Distillation column multivariable control method, apparatus, device, and medium

Abstract

The application relates to a rectifying tower multivariable control method, device, computer equipment and computer readable storage medium. The method comprises the following steps: obtaining process data of a rectifying tower, and obtaining an initial gain matrix based on the process data; correcting the initial gain matrix based on product flow to obtain a corrected gain matrix; obtaining economic parameters of operation variables and / or controlled variables, and constructing an economic objective function according to the corrected gain matrix and the economic parameters; constructing a physical constraint condition based on the physical limitation of the operation variables, and constructing a product quality constraint condition based on the quality requirement of the controlled variables; optimizing and solving the economic objective function according to the physical constraint condition and the product quality constraint condition to obtain target operation variable parameters of the rectifying tower; and adjusting the operation variables of the rectifying tower to the target operation variable parameters. The method can guarantee the stability and reliability of long-term operation of the rectifying tower, and can improve economic benefits.

Description

Technical Field

[0001] This application relates to the field of industrial production control technology, and in particular to a method, apparatus, computer equipment, and computer-readable storage medium for multivariable control of a distillation column. Background Technology

[0002] Distillation columns are the core separation equipment in process industries such as chemical and petrochemical manufacturing. They consume a great deal of energy, and optimizing their operation is crucial for improving the economic efficiency of enterprises. Currently, multivariable control is widely used in the control and optimization of distillation columns.

[0003] However, existing multivariate control methods often only focus on controlled variables such as product quality, while ignoring important variables that affect material balance, such as product flow rate. This leads to inaccurate descriptions of the system state in the constructed process model, and the relationships between variables within the model may violate basic physicochemical laws. During long-term operation, the controller may optimize to a physically infeasible operating region, affecting system stability and control quality. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for multivariable control of distillation columns that can improve the stability and reliability of long-term operation of distillation columns and enhance economic benefits, in order to address the aforementioned technical problems.

[0005] In a first aspect, this application provides a method for multivariable control of a distillation column, the method comprising:

[0006] Acquire process data from the distillation column and obtain an initial gain matrix based on the process data; the process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate;

[0007] Based on the product flow rate, the initial gain matrix is ​​corrected by material balance to obtain the corrected gain matrix;

[0008] Obtain the economic parameters of the operating variable and / or the controlled variable, and construct an economic objective function based on the corrected gain matrix and the economic parameters;

[0009] Physical constraints are constructed based on the physical limitations of the operated variables, and product quality constraints are constructed based on the quality requirements of the controlled variables.

[0010] Based on the physical constraints and the product quality constraints, the economic objective function is optimized to obtain the target operating variable parameters of the distillation column;

[0011] The operating variables of the distillation column are adjusted to the target operating variable parameters.

[0012] In one embodiment, acquiring process data from the distillation column and obtaining an initial gain matrix based on the process data includes:

[0013] A step test is performed on the distillation column to obtain process data of the distillation column; wherein, the process data includes the step change of the operating variable and the steady-state response change of the controlled variable;

[0014] The initial gain matrix is ​​obtained based on the process data.

[0015] In one embodiment, the process data is a set; obtaining an initial gain matrix based on the process data includes:

[0016] A step change of a preset magnitude is applied to each controlled variable in sequence, and the steady-state response change corresponding to each controlled variable is collected.

[0017] Based on the step change of each operated variable and the corresponding steady-state response change of each controlled variable, the initial gain coefficient between each operated variable and each controlled variable is obtained.

[0018] The initial gain matrix is ​​constructed based on the initial gain coefficients between each operated variable and each controlled variable.

[0019] In one embodiment, based on the step change of each manipulated variable and the corresponding steady-state response change of each controlled variable, the initial gain coefficient between each manipulated variable and each controlled variable is obtained, including:

[0020] For any manipulated variable and any controlled variable, calculate the ratio of the steady-state response change of the controlled variable to the step change of the manipulated variable, and use this ratio as the initial gain coefficient between the manipulated variable and the controlled variable.

[0021] In one embodiment, the process data consists of multiple sets; obtaining an initial gain matrix based on the process data includes:

[0022] Construct a linear equation with the operated variable as the independent variable and the controlled variable as the dependent variable;

[0023] Substituting the step changes of the multiple sets of the operated variables and the steady-state response changes of the controlled variables into the linear equations, a system of linear equations is obtained.

[0024] The linear equations are solved using the least squares method to obtain the initial gain coefficient between each operated variable and each controlled variable.

[0025] The initial gain matrix is ​​constructed based on the initial gain coefficients between each operated variable and each controlled variable.

[0026] In one embodiment, the initial gain matrix is ​​corrected for material balance based on the product flow rate to obtain a corrected gain matrix, including:

[0027] Material balance constraints are constructed based on the product flow rate to reflect the material balance relationship of the distillation column; wherein, the material balance constraints are used to ensure that the sum of the gain coefficients of each product flow rate corresponding to the same operating variable satisfies a preset mathematical relationship;

[0028] Under the material balance constraint, with the goal of minimizing the sum of squared prediction errors, the initial gain matrix is ​​subjected to regression correction to obtain the corrected gain matrix.

[0029] In one embodiment, the economic objective function is optimized based on the physical constraints and the product quality constraints to obtain the target operating variable parameters of the distillation column, including:

[0030] Based on the corrected gain matrix, a process steady-state model characterizing the steady-state relationship between the operated variable and the controlled variable is constructed.

[0031] Substituting the process steady-state model into the economic objective function yields an economic objective function with the operation variable as the independent variable.

[0032] Substituting the quality index boundary values ​​from the product quality constraints into the process steady-state model, the operational boundary equations characterized by the operational variables are obtained by solving the model.

[0033] Within the feasible operating region bounded by the operational boundary equation and the physical constraints, a linear programming solution is performed on the economic objective function with the operational variables as independent variables to obtain the target operational variable parameters of the distillation column.

[0034] In one embodiment, after obtaining the target operating variable parameters of the distillation column, the method further includes:

[0035] Based on the economic objective function, obtain the constraint economic value quantification index of the operational variable corresponding to the physical constraint condition; the constraint economic value quantification index includes the marginal economic value used to quantify the operational variable corresponding to the physical constraint condition.

