A hierarchical optimization control method for the coordinated production and treatment of organic pollutants in reactive distillation

Abstract

The present invention discloses a hierarchical optimization control method for the coordinated production and treatment of organic pollutants in reactive distillation, which belongs to the field of control. The control method is applied to a strongly coupled system of emission and treatment of organic pollutants in reactive distillation. The system is divided into a coordination layer and a subsystem layer, a two-layer optimization system objective function is constructed, and a multi-objective dynamic optimization algorithm based on reference point prediction is used to solve the optimal operation trajectory. The subsystem layer is subjected to correlation analysis and decomposed into a reactive distillation production subsystem X 1 and pollutant control subsystem X 2 Define the pollutant removal rate index function and establish X 1 With X 2 The master-slave relationship between X 1 Adopt the original production control system as the main control system, X 2 The synchronous optimization control (SMBOC) based on economic model is used to design the slave control system and track the optimal operation trajectory. The invention solves the problem of coordinated optimization control of the production and treatment of organic pollutants in reactive distillation, and realizes safe and stable operation of the system under ultra-clean emissions, while maximizing economic benefits.

Description

Technical Field

[0001] The present invention relates to the field of coordinated production and treatment of organic pollutants and hierarchical optimization control, and in particular to the establishment of a molar flow model for predicting pollutant emissions according to different working conditions, a multi-objective dynamic optimization algorithm (PDMOP) based on reference point prediction to solve the optimal operating trajectory of a two-level system, and a synchronous optimization control (SMBOC) based on an economic model to design a slave control system to track the optimal operating trajectory. Background Art

[0002] At present, most of the chemical production in my country still adopts the "separation of production and treatment" mode. The operation mode of "production first and treatment later" of pollutants in the chemical production process has caused the problem of complex and diverse pollutant composition and huge emission volume. The traditional terminal pollutant treatment process technology is difficult to meet the increasingly stringent environmental protection requirements. In addition, organic pollutants contain a large amount of potential available energy. How to efficiently utilize the energy of pollutants themselves is very critical to greatly reduce the energy consumption of pollutant treatment. In the traditional pollutant treatment process, on the one hand, due to the lack of effective treatment process technology for the treatment of difficult-to-degrade organic pollutants, the pollutant emission indicators and total amount exceed the standard; on the other hand, the traditional pollutant treatment equipment has low intelligence, high energy consumption, and high operating costs. In order to save costs, enterprises have repeatedly violated the law of stealing and leaking. The treatment of industrial pollutants has become the primary task of my country's environmental governance. In order to meet the strict safety and environmental protection index requirements, it is necessary to take the coordinated development of production and treatment as the goal, and carry out the dynamic coordinated optimization control research of pollutant emission and treatment large systems. It is of great significance to solve the problem of high-concentration organic pollutant treatment, achieve energy-saving, efficient and stable operation of equipment, and achieve long-term standard emission of organic pollutants. Summary of the invention

[0003] The present invention provides a hierarchical optimization control method for the coordinated production and treatment of organic pollutants in reactive distillation in response to the above-mentioned technical problems:

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] A hierarchical optimization control method for the coordinated production and treatment of organic pollutants in reactive distillation, the method comprising the following steps:

[0006] (1) Conduct a correlation analysis between the reactive distillation production process and the types and amounts of organic pollutants emitted, and define a pollutant removal rate index function;

[0007] (2) Decompose the strongly coupled system of organic pollutant emission control of reactive distillation into a synergistic layer and a subsystem layer, and decompose the subsystem layer into subsystem X 1 and subsystem X 2 , construct the optimization system objective function of the collaboration layer and the subsystem layer;

[0008] (3) Through the reactive distillation production operating condition environment detection operator, the normal operating condition and raw material fluctuation, load adjustment, excessive separation, accident occurrence, and unknown abnormal operating condition are identified online, and a pollutant emission prediction model is established to predict the molar flow rate of pollutant emissions;

[0009] (4) Coordinate subsystem X for the coupled system of pollutant emission and pollutant treatment in reactive distillation production 1 With X 2 The dynamic coordination relationship between the coordination layer and subsystem X is calculated by using the multi-objective dynamic optimization algorithm based on reference point prediction (PDMOP). 1 and subsystem X 2 The optimal operation trajectory of

[0010] (5) Establish subsystem X 1 With X 2 The master-slave relationship between X 1 As the main control system, X 1 The optimal operation trajectory is applied to the reactive distillation production control system to predict pollutant emissions; 2 For the slave control system, X 2 The optimal operating trajectory is applied to the pollutant treatment subsystem, and the simultaneous optimal control based on economic model (SMBOC) is used to track the optimal operating trajectory of the pollutant treatment process.

