CN111474856A - Concentration soft measurement method for dividing wall rectifying tower - Google Patents
Concentration soft measurement method for dividing wall rectifying tower Download PDFInfo
- Publication number
- CN111474856A CN111474856A CN202010446679.2A CN202010446679A CN111474856A CN 111474856 A CN111474856 A CN 111474856A CN 202010446679 A CN202010446679 A CN 202010446679A CN 111474856 A CN111474856 A CN 111474856A
- Authority
- CN
- China
- Prior art keywords
- concentration
- model
- temperature
- dividing wall
- control
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B11/00—Automatic controllers
- G05B11/01—Automatic controllers electric
- G05B11/36—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential
- G05B11/42—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential for obtaining a characteristic which is both proportional and time-dependent, e.g. P.I., P.I.D.
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
- G05B13/027—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks only
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D11/00—Control of flow ratio
- G05D11/02—Controlling ratio of two or more flows of fluid or fluent material
- G05D11/13—Controlling ratio of two or more flows of fluid or fluent material characterised by the use of electric means
- G05D11/135—Controlling ratio of two or more flows of fluid or fluent material characterised by the use of electric means by sensing at least one property of the mixture
- G05D11/138—Controlling ratio of two or more flows of fluid or fluent material characterised by the use of electric means by sensing at least one property of the mixture by sensing the concentration of the mixture, e.g. measuring pH value
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D11/00—Control of flow ratio
- G05D11/16—Controlling mixing ratio of fluids having different temperatures, e.g. by sensing the temperature of a mixture of fluids having different viscosities
Abstract
The invention relates to a soft concentration measurement method for a dividing wall rectifying tower, belonging to a soft measurement method in a chemical process. The concentration soft measurement method integrated with the data driving model increases concentration soft measurement on the basis of the cascade control of the temperature and the concentration of the dividing wall rectifying tower, thereby realizing the prediction of the product concentration by detecting easily detected variables. The inner layer in the cascade control adopts proportional-integral control to control the temperature of a sensitive plate of the dividing wall rectifying tower; and the outer layer adopts model prediction control to control the product concentration. By utilizing the method, the auxiliary variable easy to detect is fully utilized to predict the product concentration, so that the high price and long time delay of real-time online concentration detection are avoided; the proportional-integral control and the model prediction control are combined fully, and the quick differential-free control of the dividing wall rectifying tower is realized.
Description
Technical Field
The invention relates to a product concentration control method of a dividing wall rectifying tower, in particular to a concentration soft measurement method integrated with a data driving model, belonging to a soft measurement method of a chemical process.
Background
The dividing wall rectifying tower is a thermally coupled rectifying tower which integrates two (or more) rectifying towers into one tower shell to realize the separation of ternary (and above) systems. Compared with the traditional rectifying tower system, the dividing wall rectifying tower can generally save about 30 percent of energy consumption, about 30 percent of equipment investment and space. The four-component Kaibel dividing wall rectifying tower is an important type of dividing wall rectifying tower and consists of a pre-dividing tower and a main tower, wherein the two towers are connected with each other through two pairs of countercurrent thermally coupled vapor-liquid streams. The top of the main tower, the side extraction 1, the side extraction 2 and the bottom of the tower respectively extract four products, and the whole four-component Kaibel dividing wall rectifying tower only needs one top condenser and one bottom reboiler. Ghaddran et al, norway, studied the correlation of manipulating the gas phase split ratio with minimum energy consumption operation in a Kaibel dividing wall rectification column, indicating that changing the gas phase split ratio on-line may be a necessary condition to maintain minimum energy consumption for operation. Dwivedi et al, Norwegian proposed a control structure for a Kaibel dividing wall rectification column with variable gas phase split ratio. The Kaibel dividing wall rectifying tower for separating benzene, toluene, xylene and mesitylene is researched by Fuhao et al in China, and the obtained result shows the necessity of controlling the top and the bottom of the pre-separating rectifying tower. The results of more intensive studies on the temperature control, concentration control and cascade system of the Kaibel dividing wall rectifying tower for separating methanol, ethanol, n-propanol and n-butanol by the chien people and the like show that the pure temperature control by adopting a liquid phase division ratio (or a gas phase division ratio) as an operation variable can quickly stabilize the Kaibel dividing wall rectifying tower and realize the effect of stabilization control. However, pure temperature control has a steady state residual difference in product concentration, and therefore, control of product concentration is also required. The conventional concentration detection is prolonged, expensive and incapable of realizing continuous measurement, and according to MESH equations (material balance equation, phase balance equation, normalization equation and heat balance equation of each component of each balance stage) of the rectifying tower, variables (concentration, temperature, pressure and flow rate) in the rectifying tower are correlated, so that the product concentration can be predicted through auxiliary variables (temperature, pressure and flow rate) which are easy to detect, and the soft measurement of the product concentration of the rectifying tower with the partition wall is realized.
