations based on real-time waste liquid recovery data, thereby obtaining a comprehensive benefit score. The specific process is as follows:

[0103] (1) Based on the top extraction rate F3 and the predicted NMP concentration C at the top of the tower. pred Calculate the amount of NMP extracted from the top of the tower: NMP extracted from the top of the tower = F3 × C pred ;

[0104] (2) Based on the feed rate F1 and the feed NMP concentration C feed Calculate the NMP feed rate: NMP feed rate = F1 × C feed ;

[0105] (3) Calculate the NMP recovery rate YH, the process is as follows:

[0106]

[0107] F3 is the extraction speed at the top of the tower, (F3 × C) pred This indicates the amount of NMP extracted from the top of the tower;

[0108] F1 is the feed rate, C feed The feed NMP concentration is (F1×C) feed () indicates the NMP feed rate;

[0109] E all The relative energy consumption ratio (%) represents the proportion of the total energy consumption of the distillation system to the standard energy consumption. The total energy consumption includes the reboiler heating energy consumption and the reflux pump energy consumption.

[0110] W3 is the total energy consumption impact coefficient, with a value of 0.2, W3×E all This represents the energy consumption penalty score; the higher the value, the more energy the system consumes. Therefore, the concentration benefit score W1×C is calculated. pred The sum of the recovery benefit W2×YH minus the energy consumption penalty W3×Eall The energy consumption penalty is W3×E. all Negative impact on overall benefit score A;

[0111] (4) Obtain the tower reboiler heating consumption E from the energy consumption information. heat and return pump consumption E pump Calculate the total energy consumption: Total energy consumption = E heat +E pump ;

[0112] Among them, the heating consumption of the tower pot is E heat The calculation process is as follows;

[0113]

[0114] T1 is the reboiler temperature, and F1 is the feed rate. The reboiler temperature T1 × feed rate F1 represents the total heating load of the reboiler per unit time. When the temperature increases or the feed rate increases, the reboiler heating consumption E... heat This will increase accordingly, ensuring that the system dynamically calculates heating energy consumption based on the actual load;

[0115] k heat The heating energy consumption coefficient, with a value of 0.02 kWh / (°C·L), represents the heating energy consumption per unit reboiler temperature and per unit feed rate. The heating energy consumption coefficient k... heat By keeping the feed rate F1 constant and adjusting the reboiler temperature T1 to different values, the power consumption of the reboiler heater is measured in real time using a power meter. The average value is obtained after conducting multiple sets of experiments.

[0116] Return pump consumes E pump The calculation process is as follows:

[0117]

[0118] k pump The reflux energy consumption coefficient, 0.015 kW·h / (L), represents the reflux pump energy consumption per unit reflux ratio and per unit feed rate. The reflux pump consumes E... pump By keeping the feed rate F1 constant, adjusting the real-time reflux ratio R to different values, measuring the power consumption of the reflux pump, and taking the average value after conducting multiple sets of experiments;

[0119] R is the real-time reflux ratio, calculated as the ratio of reflux flow rate to the top-collected flow rate. The real-time reflux ratio R × feed rate F1 represents the total circulating load of the reflux pump per unit time, quantifying the energy consumption of the system's reflux process. When the reflux ratio increases or the feed rate increases, the reflux pump consumes E... pump As a result, it increases;

[0120] (5) Calculate the relative energy consumption ratio E all The process is as follows:

[0121]

[0122] Where E ref Baseline energy consumption;

[0123] E heat E is the cost of heating the reboiler. pump Consumed by the return pump;

[0124] E ref The baseline energy consumption represents the target energy consumption under standard operating conditions.