[0036] In one embodiment, after obtaining the target operating variable parameters of the distillation column, the method further includes:

[0037] Monitor the operating mode of the controller corresponding to the operation variable;

[0038] If any controller is detected to be in manual mode, the corresponding operating variable of that controller is marked as a fixed parameter in the economic objective function, and the economic objective function is re-optimized to obtain the updated target operating variable parameters of the distillation column.

[0039] Secondly, this application also provides a multivariable control device for a distillation column, the device comprising:

[0040] The model building module is used to acquire process data of the distillation column and obtain an initial gain matrix based on the process data; the process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate;

[0041] The model calibration module is communicatively connected to the model construction module and is used to perform material balance calibration on the initial gain matrix based on the product flow rate to obtain the calibrated gain matrix.

[0042] The objective function construction module is communicatively connected to the model calibration module and is used to obtain the economic parameters of the operating variables and / or the controlled variables, and construct an economic objective function based on the calibrated gain matrix and the economic parameters.

[0043] The constraint construction module is used to construct physical constraints based on the physical limitations of the operating variables, and to construct product quality constraints based on the quality requirements of the controlled variables.

[0044] The objective function solving module is communicatively connected to the objective function construction module and the constraint condition construction module, respectively, and is used to optimize and solve the economic objective function according to the physical constraints and the product quality constraints to obtain the target operating variable parameters of the distillation column;

[0045] The control module is communicatively connected to the objective function solving module and is used to adjust the operating variables of the distillation column to the target operating variable parameters.

[0046] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described in any of the preceding claims.

[0047] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0048] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the preceding claims.

[0049] The aforementioned multivariate control method, apparatus, computer equipment, computer-readable storage medium, and computer program product for distillation columns, by including product flow rate within the range of controlled variables and correcting the material balance relationship of the initially obtained initial gain matrix based on product flow rate, ensures that the corrected gain matrix strictly satisfies the physical conservation law. This solves the problem in existing technologies where inconsistencies within the model may lead to optimization results falling into physically infeasible regions, thus improving the executability of the model optimization results. Furthermore, this invention introduces an economic objective function directly related to the enterprise's real economic data, ensuring that the controller's optimization direction aligns with the goal of maximizing economic benefits. By obtaining and executing the target operating variable parameters within the feasible operating region jointly defined by physical operational constraints and product quality constraints, the distillation column can always operate at an operating point that is both safe and compliant, as well as economically optimal, guaranteeing the long-term stability and reliability of the distillation column while improving economic efficiency. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is an application environment diagram of a multivariable control method for a distillation column in one embodiment;

[0052] Figure 2 This is a flowchart illustrating a multivariable control method for a distillation column in one embodiment;

[0053] Figure 3 A system architecture diagram illustrating the application of multivariable control methods in distillation columns;

[0054] Figure 4 A schematic diagram illustrating the initialization and extended modeling of the example;

[0055] Figure 5 A schematic diagram illustrating the construction of the economic objective function and the optimization solution;

[0056] Figure 6 The diagram illustrates the constraints when solving a linear programming problem for an economic objective function, as shown in the example.

[0057] Figure 7 This is a schematic diagram of a system monitoring interface that includes information on constrained economic value quantification indicators.

[0058] Figure 8This is a flowchart illustrating the process of controller monitoring as an example.

[0059] Figure 9 This is a structural block diagram of a multivariable control device for a distillation column in one embodiment;

[0060] Figure 10 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0062] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0063] Before describing the embodiments of the present invention in detail, the key concepts involved in the present invention will be briefly explained first.

[0064] Distillation columns are core separation equipment in process industries such as chemical and petrochemical manufacturing. Their working principle utilizes the differences in volatility of components in a mixture, achieving separation of components with different boiling points through countercurrent gas-liquid contact within the column. During the operation of a distillation column, operators or automatic control systems need to adjust several manipulated variables (MV), such as reflux flow rate, reboiler heating steam rate, top product flow rate, and bottom product flow rate, to ensure the distillation column's operating status meets process requirements. The operating status of a distillation column is characterized by several controlled variables (CV), such as the content of key components in the top product, the content of key components in the bottom product, the top product flow rate, and the bottom product flow rate.

[0065] Traditional multivariate predictive control (MDC) techniques, when applied to distillation columns, typically focus only on the relationship between key controlled variables (such as component content) and operating variables (especially the gain matrix) in their internal process models, neglecting other variables like product flow rate. More critically, the model parameters obtained through fitting conventional data do not necessarily adhere to the fundamental physicochemical laws governing material and energy balance required for distillation column operation. This inconsistency within the model can lead to the controller finding a mathematically "optimal" operating point during optimization calculations, but which is physically unrealizable. This means that material inflow and outflow at this operating point are not conserved, severely impacting the stability and quality of the control system.

[0066] The multivariable control method for distillation columns provided in this application can be applied to, for example... Figure 1 The application environment shown is illustrated. Terminal 101 communicates with server 102 via a network. A data storage system can store the data that server 102 needs to process. The data storage system can be integrated onto server 102, or it can be located in the cloud or on another network server. Server 102 acquires process data of the distillation column collected by terminal 101 and obtains an initial gain matrix based on the process data. The process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate. Server 102 performs material balance correction on the initial gain matrix based on the product flow rate to obtain a corrected gain matrix. Server 102 acquires the economic parameters of the operating variables and / or the controlled variables and constructs an economic objective function based on the corrected gain matrix and the economic parameters. Server 102 constructs physical constraints based on the physical limitations of the operating variables and constructs product quality constraints based on the quality requirements of the controlled variables. Server 102 optimizes the economic objective function based on the physical constraints and the product quality constraints to obtain the target operating variable parameters of the distillation column. Server 102 controls terminal 101 to adjust the operating variables of the distillation column to the target operating variable parameters. Terminal 101 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, etc. Server 102 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0067] In one exemplary embodiment, such as Figure 2 As shown, a multivariable control method for a distillation column is provided, which is applied to... Figure 1 Taking server 102 as an example, the explanation includes the following steps 201 to 206:

[0068] Step 201: Obtain the process data of the distillation column and obtain the initial gain matrix based on the process data; the process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate.