[0011] The hierarchical optimization control method is applied to the strongly coupled system of reactive distillation organic pollutant emission control. The whole system is divided into a synergistic layer and a subsystem layer. The subsystem layer is decomposed into a reactive distillation production subsystem X 1 and pollutant control subsystem X 2 ; Construct the optimization system objective function of the synergistic layer and the subsystem layer, and use the multi-objective dynamic optimization algorithm based on reference point prediction (PDMOP) to solve the optimal operation trajectory of the synergistic layer and the subsystem layer. According to the correlation between the emission of organic pollutants in reactive distillation production and the production conditions and external disturbance factors, a pollutant emission prediction model under different production conditions of reactive distillation is established to predict the subsystem X 1 Molar flow rate of pollutant emissions; Subsystem X 1 The original production control system is used as the main control system, subsystem X 2 The synchronous optimal control based on economic model (SMBOC) is used to design the slave control system to track the optimal operation trajectory.

[0012] First, the Pearson correlation coefficient method was used to conduct a correlation analysis between the key variables of temperature, flow rate, and pressure collected during the reaction distillation production process and the types, components, and emissions of organic pollutants. The correlation coefficient ranged from -1 to 1. When the calculated result was zero, it indicated that there was no correlation between the variables undergoing the correlation coefficient analysis; when the calculated result was greater than zero, it indicated that the variables undergoing the correlation coefficient analysis were positively correlated; when the calculated result was less than zero, it indicated that the variables undergoing the correlation coefficient analysis were negatively correlated. Variables with an absolute value of 0.6 or more were taken as strongly correlated variables, and variables with strong correlation were used to establish subsequent pollutant emission prediction models.

[0013] Secondly, the types of organic pollutants emitted during the reactive distillation process include volatile organic compound waste gas (VOCs), organic wastewater (COD) and high-concentration distillation residue. The Pearson correlation coefficient method was used to analyze the correlation of different types of organic pollutant emissions. 1 and X 2 The removal rate index is established as the indicator variable of the influence relationship between subsystems;

[0014] The pollutant removal rate index function is defined as the ratio of the total amount of pollutant treatment to the total amount of emissions. It is used as the index variable of the influence relationship between subsystems to coordinate subsystem X 1 With X 2 The dynamic synergistic relationship between them, the pollutant removal rate is defined as follows:

[0015]

[0016] where F ig , F il , F is are the discharge flow rates of reactive distillation waste gas, waste water and residual liquid, v ig , v il , v is are the emission concentrations of waste gas, wastewater and residual liquid, respectively, and F og , F ol , F os is the waste gas, wastewater and residual liquid treatment flow, v og , v ol , v os To control the concentration of waste gas, wastewater and residual liquid.

[0017] Thirdly, the whole system is divided into the coordination layer and the subsystem layer. The subsystem layer is decomposed into the reaction distillation production subsystem X 1 and pollutant control subsystem X 2 ; Construct collaborative layer and subsystem layer to optimize the system objective function.

[0018] The economic optimization goal of the collaborative layer is:

[0019] J~{maxQ(X),maxη(X),minE(X)}

[0020] Where: Q(X) = P p F p -P in F in ; E(X) = X E1 +X E2

[0021]

[0022] Among them, J is the optimization objective function of the collaborative layer, Q(X) is the product revenue, η(X) is the pollutant removal rate, E(X) is the total energy consumption, P p is the product price, P in is the raw material price, F p Refers to product flow, F in Refers to the raw material flow rate, X = f (X 1 ,X 2 ) is subsystem X 1 , X 2 Decision variables, J 1 (X P ,X E1 ), J 2 (X η ,X E2 ) are subsystem X 1 and subsystem X 2 The expected trajectory of X P is the product revenue, X η is the removal rate, X E1 is the energy consumption of the reactive distillation process; X E2 is the energy consumption of the pollutant treatment process;

[0023] The subsystem X 1 The optimization objective function is:

[0024] Min:

[0025] stg j (F in (t),F p (t),F ig (t),F il (t),T i (t))≥0, (j=1,2,…,n; i=1,2,…,m)

[0026]

[0027] Among them, J 1For subsystem X 1 The optimization goal is X P is the product revenue, X E1 is the energy consumption of reactive distillation process, F in (t),F p (t),F ig (t),F il (t),T i (t) operating variables of feed flow rate, product flow rate, waste gas flow rate, waste water flow rate and temperature of the reactive distillation tower during the reactive distillation process; is the state equation of the control loop for the manipulated variable;

[0028] The subsystem X 2 The optimization objective function is:

[0029] Min:

[0030] stg k (K O2 (t),F Res (t),T(t))≥0, (k=1,2,…,n)

[0031]

[0032] Among them, J 2 For subsystem X 2 The optimization goal is X η is the removal rate, X E2 is the energy consumption of the pollutant treatment process, They represent the oxygen excess multiple, residual liquid flow rate, and fluidized bed temperature operating variables in the organic pollution control process. is the state equation of the control loop for the manipulated variable.