Disclosure of Invention
In order to improve the control effect of the dividing wall rectifying tower, the invention provides a concentration soft measurement method integrated with a data driving model, which adds concentration soft measurement on the basis of the cascade control of the temperature and the concentration of the dividing wall rectifying tower, predicts the product concentration through an easily-detected variable and replaces the actual high-frequency concentration detection.
In order to achieve the purpose, the technical scheme of the invention is realized as follows:
the invention relates to a concentration soft measurement method of a dividing wall rectifying tower, which has the innovation point that the concentration soft measurement is applied to the temperature concentration cascade control of the dividing wall rectifying tower, but not the improvement of the method of each module; the system comprises a model prediction control module, a concentration predictor and a proportional integral control module, wherein the model prediction control module is used for controlling the concentration of a product, the concentration predictor is used for predicting the concentration of the product, and the proportional integral control module is used for controlling the temperature of a sensitive plate; the collected feedback signal comprises the concentration C output by the dividing wall rectifying towerkTemperature T of sensitive platekAnd easily detectable auxiliary variables (including pressure P)kFlow rate FkAnd temperature T 'of product extraction plate'k) (ii) a The change of the concentration of the components in the rectifying tower can cause the change of the temperature, the more heavy components are, the higher the temperature is, and the more light components are, the lower the temperature is, so that the product concentration can be indirectly controlled through temperature control, and the stable operation of the rectifying tower is realized; however, the steady state residual difference of the product concentration cannot be eliminated by pure temperature control, so that the inner layer temperature control and the outer layer concentration control need to be realized by adopting temperature concentration cascade control; the detection of concentration is prolonged and expensive, so that the concentration cannot be continuously measured; according to MESH equations (material balance equation, phase balance equation, normalization equation and heat balance equation of each component of each balance stage) of the rectifying tower, variables (concentration, temperature, pressure and flow rate) in the rectifying tower are correlated, and the partition wall can be predicted through auxiliary variables (temperature, pressure and flow rate) which are easy to detectProduct concentration of the rectification column; in the temperature concentration cascade control, the inner layer controls the temperature of the sensitive plate through a proportional-integral control module, and the outer layer controls the concentration of a product through a model prediction control module; sensitive plate temperature T detected by temperature detector in inner layer temperature control loopkOn one hand, inputting a concentration predictor, on the other hand, comparing the concentration predictor with a set value TCSP of a temperature controller to obtain a deviation value, and inputting the deviation value into a proportional-integral controller, thereby realizing the control of the temperature of a sensitive plate of the dividing wall rectifying tower; the outer layer concentration control loop comprises two signal loops, wherein the first loop is a concentration prediction loop comprising a concentration predictor, and the second loop is a concentration detection loop comprising a concentration detector; the inputs to the concentration predictor are readily detectable auxiliary variables, including the sense plate temperature TkAnd temperature T 'of a product extraction plate'kPressure PkFlow rate FkAnd the concentration of the previous step<C>k-1The output of the density predictor being the predicted density
A prediction model of the concentration predictor is a nonlinear autoregressive neural network model, is used as a prediction model of soft measurement, and is subjected to system identification to obtain parameters; the input to the error corrector is the predicted density value