[0069] In this context, process data refers to the set of measurement data collected under safe operating conditions that reflects the steady-state input-output relationship of the distillation column. Unlike existing technologies, in this embodiment, the collected controlled variable data includes not only indicator data characterizing product quality (hereinafter referred to as "first-class controlled variables"), such as the content of heavy components in the top product, the content of light components in the bottom product, and the purity of key components, but also indicator data characterizing product flow rate (hereinafter referred to as "second-class controlled variables"), such as the top product outflow rate, the bottom product outflow rate, and the side stream product outflow rate.

[0070] The initial gain matrix is ​​a parametric matrix that characterizes the static influence relationship between each manipulated variable and each controlled variable. Mathematically, this matrix is ​​an m×n matrix, where m is the number of controlled variables, n is the number of manipulated variables, and the element (Kp) in the i-th row and j-th column of the matrix... ij This represents the steady-state change of the i-th controlled variable caused by a unit change in the j-th manipulated variable. This element is also the gain coefficient between the j-th manipulated variable and the i-th controlled variable. The sign (positive or negative) of the gain coefficient indicates the direction of the influence (same direction or opposite direction), and its absolute value indicates the strength of the influence.

[0071] Step 202: Perform material balance correction on the initial gain matrix based on product flow rate to obtain the corrected gain matrix.

[0072] It should be noted that although the initial gain matrix obtained by fitting conventional data in step 201 can reflect the mathematical relationship between input and output to a certain extent, it does not guarantee that it meets the physical laws of distillation column operation, especially the material balance relationship.

[0073] Material balance refers to the principle that, under steady-state operating conditions, the total amount of material entering a distillation column must equal the total amount of material leaving the column. For distillation columns that produce both top and bottom products, this relationship means that the sum of the changes in the flow rates of each product caused by a change in a certain operating variable (such as feed rate, reflux rate, or reboiler heating rate) should physically satisfy a specific conservation relationship. For example, when there are no other outlets, the sum of the gain coefficients of the flow rates of each product should be zero.

[0074] Therefore, in this embodiment, the material balance relationship corresponding to the collected product flow data is used to apply a physical consistency constraint to the initial gain matrix obtained in step 201, and the gain coefficients therein are forcibly corrected, so that the corrected matrix strictly satisfies the material balance equation.

[0075] The matrix obtained after correction is referred to as the corrected gain matrix in this paper. It can also be called the self-consistent extended gain matrix or the material balance uniformity gain matrix, which is not limited here.

[0076] Step 203: Obtain the economic parameters of the operating variables and / or controlled variables, and construct the economic objective function based on the corrected gain matrix and economic parameters.

[0077] Among these, the economic parameters related to the distillation column production process, such as operating variables and controlled variables, can be obtained through the data interface of the upper-level information system, such as Enterprise Resource Planning (ERP) systems and energy management systems. Economic parameters refer to quantifiable economic indicators that reflect operating costs and product value, such as real-time market prices or sales prices of products obtained from the ERP system, and real-time settlement unit prices or cost unit prices of utilities (such as steam, electricity, and circulating water) obtained from the energy management system.

[0078] The economic parameter associated with the controlled variable can be the unit price of a product flow rate. If the product is a qualified product, the unit price is positive, indicating revenue generation. The economic parameter associated with the operating variable can be the unit price of reboiler steam, which is negative, indicating cost generation. When constructing the objective function, since the reflux flow rate itself may not directly correspond to cost, and the key component content of the top product and the key component content of the bottom product, etc., do not generate revenue, it is not required to assign economic parameters to all operating variables.

[0079] An economic objective function is a mathematical function constructed based on the variables in the corrected gain matrix and their corresponding economic parameters, with actual production profit (or cost) as the optimization objective. For example, the economic objective function can take the form: J = Σ(P i ×CV i )+Σ(Q j ×MV j ), where P i Let CV be the economic coefficient of the i-th controlled variable. i Let Q be the steady-state value of the i-th controlled variable. j Let MV be the economic coefficient of the j-th operational variable. j Let be the value of the j-th operation variable. To facilitate subsequent solutions, a negative sign can be added before the above function to transform the economic objective function into a minimization problem.

[0080] Step 204: Construct physical constraints based on the physical limitations of the operated variables, and construct product quality constraints based on the quality requirements of the controlled variables.

[0081] Physical constraints refer to the rigid limitations on operating variables determined by the inherent capacity of the equipment or the process design within the distillation column. For example, a control valve with an opening range of 0% to 100% corresponds to a certain operating variable (such as reflux flow) that can only take values ​​between its minimum (e.g., 50 t / h, tons / hour) and maximum (e.g., 80 t / h); the maximum heating capacity of the reboiler determines the upper limit of the steam flow rate. These constraints are typically expressed in the form of inequalities, such as MV. min ≤MV j ≤MV max .

[0082] Product quality constraints refer to the range of process parameters that the controlled variable must meet to ensure product quality. For example, the content of heavy components in the top product of a column must not exceed a certain upper limit (e.g., 5%), and the content of light components in the bottom product of a column must not exceed a certain upper limit (e.g., 5%). These constraints are usually expressed in the form of inequalities, such as CV. i ≤CV i,max , or CV i ≥CV i,min .

[0083] It should be noted that physical constraints and product quality constraints together define the feasible operational region for solving the optimization problem. That is, any combination of operational variables corresponding to any point within this region can both meet the requirements for safe operation of the equipment and ensure that the product quality meets the standards.

[0084] Step 205: Based on the physical constraints and product quality constraints, optimize the economic objective function to obtain the target operating variable parameters of the distillation column.

[0085] In this embodiment, within the feasible operational region defined by the two types of constraints in step 204, the combination of operational variables that optimizes the economic objective function (i.e., maximizes profit or minimizes cost) is sought. The corresponding optimization algorithm can be, but is not limited to, numerical optimization methods such as linear programming (LP), quadratic programming (QP), sequential quadratic programming (SQP), or interior-point methods. Tools such as Matlab can also be used for calculation; this embodiment does not impose specific limitations on this approach.

[0086] The target operating variable parameters refer to the target setpoints of each operating variable that should be applied to the distillation column after optimization, such as the optimal operating variable setpoints.

[0087] Step 206: Adjust the operating variables of the distillation column to the target operating variable parameters.