[0033] Fourthly, the reaction distillation production condition environment detection operator is used to identify the production conditions online, establish a pollutant emission prediction model under different conditions, and realize the online prediction of the molar flow rate of pollutant emissions based on the established prediction model.

[0034] A large amount of organic pollutants will be generated in the reactive distillation production process, including volatile organic waste gas, organic wastewater and high-concentration distillation residual liquid. Firstly, the reactive distillation production condition environment detection operator is used to perform online working condition identification of the reactive distillation production process, which is specifically divided into normal operation, raw material fluctuation, load adjustment, excessive separation, accident occurrence, and unknown abnormal working conditions. A data set including input data of temperature, flow rate, and pressure and output data of pollutant emission components is established. Then, the data sets are divided into different working condition data sets according to the time series based on the expert marking information. A long short-term memory recurrent neural network (LSTM-RNN) prediction model is established for each working condition to make real-time prediction of the molar flow rate of organic pollutant emissions.

[0035] The specific judgment rules of the reactive distillation production working condition environment detection operator are as follows:

[0036] Different working conditions Judgment rules Raw material fluctuations The flow rate or concentration of raw materials increases or decreases by more than 10% within 1 hour Load adjustment The heat setting value is modified by more than 10% within 1 hour Excessive separation The product concentration in the tower bottom effluent was detected to be 99.9% The accident happened An accident alarm occurs when the high limit is exceeded Unknown abnormal conditions When abnormal conditions other than those mentioned above occur

[0037] Fifthly, the multi-objective dynamic optimization algorithm based on reference point prediction (PDMOP) is used to calculate the subsystem X 1 and subsystem X 2 The optimal operating trajectory.

[0038] The improved non-dominated genetic algorithm is used to perform steady-state optimization on the production-control collaborative optimization target. When the working conditions change, reference point prediction is performed based on the working condition environment detection operator, individual association rules and pollutant emission prediction model. The initial population of the Pareto optimal solution set under the new working conditions is predicted using the environmental and historical optimal solution information. The convergence speed is accelerated to quickly track the Pareto frontier. The multi-objective optimization algorithm (PDMOP) is used to optimize the calculation and update the optimal operation trajectory of the two subsystems; the system-level optimization target is quickly solved. The specific process of the PDMOP algorithm to solve this optimization problem includes:

[0039] 1) An improved non-dominated genetic algorithm is used to perform steady-state optimization on the production-control collaborative optimization target. When the environmental detection operator determines that a change in the working condition occurs, it starts to switch to a dynamic optimization mode based on reference point prediction;

[0040] 2) If it is in an unknown abnormal working condition, use linear regression or neural network method to establish the required reference point, combine individual association rules, predict the movement of the Pareto solution set containing time series information, and make Gaussian perturbations near each individual to increase the diversity of the predicted population;

[0041] 3) If it is under known abnormal conditions, the movement of the Pareto solution set is predicted based on the pollutant emission prediction model information, and the historical information about the optimization process is stored and updated iteratively;

[0042] 4) The predicted population is used as the initial population in the new environment, and it is continuously iterated until the hypervolume index meets the dynamic optimization algorithm index, thereby accelerating the convergence speed while maintaining the diversity of the population.

[0043] Sixth, the economic model-based synchronous optimization control (SMBOC) is used to track the optimal operation trajectory of the pollutant treatment process. Within a production cycle ΔT, the optimal operation trajectory is solved by the multi-objective dynamic optimization algorithm based on reference point prediction (PDMOP), and ΔT is divided into N equal parts, each with a control step of Δt. CM , that is, ΔT = NΔt CM , subsystem X 2 The optimal operation trajectory is tracked in real time using the synchronous optimal control based on economic model (SMBOC);

[0044] The synchronous optimization control based on economic model (SMBOC) is used to transform the dynamic optimization into a nonlinear programming problem with discrete control input variables. The improved interior point method based on adaptive penalty function and feasibility driven strategy is used to solve the nonlinear programming problem and obtain the optimal system state equilibrium point x * and the control input variable trajectory u ij ; Then, a rolling optimization strategy is adopted, and the optimal balance point between the waste gas, wastewater or residual liquid flow rate and the oxygen excess multiple is given through synchronous optimization and iterative optimization calculation, and the system state is controlled near the balance point at the lowest energy cost. The optimization objective function is composed of quadratic tracking indicators, process state variables and control input variable slope control, and the energy dynamic balance rate model, pollutant removal rate and operation boundary are used as equations or inequality multi-constraints. The optimization problem is described as:

[0045]

[0046] Among them, J is the optimization objective function, α k , β q , γ i Represent the diagonal elements of the coefficient matrix of the corresponding indicator item; x represents the process state variable, represents the measured value of the state variable predicted by the model at time step k; x k represents the actual value of the system process state variable when the time step is k, x * represents the optimal system state equilibrium point; u represents the operating variable, u ij Input variable value for the jth step of the ith control, u ij Input variable value for the i-th control j-1 step, dimension is N v m ; Represent the lower and upper limits of the feasible domain of the operating variable, ΔtCM Indicates the control domain period, Represent the lower and upper limits of the control time respectively, Represent the lower and upper limits of the feasible domain of the state variables, respectively, and the dimensions are HBR(x) represents the energy balance rate; y is the output of the prediction model; COD(x * ) is the expected content of organic wastewater in organic pollutant emissions, VOCs (x*) is the expected content of volatile organic waste gas in organic pollutant emissions, N CI i and Δt CI i Respectively represent the control steps and step length of the i-th control input variable;

[0047]

[0048] HBR is defined as the ratio of total energy absorbed to total energy released:

[0049] Among them, the sensible heat required for heating wastewater and residual liquid and latent heat of vaporization And the total sensible heat Q required to raise the water vapor and air to the optimal reaction temperature gas ;Q rds , and represents the heat released by the oxidation of organic matter in the distillation residue, water and carbon dioxide, respectively. wg , H ww and H ds Respectively represent the heat release corresponding to the oxygen consumption of waste gas, wastewater and residual liquid, N O2 represents the mole fraction of oxygen, V wg 、V ww and V ds Respectively represent the volume corresponding to the oxygen consumption of waste gas, wastewater and residual liquid, ρ ds represents the density of the residual liquid, Q H2o-ds Indicates the heat absorbed by the residual liquid during the purification process to generate water, Q CO2-ds It indicates the heat absorbed by the residual liquid to generate carbon dioxide during the purification process.

[0050] Seventh aspect, subsystem X 1 and subsystem X 2 Each control system operates independently.

[0051] Subsystem X 1 Through the underlying multiple PID control loops, the rapid tracking control of the reactor, reaction distillation tower, recovery tower and other units can be achieved. Similarly, subsystem X 2The bottom control loop is composed of multiple PID control loops to achieve rapid tracking control of wastewater residual liquid feed, air feed and reactor temperature. Each bottom controller feeds back the predicted state of the next moment and the coupling relationship between them to the coordinator, and transmits the actual state parameter information to the iterative prediction model for model correction, and passes the correction information to the coordinator for unified correction. Establish a hierarchical optimization control system of "production and treatment coordination" to achieve safe and stable operation of the entire large system under ultra-clean emissions and maximize economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification are used to explain the principles of the present application, and do not constitute an improper limitation on the present application.

[0053] Figure 1 It is a schematic diagram of the overall architecture of the present invention.

[0054] Figure 2 Schematic diagram of the coupling process of reactive distillation and pollutant control.

[0055] Figure 3 This is the flow chart of the pollutant emission prediction model under different working conditions.

[0056] Figure 4 This is the flow chart of the multi-objective dynamic optimization algorithm based on reference points.

[0057] Figure 5(a) shows the optimization operation results of the collaborative layer.

[0058] Figure 5(b) and (c) are subsystem X 1 , X 2 The optimization results of the run.

[0059] Figure 6 The molar flow prediction results of organic pollutants emitted from the acrylic acid production process.

[0060] Figure 7 Tracking curves for the pollution control subsystem. DETAILED DESCRIPTION

[0061] The present invention will be further described below in conjunction with embodiments, but the protection scope of the present invention is not limited thereto:

[0062] Taking the acrylic acid reactive distillation production process as an example, a hierarchical optimization control method for the coordinated production and treatment of organic pollutants in reactive distillation is introduced.

[0063] The types of organic pollutants generated in the production process of acrylic acid reactive distillation include volatile organic waste gas VOCs, organic wastewater COD and high-concentration distillation residue; the Pearson correlation coefficient method was used to conduct correlation analysis on the generation of various types of organic pollutants, among which the removal rate X η (%) is defined as the ratio of the total amount of pollutants treated to the total amount of emissions in acrylic acid production, as expressed by the following formula:

[0064]

[0065] where F ig , F il , F is are the discharge flow rates of reactive distillation waste gas, waste water and residual liquid, v ig , v il , v is are the emission concentrations of waste gas, wastewater and residual liquid, respectively, and F og , F ol , F os is the waste gas, wastewater and residual liquid treatment flow, v og , v ol , v os To control the concentration of waste gas, wastewater and residual liquid.