[0125] E heat +E pump This represents the total energy consumption, and is expressed by comparing the total energy consumption with the baseline energy consumption E. ref By comparing absolute energy consumption to relative proportions, the deviation of the current system's actual energy consumption from the standard is quantified. heat +E pump >E ref Relative energy consumption ratio E all A value greater than 1 will lower the overall benefit score A; conversely, when E is less than 1, the overall benefit score A will decrease. heat +E pump <E ref Relative energy consumption ratio E all If the score is less than 1, the overall benefit score A will increase. This is achieved by adding an energy consumption penalty score W3×E to the overall benefit score A. all This can help the system optimize energy consumption control while ensuring purity and recovery rate, which meets the development needs of energy conservation and environmental protection.

[0126] (6) The predicted NMP concentration C at the top of the tower pred NMP recovery rate YH and relative energy consumption ratio E all The overall benefit score A is obtained by weighting and merging the results using weighted coefficients W1, W2, and W3. The specific process is as follows:

[0127]

[0128] A represents the comprehensive benefit score, which is the core indicator for measuring the overall effect. The larger the value, the better the comprehensive benefit of the current parameter combination, including purity, yield, and energy consumption balance.

[0129] C pred The predicted NMP concentration at the top of the column (%) represents the predicted NMP concentration per unit time in the future. The predicted NMP concentration at the top of the column is C. pred The purity of the NMP product at the top of the tower reflects its contribution; the higher the purity, the greater the overall benefit score A.

[0130] W1 is the concentration prediction influence coefficient, with a value of 0.5, representing the weight set by the system according to process requirements;

[0131] W1×C pred The concentration benefit score is represented by NMP concentration, which is a core benefit item of the system. The level of NMP concentration can significantly affect the comprehensive benefit score A.

[0132] YH represents the NMP recovery rate (%), which is obtained by comparing the amount of NMP recovered with the total amount of NMP in the feed. It reflects the recovery efficiency of NMP in the raw material. The higher the recovery rate, the greater the proportion of NMP effectively recovered from the feed and the higher the raw material utilization rate.

[0133] W2 is the recovery rate influence coefficient, with a value of 0.3. W2×YH represents the recovery rate benefit score. The higher the NMP recovery rate YH, the higher the overall system benefit score A. Therefore, it is related to the concentration benefit score W1×C. pred The summation can fully reflect the NMP recovery rate YH and the concentration benefit W1×C pred Positive impact on overall benefit score A;

[0134] Where W1 + W2 + W3 = 1, and W1, W2 and W3 are preset values.

[0135] Specifically, by introducing a concentration benefit score W1×C into the overall benefit score A. pred The recovery rate benefit is W2×YH and the energy consumption penalty is W3×E. all The energy consumption penalty is W3×E. all The negative impact on the overall benefit score A should be avoided. Focusing solely on optimizing a single objective can lead to reduced benefits for other parameters. For example, considering only NMP purity requires continuously increasing the reboiler temperature or reflux ratio, which significantly increases heating energy consumption, reduces yield, and increases production costs. The benefit assessment module uses a weighted combination of key indicators in the NMP waste liquid recovery process to obtain the overall benefit score A. This facilitates quantitative comparison of key indicators for various parameters, allowing for the priority selection of the parameter combination with the highest overall benefit. Furthermore, by setting influence coefficients for each parameter, users can adjust them according to actual conditions, improving system applicability.

[0136] Furthermore, when the overall benefit score is less than the target benefit score, and the difference between the overall benefit score and the target benefit score is greater than a set value, the target benefit score is set to 90. According to historical data, when the score is 2 to 3 units lower, the key parameters of the NMP waste liquid recovery distillation system show significant deterioration. Therefore, the set value is set to 3. When the target benefit score - overall benefit score A > 3, the optimization and adjustment module is automatically activated to adjust the system parameters. The specific process is as follows:

[0137] First, to ensure process safety and equipment limitations, the optimization variables need to be constrained, as follows:

[0138]

[0139] Reboiler temperature T1 constraint: NMP is easily decomposed at high temperatures, with a decomposition temperature of approximately 230°C. If the temperature is too low, gas-liquid separation cannot be achieved. Therefore, the reboiler temperature T1 must be limited to a safe and effective range.