[0088] For example, the optimal setpoint values ​​of the operating variables obtained in step 205 are sent to the underlying basic control layer through a standard industrial communication protocol. For example, the basic control layer may be a PID (Proportional Integral Derivative) controller, an AC controller (Wireless Access Point Controller), a DCS (Distributed Control System), or a PLC controller (Programmable Logic Controller). The basic control layer performs closed-loop regulation to adjust the actual values ​​of each operating variable to the target values, thereby achieving economically optimal operation.

[0089] In the aforementioned multivariate control method for distillation columns, by including product flow rate within the range of controlled variables and correcting the initial gain matrix based on product flow rate using material balance relationships, the corrected gain matrix can strictly satisfy the physical conservation law. This solves the problem in existing technologies where inconsistencies within the model may lead to optimization results falling into physically infeasible regions, thus improving the executability of the model optimization results. Furthermore, this embodiment introduces an economic objective function directly related to the enterprise's real economic data, ensuring that the controller's optimization direction aligns with the goal of maximizing economic benefits. By obtaining and executing the target operating variable parameters within the feasible operating region jointly defined by physical operational constraints and product quality constraints, the distillation column can always operate at an operating point that is both safe and compliant, as well as economically optimal, guaranteeing the long-term stability and reliability of the distillation column while improving economic efficiency.

[0090] In an exemplary embodiment, process data of the distillation column is acquired, and an initial gain matrix is ​​obtained based on the process data, including:

[0091] Step 2011: Perform a step test on the distillation column to obtain process data of the distillation column; wherein, the process data includes the step change of the operating variable and the steady-state response change of the controlled variable.

[0092] Step testing is a standard active excitation experimental method for control system analysis. It involves applying a unit step input to the system and observing the output response curve to evaluate the system's dynamic and steady-state performance. In this embodiment, step testing is performed on the distillation column. Specifically, while the distillation column is in steady-state operation, a known-amplitude step change is applied sequentially to each controlled variable (e.g., instantaneously increasing the reflux setpoint from 60 t / h to 65 t / h, a step change of +5 t / h), while simultaneously recording the response curves of all controlled variables in real time. Between step tests of each controlled variable, the distillation column must be allowed to reach a new steady state before proceeding to the next controlled variable. It should be noted that steady state here refers to a state where the fluctuation amplitude of the process variable has entered its steady-state noise level range and no longer exhibits a significant trend change.

[0093] For each controlled variable, the difference between its steady-state value before the step and its new steady-state value after the step is taken as the steady-state response change of the controlled variable to the step of the manipulated variable. For example, if the product flow rate at the top of the tower is stable at 12.0 t / h before the step and the new steady-state value after the step is 12.8 t / h, then its steady-state response change is +0.8 t / h.

[0094] Step 2012: Obtain the initial gain matrix based on the process data.

[0095] In this embodiment of the application, the initial gain matrix is ​​calculated by a preset data processing logic based on the step change of the operated variable and the steady-state response change of the controlled variable collected by the step test.

[0096] In this embodiment of the application, the step test can accurately capture the true static gain characteristics of the distillation column under the action of various operating variables, which is beneficial to improving the accuracy of the initial model identification and the reliability of the data.

[0097] In an exemplary embodiment, the process data is a set, i.e., a single step test is performed for each operated variable; correspondingly, in step 2012, obtaining the initial gain matrix based on the process data includes:

[0098] a1. Apply a step change of a preset magnitude to each operated variable in sequence, and collect the steady-state response change of each controlled variable.

[0099] a2. Based on the step change of each operated variable and the corresponding steady-state response change of each controlled variable, the initial gain coefficient between each operated variable and each controlled variable is obtained.

[0100] a3. Construct the initial gain matrix based on the initial gain coefficients between each operated variable and each controlled variable.

[0101] In this embodiment, a gain matrix construction scheme is given when the process data is a set of step test data. This method is simple to calculate and is suitable for scenarios where step test execution specifications are followed. By calculating the ratios one by one to obtain the gain coefficients and combining them into a matrix, an initial model reflecting the steady-state input-output relationship of the process can be quickly obtained.

[0102] In an exemplary embodiment, step a2, based on the step change of each operated variable and the corresponding steady-state response change of each controlled variable, obtains the initial gain coefficient between each operated variable and each controlled variable, including:

[0103] For any manipulated variable and any controlled variable, calculate the ratio of the steady-state response change of the controlled variable to the step change of the manipulated variable, and use this ratio as the initial gain coefficient between the manipulated variable and the controlled variable.

[0104] The initial gain coefficient between the j-th operated variable and the i-th controlled variable is also:

[0105] (Kp) ij =ΔCV i / ΔMV j

[0106] Where, ΔMV j Let ΔCV be the step change of the j-th operated variable. i Let be the steady-state response change of the i-th controlled variable, which is the difference between the steady-state values ​​before and after the step.

[0107] The initial gain coefficient is used to quantify the degree and direction of the steady-state influence of the manipulated variable on the controlled variable. A positive ratio indicates that an increase in the manipulated variable leads to an increase in the controlled variable; a negative ratio indicates that an increase in the manipulated variable leads to a decrease in the controlled variable. The larger the absolute value of the ratio, the more significant the change in the controlled variable caused by a unit change in the manipulated variable.

[0108] This embodiment discloses a method for calculating the gain coefficient under a single step test, which is to directly take the ratio of the steady-state response change to the step change. This method is intuitive and does not require solving a set of equations. It is particularly suitable for use in the distillation column operating area where the signal-to-noise ratio is high and the nonlinearity is low during step testing. It has the advantages of high computational efficiency and ease of engineering implementation.

[0109] In an exemplary embodiment, the process data is in multiple sets, meaning that multiple step tests are performed for each operated variable; correspondingly, in step 2012, obtaining the initial gain matrix based on the process data includes:

[0110] b1. Construct a linear equation with the operated variable as the independent variable and the controlled variable as the dependent variable.

[0111] b2. Substitute the step changes of multiple sets of manipulated variables and the steady-state response changes of the controlled variables into the linear equations to obtain a system of linear equations.

[0112] b3. Solve the linear equation system using the least squares method to obtain the initial gain coefficient between each operated variable and each controlled variable.

