Self-adaptive storage tank concentration quantification tracking control method for continuous stirring reaction kettle system

Abstract

The invention relates to the field of automatic control, in particular to a self-adaptive storage tank concentration quantitative tracking control method and device for a continuous stirring reaction kettle system. Firstly, a mechanism model of a continuous stirring reaction kettle system is established; converting the mechanism model into a state constrained non-strict feedback nonlinear system model; establishing a neural network system to carry out approximation on a nonlinear term and a coupling term in the state-constrained non-strict feedback nonlinear system model, and designing a virtual controller in combination with an adaptive adjustment algorithm; based on the virtual controller, designing an actual neural network adaptive quantization controller; based on the actual neural network, a controller is adaptively quantized, and a control signal is obtained; the control signal acts on the continuous stirring reaction kettle system, the expected storage tank concentration is tracked through the storage tank concentration output by the continuous stirring reaction kettle system, and the control target is achieved. According to the invention, good performance and stability can be provided during the operation period of the continuous stirring reaction kettle system.

Description

Technical Field

[0001] The invention relates to the field of automatic control technology, and in particular to a method and a device for self-adapting storage tank concentration quantitative tracking control of a continuous stirring reactor system. Background Art

[0002] Among the many chemical reactors, they can be divided into kettle type, tower type, fluidized bed type and other types according to the reactor structure. Among them, the continuous stirred reactor can realize flexible adjustment of operating temperature, concentration and residence time, can handle Newtonian and non-Newtonian fluids, and has flexible operation mode, which can be operated indirectly or continuously. It is especially suitable for occasions that require small batch and multi-variety production. The continuous stirred reactor has the characteristics of wide controllable operating conditions, adjustable volume, strong adaptability, and can evenly stir reactants in various forms. The efficiency is relatively ideal. It has a wide range of applications in chemical production, especially in the production of three major synthetic materials. It occupies an important position, accounting for about 90% of the total number of reactors. It is also widely used in other industries such as dyes, pharmaceuticals, pesticides, etc., and has great academic research value and room for improvement.

[0003] The continuous stirred reactor can be used in a variety of different industrial processes due to its adaptability. It can be operated at a lower reaction rate, which helps to eliminate hot spots and achieve isothermal operation, which is especially important for reactions with high activation energy or strong exothermic reactions. One of the most significant advantages of the continuous stirred reactor is the easy control of temperature, making it an important and flexible reaction equipment in chemical production.

[0004] In the control of continuous stirred tank reactor systems, the application of adaptive control technology is aimed at dealing with system dynamic changes and parameter uncertainties. Adaptive control can cope with system complexity and nonlinear characteristics by adjusting the parameters of the controller in real time to adapt it to the changing system characteristics, ensuring the stability and performance of the system under different operating conditions, thereby improving the system robustness and performance.

[0005] State constraint control is necessary in a continuous stirred tank reactor system. State constraints usually refer to constraints or restrictions on the system's operating state, such as temperature constraints, concentration constraints, pH value constraints, etc. Since a continuous stirred tank reactor system is usually subject to various state constraints, if these constraints are not effectively handled, especially concentration constraints and temperature constraints, the system and equipment may exceed the safe operating range, affecting product quality and reaction efficiency, thereby increasing the risk of accidents.

[0006] Quantitative control focuses on the precise monitoring and measurement of the reaction process in a continuous stirred reactor system. If the quantitative control in a continuous stirred reactor system is not effective enough, that is, if the operating variables in the system such as temperature and concentration cannot be accurately controlled, the system cannot respond to changes accurately, affecting the efficiency and stability of the reaction process, and thus causing fluctuations in product quality. Without effective quantitative control, it means that the energy efficiency of the system cannot be optimized to the greatest extent, and even the safety of the system may be reduced.

[0007] During the chemical reaction process, the continuous stirred reactor control system needs to continuously consume energy to maintain the stirring process, especially for large reactors, the energy consumption is relatively high. Combining the existing continuous stirred reactor control system with quantitative control technology can significantly improve the performance of the entire continuous stirred reactor, reduce the energy consumption during heating, cooling and stirring, improve the energy efficiency of the system, and optimize the reaction conditions and operation strategies under the premise of ensuring product quality, improve production efficiency, and enhance the robustness and adaptability of the system. Quantization can be regarded as a mapping process that converts continuous signals such as temperature signals into discrete finite value signal sets.

