CN113110060B - Real-time optimization method of reforming device - Google Patents
Real-time optimization method of reforming device Download PDFInfo
- Publication number
- CN113110060B CN113110060B CN202110475629.1A CN202110475629A CN113110060B CN 113110060 B CN113110060 B CN 113110060B CN 202110475629 A CN202110475629 A CN 202110475629A CN 113110060 B CN113110060 B CN 113110060B
- Authority
- CN
- China
- Prior art keywords
- data
- cur
- independent variable
- variable data
- dependent variable
- 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.)
- Active
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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to a real-time optimization method of a reforming device, which comprises the following steps: acquiring historical operating data corresponding to constant parameters, independent variable parameters and dependent variable parameters of a reforming device in a period of time; calculating a Jacobian matrix A of the independent variable data to the dependent variable data; fitting each element in the A by adopting a multiple linear regression method, and extracting a fitting coefficient; collecting operation data in the current time as current working condition data; calculating a Jacobian matrix of the current working condition according to the fitting coefficient; and finally, determining an objective function related to the independent variable parameter MV or/and the dependent variable parameter CV, and searching a proper MV to maximize the objective function in the upper and lower limit ranges of the CV and the MV by adopting an optimization algorithm and taking the equality constraint between the CV and the MV as a constraint condition. The method reduces the calculation amount and the optimization calculation time, and the calculation efficiency and the optimization precision completely meet the real-time optimization requirement, so that the method is more practical than the existing mechanism model.
Description
Technical Field
The invention relates to the field of data optimization, in particular to a real-time optimization method for a reforming device.
Background
The real-time optimization (RTO) technology is the most advanced optimization technology at the present stage, and automatically seeks optimal operation parameters based on a device model and an online optimization algorithm under the condition of meeting production technical indexes, and transmits the optimal operation parameters to advanced control software (APC), so that the whole production device is maintained in an optimal operation state. The core of the real-time optimization technology lies in a device model and an optimization algorithm, the device model is required to accurately reflect the actual operation condition of the device, and the device model and the optimization algorithm are required to have high calculation efficiency.
The reforming reaction process is complex, and the main chemical reactions are as follows: dehydrogenation of six-membered ring alkane, isomerization and dehydrogenation of five-membered ring alkane, dehydrocyclization of alkane, isomerization of straight chain alkane, hydrogenolysis and hydrocracking of hydrocarbon, demethylation, dealkylation of aromatic hydrocarbon, carbon deposition reaction and the like. The reforming reaction components which can be analyzed by a gas chromatograph are more than 300, and the coupling among the components is strong, so that the whole reaction system is combined into a complex reaction network by a plurality of reactions such as parallel, serial, reversible and irreversible reactions, and the reforming reaction belongs to a complex reaction system. At present, most of models for real-time optimization of the reforming device are mechanism models, the mechanism models are built based on chemical principles, reaction engineering, chemical thermodynamics and actual production conditions of the device, and the production and operation conditions of the reaction device can be real and accurate. The mechanism model has the problem of low calculation efficiency, and can be solved by establishing an EO (simultaneous equation) mechanism model. However, for a mechanism model involving a complex chemical reaction system, even though the mechanism model is based on an EO (simultaneous equations), it is difficult to improve the calculation efficiency to the level required by real-time optimization, and the real-time optimization utilization rate is affected. Therefore, the real-time optimization of the reforming device by adopting the mechanism model has low calculation efficiency and influences the utilization rate of the real-time optimization.
In order to solve the technical problem, chinese patent application No. CN201910285487.5 (application publication No. CN 110187635A) discloses a real-time optimization method and apparatus for a continuous reformer, in which although the advantages of a mechanism model and a data experience model are combined, a deep learning network model is used for off-line learning and estimation of a correlation coefficient in a simplified model, so that a correlation mathematical model of the continuous reformer is established for on-line real-time calculation, and a network parameter of the deep learning network module directly affects an optimization result, so that a deep learning network module of the optimization method needs to be trained before estimating the correlation coefficient, the training time is long, and in addition, a constraint condition in a solving process of the target optimization function is not clear, so the optimization method is complex, and the computation time of the deep learning network module is long.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a real-time optimization method of a reforming device, which can meet the requirement on precision, reduce the calculated amount and improve the calculation efficiency aiming at the current situation of the prior art.