[0113] b4. Construct the initial gain matrix based on the initial gain coefficients between each operated variable and each controlled variable.

[0114] For example, we can first assume that the distillation column can be approximated by a linear steady-state model near the test operating point, and construct a linear equation of the following form:

[0115] CV i =Σ(Kp ij ×MV j )+bias i

[0116] Among them, bias i The bias term for the i-th controlled variable is used to characterize the steady-state deviation between the process model and the current actual operating conditions, so as to achieve zero steady-state error control (feedback correction).

[0117] Then, the collected input and output data samples are substituted into the above linear equations, with each set of data corresponding to one equation, thus constructing a system of linear equations.

[0118] Subsequently, the least squares method is used to process multiple sets of data. Based on this, the random noise interference in a single measurement can be effectively smoothed, and the statistically optimal gain coefficient estimate can be obtained. Based on this, the initial gain matrix is ​​constructed.

[0119] This embodiment presents a matrix construction scheme when multiple sets of process data are available. By constructing a system of linear equations and solving it using the least squares method, the data information from multiple tests can be fully utilized, effectively suppressing the interference of random measurement noise on the estimated gain coefficient, and improving the statistical reliability and robustness of the initial gain matrix. It is particularly suitable for application scenarios where the dynamic characteristics of the distillation column are relatively complex, and can further improve the identification accuracy and noise resistance of the initial model under multiple data conditions.

[0120] In an exemplary embodiment, step 202 involves performing material balance correction on the initial gain matrix based on product flow rate to obtain a corrected gain matrix, including:

[0121] Step 2021: Construct material balance constraints based on product flow rates to reflect the material balance relationship of the distillation column; wherein, the material balance constraints are used to ensure that the sum of the gain coefficients of each product flow rate corresponding to the same operating variable satisfies a preset mathematical relationship.

[0122] The material balance constraint is an equation-related constraint on product flow rate derived from the fundamental physical conservation laws of distillation columns. Taking a typical distillation column with two product streams (top and bottom) and no other material outlets as an example, under steady-state conditions, the total change in material entering or leaving the system caused by a change in any operating variable should be conserved. For controlled variables such as product flow rate, this physical law is expressed as follows: for any operating variable MV... j The sum of the gain coefficients of each product's flow CV should be zero, i.e., Σ(Kp). ij =0, where the value of i iterates through all product traffic control variables.

[0123] Step 2022: Under the constraint of material balance, with the goal of minimizing the sum of squared prediction errors, the initial gain matrix is ​​subjected to regression correction to obtain the corrected gain matrix.

[0124] In this embodiment, an optimization algorithm with equality constraints is invoked to minimize the sum of squares of the errors between the model's predicted values ​​and the actual measured values, while forcibly satisfying the material balance equality constraints constructed in step 2021. The corrected gain matrix is ​​the optimal gain matrix that minimizes the sum of squares of the prediction errors under the material balance constraints. This matrix retains the accuracy of the data fitting while strictly satisfying the physical conservation laws.

[0125] This embodiment discloses a scheme for material balance correction of the initial gain matrix. Specifically, it constructs physical equality constraints based on product flow rate, and then performs regression correction under these constraints with the goal of minimizing prediction error. Based on this, it can automatically correct material imbalance deviations caused by factors such as measurement noise, nonlinear effects, or non-standard testing in the initial identification, so that the final gain matrix is ​​physically self-consistent. This fundamentally avoids the problem of the controller optimizing to a false operating point due to internal model contradictions, thereby further improving the long-term operational stability of the model.

[0126] In an exemplary embodiment, in step 205, the economic objective function is optimized based on physical constraints and product quality constraints to obtain the target operating variable parameters of the distillation column, including:

[0127] Step 2051: Based on the corrected gain matrix, construct a process steady-state model that characterizes the steady-state relationship between the operating variable and the controlled variable.

[0128] For example, based on the corrected gain matrix and the gain coefficients of each controlled variable, the following steady-state model of the process is constructed:

[0129] CV = Kp × MV + Bias

[0130] Where Kp* represents the corrected gain matrix, MV represents the manipulated variable vector, CV represents the controlled variable vector, and Bias* represents the bias vector. This model can describe the steady-state mapping relationship between the manipulated variable MV and the controlled variable CV.

[0131] Step 2052: Substitute the process steady-state model into the economic objective function to obtain the economic objective function with the operation variables as independent variables.

[0132] In this embodiment of the application, the linear equation system constructed in step 2051 is substituted into the economic objective function J=f(CV,MV) constructed in step 203, such as J=Σ(P i ×CV i )+Σ(Q j ×MV j This is done by eliminating the controlled variable CV, thereby transforming the objective function into a function with only the operation variable MV as its independent variable, i.e., J=g(MV). This transformation simplifies the variable space for subsequent optimization solutions.

[0133] Step 2053: Substitute the quality index boundary values ​​in the product quality constraints into the process steady-state model to obtain the operational boundary equations characterized by the operational variables.

[0134] In this embodiment, the upper or lower limit of the quality specification of the controlled variable (e.g., CV1 = 5%) is substituted as a known quantity into the steady-state process model constructed in step 2051, and the linear equation for the manipulated variable MV is obtained. This equation is the operational boundary equation. In the coordinate system formed by the manipulated variables, this equation is represented by a straight line (for two manipulated variables) or a hyperplane (for multiple manipulated variables). Its physical meaning is: on one side of the straight line or hyperplane, the product quality meets the requirements; on the other side, the product quality does not meet the requirements. For example, substituting the upper limits of the product quality constraints CV1 = 5% and CV2 = 5% into the model respectively yields two operational boundary lines.

[0135] Step 2054: Within the feasible operating region bounded by the operating boundary equation and physical constraints, perform linear programming to solve the economic objective function with the operating variables as independent variables, and obtain the target operating variable parameters of the distillation column.

[0136] For example, within the feasible operational region bounded by the operational boundary equation and physical constraints, a linear programming (LP) solver is invoked to optimize the economic objective function with MV as the independent variable, finding the vertex or boundary point that makes the objective function optimal (maximum profit or minimum cost). The value of the operational variable corresponding to this point is the objective operational variable parameter.