[0008] Through advanced quantitative control technology, noise and jitter in the input signal can be effectively suppressed, signal transmission frequency can be reduced, and the effective utilization rate of network resources can be improved. Information on key system parameters can be obtained in real time, thereby optimizing reaction conditions and production processes, improving control strategies, and enabling the control system to respond more quickly to input changes such as feed concentration, flow rate, temperature, etc., thereby improving the efficiency and stability of the reaction process, maintaining the reaction temperature within a predetermined range, ensuring a stable reaction rate, avoiding runaway or inefficiency caused by overheating or overcooling, and maintaining stable operation of the continuous stirred reactor, thereby improving economic benefits and reducing environmental and safety risks.

[0009] However, in the existing control methods for continuous stirred tank reactor systems, few consider combining quantitative control to optimize the entire control strategy, resulting in low stability and reliability of the existing continuous stirred tank reactor systems. Therefore, in the existing control methods for continuous stirred tank reactor systems, effective state constraint control and quantitative control issues still need to be solved. Summary of the invention

[0010] To this end, the technical problem to be solved by the present invention is to overcome the problem in the prior art that state constraint control and quantization control are not effective enough, thereby affecting the performance of the system and the controller.

[0011] In order to solve the above technical problems, the present invention provides a method for quantifying and tracking the concentration of a continuous stirred tank reactor system, which comprises the following steps:

[0012] Establish the mechanism model of continuous stirred tank reactor system;

[0013] Converting the mechanism model into a non-strict feedback nonlinear system model subject to state constraints;

[0014] Establish a neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and design a virtual controller in combination with an adaptive adjustment algorithm; design an actual neural network adaptive quantization controller based on the virtual controller;

[0015] Based on the actual neural network adaptive quantization controller, a control signal is obtained;

[0016] A control signal is applied to the continuous stirred tank reactor system, and the desired tank concentration is tracked by the tank concentration output by the continuous stirred tank reactor system, and the control signal is iteratively updated until the preset running time ends.

[0017] Preferably, the mechanism model of the continuous stirred tank reactor system is:

[0018]

[0019] in And Q a Indicates the tank concentration, H a represents the tank temperature, q c represents the coolant flow rate, q represents the process flow rate, Q a0 represents the component concentration, H f Indicates the feed temperature, H cf represents the inlet coolant temperature, V represents the tank volume, h a represents the heat transfer coefficient, a 0 represents the pre-exponential factor, represents the activation energy, -ΔH c represents the heat of reaction, ρ 1 and ρ c Indicates the liquid density, C P and C Pc Represents heat capacity.

[0020] Preferably, the state-constrained non-strict feedback nonlinear system model is described as:

[0021]

[0022]

[0023] Where y represents the output of the continuous stirred tank reactor system, 1 =Q a represents the tank concentration, χ 2 =H a Indicates the tank temperature, Represents χ1 The derivative of Represents χ 2 The derivative of q(u) represents the hysteresis quantizer and serves as the input of the non-strict feedback nonlinear system model subject to state constraints, u = q c As input to the hysteresis quantizer q(u), the process flow q = 100 l / min and the component concentration Q a0 =1mol / l, feed temperature H f =350K, inlet coolant temperature H cf =350K, oil tank capacity V = 100l, heat transfer coefficient h a =7×10 5 J / min·K, pre-exponential factor a 0 =7.2×10 10 min -1 ,activation energy Heat of reaction - ΔH c =2×10 4 cal / mol, liquid density ρ 1 =ρ c =1×10 3 g / l, heat capacity C P =C Pc =1cal / g·K,u i =δ 1-i u min , 0<δ<1 represents the measure of quantization density, q(u) in the set in, u min >0 indicates the dead zone range of q(u), and sign() indicates the sign function.