[0140] Reflux ratio R constraint: If the reflux ratio is too small, the gas-liquid contact will be insufficient, resulting in insufficient purity at the top of the column. If the reflux ratio is too large, the energy consumption will increase. Therefore, it is necessary to balance purity and energy consumption.

[0141] Feed rate F1 constraint: Too slow a feed rate will reduce the waste liquid treatment efficiency, while too fast a feed rate will easily lead to excessive gas-liquid load in the column, which will easily cause flooding and mist entrainment, reducing the stability of distillation. Therefore, the feed rate F1 needs to be controlled.

[0142] After the constraints are determined, at least 50 sets of candidate parameters are randomly generated. Each set of candidate parameters includes the reboiler temperature T1, reflux ratio R, and feed rate F1 that meet the constraints. Before the system is run, the relevant system parameters need to be set, and the parameters also need to meet the constraints.

[0143] Each set of candidate parameters is substituted into the concentration prediction module and the benefit evaluation module respectively, and a comprehensive benefit score is obtained. When substituting each set of candidate parameters, the safety process requirements must be met, that is, the difference between the two sets of candidate parameters should not be too large to avoid damaging the system equipment.

[0144] Compare all comprehensive benefit scores, take the comprehensive benefit score with the largest value as the adjustment reference, and adjust the system parameters according to the candidate parameters corresponding to the comprehensive benefit score with the largest value;

[0145] When the difference between the comprehensive benefit score and the target benefit score corresponding to the candidate parameter with the highest value is greater than the set value, the management personnel will be notified through the alarm module.

[0146] Specifically, by optimizing and adjusting the module, various combinations of key parameters in NMP waste liquid recovery are compared, and the optimal system parameters for waste liquid recovery distillation are automatically selected without manual adjustment, reducing human intervention, improving the system's intelligence level, ensuring the real-time performance and accuracy of waste liquid recovery and treatment, and improving the purity and recovery efficiency of NMP waste liquid recovery.

[0147] The optimization and adjustment module includes a control terminal, which is used for system parameter setting and data visualization. By setting up the control terminal, system parameters can be adjusted. In emergency situations, it can be manually controlled by staff to reduce losses. Through data visualization, the system's operating status is displayed intuitively in the form of charts, helping managers to quickly judge process trends.

[0148] The alarm module includes an audible and visual alarm unit and a pop-up alarm unit. The audible and visual alarm unit is installed on the system equipment and includes an alarm light and a buzzer. The pop-up alarm unit is set in the control terminal for remote early warning.

[0149] By setting up alarm units, on-site personnel and remote management personnel can be notified in a timely manner, thereby enabling rapid response measures to reduce casualties and property losses.

[0150] When key system parameters such as reboiler temperature T1, reflux ratio R, and feed rate F1 do not meet the constraints, the alarm module can be activated. In practical applications, managers can set equipment parameter thresholds according to actual process requirements to ensure safety while carrying out production work.