The technical scheme adopted by the invention for solving the technical problems is as follows: a method for real-time optimization of a reformer, comprising: the method comprises the following steps:
step 1, dividing operation parameters of a reforming device into constant parameters, independent variable parameters and dependent variable parameters, collecting historical operation data corresponding to the constant parameters, the independent variable parameters and the dependent variable parameters of the reforming device within a period of time, sorting the historical operation data into a plurality of working condition data, and expressing each working condition data by using a vector case _ 1;
case_1＝[Mf 1 ，Mf 2 ，...，Mf k ，Mv 1 ，Mv 2 ，...，Mv m ，Cv 1 ，Cv 2 ，...，Cv n ]；
wherein Mf 1 ，Mf 2 ，...，Mf k The first constant data and the second constant data are respectively 1 st constant data and 2 nd constant data; mv 1 ，Mv 2 ，...，Mv m The mth independent variable data is the 1 st independent variable data and the 2 nd independent variable data respectively; cv 1 ，Cv 2 ，...，Cv n The data are respectively the 1 st dependent variable data, the 2 nd dependent variable data \8230andthe nth dependent variable data;
step 2, respectively calculating a Jacobian matrix A of independent variable data to dependent variable data in each working condition data;
wherein the content of the first and second substances,
is Cv 1 For Mv 1 Partial derivatives of (d);
is Cv 1 For Mv 2 Partial derivatives of (d);
is Cv 1 For Mv m Partial derivatives of (d);
is Cv n For Mv 1 Partial derivatives of (d);
is Cv n For Mv 2 Partial derivatives of (d);
is Cv n For Mv m Partial derivatives of (d);
and 3, fitting each element in the A by using the constant data, the independent variable data and the dependent variable data in the step 1 by adopting a multiple linear regression method, namely:
wherein i =1, 2.. N; j =1, 2 \ 8230m; c. C i，j Is a pair of
Fitting the obtained constant coefficient; a is a i，j，1 、a i，j，2 …a i，j，k Are respectively a pair
Analogously obtained Mf 1 、Mf 2 …Mf k The corresponding coefficients; b is a mixture of i，j，1 、b i，j，2 ...b i，j，m Are respectively a pair
Analogously obtained Mv 1 、Mv 2 ...Mv m The corresponding coefficients;
and will be
And the coefficient obtained by fitting forms a matrix B i，j (ii) a The method comprises the following specific steps:
B i，j ＝[c i，j a i，j，1 a i，j，2 …a i，j，k b i，j，1 b i，j，2 ...b i，j，m ](ii) a (formula 2)
Sequentially let i =1, 2.. N; j =1, 2.. M, obtained and
one-to-one corresponding fitting coefficient matrix B 1，1 、B 1，2 、...、B i，j 、...B n，m ；
Step 4, collecting the operation data corresponding to the constant parameter, the independent variable parameter and the dependent variable parameter which are the same as those in the step 1 in the current time of the reforming device, and obtaining a matrix case _ cur corresponding to the current working condition data:
case_cur＝
[Mf 1，cur ，Mf 2，cur ，...，Mf k，cur ，Mv 1，cur ，Mv 2，cur ，...，Mv m，cur ，Cv 1，cur ，Cv 2，cur ，...，Cv n，cur ]；
wherein Mf 1，cur ，Mf 2，cur ，...，Mf k，cur Respectively being the 1 st constant data and the 2 nd constant data of the current working condition. Mv 1，cur ，Mv 2，cur ，...，Mv m，cur Respectively 1 st independent variable data and 2 nd independent variable data of the current working condition; cv 1，cur ，Cv 2，cur ，...，Cv n，cur The data of the 1 st dependent variable and the 2 nd dependent variable of the previous working condition are respectively.