[0024] Preferably, the neural network system is established as follows:

[0025] For any continuous function P 1 (S 1 ):R 5 →R, any given compact set and any positive constant There exists an ideal constant weight vector and a radial basis function vector So that:

[0026]

[0027] Established, of which l>1 is the number of nodes in the neural network, is defined as a Gaussian function of the form:

[0028]

[0029] Among them, μ 1j and σ 1j They are Gaussian functions The width and center of the approximation error ε 1 (S 1 )satisfy ||S 1 -μ 1j || indicates S 1 -μ 1j The Euclidean norm of the optimal weight vector The definition is as follows:

[0030]

[0031] Preferably, the design of the virtual controller and the actual neural network adaptive quantization controller comprises the following steps:

[0032] Step 1: Define s 1 Define s as the tank concentration state after conversion by nonlinear mapping technology in the non-strict feedback nonlinear system model subject to state constraints. 2 The expected tank concentration of the state-constrained non-strict feedback non-linear system model is selected as the reference signal y d ;

[0033] Step 2: Set the desired tank concentration y d The expected tank concentration is obtained by nonlinear mapping technology. The desired tank concentration after conversion and the tank concentration state s 1 The difference between the two is obtained by 1 , and perform differential solutions to the dynamic equations;

[0034] Step 3: Use the neural network system to approximate the nonlinear terms and coupling terms in the non-strict feedback nonlinear system model subject to state constraints, and convert the tank concentration state s 1 , the expected tank concentration after conversion and its first-order derivative As the input of the neural network system, the output is the neural network system corresponding to the tank concentration dynamic subsystem and contains the corresponding accuracy level error ε 1 (S 1 );

[0035] Step 4: Use the two signals output by the neural network system and Establish the neural network system weights corresponding to the tank concentration dynamic subsystem Virtual adaptive adjustment algorithm in is λ 1 The estimated value of yes The derivative of

[0036] Step 5: Based on the virtual adaptive adjustment algorithm Combined with the given corresponding positive parameter κ 1 、a 1 , Υ 1 , σ 1 , design the virtual controller α of the tank concentration dynamic subsystem 1 ;

[0037] Step 6: According to the virtual controller α 1 and virtual adaptive adjustment algorithm The tank temperature state s in the non-strict feedback nonlinear system model with state constraints 2 Processing, get the virtual controller α of the tank temperature dynamic subsystem 1 The corresponding control error signal e 2 , input the neural network system corresponding to the dynamic subsystem of the tank temperature state, and establish the weights of the neural network system The actual adaptive adjustment algorithm in is λ 2 The estimated value of yes The derivative of; combined with the corresponding positive parameter κ 2 、a 2 Γ, Υ 2 , σ 2 , design a practical neural network adaptive quantized controller u for the tank temperature state dynamic subsystem.

[0038] Preferably, the tank concentration state s 1 and tank temperature status s 2 Described as:

[0039]

[0040]

[0041]

[0042] χ i represents the state of each order subsystem in the state-constrained non-strict feedback nonlinear system model before the nonlinear mapping technology conversion, χ 1represents the tank concentration before nonlinear mapping technology conversion, χ 2 represents the tank temperature before nonlinear mapping technology conversion; i is constrained to be within the interval (-k 1 (t),k 2 (t)) on, k 1 (t) and k 2 (t) is a known positive function, 0<K 1 <k 1 (t), 0<K 2 <k 2 (t), is a known positive constant, It is k 1 The derivative of (t), It is k 2 The derivative of (t).

[0043] Preferably, the desired tank concentration after conversion is Described as:

[0044]

[0045]

[0046]

[0047] Among them, ω 1 represents the desired tracking trajectory after conversion by nonlinear mapping technology, where Preferably, the virtual adaptive adjustment algorithm and virtual controller α 1 , actual adaptive adjustment algorithm And the actual neural network adaptive quantization controller u is as follows:

[0048]

[0049]

[0050] in X 1 =[s 1 ,ω 1 ] T , And a 1 , Υ 1 , σ 1 and κ 1 It indicates that the design is a positive constant;

[0051]

[0052]

[0053]

[0054] where e 2 Represents the error of the first-order filter output of the tank concentration and tank temperature dynamic subsystem; α 2 Represents the controller intermediate variable, Υ 2 、a 2 , Γ, σ 2 and κ 2 It represents a designed positive constant, tanh() represents a hyperbolic tangent function, X 2 =[s 1 ,s 2 ,ω 2 ] T .

[0055] Preferably, applying the control signal to the continuous stirred tank reactor system, tracking the expected tank concentration through the tank concentration output by the continuous stirred tank reactor system, and iteratively updating the control signal until the preset running time ends comprises:

[0056] Applying a control signal to the continuous stirred reactor system to determine whether a preset operating time has ended;

[0057] If the preset running time has not ended, the tank concentration output by the continuous stirred tank reactor system is tracked and monitored during the running time, and the process returns to step 1 to continue the cycle, and the control signal is iteratively updated so that the error between the tank concentration output by the continuous stirred tank reactor system and the expected tank concentration remains around 0 until the preset running time ends, that is, the control target is achieved;

[0058] If the preset running time is over, the control of the reaction process is stopped.