And the concentration detection value C of the detectork(once per hour), the output of the error corrector is the corrected concentration value<C>kThe error corrector adopts extended Kalman filtering to correct errors, and the core idea is to utilize the local linear characteristic of a nonlinear prediction model to carry out local linearization on the nonlinear prediction model at each moment and near different concentrations so as to obtain corrected concentration; the specific calculation steps of the extended Kalman filtering for error correction comprise: (1) the method comprises the steps of (1) correcting a nonlinear autoregressive neural network prediction model near a concentration point at the previous moment, (2) calculating a covariance matrix of a one-step concentration prediction error, (3) calculating a Kalman gain, (4) calculating a correction concentration, (5) updating a covariance matrix of the concentration prediction errorThe array is used for the extended Kalman filtering correction at the next moment; correcting density value<C>kComparing with a product concentration set point, inputting the obtained deviation into a model predictive controller, and outputting a temperature controller set value TCSP by the model predictive controller, thereby realizing the cascade control of the temperature concentration of the dividing wall rectifying tower; the model prediction control comprises a prediction model, feedback correction and online optimization; the model prediction controller adopts a state space model as a prediction model, and is obtained by strict model linearization of a dividing wall rectifying tower; the feedback correction adopts the simplest deviation based on the model calculation value and the detection value, and corrects the model calculation value to obtain a model prediction value; the objective function of the on-line optimization is a multi-objective function including a square representing the deviation of the controlled variable from the set value and a square representing the variation of the manipulated variable, and the objective function of the on-line optimization is to minimize the objective function.
The technical scheme shows that the invention has the following beneficial effects.
(1) By utilizing the method, the product concentration can be predicted by utilizing auxiliary variables easy to detect, high-frequency concentration detection is realized, and high price and long time delay of real-time online concentration detection are avoided;
(2) by utilizing the method, the combination of proportional-integral control and model predictive control can be fully utilized, and the quick and error-free control of the dividing wall rectifying tower is realized.
Drawings
FIG. 1 is a temperature concentration cascade control structure combining proportional-integral control and model predictive control of a Kaibel dividing wall rectifying tower provided by the invention; FIG. 1 shows a pre-separation section; 2 is a main separation section; 3 is a condenser; 4 is a reboiler;
FIG. 2 is a flow chart of cascade control of a Kaibel dividing wall rectifying column after soft measurement is added.
Detailed Description
The invention provides a concentration soft measurement method integrated with a data driving model.
In order that the objects, aspects and advantages of the present invention will become more apparent, the present invention will be further described in detail with reference to the accompanying drawings.
The Kaibel dividing wall rectifying tower used in the following is a single tower and comprises a pre-separation section 1, a main separation section 2, a condenser 3 and a reboiler; a condenser is arranged at the top end of the tower, and a reboiler is arranged at the bottom end of the tower. The inner layer adopts proportional-integral control to control the temperature of the sensitive plate, so that the rapid and stable performance of the dividing wall rectifying tower is ensured; the outer layer adopts model prediction control to control the product concentration, and the product purity requirement is met. The inputs to the concentration predictor are easily detectable auxiliary variables (e.g., temperature, pressure, flow). The prediction model selects a nonlinear autoregressive neural network model as a soft-measurement concentration prediction model. And the error correction adopts extended Kalman filtering.
Example 1: the concentration soft measurement method integrated with the data driving model is utilized to predict and control the concentration of the Kaibel dividing wall rectifying tower.