Step 5, calculating a Jacobian matrix D of the current working condition;
wherein, the calculation formula of the corresponding element in the ith row and the jth column in D is as follows:
[] T transposing symbols for the matrix;
sequentially leading i =1 and 2 \8230n; j =1, 2 \ 8230m; obtaining the numerical value of each element in D;
step 6, establishing equality constraint between the dependent variable data CV and the independent variable data MV;
wherein, Δ Mv p ＝Mv p -Mv p，cur ；p＝1、2...m；
Is a matrix R q Transpose of (2), matrix R q For the q-th row in D,
ΔCv q ＝Cv q -Cv q，cur ；
that is, equation 4 is:
thus obtaining:
step 7, setting an upper limit L and a lower limit U of each independent variable data MV and dependent variable data CV;
L＝[Mv 1，L ，Mv 2，L ，...，Mv m，L ，Cv 1，L ，Cv 2，L ，...，Cv n，L ]；
U＝[Mv 1，U ，Mv 2，U ，...，Mv m，U ，Cv 1，U ，Cv 2，u ，...，Cv n，U ]；
wherein Mv 1，L ，Mv 2，L ，...，Mv m，L Upper limit values of the 1 st independent variable data and the 2 nd independent variable data; cv 1，L ，Cv 2，L ，...，Cv n，L Respectively 1 st dependent variable data, 2 nd dependent variable data \8230andthe nth dependent variable data 1，U ，Mv 2，U ，...，Mv m，u The lower limit values of the 1 st independent variable data, the 2 nd independent variable data \8230andthe mth independent variable data are respectively; cv 1，U ，Cv 2，U ，...，Cv n，U The lower limit values of the 1 st dependent variable data, the 2 nd dependent variable data \8230andthe nth dependent variable data are respectively;
step 8, determining a target function related to the independent variable parameters or/and the dependent variable parameters, adopting an optimization algorithm, taking equation constraint of CV and MV, namely formula 6 as a constraint condition, and searching optimal independent variable data MV within the upper and lower limit ranges of CV and MV to maximize the target function;
and 9, modifying the independent variable data MV of the current working condition of the reforming device into the optimal independent variable data MV obtained in the step 8, and simultaneously changing the dependent variable data CV of the current working condition along with the optimal independent variable data MV.
In the scheme, the operation parameters of the reforming device in the step 1 are determined according to a mechanism model.
In order to reduce data noise interference, in step 1, the working condition data is divided by taking the interval time t as a period, and an average value of each parameter data in the interval time t is taken as each parameter data in each working condition data.
Preferably, the optimization algorithm in step 8 is an SQP optimization algorithm.
Compared with the prior art, the invention has the advantages that: the Jacobian matrix of independent variable data to dependent variable data is calculated according to historical operating data, and the Jacobian matrix is fitted through a multiple linear regression method, so that the Jacobian matrix of the current working condition can be calculated according to the fitting coefficient, the Jacobian matrix of the current working condition does not need to be repeatedly calculated according to a partial derivative mode, a complex mechanism model does not need to be solved, and the calculated amount and the optimized calculation time are reduced; in addition, compared with the optimization result of the mechanism model, the error of the optimization result is less than 3%, so that the calculation efficiency and the optimization precision completely meet the real-time optimization requirement, and the method is more practical than the existing mechanism model.
Drawings
FIG. 1 is a comparison graph of the real-time BTX yield of a reformer and the BTX yield calculated by the optimization method in an embodiment of the present invention;
FIG. 2 is a graph showing the BTX yield comparison between before and after the optimization method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the accompanying examples.