[0059] Preferably, a continuous stirred tank reactor system adaptive storage tank concentration quantitative tracking control device comprises:

[0060] A model building module, wherein the model building module is used to establish a mechanism model of a continuous stirred tank reactor system;

[0061] A model conversion module, the model conversion module is used to convert the mechanism model into a state-constrained non-strict feedback nonlinear system model;

[0062] A quantized controller building module, wherein the quantized controller building module is used to establish a neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and design a virtual controller in combination with an adaptive adjustment algorithm; based on the virtual controller, design an actual neural network adaptive quantized controller;

[0063] A control signal calculation module, the control signal calculation module is used to obtain a control signal based on the actual neural network adaptive quantization controller;

[0064] A tracking control module is used to apply a control signal to the continuous stirred tank reactor system, and track the expected tank concentration through the tank concentration output by the continuous stirred tank reactor system, and iteratively update the control signal until the preset running time ends.

[0065] The above technical solution of the present invention has the following beneficial effects compared with the prior art:

[0066] The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system described in the present invention is based on a state-constrained non-strict feedback nonlinear system model after the continuous stirred tank reactor system is converted, a neural network system is established, and nonlinear mapping technology is used to effectively solve the problem of state constraints in the continuous stirred tank reactor system. At the same time, an adaptive adjustment algorithm and an actual neural network adaptive quantitative controller are designed, so that the continuous stirred tank reactor system can overcome disturbances and achieve self-regulation when subjected to asymmetric time-varying state constraints and coupling terms, and can still maintain operation on the control target, thereby improving system performance and work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0068] Figure 1 It is a schematic diagram of the principle flow of the present invention;

[0069] Figure 2 It is a schematic diagram of the structure of the continuous stirred reactor system of the present invention;

[0070] Figure 3 is the mapping relationship of q(u) of the hysteresis quantizer of the continuous stirred reactor system of the present invention when u>0;

[0071] Figure 4 It is a tracking effect diagram between the output tank concentration and the expected tank concentration of the continuous stirred reactor system of the present invention;

[0072] Figure 5This is a storage tank temperature effect diagram of the storage tank temperature dynamic subsystem of the continuous stirred reactor system of the present invention;

[0073] Figure 6 It is a coolant flow rate and quantified signal effect diagram of the continuous stirred reactor system of the present invention;

[0074] Figure 7 The continuous stirred reactor system adjustment parameter of the present invention is and Adaptive change curve graph. DETAILED DESCRIPTION

[0075] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0076] Example 1

[0077] Reference Figure 1 As shown, it is a schematic diagram of the principle flow of the present invention, which is a method for quantitative tracking and control of concentration of a continuous stirred tank system adaptive storage tank, comprising the following steps:

[0078] S1: Establish the mechanism model of continuous stirred tank reactor system;

[0079] Reference Figure 2 As shown, it is a schematic diagram of the structure of the continuous stirred reactor system of the present invention, and the mechanism model of the continuous stirred reactor system is as follows:

[0080]

[0081] in And Q a Indicates the tank concentration, H a represents the tank temperature, q c represents the coolant flow rate, q represents the process flow rate, Q a0 represents the component concentration, H f Indicates the feed temperature, H cf represents the inlet coolant temperature, V represents the tank volume, h a represents the heat transfer coefficient, a 0 represents the pre-exponential factor, represents the activation energy, -ΔH c represents the heat of reaction, ρ 1 and ρ c Indicates the liquid density, C P and C Pc Represents heat capacity.