In FIG. 1, the total number of the pre-separation section is 40, and the number of the feeding plates is 20; the total number of plates in the main separation section is 70, the number of gas-phase feed plates is 21, the number of liquid-phase feed plates is 60, the number of side line 1 take-off plates is 30, and the number of side line 2 take-off plates is 50. The feeding materials are saturated liquid feeding materials with mole fractions of methanol, ethanol, n-propanol, n-butanol and the like, and the feeding flow rate is 1000 mol/h. The heat duty of the condenser at the top of the tower is 26.007kW, and the heat duty of the reboiler at the bottom of the tower is 25.107 kW. The operation pressure at the top of the tower is 1atm, and the pressure drop of the single plates in the tower is 0.0068 atm. The overhead molar reflux ratio was 9.604. The gas phase separation ratio was 0.515, and the liquid phase separation ratio was 0.352. The product purity in the four product streams was 99% mole fraction.
The inner layer in the temperature concentration cascade control adopts proportional-integral control to control the temperature of the sensitive plate, so that the rapid stability of the dividing wall rectifying tower is ensured; the outer layer adopts model prediction control to control the product concentration, and the product purity requirement is met. The 4 inputs (controlled variables) of the inner proportional-integral control are the temperature of the 12 th sensitive plate in the pre-separation section and the temperatures of the 12 th, 43 th and 64 th sensitive plates in the main separation section, which respectively correspond to the detection quantity inputs of the temperature controllers TCP, TC1, TC2 and TC3 in FIG. 1. The 4 outputs (manipulated variables) of the inner layer proportional-integral control are the liquid phase coupling stream flowing back into the pre-separation section from the main separation section, the tower top reflux quantity of the main separation section, the side line 1 product flow and the side line 2 product flow, and are respectively corresponding to the manipulated variable outputs of the temperature controllers TCP, TC1, TC2 and TC3 in the graph 1. The 6 inputs (controlled variables) for the outer layer model predictive control are the four product stream product purities and the two pre-splitter top and bottom impurity concentrations obtained by the soft measurement module, which correspond to the measured quantity inputs of the MPC in fig. 1, respectively. The 6 outputs (manipulated variables) of the outer model predictive control are the set points for TCP, TC1, TC2, and TC3 (collectively referred to as TCSP), and the main column reboiler heat duty, the flow of the gas coupled stream from the main column reflux to the pre-splitter, corresponding to the manipulated variable outputs of the MPC in fig. 1, respectively.
Sensitive plate temperature T detected by temperature detector in inner layer temperature control loopkThe temperature of the 12 th sensitive plate in the pre-separation section and the temperatures of the 12 th, 43 th and 64 th sensitive plates in the main separation section respectively correspond to the detection quantity input of a temperature controller TCP, TC1, TC2 and TC3 in the figure 1, an outer-layer concentration control loop comprises two signal loops, the first loop is a concentration prediction loop comprising a concentration predictor, and the second loop is a concentration detection loop comprising a concentration detector; auxiliary variables easy to detect in the first pass include product panel temperature T'kI.e. temperature of the 20 th, 30 th, 50 th and 60 th tray of the main column and reboiler temperature and heat load at the bottom of the column, pressure PkI.e. the top pressure, flow rate F of the main separation sectionkNamely the feeding flow, the liquid phase reflux quantity, the gas phase reflux quantity, the top product flow of the main separation section, the side line 1 product flow, the side line 2 product flow and the bottom product flow; c detected by the concentration detector in the second pathkNamely the product purity of the four product streams and the impurity concentration at the top and bottom of the two pre-splitters. The manipulated variables output by the proportional integral controller are the liquid phase coupling stream flowing back into the pre-separation section from the main separation section, the overhead reflux quantity of the main separation section, the product flow quantity of the side line 1 and the product flow quantity of the side line 2, and are respectively output corresponding to the manipulated variables of the temperature controllers TCP, TC1, TC2 and TC3 in the graph 1.