[0151] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart control system for NMP waste liquid recovery and distillation, characterized in that: It includes a data acquisition module, a concentration prediction module, a benefit evaluation module, an optimization and adjustment module, an alarm module, and a data storage module; The data acquisition module collects waste liquid recovery data in real time, including feed information, column bottom temperature, column top pressure, reflux information, energy consumption information, and column top NMP concentration. After dynamic calibration, the data is sent to the data storage module, which includes real-time waste liquid recovery data and historical waste liquid recovery data. The concentration prediction module performs weighted analysis on real-time waste liquid recovery data and historical waste liquid recovery data to predict the NMP concentration in the future. The concentration prediction module includes a real-time analysis unit and a historical analysis unit. The real-time analysis unit is used to integrate and analyze various real-time parameters of waste liquid recovery, and the historical analysis unit is used to analyze the deviation between historical waste liquid recovery data and real-time waste liquid recovery data. The benefit assessment module predicts the expected effects of different parameter combinations based on future NMP concentrations and real-time waste liquid recovery data, and obtains a comprehensive benefit score. The benefit assessment module calculates the overall benefit score according to the following steps: (1) Based on the top extraction rate F3 and the predicted NMP concentration C at the top of the tower. pred Calculate the amount of NMP extracted from the top of the tower: NMP extracted from the top of the tower = F3 × C pred ; (2) Based on the feed rate F1 and the feed NMP concentration C feed Calculate the NMP feed rate: NMP feed rate = F1 × C feed ; (3) Calculate the NMP recovery rate YH, the process is as follows: ； (4) Obtain the tower reboiler heating consumption E from the energy consumption information. heat and return pump consumption E pump Calculate the total energy consumption: Total energy consumption = E heat +E pump ; (5) Calculate the relative energy consumption ratio E all The process is as follows: ； Where E ref Baseline energy consumption; (6) The predicted NMP concentration C at the top of the tower pred NMP recovery rate YH and relative energy consumption ratio E all The overall benefit score A is obtained by weighting and merging the results using weighted coefficients W1, W2, and W3. The specific process is as follows: ； Where W1 + W2 + W3 = 1, and W1, W2 and W3 are preset values; A target benefit score is set. When the difference between the target benefit score and the comprehensive benefit score exceeds a set value, the system parameters are automatically adjusted by the optimization and adjustment module to optimize the system. The process is as follows: The reboiler temperature T1, reflux ratio R, and feed rate F1 are obtained from the real-time waste liquid recovery data and constrained as follows: ； Multiple sets of candidate parameters are generated through a preset parameter combination algorithm. Each set of candidate parameters includes the reboiler temperature T1, reflux ratio R, and feed rate F1 that meet the constraints. Each set of candidate parameters is substituted into the concentration prediction module and the benefit evaluation module to obtain multiple sets of comprehensive benefit scores A; Compare all comprehensive benefit scores A, and adjust the system parameters according to the candidate parameters with the highest comprehensive benefit score; When the difference between the comprehensive benefit score A corresponding to the candidate parameter with the highest value and the target benefit score is greater than the preset threshold, the management personnel will be notified through the alarm module.

2. The NMP waste liquid recovery and distillation intelligent control system according to claim 1, characterized in that: The concentration prediction module's prediction process is as follows: ； ； C pred The predicted NMP concentration at the top of the tower; C real These are real-time parameter values; C hist This is a historical deviation correction term, calculated based on the average deviation between historical predicted concentrations and actual concentrations. D tot The total fluctuation of real-time parameters is obtained by calculating the sum of the standard deviations of the top temperature, top pressure, and feed rate over the past unit time. D ref The standard fluctuation threshold is set to 0.5, which is used to determine whether the system is in a stable state. The unit of time is 1 hour.

3. The NMP waste liquid recovery and distillation intelligent control system according to claim 2, characterized in that: The real-time analysis unit first obtains the real-time temperature, pressure, feed rate, and reflux ratio at the top of the column from the real-time waste liquid recovery data. It then compares these values ​​with their standard values ​​and weights and merges them to analyze the changing trend of NMP concentration under the current process conditions, obtaining the real-time parameter value C. real .

4. The NMP waste liquid recovery and distillation intelligent control system according to claim 3, characterized in that: The historical analysis unit calculates the difference between the average actual concentration and the average predicted concentration over a past unit of time, and introduces a correction coefficient to obtain the historical deviation correction term C. hist According to the historical deviation correction term C hist Reverse correction of real-time parameter value C real .

5. The NMP waste liquid recovery and distillation intelligent control system according to claim 4, characterized in that: The optimization and adjustment module includes a control terminal for setting system parameters and visualizing data.

6. The NMP waste liquid recovery and distillation intelligent control system according to claim 5, characterized in that: The alarm module includes an audible and visual alarm unit and a pop-up alarm unit. The audible and visual alarm unit is installed on the system equipment and includes an alarm light and a buzzer. The pop-up alarm unit is located in the control terminal and is used for remote early warning.

Images