[0082] S2: converting the mechanism model into a non-strict feedback nonlinear system model subject to state constraints;

[0083] The state-constrained non-strict feedback nonlinear system model is as follows:

[0084]

[0085]

[0086] where χ 1 =Q a represents the tank concentration, χ 2 =H a Indicates the tank temperature, Represents χ 1 The derivative of Represents χ 2 The derivative of q(u) represents the hysteresis quantizer and serves as the input of the non-strict feedback nonlinear system model subject to state constraints, u = q c As input to the hysteresis quantizer q(u), the process flow q = 100 l / min and the component concentration Q a0 =1mol / l, feed temperature H f =350K, inlet coolant temperature H cf =350K, oil tank capacity V = 100l, heat transfer coefficient h a =7×10 5 J / min·K, pre-exponential factor a 0 =7.2×10 10 min -1 ,activation energy Heat of reaction - ΔH c =2×10 4 cal / mol, liquid density ρ 1 =ρ c =1×10 3 g / l, heat capacity C P =C Pc =1cal / g·K,u i =δ 1-i u min , 0<δ<1 represents the measure of quantization density, q(u) in the set in, u min >0 indicates the dead zone range of q(u), sign() indicates the sign function; refer to Figure 3 As shown, it is the mapping relationship of q(u) of the hysteresis quantizer of the continuous stirred reactor system when u>0.

[0087] S3: Establish a neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and design a virtual controller in combination with an adaptive adjustment algorithm; design an actual neural network adaptive quantization controller based on the virtual controller;

[0088] The neural network system is established as follows: For any continuous function P 1 (S 1 ):R 5 →R, any given compact set and any positive constant There exists an ideal constant weight vector and a radial basis function vector So that:

[0089]

[0090] Established, of which l>1 is the number of nodes in the neural network, is defined as a Gaussian function of the form:

[0091]

[0092] Among them, μ 1j and σ 1j They are Gaussian functions The width and center of the approximation error ε 1 (S 1 )satisfy ||S 1 -μ 1j || indicates S 1 -μ 1j The Euclidean norm of the optimal weight vector The definition is as follows:

[0093]

[0094] Definitions 1 Define s as the tank concentration state after conversion by nonlinear mapping technology in the non-strict feedback nonlinear system model subject to state constraints. 2 The expected tank concentration of the state-constrained non-strict feedback non-linear system model is selected as the reference signal y d ;

[0095] Among them, the tank concentration state s 1 and tank temperature status s 2 Described as:

[0096]

[0097]

[0098]

[0099] χ i represents the state of each order subsystem in the state-constrained non-strict feedback nonlinear system model before the nonlinear mapping technology conversion, χ 1 represents the tank concentration before nonlinear mapping technology conversion, χ 2 represents the tank temperature before nonlinear mapping technology conversion; i is constrained to be within the interval (-k 1 (t),k 2 (t)) on, k 1 (t) and k 2 (t) is a known positive function, 0<K 1 <k 1 (t), 0<K 2 <k 2 (t)0<K 2 <k 2 (t), is a known positive constant, It is k 1 The derivative of (t), It is k 2 The derivative of (t);

[0100] The desired tank concentration y d The expected tank concentration is obtained by nonlinear mapping technology. The desired tank concentration after conversion and the tank concentration state s 1 The difference between the two is obtained by 1 , and perform differential solutions to the dynamic equations;

[0101] Expected tank concentration after conversion Described as follows, ω 1 represents the desired tracking trajectory after conversion by nonlinear mapping technology, where

[0102]

[0103]

[0104]

[0105] The nonlinear terms and coupling terms in the non-strict feedback nonlinear system model with state constraints are approximated by the neural network system, and the tank concentration state s 1 , the expected tank concentration after conversion and its first-order derivative As the input of the neural network system, the output is the neural network system corresponding to the tank concentration dynamic subsystem and contains the corresponding accuracy level error ε 1 (S 1 );

[0106] Due to the particularity of non-strict feedback systems, the unmodeled dynamic nonlinear terms and coupling terms include all states. X 1 =[s 1 ,ω 1 ] T ,ω 1 represents the desired tracking trajectory after nonlinear mapping transformation, Represents ω 1 The derivative of 2 represents the output of the first-order filter in the tank concentration dynamic subsystem, S 1j Indicates S 1 The jth element in X 1j Represents X 1 Based on the inequality relationship, the properties of the Gaussian function can be applied as follows, where μ 1j and σ 1j Gaussian basis functions Width and center:

[0107]

[0108] Using two signals output by the neural network system and Establish the neural network system weights corresponding to the tank concentration dynamic subsystem Virtual adaptive adjustment algorithm in is λ 1 The estimated value of yes Derivative of; Based on the virtual adaptive adjustment algorithm Combined with the given corresponding positive parameter κ 1 、a 1 , Υ 1 , σ 1 , design the virtual controller α of the tank concentration dynamic subsystem 1 ;