[0066] The entire system is decomposed into a synergistic layer and a subsystem layer. The economic optimization target indicators of the synergistic layer include output, energy consumption, and pollutant removal rate. The economic optimization target of the synergistic layer is as follows:

[0067] J~{maxQ(X),maxη(X),minE(X)}

[0068] Where: Q(X) = P p F p -P in F in ; E(X) = X E1 +X E2

[0069]

[0070] Among them, J is the optimization objective function of the collaborative layer, Q(X) is the acrylic acid revenue, η(X) is the pollutant removal rate, E(X) is the total energy consumption, P p is the acrylic price, P in is the price of propylene, F p Refers to acrylic acid flow, F in Refers to the propylene flow rate, X = f (X 1 ,X 2 ) is acrylic acid subsystem X 1 , X 2 Decision variables, J 1 (XP ,X E1 ), J 2 (X η ,X E2 ) are acrylic acid subsystem X 1 and acrylic subsystem X 2 The expected trajectory of X P For acrylic acid yield, X η is the removal rate, X E1 is the energy consumption of the distillation process of acrylic acid production; X E2 is the energy consumption of acrylic acid pollutant treatment process;

[0071] The optimization problem and multi-objective optimization function of the subsystem layer are obtained by analyzing the optimization problem of the collaborative layer, that is, the strongly coupled subsystem layer of acrylic acid reactive distillation organic pollutant emission control consists of two parts, namely, acrylic acid reactive distillation production subsystem X 1 Acrylic acid pollution control subsystem X 2 , construct the optimization objective function of the two subsystems, as shown in the following formula:

[0072] (1) Acrylic acid reactive distillation production subsystem X 1 The optimization target is acrylic acid yield X P (%), total energy consumption of production subsystem X E1 (W), the specific optimization model is:

[0073] Min:J 1 =X P , J 2 =X E1

[0074] st(F 1 ,F 2 ,F 3 ,R,T 1 )≥0

[0075]

[0076] Among them, J 1 and J 2 Acrylic acid reactive distillation production subsystem X 1 Objective functions of two multi-objective optimization problems; F 1 is the propylene flow rate, F 2 is the water flow rate, F 3 is the air flow rate of the production system, R is the distillation reflux ratio flow rate and T 1 Reactive distillation tower temperature; For subsystem X 1 Manipulated variable control loop state equation,

[0077] F 1、F 2 、F 3 , R and T 1 is the operation variable.

[0078] (2) Acrylic acid pollutant treatment subsystem X 2 The optimization goal is the organic pollutant removal rate X η (%), total energy consumption of pollution control subsystem X E2 (W), the specific optimization model is:

[0079] Min:J 3 =X η ,J 4 =X E2

[0080]

[0081] Among them, J 3 and J 4 Acrylic acid reactive distillation treatment subsystem X 2 Objective functions of two multi-objective optimization problems; F 4 is the air flow rate, T 2 is the water flow rate, T 3 is the air flow rate of the treatment system, and P is the pressure of the pollution control reactor; For subsystem X 2 Manipulated variable control loop state equation, F 4 , T 2 , T 3 , P is the operating variable.

[0082] The above-mentioned collaborative layer and the multi-objective model are calculated using the multi-objective dynamic optimization algorithm based on reference point prediction (PDMOP), and the optimization operation results are shown in Figure 5(a).

[0083] The above subsystem X is transformed into 1 and subsystem X 2 The multi-objective optimization model is used for calculation, and the optimization operation results are shown in Figure 5(b) and Figure 5(c).

[0084] You can select an optimal solution from the Pareto solution set in the figure and adjust the decision variables to the corresponding parameters, so that the acrylic acid reaction distillation production subsystem X 1 Acrylic acid yield X P (%), total energy consumption of production subsystem X E1 (W) and acrylic acid pollutant treatment subsystem X 2 Organic pollutant removal rate X η (%), total energy consumption of pollution control subsystem X E2(W) Optimization goal, ultimately achieve the optimization task of the collaborative layer, achieve the best expected value, and achieve the purpose of improving production efficiency and energy conservation and emission reduction.