On the basis of a temperature concentration cascade control structure, a soft measurement technology consisting of a concentration predictor and an error corrector is further adopted to replace the measurement of the concentration. FIG. 2 is a flow chart of cascade control of a Kaibel dividing wall rectifying column after soft measurement is added.
When the state change of the rectifying tower is large, the linear concentration prediction model causes large errors, and the product concentration has large steady-state residual difference. Therefore, a non-linear autoregressive neural network model is selected as the concentration prediction model in the soft-measure concentration predictor. The input is an auxiliary variable (in T) at the current momentkFor example), and the corrected density at the previous time<C>k-1. And the output is the predicted concentration of the current time
Therefore, the non-linear autoregressive neural network model is defined by the equation:
the formula of the nonlinear autoregressive neural network model is as follows:
nk＝w1Tk+w2<C>k-1+b1(2)
in the formula, w1And w2To input the weight, w3Is the layer weight; b1And b2The number of hidden layer nodes of the neural network is 21, the parameters (weights and bias) of the prediction model are obtained by system identification, and the mean square error and regression value of the identification in the example are 3.18 × 10 respectively-10And 0.9999, which means that the recognition results of the nonlinear autoregressive neural network model are good.
Because the prediction model in the concentration predictor is necessarily deviated from the actual rectifying tower model, when the deviation is accumulated for a period of time, the predicted concentration needs to be corrected through an error corrector. The prediction error of the concentration can be expressed as the detected concentration and the predicted concentrationA difference of (i) that
The purpose of the correction is to derive a corrected concentration from the prediction error<C>k. The formula for the correction is:
wherein, KkIs the kalman gain, the magnitude of which is obtained by the extended kalman filter algorithm.
When a linear prediction model is used, a linear kalman filter may be used for the correction. When the prediction model is a nonlinear model, extended kalman filtering may be used for the correction. In this example, the concentration detection is realized only once every 1 hour by the conventional nonlinear extended kalman filtering. The extended Kalman filtering is based on Kalman filtering, and the core idea is to utilize the local linear characteristics of a nonlinear prediction model to carry out local linearization on the nonlinear prediction model at each moment and around different concentrations. That is to say in<C>k-1Concentration points, versus a nonlinear autoregressive neural network model f (T)k，<C>k-1) Taylor expansion is performed and only the first order term is retained.
Extended Kalman Filter calculation KkAnd the specific calculation steps for error correction are as follows:
the first step is as follows: linearizing the non-linear prediction model near the last time correction concentration point:
the second step is that: calculating a covariance matrix of the concentration prediction error:
where Q is a covariance matrix of the process noise, representing the perturbation of the prediction model by external conditions, and is usually set to a constant value.
The third step: calculating the Kalman gain Kk：
Wherein HkObtained by linearizing the measurement model. In this example, since the output value and the measured value of the prediction model are both 6 strand concentrations, H iskAn identity matrix of 6 × 6, R is a covariance matrix of measurement noise, and may be set to a non-zero constant value, indicating a measurement error between a detection value and a true value of the sensor.
The fourth step: calculating a correction concentration:
the fifth step: and updating the covariance matrix of the concentration prediction error for the extended Kalman filter correction at the next moment:
Pk＝(I-KkHk)P′k(9)
it can be seen that in the concentration correction process, PkIs a time-varying value, and its initial value is set as the covariance matrix of the initial time density prediction error.
As shown in fig. 2, the corrected density value<C>kThe resulting deviation is input to the model predictive controller, compared to the concentration set point. The model prediction controller comprises three parts: prediction models, feedback correction and online optimization. In this example, the prediction model in the model prediction controller is a state space model obtained by linearizing a strict dividing wall rectifying column model. Considering the complexity and the amount of computation, we choose the order of the discrete state space model to be 60. Feedback correction for correcting state space model calculation values and concentrations<C>kThe deviation of (2). The objective function of the online optimization is a linear combination of two control performance indicators: the sum of the squares of the deviations of the controlled variable from the set point and the sum of the squares of the variations of the manipulated variable.And the online optimization enables the objective function to reach the minimum value, so that the model predictive controller has the optimal control performance index.