[0109] Virtual Adaptive Adjustment Algorithm and virtual controller α 1 As follows, X 1 =[s 1 ,ω 1 ] T , And a 1 , Υ 1 , σ 1 and κ 1 Represents a design positive constant:

[0110]

[0111]

[0112] According to the virtual controller α 1 and adaptive adjustment algorithm The tank temperature state s in the non-strict feedback nonlinear system model with state constraints 2 Processing, get the virtual controller α of the tank temperature dynamic subsystem 1 The corresponding control error signal e 2 , input the neural network system corresponding to the dynamic subsystem of the tank temperature state, and output the neural network system of the last-order subsystem and contains the corresponding accuracy level error ε 2 (S 2 );

[0113] Because of the particularity of non-strict feedback systems, the strong nonlinear terms and coupling terms include all states. X 2 =[s 1 ,s 2 ,ω 2 ] T ,ω 2 represents the output of the first-order filter in the tank concentration dynamics subsystem, Represents ω 2 The derivative of S 2j Indicates S 2 The jth element in X 2j Represents X 2 Based on the inequality relationship, the properties of the Gaussian function can be applied as follows, where μ 2j and σ 2j Gaussian basis functions Width and center:

[0114]

[0115] Using the signal output by the neural network system Building a practical adaptive regulation algorithm in is λ 2 The estimated value of yes The derivative of; combined with the corresponding positive parameter κ 2 、a 2 Γ, Υ 2 , σ 2 , design the practical neural network adaptive quantized controller u for the dynamic subsystem of the tank temperature state;

[0116] Practical adaptive adjustment algorithm And the actual neural network adaptive quantization controller u is as follows, where e 2 Represents the error of the first-order filter output of the tank concentration and tank temperature dynamic subsystem; α 2 Represents the controller intermediate variable, Υ 2 、a 2 , Γ, σ 2 and κ 2 It represents a designed positive constant, tanh() represents a hyperbolic tangent function, X 2 =[s 1 ,s 2 ,ω 2 ] T . :

[0117]

[0118]

[0119]

[0120] S4: obtaining a control signal based on the actual neural network adaptive quantization controller u;

[0121] S5: applying a control signal to the continuous stirred tank reactor system, tracking the expected tank concentration through the tank concentration output by the continuous stirred tank reactor system, and iteratively updating the control signal until the preset running time ends;

[0122] Applying a control signal to the continuous stirred reactor system to determine whether a preset operating time has ended;

[0123] If the preset running time has not ended, the tank concentration output by the continuous stirred tank reactor system is tracked and monitored during the running time, and the process returns to step 1 to continue the cycle, and the control signal is iteratively updated so that the error between the tank concentration output by the continuous stirred tank reactor system and the expected tank concentration remains around 0 until the preset running time ends, that is, the control target is achieved;

[0124] If the preset operation time is over, the control of the reaction process is stopped, wherein the preset operation time is set based on the time required for the actual reaction of the reactants and the actual production needs.

[0125] In order to verify the effectiveness of the method proposed in the present invention, a simulation experiment is carried out on this embodiment, and the simulation results are as follows: Figures 4 to 7 As shown. Figure 4 It can be obtained that the output satisfies the asymmetric time-varying constraints in a probabilistic sense and can achieve the control goal well. The solid line represents the actual trajectory and the dotted line represents the tracking trajectory. Figure 5 The tank temperature of the dynamic subsystem representing the continuous stirred reactor system also satisfies the asymmetric time-varying state constraints in the probabilistic sense. In addition, Figure 6 It shows that the control signal coolant flow rate u is bounded; Figure 6 Shows the adaptive tuning algorithm and From the above simulation experimental results, it can be concluded that the adaptive quantitative control scheme proposed in the present invention has been effectively simulated on the most common chemical kinetics nonlinear oscillation model - the continuous stirred reactor system mechanism model, and can achieve the expected goal well, which means that this method also has good prospects in specific practical applications.