[0137] This embodiment discloses the steps for solving the economic objective function. By converting the economic objective function into a form where only the operational variables are independent variables, the dimensionality of the optimization problem can be reduced, thereby improving the solution efficiency. Furthermore, by constructing operational boundary equations, product quality constraints can be visualized as geometric boundaries in the operational space, making the optimization process more intuitive and controllable. In addition, using a linear programming algorithm ensures that a globally optimal solution is found within the feasible region constituted by linear constraints, avoiding local optima problems.

[0138] In an exemplary embodiment, after obtaining the target operating variable parameters of the distillation column, the method further includes:

[0139] c1. Based on the economic objective function, obtain the quantitative indicators of the constrained economic value of the operational variables corresponding to the physical constraints; the quantitative indicators of constrained economic value include the marginal economic value used to quantify the operational variables corresponding to the physical constraints.

[0140] Among them, the quantitative indicators of the economic value of constraints are indicators for the monetization of operational constraints, which include, but are not limited to, the following two specific indicator forms:

[0141] (1) Shadow price: At the optimal solution, the change in the value of the optimal economic objective function caused by relaxing a currently effective constraint (i.e., tight constraint) by one unit. It quantifies the real-time economic loss or marginal value caused by the tight constraint. For example, if the current upper limit constraint of reboiler steam output is 16 t / h, the corresponding shadow price is 2990 yuan / hour / unit, which means that if the upper limit of steam output can be increased by 1 t / h, theoretically the profit can be increased by 2990 yuan per hour.

[0142] (2) Cost Reduction: At the optimal solution, the maximum potential economic benefit that may be brought about by a unit change in the constraint value of a currently inactive constraint (i.e., a loose constraint) or a non-basic variable at the boundary. It is used to evaluate the economic potential that can be released after the constraint is removed or relaxed. For example, if the current lower limit constraint of the return flow is 50t / h, the corresponding cost reduction is 600 yuan / hour / unit, which means that if the lower limit of the return flow can be reduced by 1t / h, theoretically the profit can be increased by 600 yuan per hour.

[0143] Marginal economic value is the economic meaning jointly represented by shadow price and reduction cost, that is, the change in the value of the optimal objective function caused by a unit change in the constraints.

[0144] Based on optimized control, this embodiment further transforms the originally implicit and abstract operational constraints into explicit and monetized economic information by calculating and outputting quantitative indicators of the economic value of constraints (shadow prices and / or cost reduction). This can provide operators with clear operational guidance, such as identifying which constraints are the key bottlenecks currently restricting efficiency improvement. It can also provide management with a scientific basis for equipment modification investment decisions, such as quantifying the expected returns and investment payback period of "bottleneck removal" through shadow prices, thus enabling this embodiment to have decision support capabilities.

[0145] In an exemplary embodiment, after obtaining the target operating variable parameters of the distillation column, the method further includes:

[0146] Step 207: Monitor the operating mode of the controller corresponding to the operation variable.

[0147] Step 208: If it is detected that the working mode of any controller is in manual mode, mark the corresponding operating variable of the controller as a fixed parameter in the economic objective function, and re-optimize the economic objective function to obtain the updated target operating variable parameters of the distillation column.

[0148] The working mode includes at least two states: automatic mode and manual mode. In automatic mode, the underlying controller accepts external remote setpoint instructions and automatically adjusts the actuator to track the setpoint. In manual mode, the underlying controller does not accept external setpoint instructions, and its output value is manually set by the operator.

[0149] In this embodiment, by continuously monitoring the current working state of the controllers associated with each operand, when it is detected that a certain underlying controller has been switched to manual mode, a security collaboration mechanism is immediately triggered. During the optimization problem construction process, the operand MV corresponding to this controller is... k Instead of being considered as optimization variables, they are marked as fixed parameters, with their values ​​taken from the current actual measurements. At this point, the dimensionality of the optimization problem is reduced accordingly, i.e., dimensionality reduction is performed. Subsequently, the same optimization solution process as in step 205 above is invoked to resolve the dimensionality-reduced optimization problem, obtaining the updated optimal setpoints of the remaining adjustable operational variables, i.e., the updated target operational variable parameters.

[0150] In this embodiment, by introducing a monitoring step for the underlying controller's operating mode, when some controllers are switched to manual mode due to maintenance, troubleshooting, or operator intervention, the system will not blindly continue to output unexecutable setpoints for the variable. Instead, it will "lock" the variable to its current actual value and automatically perform dimensionality reduction optimization. Based on this embodiment, it can ensure that the remaining controllable operational variables can still be effectively optimized under abnormal operating conditions, rather than completely exiting operation. It can also avoid conflicts with manual operation, thereby improving the robustness and safety of this embodiment in complex industrial environments and ensuring that the system can still operate safely and effectively under abnormal operating conditions.

[0151] The following is an implementation case of a multivariable control method for distillation columns:

[0152] A petrochemical company's butane-propane (C4 / C3) distillation column originally used a traditional multivariate predictive control method with empirical weighting coefficients. Operation revealed that the system consistently set the reflux flow rate (MV1) to 56.5 and the reboiler steam (MV2) to 14.0, at which point the product quality just met the standards (C4 content at the top of the column = 5%, C3 content at the bottom of the column = 5%). Analysis showed that although the system operated stably and achieved "edge-limit" operation, the annual profit consistently fell below the industry's advanced level, raising suspicion that the operating point was not truly optimal.

[0153] To this end, the petrochemical company deployed the multivariable control method for distillation columns described in this invention, and its corresponding system architecture diagram is as follows: Figure 3 As shown, it includes an enterprise information system layer, a multivariable economic optimization control core system layer, and a basic control layer, and executes the multivariable control method of the distillation column based on the multivariable economic optimization control core system.

[0154] 1. Initialization and Extended Modeling (Reference) Figure 4 )

[0155] First, execute Figure 4 The extended modeling process is shown.