In the objective function, the setting of the weight parameters of the linear combination is completed through multi-objective optimization: on a Pareto front edge formed by a multi-target optimal solution set, two control performance indexes (the sum of squares of deviations of controlled variables and set values and the sum of squares of variable quantities of manipulated variables) are comprehensively considered, and an optimal solution on an inflection point with smaller indexes is selected as a weight.
As shown in fig. 2, the output of the model predictive controller is a set value TCSP of a temperature controller (proportional integral controller), thereby realizing cascade control of the temperature concentration of the dividing wall rectifying column.
The temperature concentration cascade control structure for predicting the concentration by adopting the soft measurement technology can obtain a very good dynamic response effect, has the advantages of minimum oscillation, shortest time required for stabilization and very small maximum deviation and steady-state deviation.
The method fully utilizes auxiliary variables easy to detect to predict the product concentration, and avoids high price and long time delay of real-time online concentration detection; the proportional-integral control and the model prediction control are combined fully, and the quick differential-free control of the dividing wall rectifying tower is realized.
The purpose, technical scheme and beneficial effects of the invention are further explained by the specific implementation examples. It should be noted that the above description is only a specific embodiment of the present invention, and they are not intended to limit the present invention. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. A concentration soft measurement method of a dividing wall rectifying tower is characterized in that: temperature concentration cascade control structure based on dividing wall rectifying tower and used for controlling product concentration of dividing wall rectifying towerThe inner layer controls the temperature of the sensitive plate through a proportional-integral control module, and the outer layer controls the concentration of a product through a model prediction control module; the system comprises a model prediction control module, a concentration predictor and a proportional integral control module, wherein the model prediction control module is used for controlling the concentration of a product, the concentration predictor is used for predicting the concentration of the product, and the proportional integral control module is used for controlling the temperature of a sensitive plate; the collected feedback signal comprises the concentration C output by the dividing wall rectifying towerkTemperature T of sensitive platekAnd auxiliary variables that are easy to detect; sensitive plate temperature T detected by temperature detector in inner layer temperature control loopkOn one hand, inputting a concentration predictor, on the other hand, comparing the concentration predictor with TCSP output by the model prediction control module to obtain a deviation value, and inputting the deviation value into a proportional-integral controller, thereby realizing the control of the temperature of a sensitive plate of the isolation wall rectifying tower; the outer layer concentration control loop comprises two signal loops, wherein the first loop is a concentration prediction loop comprising a concentration predictor, and the second loop is a concentration detection loop comprising a concentration detector; the input of the concentration predictor is the detected auxiliary variable, and the temperature T of the sensitive plate detected by the temperature detectorkAnd the output density of the error corrector in the previous step<C>k-1The output of the density predictor is the predicted density at the current time
The input to the error corrector is the predicted density value
And the concentration detection value C of the detectorkThe output of the error corrector is the corrected density value<C>k(ii) a Correcting density value<C>kAnd comparing with a product concentration set point, inputting the obtained deviation into a model predictive controller, and outputting a temperature controller set value TCSP by the model predictive controller, thereby realizing the cascade control of the temperature concentration of the dividing wall rectifying tower.
2. Concentration of a dividing wall column according to claim 1The soft measurement method is characterized in that: said auxiliary variable comprising pressure PkFlow rate FkAnd temperature T 'of product extraction plate'k。
3. The method for soft measurement of concentration of a dividing wall rectifying column according to claim 1, wherein: the prediction model of the concentration predictor is a nonlinear autoregressive neural network model, is used as a prediction model of soft measurement, and is used for obtaining parameters through system identification.