The real-time optimization method for the reforming device in the embodiment comprises the following steps of:
step 1, dividing operation parameters of a reforming device into constant parameters, independent variable parameters and dependent variable parameters, collecting historical operation data corresponding to the constant parameters, the independent variable parameters and the dependent variable parameters of the reforming device within a period of time, then sorting the historical operation data into a plurality of working condition data, and representing each working condition data by using a vector case _ 1;
case_1＝[Mf 1 ，Mf 2 ，...，Mf k ，Mv 1 ，Mv 2 ，...，Mv m ，Cv 1 ，Cv 2 ，...，Cv n ]；
wherein Mf 1 ，Mf 2 ，...，Mf k The first constant data and the second constant data are respectively 1 st constant data and 2 nd constant data; mv 1 ，Mv 2 ，...，Mv m Are respectively the 1 stIndependent variable data, 2 nd independent variable data. Cv 1 ，Cv 2 ，...，Cv n The 1 st dependent variable data and the 2 nd dependent variable data are respectively.
In this embodiment, the operation parameters of the reformer are determined according to a mechanism model; in addition, the working condition data is divided by taking the interval time t as a period, and the average value of each parameter data in the interval time t is taken as the data of each parameter in each working condition data; the constant parameter is fixed data, the independent variable parameter is optimizable data, and the dependent variable parameter is data which changes along with the change of the independent variable parameter;
step 2, respectively calculating a Jacobian matrix A of independent variable data to dependent variable data in each working condition data;
wherein the content of the first and second substances,
is Cv 1 For Mv 1 Partial derivatives of (d);
is Cv 1 For Mv 2 Partial derivatives of (d);
is Cv 1 For Mv m Partial derivatives of (d);
is Cv n For Mv 1 Partial derivatives of (d);
is Cv n For Mv 2 Partial derivatives of (d);
is Cv n For Mv m Partial derivatives of (d);
and 3, fitting each element in the A by using the constant data, the independent variable data and the dependent variable data in the step 1 by adopting a multiple linear regression method, namely:
wherein, i =1, 2 \ 8230n; j =1, 2.. M; c. C i，j Is a pair of
A constant coefficient obtained by similarity; a is i，j，1 、a i，j，2 …a i，j，k Are respectively a pair
Mf obtained by fitting 1 、Mf 2 ...Mf k The corresponding coefficients; b i，j，1 、b i，j，2 …b i，j，m Are respectively a pair
Mv obtained by fitting 1 、Mv 2 ...Mv m The corresponding coefficients;
and will be
The coefficient obtained by the middle fitting forms a matrix B i，j (ii) a The method comprises the following specific steps:
B i，j ＝[c i，j a i，j，1 a i，j，2 …a i，j，k b i，j，1 b i，j，2 …b i，j，m ](ii) a (formula 2)
Sequentially let i =1, 2.. N; j =1, 2.. M, obtained and
one-to-one corresponding fitting coefficient matrix B 1，1 、B 1，2 、...、B i，j 、...B n，m ；
Step 4, collecting the operation data corresponding to the constant parameter, the independent variable parameter and the dependent variable parameter which are the same as those in the step 1 in the current time of the reforming device, and obtaining a matrix case _ cur corresponding to the current working condition data:
case_cur＝
[Mf 1，cur ，Mf 2，cur ，...，Mf k，cur ，Mv 1，cur ，Mv 2，cur ，...，Mv m，cur ，Cv 1，cur ，Cv 2，cur ，...，Cv n，cur ]；
wherein Mf 1，cur ，Mf 2，cur ，...，Mf k，cur The current working condition is the 1 st constant data and the 2 nd constant data of the current working condition; mv 1，cur ，Mv 2，cur ，...，Mv m，cur The current working condition is the 1 st independent variable data and the 2 nd independent variable data; cv 1，cur ，Cv 2，cur ，...，Cv n，cur The method comprises the following steps of (1) dependent variable data, 2 nd dependent variable data (823082) and nth dependent variable data of previous working conditions respectively;
step 5, calculating a Jacobian matrix D of the current working condition;
wherein, the calculation formula of the corresponding element in the ith row and the jth column in D is as follows:
[] T transposing symbols for the matrix;
sequentially let i =1, 2.. N; j =1, 2.. M; obtaining the numerical value of each element in D;