[0156] Under safe operating conditions, a step test was conducted on two operating variables (MV1: return flow rate t / h; MV2: reboiler steam flow rate t / h), and data for four key controlled variables (CV) were collected:

[0157] CV1: C4 (butane) content at the top of the column (%)

[0158] CV2: C3 (propane) content in the bottom of the column (%)

[0159] CV3: C3 product flow rate at the top of the tower (t / h, tons per hour)

[0160] CV4: C4 product flow rate in the reboiler (t / h, tons per hour)

[0161] The initial gain matrix Kp obtained from the process data obtained through the step test is shown in Table 1:

[0162] Table 1: Initial Gain Matrix (Kp)

[0163]

[0164] Subsequently, an examination of the material balance revealed the initial gain coefficient (Kp) in the initial gain matrix. 31 +(Kp) 41 =0.02, which deviates from the theoretical value of 0. Therefore, a key step is to employ a constrained regression algorithm, forcibly adding an equality constraint (Kp). 31 +(Kp) 41 =0 and (Kp) 32 +(Kp) 42 =0. After correction, the self-consistent extended gain matrix is ​​obtained, which is the corrected gain matrix Kp*, as shown in Table 2:

[0165] Table 2: Corrected gain matrix (Kp*)

[0166]

[0167] Note: Prices for CV3 / CV4 products are from the company's ERP system; steam costs for Q2 are from the Energy Management Center.

[0168] 2. Construction and optimization of economic objective function (reference) Figure 5 )

[0169] The system automatically reads the latest economic parameters from the enterprise database (propane price P3 = 6800 yuan / ton, butane price P4 = 6200 yuan / ton, steam cost Q2 = 250 yuan / ton) and constructs a data structure as follows: Figure 5 The actual economic objective function shown (converted into a minimization problem):

[0170] J=-(6800×CV3+6200×CV4-250×MV2)

[0171] At the same time, define the constraints:

[0172] ① Product quality constraints: CV1≤5%, CV2≤5%

[0173] ② Physical constraints: 50≤MV1≤80, 14≤MV2≤16

[0174] 3. Optimization of Solution and Result Analysis (Reference) Figure 6 )

[0175] Obtaining the steady-state model of the process based on the process gain matrix:

[0176] CV1 = -0.962 * MV1 + 4.17 * MV2 + bias1

[0177] CV2 = 0.806 * MV1 - 5.32 * MV2 + bias2

[0178] CV3 = -1.000*MV1 + 5.40*MV2 + bias3

[0179] CV4 = 1.000 * MV1 - 5.40 * MV2 + bias4

[0180] Among them, bias1, bias2, bias3 and bias4 are the deviation quantities.

[0181] We select initial values, that is, the boundary values ​​of the quality indicators CV1 and CV2 are both equal to 5%, MV1=56.5, MV2=14;

[0182] Substituting into the steady-state model of the process, the two operational boundary line equations that satisfy the quality requirements are obtained as follows:

[0183] -0.962*MV1 + 4.17*MV2 = 4.027

[0184] 0.806*MV1 - 5.32*MV2 = -28.941

[0185] Based on this, the objective economic function can be transformed into:

[0186] J=600*MV1-2990*MV2-6800*bias3-6200*bias4

[0187] By combining boundary lines and constraints, we can obtain, for example Figure 6 The aforementioned condition constraint diagram:

[0188] Within the feasible region, the linear programming (LP) algorithm is used to automatically search for the most efficient operation point, that is, to calculate the J value in real time and move it in the direction of minimization.

[0189] The system's performance optimization engine compares two scenarios:

[0190] Scenario A (Initial Plan): The initial results are MV1=56.50, MV2=14.0, and the calculated hourly profit is 147,960 yuan.

[0191] Scenario B (Optimization Scheme): The optimization results are MV1=65.17, MV2=16.0, and the calculated hourly profit is 148,738 yuan.

[0192] 4. Decision support for the monitoring module (see reference) Figure 7 )

[0193] The system monitoring interface outputs the following key economic indicators in real time:

[0194] Shadow Price / Cost Reduction Analysis: The cost reduction for the "Recirculation Lower Limit ≥ 50" constraint is 600 yuan / hour / unit. This means that relaxing this constraint by 1 unit, i.e., "Recirculation Lower Limit ≥ 49", yields a potential revenue of 600 yuan per hour. The shadow price for the "Steam Upper Limit ≤ 16" constraint is 2990 yuan / hour / unit. This means that relaxing this constraint by 1 unit, i.e., "Steam Upper Limit ≤ 17", can increase profits by 2990 yuan per hour.

[0195] Constraints on Economic Ranking Report: Automatically generated and sorted, it clearly points out that the "steam ceiling" is the primary bottleneck currently constraining the economy.

[0196] 5. Implementation of secure collaboration logic (refer to) Figure 8 )

[0197] When the operator switches the bottom tray temperature controller (associated with MV2) to "manual" mode, the system immediately executes... Figure 8 The process shown is as follows:

[0198] ① An abnormal state of MV2 (steam) was detected.

[0199] ② Trigger the security collaboration mechanism: Fix MV2 to the current actual value in the optimization problem, and the optimization problem will automatically reduce the dimensionality, optimizing only MV1 (backflow).

[0200] ③ A prompt will pop up on the control interface to ensure safe operation.

[0201] 6. Implementation Results and Economic Benefits

[0202] Calculations show that after adopting this invention, the average hourly profit of the distillation column increased from RMB 147,900 to RMB 148,700, an increase of RMB 778 per hour. Based on 8,000 hours of operation per year, the annualized profit growth reaches RMB 6.22 million. Meanwhile, the company has launched a "bottlenecks removal" project to upgrade the reboiler, with an estimated investment payback period of only 3.4 years.

[0203] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0204] Based on the same inventive concept, this application also provides a distillation column multivariate control device for implementing the above-described distillation column multivariate control method. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more distillation column multivariate control device embodiments provided below can be found in the limitations of the distillation column multivariate control method described above, and will not be repeated here.

[0205] In one exemplary embodiment, such as Figure 9 As shown, a multivariable control device for a distillation column is provided, including: a model building module for acquiring process data of the distillation column and obtaining an initial gain matrix based on the process data; the process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate;

[0206] The model calibration module communicates with the model building module and is used to perform material balance calibration on the initial gain matrix based on product flow to obtain the calibrated gain matrix.

[0207] The objective function construction module communicates with the model calibration module to obtain the economic parameters of the operating variables and / or controlled variables, and constructs the economic objective function based on the calibrated gain matrix and economic parameters.