4. The method for soft measurement of concentration of a dividing wall rectifying column according to claim 1, wherein: and the error corrector adopts extended Kalman filtering to carry out error correction.
5. The method for soft measurement of concentration of a dividing wall rectifying tower according to claim 4, wherein the method comprises the following steps: the specific calculation step of the extended Kalman filtering for error correction comprises the following steps: (1) the method comprises the steps of (1) correcting a nonlinear autoregressive neural network prediction model near a concentration point at the previous moment, (2) calculating a covariance matrix of a one-step concentration prediction error, (3) calculating Kalman gain, (4) calculating correction concentration, and (5) updating the covariance matrix of the concentration prediction error for extended Kalman filtering correction at the next moment.
6. The method for soft measurement of concentration of a dividing wall rectifying column according to claim 1, wherein: the model prediction control comprises a prediction model, feedback correction and online optimization; concentration by error corrector<C>kPerforming feedback correction on the original state space model to obtain a corrected prediction model according to the difference between the predicted value predicted by the original state space model and the original state space model; inputting the given initial TCSP into a correction prediction model to obtain a predicted concentration time sequence; inputting the concentration time sequence, the concentration set point and the manipulated variable quantity into online optimization to obtain an optimized manipulated variable TCSP which is used as the output of model predictive control; the original state space model passes through the partition wall in the above stepsThe strict model of the rectifying tower is obtained in a linear mode, the feedback correction is completed through linear Kalman filtering, and the minimum objective function of online optimization is linear summation of the square of the deviation of a controlled variable and a set value and the square of the variable quantity of a manipulated variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010446679.2A CN111474856B (en) | 2020-05-25 | 2020-05-25 | Concentration soft measurement method for dividing wall rectifying tower |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010446679.2A CN111474856B (en) | 2020-05-25 | 2020-05-25 | Concentration soft measurement method for dividing wall rectifying tower |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111474856A true CN111474856A (en) | 2020-07-31 |
| CN111474856B CN111474856B (en) | 2022-12-27 |
Family
ID=71762665
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010446679.2A Active CN111474856B (en) | 2020-05-25 | 2020-05-25 | Concentration soft measurement method for dividing wall rectifying tower |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111474856B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112295255A (en) * | 2020-10-24 | 2021-02-02 | 四川泸天化创新研究院有限公司 | Intelligent control system and control method for methanol rectification device |
| CN114967600A (en) * | 2021-10-22 | 2022-08-30 | 天津理工大学 | Process dynamic control scheme based on paraffin separation dividing wall rectifying tower |
| CN117570512A (en) * | 2024-01-16 | 2024-02-20 | 成都航空职业技术学院 | Building temperature control system based on temperature self-adaptive control model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101590331A (en) * | 2009-04-15 | 2009-12-02 | 华东理工大学 | Ethylene rectification tower sensitive plate temperature control system and control method thereof |
| CN102339040A (en) * | 2010-07-15 | 2012-02-01 | 清华大学 | Optimization and control method of rectification column |
| CN106680428A (en) * | 2016-12-19 | 2017-05-17 | 华北电力大学（保定） | Soft measuring method for denitration control system |
| CN109725525A (en) * | 2017-10-31 | 2019-05-07 | 中国石油化工股份有限公司 | The method that a kind of pair of plate distillation column is controlled |
| CN110960881A (en) * | 2018-09-28 | 2020-04-07 | 中国石油化工股份有限公司 | Control structure and control method of Kaibel bulkhead rectifying tower |
-
2020
- 2020-05-25 CN CN202010446679.2A patent/CN111474856B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101590331A (en) * | 2009-04-15 | 2009-12-02 | 华东理工大学 | Ethylene rectification tower sensitive plate temperature control system and control method thereof |