[0085] After obtaining the optimization results, the reaction distillation production condition environment detection operator is used to identify the online working conditions of the acrylic acid reaction distillation production process, which is specifically divided into normal operation, raw material fluctuation, load adjustment, excessive separation, accident occurrence, and unknown abnormal working conditions. The factors affecting pollutant emissions under different working conditions are analyzed and the corresponding operating status time series data are collected to construct a sample set. A long short-term memory recurrent neural network (LSTM-RNN) prediction model is established for each working condition to perform real-time prediction of the molar flow rate of organic pollutants emitted in the acrylic acid production process. The prediction results are as follows: Figure 6 As shown in Figure 2, it can be seen that the predicted curve is very close to the pollutant amount under actual working conditions.

[0086] The specific judgment rules of the acrylic acid reactive distillation production working condition environment detection operator are as follows:

[0087] Different working conditions Judgment rules Raw material fluctuations The flow rate or concentration of raw materials increases or decreases by more than 10% within 1 hour Load adjustment The heat setting value is modified by more than 10% within 1 hour Excessive separation The product concentration in the tower bottom effluent was detected to be 99.9% The accident happened An accident alarm occurs when the high limit is exceeded Unknown abnormal conditions When abnormal conditions other than those mentioned above occur

[0088] Acrylic acid reactive distillation organic pollutant emission control strongly coupled system, acrylic acid reactive distillation production subsystem X 1 The proportional gain and integral time settings of the underlying PID controller are shown in the table.

[0089] Proportional Gain Integration time (min) Direction of action FC101 0.5 0.3 opposite FC102 0.5 0.3 opposite PC201 20 12 just PC202 20 12 just TC301 5 0.5 opposite TC302 5 0.5 opposite TC303 5 0.5 opposite LC401 2 9999 just

[0090] This table includes feed control, reactor pressure control, temperature control, and liquid level control in the acrylic acid reactive distillation production subsystem.

[0091] The acrylic acid organic pollutant treatment subsystem adopts the synchronous optimization control (SMBOC) based on the economic model, whose input is the output of the prediction model, and tracks the optimal operation trajectory of the pollutant treatment process. The dynamic optimization is converted into a nonlinear programming problem of discrete control input variables. The improved interior point method based on adaptive penalty function and feasibility driven strategy is used to solve the nonlinear programming problem and obtain the optimal system state equilibrium point x * and the control input variable trajectory u ij ; Then, a rolling optimization strategy is adopted to iteratively calculate the optimal solution of waste gas, wastewater, residual liquid flow rate and oxygen excess multiple; the optimization objective function is composed of quadratic tracking index, process state variable and control input variable slope control, and the energy dynamic balance rate model, pollutant removal rate and operation boundary are used as equalities or inequalities multi-constraints; the optimization problem is described as:

[0092]

[0093] Among them, J is the optimization objective function, αk , β q , γ i Represent the diagonal elements of the coefficient matrix of the corresponding indicator item; x represents the process state variable, represents the measured value of the state variable predicted by the model at time step k; x k represents the actual value of the system process state variable when the time step is k, x * represents the optimal system state equilibrium point; u represents the operating variable, u ij Input variable value for the jth step of the ith control, u ij Input variable value for the i-th control j-1 step, dimension is N v m ; Represent the lower and upper limits of the feasible domain of the operating variable, Δt CM Indicates the control domain period, Represent the lower and upper limits of the control time respectively, Represent the lower and upper limits of the feasible domain of the state variables, respectively, and the dimensions are HBR(x) represents the energy balance rate; y is the output of the prediction model; COD(x * ) is the expected content of organic wastewater in organic pollutant emissions, VOCs (x*) is the expected content of volatile organic waste gas in organic pollutant emissions, N CI i and Δt CI i Respectively represent the control steps and step length of the i-th control input variable;

[0094] Through the above SMBOC, the control loop of the pollutant treatment subsystem can be quickly tracked, and the wastewater residual liquid feed, air feed and reactor temperature can be quickly tracked and controlled. The tracking curve of the pollution treatment subsystem is as follows: Figure 7 shown.

[0095] The control performance of the overall system recovers to a steady state in a short time, and the overshoot is reduced to a certain extent, realizing rapid tracking and control of wastewater residual liquid feed, air feed and reactor temperature.

Claims

1. A hierarchical optimization control method for the coordinated production and treatment of organic pollutants in reactive distillation, characterized in that: The control method is applied to the strongly coupled system of organic pollutant emission and treatment in reactive distillation, and the system is decomposed into a collaborative layer and a subsystem layer, a two-layer optimization objective function is constructed, and the optimal solution set Pareto solution set of the two-layer system is obtained respectively by the PDMOP algorithm; the subsystem layer is decomposed into subsystem X1 and subsystem X2, the collaborative relationship between the two subsystems is coordinated, and the optimal operation trajectory of subsystem X2 is tracked by synchronous optimization control based on the economic model; The method comprises the following steps: (1) Conduct a correlation analysis between the reactive distillation production process and the types and amounts of organic pollutants emitted, and define a pollutant removal rate index function; (2) The strongly coupled system of organic pollutant emission control in reactive distillation is decomposed into a synergistic layer and a subsystem layer, and the subsystem layer is decomposed into subsystem X1 and subsystem X2. The optimization system objective function of the synergistic layer and the subsystem layer is constructed; the economic optimization target indicators of the synergistic layer include output, energy consumption, removal rate, and standard emission; the economic optimization objective function of the synergistic layer is: J~{max Q(X),maxη(X),min E(X)} Where: Q(X) = P p F p -P in F in ; E(X) = X E1 +X E2 Among them, J is the optimization objective function of the collaborative layer, Q(X) is the product revenue, η(X) is the pollutant removal rate, E(X) is the total energy consumption, P p is the product price, P in is the raw material price, F p Refers to product flow, F in refers to the raw material flow rate, X = f (X1, X2) is the decision variable of subsystem X1 and X2, J1 (X P ,X E1 ), J2(X η ,X E2 ) are the expected trajectories of subsystem X1 and subsystem X2, respectively. P is the product revenue, X η is the removal rate, X E1 is the energy consumption of the reactive distillation process; X E2 is the energy consumption of the pollutant treatment process; The optimization objective function of subsystem X1 is: s.t.g j (F in (t),F p (t),F ig (t),F il (t),T i (t))≥0,(j=1,2,…,n;i=1,2,...,m) Among them, J1 is the optimization target of subsystem X1, F in (t),F p (t),F ig (t),F il (t),T i (t) represent the operating variables of raw material flow rate, product flow rate, waste gas flow rate, waste water flow rate and reaction distillation tower temperature in the reaction distillation process, respectively. The state equation of the control loop of the subsystem X1 operating variable; The optimization objective function of subsystem X2 is: s.t.g k (K O2 (t),F Res (t),T(t))≥0,(k=1,2,…,n) Among them, J2 is the optimization target of subsystem X2, K O2 (t),F Res (t), T(t) represent the oxygen excess multiple, residual liquid flow rate, and fluidized bed temperature operating variables in the organic pollution control process, respectively. is the state equation of the manipulated variable control loop of subsystem X2; (3) Through the reactive distillation production operating condition environment detection operator, the normal operating condition and raw material fluctuation, load adjustment, excessive separation, accident occurrence, and unknown abnormal operating condition are identified online, and a pollutant emission prediction model is established to predict the molar flow rate of pollutant emissions; (4) For the coupled system of pollutant emission and pollutant treatment in reactive distillation production, the dynamic synergistic relationship between subsystems X1 and X2 is coordinated, and the optimal operation trajectory of the synergistic layer, subsystem X1 and subsystem X2 is calculated using a multi-objective dynamic optimization algorithm based on reference point prediction; (5) Establish a master-slave relationship between subsystems X1 and X2. X1 is the master control system. The optimal operation trajectory of X1 is applied to the reactive distillation production control system to predict pollutant emissions. X2 is the slave control system. The optimal operation trajectory of X2 is applied to the pollutant treatment subsystem. Synchronous optimization control based on the economic model is used to track the optimal operation trajectory of the pollutant treatment process. Within a production cycle ΔT, the optimal operation trajectory is obtained by using a multi-objective dynamic optimization algorithm based on reference point prediction. ΔT is divided into N equal parts, and the control domain period of each part is Δt. CM , that is, ΔT = NΔt CM ,Subsystem X2 uses synchronous optimization control based on economic model to track the optimal operation trajectory in real time; The synchronous optimization control based on the economic model is adopted to transform the dynamic optimization into a nonlinear programming problem of discrete control input variables. The improved interior point method based on adaptive penalty function and feasibility driven strategy is adopted to solve the nonlinear programming problem and obtain the optimal system state equilibrium point x * and the control input variable trajectory u ij ; Then, a rolling optimization strategy is adopted to iteratively calculate the optimal solution of waste gas, wastewater, residual liquid flow rate and oxygen excess multiple; the optimization objective function is composed of quadratic tracking index, process state variable and control input variable slope control, and the energy dynamic balance rate model, pollutant removal rate and operation boundary are used as equalities or inequalities multi-constraints; the optimization problem is described as: Among them, J is the optimization objective function, α k , β k , γ i Represent the diagonal elements of the coefficient matrix of the corresponding indicator item; x represents the process state variable, represents the measured value of the state variable predicted by the model at time step k; x k represents the actual value of the system process state variable when the time step is k, x * represents the optimal system state equilibrium point; u represents the operating variable, u ij Input variable value for the jth step of the ith control, u i,j-1 Input variable value for the i-th control j-1 step, dimension is Represent the lower and upper limits of the feasible domain of the operating variable, Δt CM Indicates the control domain period of each copy, Represent the lower and upper limits of the control time respectively, Represent the lower and upper limits of the feasible domain of the state variables, respectively, and the dimensions are HBR(x) represents the energy balance rate; y is the output of the prediction model; COD(x * ) is the expected content of organic wastewater in organic pollutant emissions, VOCs(x * ) is the expected content of volatile organic matter in organic pollutant emissions, N CI i and Δt CI i They represent the control steps and step length of the i-th control input variable respectively.