[0208] The constraint construction module is used to construct physical constraints based on the physical limitations of the operating variables, and to construct product quality constraints based on the quality requirements of the controlled variables.

[0209] The objective function solving module is connected to the objective function construction module and the constraint construction module respectively. It is used to optimize the economic objective function based on physical constraints and product quality constraints to obtain the target operating variable parameters of the distillation column.

[0210] The control module, which communicates with the objective function solving module, is used to adjust the operating variables of the distillation column to the target operating variable parameters.

[0211] Each module in the aforementioned multivariable control device for the distillation column can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0212] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 10 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data related to multivariate control of the distillation column. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a multivariate control method for a distillation column.

[0213] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0214] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0215] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0216] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0217] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0218] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0219] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0220] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A multivariable control method for a distillation column, characterized in that, The method includes: Acquire process data from the distillation column and obtain an initial gain matrix based on the process data; the process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate; Based on the product flow rate, the initial gain matrix is ​​corrected by material balance to obtain the corrected gain matrix; Obtain the economic parameters of the operating variable and / or the controlled variable, and construct an economic objective function based on the corrected gain matrix and the economic parameters; Physical constraints are constructed based on the physical limitations of the operated variables, and product quality constraints are constructed based on the quality requirements of the controlled variables. Based on the physical constraints and the product quality constraints, the economic objective function is optimized to obtain the target operating variable parameters of the distillation column; The operating variables of the distillation column are adjusted to the target operating variable parameters.

2. The method according to claim 1, characterized in that, Acquire process data from the distillation column and obtain an initial gain matrix based on the process data, including: A step test is performed on the distillation column to obtain process data of the distillation column; wherein, the process data includes the step change of the operating variable and the steady-state response change of the controlled variable; The initial gain matrix is ​​obtained based on the process data.

3. The method according to claim 2, characterized in that, The process data is a set; The initial gain matrix is ​​obtained based on the process data, including: A step change of a preset magnitude is applied to each controlled variable in sequence, and the steady-state response change corresponding to each controlled variable is collected. Based on the step change of each operated variable and the corresponding steady-state response change of each controlled variable, the initial gain coefficient between each operated variable and each controlled variable is obtained. The initial gain matrix is ​​constructed based on the initial gain coefficients between each operated variable and each controlled variable. Specifically, based on the step change of each manipulated variable and the corresponding steady-state response change of each controlled variable, the initial gain coefficient between each manipulated variable and each controlled variable is obtained, including: For any manipulated variable and any controlled variable, calculate the ratio of the steady-state response change of the controlled variable to the step change of the manipulated variable, and use this ratio as the initial gain coefficient between the manipulated variable and the controlled variable.

4. The method according to claim 2, characterized in that, The process data consists of multiple sets; The initial gain matrix is ​​obtained based on the process data, including: Construct a linear equation with the operated variable as the independent variable and the controlled variable as the dependent variable; Substituting the step changes of the multiple sets of the operated variables and the steady-state response changes of the controlled variables into the linear equations, a system of linear equations is obtained. The linear equations are solved using the least squares method to obtain the initial gain coefficient between each operated variable and each controlled variable. The initial gain matrix is ​​constructed based on the initial gain coefficients between each operated variable and each controlled variable.

5. The method according to claim 1, characterized in that, Based on the product flow rate, the initial gain matrix is ​​corrected using material balance to obtain the corrected gain matrix, including: Material balance constraints are constructed based on the product flow rate to reflect the material balance relationship of the distillation column; wherein, the material balance constraints are used to ensure that the sum of the gain coefficients of each product flow rate corresponding to the same operating variable satisfies a preset mathematical relationship; Under the material balance constraint, with the goal of minimizing the sum of squared prediction errors, the initial gain matrix is ​​subjected to regression correction to obtain the corrected gain matrix.

6. The method according to claim 1, characterized in that, Based on the physical constraints and the product quality constraints, the economic objective function is optimized to obtain the target operating variable parameters of the distillation column, including: Based on the corrected gain matrix, a process steady-state model characterizing the steady-state relationship between the operated variable and the controlled variable is constructed. Substituting the process steady-state model into the economic objective function yields an economic objective function with the operation variable as the independent variable. Substituting the quality index boundary values ​​from the product quality constraints into the process steady-state model, the operational boundary equations characterized by the operational variables are obtained by solving the model. Within the feasible operating region bounded by the operational boundary equation and the physical constraints, a linear programming solution is performed on the economic objective function with the operational variables as independent variables to obtain the target operational variable parameters of the distillation column.

7. The method according to claim 1, characterized in that, After obtaining the target operating variable parameters of the distillation column, the method further includes: Based on the economic objective function, obtain the constraint economic value quantification index of the operational variable corresponding to the physical constraint condition; the constraint economic value quantification index includes the marginal economic value used to quantify the operational variable corresponding to the physical constraint condition. And / or, after obtaining the target operating variable parameters of the distillation column, the method further includes: Monitor the operating mode of the controller corresponding to the operation variable; If any controller is detected to be in manual mode, the corresponding operating variable of that controller is marked as a fixed parameter in the economic objective function, and the economic objective function is re-optimized to obtain the updated target operating variable parameters of the distillation column.

8. A multivariable control device for a distillation column, characterized in that, The device includes: The model building module is used to acquire process data of the distillation column and obtain an initial gain matrix based on the process data; the process data includes operating variables and controlled variables, and the controlled variables include product quality data and product flow rate; The model calibration module is communicatively connected to the model construction module and is used to perform material balance calibration on the initial gain matrix based on the product flow rate to obtain the calibrated gain matrix. The objective function construction module is communicatively connected to the model calibration module and is used to obtain the economic parameters of the operating variables and / or the controlled variables, and construct an economic objective function based on the calibrated gain matrix and the economic parameters. The constraint construction module is used to construct physical constraints based on the physical limitations of the operating variables, and to construct product quality constraints based on the quality requirements of the controlled variables. The objective function solving module is communicatively connected to the objective function construction module and the constraint condition construction module, respectively, and is used to optimize and solve the economic objective function according to the physical constraints and the product quality constraints to obtain the target operating variable parameters of the distillation column; The control module is communicatively connected to the objective function solving module and is used to adjust the operating variables of the distillation column to the target operating variable parameters.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 9.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.

Images