step 6, establishing equality constraint between the dependent variable data CV and the independent variable data MV;
wherein, Δ Mv p ＝Mv p -Mv p，cur ；p＝1、2...m；
Is a matrix R q Transpose of (2), matrix R q For the q-th row in D,
ΔCv q ＝Cv q -Cv q，cur ；
that is, equation 4 is:
thus obtaining:
step 7, setting an upper limit L and a lower limit U of each independent variable data MV and dependent variable data CV;
L＝[Mv 1，L ，Mv 2，L ，...，Mv m，L ，Cv 1，L ，Cv 2，L ，...，Cv n，L ]；
U＝[Mv 1，U ，Mv 2，U ，...，Mv m，U ，Cv 1，U ，Cv 2，U ，...，Cv n，U ]；
wherein Mv 1，L ，Mv 2，L ，...，Mv m，L Upper limit values of the 1 st argument data and the 2 nd argument data, respectively; cv 1，L ，Cv 2，L ，...，Cv n，L Upper limit value, mv, of the 1 st dependent variable data and the 2 nd dependent variable data, respectively 1，U ，Mv 2，U ，...，Mv m，U Lower limit values of the 1 st and 2 nd independent variable data, respectively; cv 1，U ，Cv 2，U ，...，Cv n，U The lower limit values of the 1 st dependent variable data, the 2 nd dependent variable data \8230andthe nth dependent variable data are respectively;
step 8, determining an objective function related to the independent variable parameters or/and the dependent variable parameters, adopting an optimization algorithm, taking equation constraint of CV and MV, namely formula 6, as a constraint condition, and searching a proper MV within the upper and lower limits of the CV and the MV to maximize the objective function; in the embodiment, the optimization algorithm is an SQP optimization algorithm;
and 9, modifying the independent variable data MV of the current working condition of the reforming device into the optimal independent variable data MV obtained in the step 8, and simultaneously changing the dependent variable data CV of the current working condition along with the optimal independent variable data MV.
The objective function may also include constant parameters, but the constant parameters are fixed values, so that the constant parameters do not need to be adjusted, and the purpose of this step is to obtain the optimal independent variable parameters through calculation to ensure that the objective function is maximum.
The objective function Object for reformer optimization can be set according to production requirements, taking as an example a reformer that maximizes the production of BTX (B, benzene; T, toluene; X, xylene), the parameters to the right of the equation for the objective function Object are from the reformer.
The reforming feed flow is a constant parameter, so the objective function Object is irrelevant to the denominator and is only relevant to the numerator, the reforming gasoline flow is a dependent variable parameter, the benzene content, the toluene content and the xylene content refer to the benzene, the toluene and the xylene content in the reforming gasoline, are products and are dependent variable parameters, and finally, the most appropriate independent variable parameter MV is found through an optimization algorithm, so that the reforming gasoline flow (benzene content + toluene content + xylene content) is the maximum.
The optimization method can be used for carrying out primary optimization at equal intervals or primary optimization according to actual needs, and can be used for optimizing according to current working conditions at any time to obtain optimal self-variable data so as to ensure the maximum objective function.
To illustrate the effect of the optimization method in the present method, the present embodiment uses the plant operation data of a certain continuous reforming plant as a reference, namely: acquiring half-year stable operation data of the reforming device, dividing the stable operation data by taking 48 hours as interval time, and carrying out average value of each parameter data within 48 hours to form a plurality of working condition files; during optimization, stable operation data in the current 1 hour is used as the current working condition, optimization calculation is carried out by using the method and the mechanism model at the same time, the optimization result and deviation statistics are shown in table 1, and compared with the optimization result of the mechanism model, the optimization result of the method is within 3% of relative error.
TABLE 1 comparison of optimization results of the method and mechanism model of the present invention
The method of the invention is applied to the continuous reforming device for real-time optimization, and the optimization result is highly consistent with the actual operation result of the device, as shown in figure 1; the yield of BTX aromatic hydrocarbon after the application is improved by 0.58 percent compared with that before the application, as shown in figure 2. Therefore, the method provided by the invention is based on the reformer data optimization model, adopts the nonlinear programming solving technology to optimize the reformer, does not need to solve a complex mechanism model, reduces the calculated amount and the optimization calculation time, and has an error smaller than 3% compared with the optimization result of the mechanism model, so that the calculation efficiency and the optimization precision completely meet the real-time optimization requirement, and the method is more practical than the existing mechanism model.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the technical principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (4)
1. A method for real-time optimization of a reformer, comprising: the method comprises the following steps:
step 1, dividing operation parameters of a reforming device into constant parameters, independent variable parameters and dependent variable parameters, collecting historical operation data corresponding to the constant parameters, the independent variable parameters and the dependent variable parameters of the reforming device within a period of time, sorting the historical operation data into a plurality of working condition data, and expressing each working condition data by using a vector case _ 1;
case_1＝[Mf 1 ，Mf 2 ，...，Mf k ，Mv 1 ，Mv 2 ，...，Mv m ，Cv 1 ，Cv 2 ，...，Cv n ]；
wherein Mf 1 ，Mf 2 ，...，Mf k The first constant data and the second constant data are respectively 1 st constant data and 2 nd constant data; mv 1 ，Mv 2 ，...，Mv m The mth independent variable data is the 1 st independent variable data and the 2 nd independent variable data respectively; cv 1 ，Cv 2 ，...，Cv n The 1 st dependent variable data and the 2 nd dependent variable data are respectively.
Step 2, respectively calculating a Jacobian matrix A of independent variable data to dependent variable data in each working condition data;
wherein the content of the first and second substances,
is Cv 1 For Mv 1 Partial derivatives of (d);
is Cv 1 For Mv 2 Partial derivatives of (d);
is Cv 1 For Mv m Partial derivatives of (d);
is Cv n For Mv 1 Partial derivatives of (d);
is Cv n For Mv 2 Partial derivatives of (d);
is Cv n For Mv m Partial derivatives of (d);
and 3, fitting each element in the A by using the constant data, the independent variable data and the dependent variable data in the step 1 by adopting a multiple linear regression method, namely:
wherein i =1, 2.. N; j =1, 2.. M; c. C i，j Is a pair of
Fitting the obtained constant coefficient; a is a i，j，1 、a i，j，2 ...a i，j，k Are respectively a pair
Mf obtained by fitting 1 、Mf 2 ...Mf k The corresponding coefficients; b is a mixture of i，j，1 、b i，j，2 ...b i，j，m Are respectively a pair
Analogously obtained Mv 1 、Mv 2 ...Mv m The corresponding coefficients;
and will be
The coefficient obtained by the middle fitting forms a matrix B i，j (ii) a The method specifically comprises the following steps:
B i，j ＝[c i，j a i，j，1 a i，j，2 … a i，j，k b i，j，1 b i，j，2 … b i，j，m ](ii) a (formula 2)
Sequentially let i =1, 2.. N; j =1, 2.. M, obtained and
one-to-one corresponding fitting coefficient matrix B 1，1 、B 1，2 、...、B i，j 、...B n，m ；
Step 4, collecting the operation data corresponding to the constant parameter, the independent variable parameter and the dependent variable parameter which are the same as those in the step 1 in the current time of the reforming device, and obtaining a matrix case _ cur corresponding to the current working condition data:
case_cur＝
[Mf 1，cur ，Mf 2，cur ，...，Mf k，cur ，Mv 1，cur ，Mv 2，cur ，...，Mv m，cur ，Cv 1，cur ，Cv 2，cur ，...，Cv n，cur ]；
wherein Mf 1，cur ，Mf 2，cur ，...，Mf k，cur Respectively being the 1 st constant data and the 2 nd constant data of the current working condition. Mv 1，cur ，Mv 2，cur ，...，Mv m，cur Respectively 1 st independent variable data and 2 nd independent variable data of the current working condition; cv 1，cur ，Cv 2，cur ，...，Cv n，cur The data of the 1 st dependent variable and the 2 nd dependent variable of the previous working condition are respectively.