[0126] Example 2

[0127] A continuous stirred reactor system adaptive storage tank concentration quantitative tracking control device, comprising:

[0128] A model building module, wherein the model building module is used to establish a mechanism model of a continuous stirred tank reactor system;

[0129] A model conversion module, the model conversion module is used to convert the mechanism model into a state-constrained non-strict feedback nonlinear system model;

[0130] A quantized controller building module, wherein the quantized controller building module is used to establish a neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and design a virtual controller in combination with an adaptive adjustment algorithm; based on the virtual controller, design an actual neural network adaptive quantized controller;

[0131] A control signal calculation module, the control signal calculation module is used to obtain a control signal based on the actual neural network adaptive quantization controller;

[0132] A tracking control module is used to apply a control signal to the continuous stirred tank reactor system, and track the expected tank concentration through the tank concentration output by the continuous stirred tank reactor system, and iteratively update the control signal until the preset running time ends.

[0133] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0134] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0135] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0136] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0137] Obviously, the above embodiments are merely examples for the purpose of clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A method for quantitative tracking and control of concentration of an adaptive storage tank in a continuous stirred tank reactor system, characterized in that: The following steps are involved: Establish the mechanism model of continuous stirred tank reactor system; Converting the mechanism model into a non-strict feedback nonlinear system model subject to state constraints; Establishing a neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and designing a virtual controller in combination with an adaptive adjustment algorithm; Based on the virtual controller, designing a practical neural network adaptive quantization controller; Based on the actual neural network adaptive quantization controller, a control signal is obtained; A control signal is applied to the continuous stirred tank reactor system, and the desired tank concentration is tracked by the tank concentration output by the continuous stirred tank reactor system, and the control signal is iteratively updated until the preset running time ends.

2. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 1, characterized in that: The mechanism model of the continuous stirred reactor system is: in And Q a Indicates the tank concentration, H a represents the tank temperature, q c represents the coolant flow rate, q represents the process flow rate, Q a0 represents the component concentration, H f Indicates the feed temperature, H cf represents the inlet coolant temperature, V represents the tank volume, h a represents the heat transfer coefficient, a0 represents the pre-exponential factor, represents the activation energy, -ΔH c represents the heat of reaction, ρ1 and ρ c Indicates the liquid density, C P and C Pc Represents heat capacity.

3. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 1, characterized in that: The state-constrained non-strict feedback nonlinear system model is described as: Where y represents the output of the continuous stirred tank reactor system, χ1 = Q a Represents the tank concentration, χ2=H a Indicates the tank temperature, represents the derivative of χ1, represents the derivative of χ2; q(u) represents the hysteresis quantizer and serves as the input of the non-strict feedback nonlinear system model subject to state constraints, u = q c As input to the hysteresis quantizer q(u), the process flow q = 100 l / min and the component concentration Q a0 =1mol / l, feed temperature H f =350K, inlet coolant temperature H cf =350K, oil tank capacity V = 100l, heat transfer coefficient h a =7×10 5 J / min·K, pre-exponential factor a0=7.2×10 10 min -1 ,activation energy Heat of reaction - ΔH c =2×10 4 cal / mol, liquid density ρ1=ρ c =1×10 3 g / l, heat capacity C P =C Pc =1cal / g·K,u i =δ 1-i u min , 0<δ<1 represents the measure of quantization density, q(u) in the set in, u min >0 indicates the dead zone range of q(u), and sign() indicates the sign function.

4. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 1, characterized in that: The neural network system is established as follows: For any continuous function P1(S1):R 5 →R, any given compact set and any positive constant There exists an ideal constant weight vector and a radial basis function vector So that: Established, of which l>1 is the number of nodes in the neural network, is defined as a Gaussian function of the form: Among them, μ 1j and σ 1j They are Gaussian functions The width and center of the approximation error ε1(S1) satisfies ||S1-μ 1j || represents S1-μ 1j The Euclidean norm of the optimal weight vector The definition is as follows:

5. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 1, characterized in that: The design of the virtual controller and the actual neural network adaptive quantization controller includes the following steps: Step 1: Define s1 as the tank concentration state after the nonlinear mapping technology is used to convert the nonlinear system model with state constraints and define s2 as the tank temperature state after the nonlinear mapping technology is used to convert the nonlinear system model with state constraints and select the expected tank concentration of the nonlinear system model with state constraints as the reference signal y d ; Step 2: Set the desired tank concentration y d The expected tank concentration is obtained by nonlinear mapping technology. The desired tank concentration after conversion Subtract the concentration state s1 of the tank to obtain the error e1 between the two, and perform differential solution of the kinetic equation; Step 3: Use the neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and convert the tank concentration state s1 to the desired tank concentration after conversion. and its first-order derivative As the input of the neural network system, the output is the neural network system corresponding to the tank concentration dynamic subsystem And contains the corresponding accuracy level error ε1(S1); Step 4: Use the two signals output by the neural network system and Establish the neural network system weights corresponding to the tank concentration dynamic subsystem Virtual adaptive adjustment algorithm in is the estimated value of λ1, yes The derivative of Step 5: Based on the virtual adaptive adjustment algorithm Combined with the given corresponding positive parameters κ1, a1, γ1, σ1, the virtual controller α1 of the tank concentration dynamic subsystem is designed; Step 6: According to the virtual controller α1 and the virtual adaptive adjustment algorithm The tank temperature state s2 in the state-constrained non-strict feedback nonlinear system model is processed to obtain the control error signal e2 corresponding to the virtual controller α1 of the tank temperature dynamic subsystem, which is input into the neural network system corresponding to the tank temperature state dynamic subsystem to establish the neural network system weights. The actual adaptive adjustment algorithm in is the estimated value of λ2, yes The derivative of; combined with the given corresponding positive parameters κ2, a2, Γ, Υ2, σ2, design the actual neural network adaptive quantized controller u of the tank temperature state dynamic subsystem.

6. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 5, characterized in that: The tank concentration state s1 and the tank temperature state s2 are described as: χ i represents the state of each order subsystem in the non-strict feedback nonlinear system model subject to state constraints before the nonlinear mapping technology conversion, χ1 represents the tank concentration before the nonlinear mapping technology conversion, χ2 represents the tank temperature before the nonlinear mapping technology conversion; χ i is constrained to the interval (-k1(t), k2(t)), k1(t) and k2(t) represent known positive functions, 0<K1<k1(t), 0<K2<k2(t), K1, K2, is a known positive constant, is the derivative of k1(t), is the derivative of k2(t).

7. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 5, characterized in that: Expected tank concentration after conversion Described as: Among them, ω1 represents the expected tracking trajectory after conversion by nonlinear mapping technology, where 8. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 5, characterized in that: Virtual Adaptive Adjustment Algorithm and virtual controller α1, actual adaptive adjustment algorithm And the actual neural network adaptive quantization controller u is as follows: in X1=[s1,ω1] T , And a1, Υ1, σ1 and κ1 represent design positive constants; Where e2 represents the error of the first-order filter output of the dynamic subsystem of the tank concentration and tank temperature; α2 represents the intermediate variable of the controller, γ2, a2, Γ, σ2 and κ2 represent the designed positive constants, tanh() represents the hyperbolic tangent function, X2=[s1,s2,ω2] T .

9. The method for adaptive storage tank concentration quantitative tracking control of a continuous stirred tank reactor system according to claim 5, characterized in that: The step of applying the control signal to the continuous stirred tank reactor system, tracking the expected tank concentration through the tank concentration output by the continuous stirred tank reactor system, and iteratively updating the control signal until the preset running time ends includes: Applying a control signal to the continuous stirred reactor system to determine whether a preset operating time has ended; If the preset running time has not ended, the tank concentration output by the continuous stirred tank reactor system is tracked and monitored during the running time, and the process returns to step 1 to continue the cycle, and the control signal is iteratively updated so that the error between the tank concentration output by the continuous stirred tank reactor system and the expected tank concentration remains around 0 until the preset running time ends, that is, the control target is achieved; If the preset running time is over, the control of the reaction process is stopped.

10. A continuous stirred tank reactor system adaptive storage tank concentration quantitative tracking control device, characterized in that: include: A model building module, wherein the model building module is used to establish a mechanism model of a continuous stirred tank reactor system; A model conversion module, the model conversion module is used to convert the mechanism model into a state-constrained non-strict feedback nonlinear system model; A quantized controller building module, wherein the quantized controller building module is used to establish a neural network system to approximate the nonlinear terms and coupling terms in the state-constrained non-strict feedback nonlinear system model, and design a virtual controller in combination with an adaptive adjustment algorithm; Based on the virtual controller, designing a practical neural network adaptive quantization controller; A control signal calculation module, the control signal calculation module is used to obtain a control signal based on the actual neural network adaptive quantization controller; A tracking control module is used to apply a control signal to the continuous stirred tank reactor system, and track the expected tank concentration through the tank concentration output by the continuous stirred tank reactor system, and iteratively update the control signal until the preset running time ends.

Images