2. The hierarchical optimization control method for the coordinated production and treatment of organic pollutants by reactive distillation according to claim 1 is characterized by: The types of organic pollutants emitted during the reactive distillation production process include volatile organic compound waste gas VOCs, organic wastewater COD and high-concentration distillation residual liquid; the Pearson correlation coefficient method is used to conduct correlation analysis on the emissions of different types of organic pollutants, and a removal rate index is established between subsystems X1 and X2 as an indicator variable for the influencing relationship between subsystems; The pollutant removal rate is defined as the ratio of the total amount of pollutant treatment to the total amount of emissions, as expressed in the following formula: where F ig , F il , F is are the discharge flow rates of reactive distillation waste gas, waste water and residual liquid, v ig , v il , v is are the emission concentrations of waste gas, wastewater and residual liquid, respectively, and F og , F ol , F os is the waste gas, wastewater and residual liquid treatment flow, v og , v ol , v os To control the concentration of waste gas, wastewater and residual liquid.

3. The hierarchical optimization control method for the coordinated production and treatment of organic pollutants by reactive distillation according to claim 1 is characterized by: In step (3), a large amount of organic pollutants are generated during the reaction distillation production process, including volatile organic waste gas, organic wastewater and high-concentration distillation residual liquid. First, the reaction distillation production process is identified online through the reaction distillation production condition environment detection operator, which is specifically divided into normal operation, raw material fluctuation, load adjustment, excessive separation, accident occurrence, and unknown abnormal conditions. The factors affecting pollutant emissions under different conditions are analyzed and the corresponding operating status time series data are collected to construct a sample set. A long short-term memory recurrent neural network prediction model is established according to the working condition to perform real-time prediction of the molar flow rate of organic pollutant emissions.

4. The hierarchical optimization control method for the production and treatment of organic pollutants by reactive distillation according to claim 1 is characterized by: In step (4), the improved non-dominated genetic algorithm is first used to solve the steady-state multi-objective optimization problem. When the operating conditions change, the reference point prediction is performed based on the operating environment detection operator, individual association rules and pollutant emission prediction model, and the multi-objective dynamic optimization algorithm is used to optimize and calculate the update of the optimal operation trajectory of the two subsystems. The specific process of using the PDMOP algorithm to solve the optimization problem includes: 1) An improved non-dominated genetic algorithm is used to perform steady-state optimization on the production-control collaborative optimization target. When the environmental detection operator determines that a change in the working condition occurs, it starts to switch to a dynamic optimization mode based on reference point prediction; 2) If it is in an unknown abnormal working condition, use linear regression or neural network method to establish the required reference point, combine individual association rules, predict the movement of the Pareto solution set containing time series information, and make Gaussian perturbations near each individual to increase the diversity of the predicted population; 3) If it is under known abnormal conditions, the movement of the Pareto solution set is predicted based on the pollutant emission prediction model information, and the historical information about the optimization process is stored and updated and iterated; 4) The predicted population is used as the initial population in the new environment, and it is continuously iterated until the hypervolume index meets the dynamic optimization algorithm index, thereby accelerating the convergence speed while maintaining the diversity of the population.

5. The hierarchical optimization control method for the coordinated production and treatment of organic pollutants by reactive distillation according to claim 1 is characterized by: The optimal operation trajectory of subsystem X1 and subsystem X2 is obtained according to multi-objective dynamic optimization calculation, and the two subsystems operate their own control systems independently; Subsystem X1 uses multiple PID control loops at the bottom layer to realize fast tracking control of the reactor, reaction distillation tower, and recovery tower units. Similarly, the bottom layer control loop of subsystem X2 uses multiple PID control loops to realize fast tracking control of wastewater residual liquid feed, air feed, and reactor temperature. Each bottom layer controller feeds back the predicted state of the next moment and the coupling relationship between each other to the coordinator, and transmits the actual state parameter information to the iterative prediction model for model correction, and passes the correction information to the coordinator for unified correction, so as to establish a production and treatment collaborative hierarchical optimization control system.

Images