Step 5, calculating a Jacobian matrix D of the current working condition;
wherein, the calculation formula of the corresponding element in the ith row and the jth column in D is as follows:
[] T transposing symbols for the matrix;
sequentially let i =1, 2.. N; j =1, 2.. M; obtaining the numerical value of each element in D;
step 6, establishing equality constraint between the dependent variable data CV and the independent variable data MV;
wherein, Δ Mv p ＝Mv p -Mv p，cur ；p＝1、2...m；
Is a matrix R q Transpose of (2), matrix R q For the q-th row in D,
q＝1、2、...n；ΔCv q ＝Cv q -Cv q，cur ；
that is, equation 4 is:
thus obtaining:
…
step 7, setting an upper limit L and a lower limit U of each independent variable data MV and dependent variable data CV;
L＝[Mv 1，L ，Mv 2，L ，...，Mv m，L ，Cv 1，L ，Cv 2，L ，...，Cv n，L ]；
U＝[Mv 1，U ，Mv 2，U ，...，Mv m，U ，Cv 1，U ，Cv 2，U ，...，Cv n，U ]；
wherein Mv 1，L ，Mv 2，L ，...，Mv m，L Upper limit values of the 1 st independent variable data and the 2 nd independent variable data; cv 1，L ，Cv 2，L ，...，Cv n，L Upper limit value, mv, of the 1 st dependent variable data and the 2 nd dependent variable data, respectively 1，U ，Mv 2，U ，...，Mv m，U Lower limit values of the 1 st argument data and the 2 nd argument data, respectively; cv 1，U ，Cv 2，U ，...，Cv n，U Lower limit values of the 1 st dependent variable data and the 2 nd dependent variable data, respectively;
step 8, determining an objective function related to the independent variable parameters or/and the dependent variable parameters, adopting an optimization algorithm, taking equation constraint of CV and MV, namely formula 6 as a constraint condition, and searching optimal independent variable data MV within the upper and lower limit ranges of CV and MV to enable the objective function to be maximum;
and 9, modifying the independent variable data MV of the current working condition of the reforming device into the optimal independent variable data MV obtained in the step 8, and simultaneously changing the dependent variable data CV of the current working condition along with the optimal independent variable data MV.
2. The real-time optimization method of a reformer unit according to claim 1, characterized in that: the operation parameters of the reforming device in the step 1 are determined according to a mechanism model.
3. The method of real-time optimization of a reformer as set forth in claim 1, wherein: in the step 1, the working condition data is divided by taking the interval time t as a period, and the average value of each parameter data in the interval time t is taken as the data of each parameter in each working condition data.
4. A method for real-time optimization of a reformer unit as claimed in any one of claims 1 to 3, characterized in that: the optimization algorithm in the step 8 is an SQP optimization algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110475629.1A CN113110060B (en) | 2021-04-29 | 2021-04-29 | Real-time optimization method of reforming device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110475629.1A CN113110060B (en) | 2021-04-29 | 2021-04-29 | Real-time optimization method of reforming device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113110060A CN113110060A (en) | 2021-07-13 |
| CN113110060B true CN113110060B (en) | 2023-01-06 |
Family
ID=76720924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110475629.1A Active CN113110060B (en) | 2021-04-29 | 2021-04-29 | Real-time optimization method of reforming device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113110060B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004040386A1 (en) * | 2002-10-28 | 2004-05-13 | Siemens Aktiengesellschaft | Method for controlling the mode of operation of the equipment comprised in a refinery or a petrochemical industrial complex |
| CN109814513A (en) * | 2019-03-20 | 2019-05-28 | 杭州辛孚能源科技有限公司 | A kind of catalytic cracking unit optimization method based on data model |
| CN110187635A (en) * | 2019-04-10 | 2019-08-30 | 浙江中控软件技术有限公司 | Real-time optimization method and apparatus for continuous reformer |
| CN111914382A (en) * | 2019-05-07 | 2020-11-10 | 宁波大学 | Constraint evolution optimization method of atmospheric and vacuum pressure device based on proxy model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100152900A1 (en) * | 2008-10-10 | 2010-06-17 | Exxonmobil Research And Engineering Company | Optimizing refinery hydrogen gas supply, distribution and consumption in real time |
-
2021
- 2021-04-29 CN CN202110475629.1A patent/CN113110060B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004040386A1 (en) * | 2002-10-28 | 2004-05-13 | Siemens Aktiengesellschaft | Method for controlling the mode of operation of the equipment comprised in a refinery or a petrochemical industrial complex |
| CN109814513A (en) * | 2019-03-20 | 2019-05-28 | 杭州辛孚能源科技有限公司 | A kind of catalytic cracking unit optimization method based on data model |
| CN110187635A (en) * | 2019-04-10 | 2019-08-30 | 浙江中控软件技术有限公司 | Real-time optimization method and apparatus for continuous reformer |
| CN111914382A (en) * | 2019-05-07 | 2020-11-10 | 宁波大学 | Constraint evolution optimization method of atmospheric and vacuum pressure device based on proxy model |
Non-Patent Citations (2)
| Title |
|---|
| 催化重整反应部分的参数优化分析;张磊等;《现代化工》;20161231;第36卷(第12期);第158-161页 * |
| 基于状态观测器的催化重整芳烃收率在线估计;郭颖等;《计算机与应用化学》;20171231;第34卷(第04期);第280-284页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113110060A (en) | 2021-07-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2118885C (en) | Process control system | |
| CN110320806B (en) | Sewage treatment process self-adaptive prediction control method based on integrated instant learning | |
| US11214850B2 (en) | Prediction control method and system for component contents in rare earth extraction process | |
| CN102540879A (en) | Multi-target evaluation optimization method based on group decision making retrieval strategy | |
| CN113724800A (en) | Catalytic cracking unit simulation prediction method based on molecular level mechanism model and big data technology | |
| CN111650834B (en) | Sewage treatment process prediction control method based on extreme learning machine | |
| CN104463343A (en) | Method for predicting catalytic cracking light oil yield | |
| CN109670625A (en) | NOx emission concentration prediction method based on Unscented kalman filtering least square method supporting vector machine | |
| CN113012766B (en) | Self-adaptive soft measurement modeling method based on online selective integration | |
| CN109492335B (en) | Method and system for predicting furnace temperature of annealing furnace | |
| CN105159071A (en) | Method for estimating economic performance of industrial model prediction control system in iterative learning strategy | |
| Zhang et al. | Multi‐variable direct self‐organizing fuzzy neural network control for wastewater treatment process | |
| CN113110060B (en) | Real-time optimization method of reforming device | |
| US7848831B2 (en) | Real-time operating optimized method of multi-input and multi-output continuous manufacturing procedure | |
| CN108287474B (en) | Robust operation optimization method of catalytic reforming reactor based on uncertainty of raw materials | |
| CN109101683B (en) | Model updating method for pyrolysis kettle of coal quality-based utilization and clean pretreatment system | |
| CN114187975A (en) | Method and system for optimizing and automatically updating data model of diesel hydrogenation device | |
| CN102609601B (en) | Method for estimating parameters of residual oil hydrogenation reaction kinetic model based on similar endoplasmic reticulum body membrane calculation | |
| CN114861759A (en) | Distributed training method of linear dynamic system model | |
| CN114926075A (en) | Mechanical part production scheduling method based on man-hour prediction | |
| CN114186771A (en) | Hybrid regularization random configuration network industrial process operation index estimation method | |
| Kano et al. | Just-in-time statistical process control: adaptive monitoring of vinyl acetate monomer process | |
| CN108197369A (en) | A kind of modeling and Multipurpose Optimal Method for catalytic reforming reactor | |
| CN109948302A (en) | Cracking reaction modeling method and device based on artificial neural network and modelling by mechanism | |
| Muravyova | Control of the Pyrolysis Fraction Cleaning Process Using a Neural Network |
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 |