| CN102339040A (en) * | 2010-07-15 | 2012-02-01 | 清华大学 | Optimization and control method of rectification column |
| CN106680428A (en) * | 2016-12-19 | 2017-05-17 | 华北电力大学（保定） | Soft measuring method for denitration control system |
| CN109725525A (en) * | 2017-10-31 | 2019-05-07 | 中国石油化工股份有限公司 | The method that a kind of pair of plate distillation column is controlled |
| CN110960881A (en) * | 2018-09-28 | 2020-04-07 | 中国石油化工股份有限公司 | Control structure and control method of Kaibel bulkhead rectifying tower |
Non-Patent Citations (3)
| Title |
|---|
| X.QIAN ET AL.: ""MPC-PI cascade control for the Kaibel dividing wall column integrated with data-driven soft sensor model"", 《CHEMICAL ENGINEERING SCIENCE》 * |
| XING QIAN ET AL.: ""Comparison of Composition Soft Sensor Correction Methods for a Kaibel Dividing-Wall Column"", 《INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH》 * |
| 徐凤 等: ""基于软测量的醋酸精馏过程串级预测控制策略"", 《计算机与应用化学》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112295255A (en) * | 2020-10-24 | 2021-02-02 | 四川泸天化创新研究院有限公司 | Intelligent control system and control method for methanol rectification device |
| CN114967600A (en) * | 2021-10-22 | 2022-08-30 | 天津理工大学 | Process dynamic control scheme based on paraffin separation dividing wall rectifying tower |
| CN117570512A (en) * | 2024-01-16 | 2024-02-20 | 成都航空职业技术学院 | Building temperature control system based on temperature self-adaptive control model |
| CN117570512B (en) * | 2024-01-16 | 2024-03-19 | 成都航空职业技术学院 | Building temperature control system based on temperature self-adaptive control model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111474856B (en) | 2022-12-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111474856B (en) | Concentration soft measurement method for dividing wall rectifying tower | |
| Skogestad | Simple analytic rules for model reduction and PID controller tuning | |
| US7292899B2 (en) | Model predictive control having application to distillation | |
| Minh | Modeling and control of distillation column in a petroleum process | |
| Halvorsen et al. | Optimal operation of Petlyuk distillation: steady-state behavior | |
| EP0609814B1 (en) | Process for maximizing the recovery of argon from an air separation system | |
| Kim | Structural design and operation of a fully thermally coupled distillation column | |
| JPH0869302A (en) | Model-prediction control method for air separation system | |
| JPH0363081B2 (en) | ||
| CN111537679B (en) | Soft measurement method for concentration of dividing wall rectifying tower based on nonlinear filtering | |
| Hurowitz et al. | Distillation control configuration selection | |
| Shehu et al. | Applications of MPC and PI controls for liquid level control in coupled-tank systems | |
| Sivakumar et al. | Design and development of model predictive controller for binary distillation column | |
| Dutta et al. | Application of neural network control to distillation and an experimental comparison with other advanced controllers | |
| Huang et al. | Comparison of PI and MPC for control of a gas recovery unit | |
| Daosud et al. | Neural network based model predictive control of batch extractive distillation process for improving purity of acetone | |
| Karacan et al. | Application of optimal adaptive generalized predictive control to a packed distillation column | |
| Karacan | Application of a non-linear long range predictive control to a packed distillation column | |
| Cho et al. | Operation of divided wall column with vapor sidedraw using profile position control | |
| CN113304497B (en) | Control method and control system for rectifying tower under non-constant pressure condition | |
| TW202215179A (en) | Methods of operating process technology equipment, process technology equipment and methods of modifying process technology equipment The control process is a self-optimizing control process | |
| Pop et al. | Feedback linearization control design for the 13C cryogenic separation column | |
| Huang et al. | Offset-free nonlinear model predictive control based on moving horizon estimation for an air separation unit | |
| Han et al. | Nonlinear model based control of two-product reactive distillation column | |
| Chandra et al. | Multistep model predictive control of ethyl acetate reactive distillation column |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |