CN101452261A - Polypropylene apparatus grade switching and controlling method - Google Patents

Abstract

The invention relates to a method for switching and controlling polypropylene device mark. The method comprises: establishing a linear subsystem optimal controller according to the systematic linear performance index by establishing a polypropylene mark switching process discrete time state space Hammerstein model; designing state observation through a pole assignment method according to an expected pole of a systematic state observer; establishing a model forecasting controller based on the state observer according to a systematic quadratic performance index; updating initial conditions of the optimal control problem; and then carrying out the rolling optimization calculation of the current forecasting control variable repeatedly till the mark switching process is switched to a target market and stable production. The method has simple design, easy understanding, easy realization of online commissioning, and good practicability. The method can realize the continuous automatic switching of polypropylene marks, particularly marks of different categories on a wide range, thereby greatly shortening the mark switching time, reducing the emission of unqualified products, and improving the production economic benefits and market competitiveness.

Description

A kind of polypropylene apparatus grade switching and controlling method

Technical field

The present invention relates to a kind of polypropylene apparatus grade switching and controlling method.

Background technology

In order to satisfy market to the diversity demand of polypropylene product and the quality requirements of Geng Gao, and for pursuing higher business economic benefit, polypropylene plant need carry out the trade mark continually and switch and produce.While is along with the development of polypropylene industrial and production technology, the scale of producing is increasing, the switching frequency of the trade mark is more and more higher, and each switching always is accompanied by a large amount of transit times and defective material is the generation of waste material, finally influences the production economic benefit and the energy-saving and cost-reducing index of enterprise.Large-scale polypropylene apparatus grade handoff procedure is " essence " non-linear process of complicated multivariate, strong coupling, multiple constraint, its essence is that a kind of product quality (trade mark) is converted into another kind of product quality, be reflected on the polymerization process condition, under the prerequisite that guarantees the polyplant safety in production, switch to the polymerization process condition of another kind of expectation exactly, require handoff procedure to have composite targets such as good stationarity, rapidity and economy simultaneously by a kind of polymerization process condition.Therefore, trade mark blocked operation has become a crucial research task of current polypropylene industrial advanced production technology.The method of the trade mark blocked operation that adopts in industrial practice in the past and the scientific research mainly contains: the empirical method of switching according to the typical trade mark, computer simulation emulation method, based on the optimum changing method of optimisation technique, based on the control method of advanced control strategy etc.In these trade mark changing methods, the empirical method of switching according to the typical trade mark is unfavorable for the blocked operation between the new trade mark, many trades mark only at specific several trade mark blocked operations; Though computer simulation emulation method and can realize in theory that based on the optimum changing method of optimisation technique the optimum trade mark switches all belongs to the off-line blocked operation, can't automatically switch according to the actual realization of the production of the polypropylene plant trade mark; Based on the method for handover control of advanced control strategy, find through retrieval the prior art document, mainly comprise McAuley, 1993; Gobin, 1994; Meziou, 1996; Ozkan, 1998; Ali, 1998; Seki, 2001; Bindlish ﹠amp; Rawlings, 2003; BenAmor, 2004; Feather ﹠amp; Lieberman, 2004; Ali, 2007, usually in conjunction with the polymerization reaction mechanism model, design of Controller is complicated and used relevant speciality knowwhy is more, often causes the online difficulty that puts into operation of controller, and is not easy to be grasped and promote the use of by engineering technical personnel.Therefore, the nearly more than ten years, the challenging hereto important control difficult problem of many scholars in the process control field and engineering specialist has carried out in depth studying in a large number and inquiring into, to satisfy current polypropylene production practices for an urgent demand of regulating and control trade mark handoff procedure effectively, easily.

Summary of the invention

For the deficiency of the design complexity that overcomes existing polypropylene apparatus grade switching and controlling method, the online difficulty that puts into operation, poor practicability, the invention provides a kind of simplicity of design, understand easily, realize onlinely to put into operation, practical polypropylene apparatus grade switching and controlling method.

The technical solution adopted for the present invention to solve the technical problems is:

A kind of polypropylene apparatus grade switching and controlling method, described control method comprises the steps: 1), set up polypropylene trade mark handoff procedure discrete time state-space model, referring to formula (1a) and formula (1b):

$\left\{\begin{array}{c}x(t+1)=A_{1}x\left(t\right)+B_{1}v\left(t\right),v\left(t\right)=g_1(u\left(t\right),t),y\left(t\right)=C_{1}x\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A_1=\left[\begin{array}{cccc}1-T_s/{\mathrm{\&tau;}}_1& 0& 0& 0\\ T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2& 0& 0\\ 0& 0& 1-T_s/{\mathrm{\&tau;}}_1& 0\\ 0& 0& T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2\end{array}\right]\\ {\mathrm{<figure-callout\; id="1"\; label="B"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">B</figure-callout>}}_1=\left[\begin{array}{cccc}{T}_s/{\mathrm{\&tau;}}_1& & & 0\\ & T_s/{\mathrm{\&tau;}}_2& & \\ & & T_s/{\mathrm{\&tau;}}_1& \\ 0& & & T_s/{\mathrm{\&tau;}}_2\end{array}\right],g_1(u,t)=\left[\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ {\mathrm{Et}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\\ {\mathrm{Et}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\end{array}\right]\\ {\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_1=\left[\begin{array}{cccc}0& 1& 0& 0\\ 0& 0& 0& 1\end{array}\right]\end{array}\right.$

(1a)

$\left\{\begin{array}{c}x(t+1)=A_{2}x\left(t\right)+B_{2}v\left(t\right),v\left(t\right)=g_2(u\left(t\right),t),y\left(t\right)=C_{2}x\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A_2=\left[\begin{array}{cccc}1-T_s/{\mathrm{\&tau;}}_1& 0& 0& 0\\ T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2& 0& 0\\ 0& T_s/{\mathrm{\&tau;}}_2& 1-T_s/{\mathrm{\&tau;}}_2-T_s/{\mathrm{\&tau;}}_3& 0\\ 0& 0& 0& 1-T_s/{\mathrm{\&tau;}}_3\end{array}\right],\\ {\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">B</figure-callout>}}_2=\left[\begin{array}{cccc}{T}_s/{\mathrm{\&tau;}}_1& & & 0\\ & T_s/{\mathrm{\&tau;}}_2& & \\ & & T_s/{\mathrm{\&tau;}}_3& \\ 0& & & T_s/{\mathrm{\&tau;}}_3\end{array}\right],g_2(u,t)=\left[\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ \ln ({\mathrm{MI}}_{i,3}({\mathrm{\&theta;}}_3,u_3))\\ {\mathrm{Et}}_{i,3}({\mathrm{\&theta;}}_3,u_3)\end{array}\right]\\ {\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_2=\left[\begin{array}{cccc}0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]\end{array}\right.---\left(1b\right)$

(1b)

Wherein, in the formula (1a), x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), Et _{C, 1}, Et _{C, 2}] ', u=[u ₁, _U2]=[T ₁, C _1hm, C _1mmT ₂, C _2hm, C _2mm] ' and y=[y ₁, y ₂[ln (the MI of] '= _{C, 2}), Et _{C, 2}State variable, input variable and output variable when] ' be respectively switches to the homopolymer or the random copolymers trade mark; In the formula (1b), x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), ln (MI _{C, 3}), Et _{C, 3}] ', u=[u ₁, u ₂, u ₃]=[T ₁, C _1hm, C _1mmT ₂, C _2hm, C _2mmT ₃, C _3hm, C _3mm] ' and y=[y ₁, y ₂[ln (the MI of] '= _{C, 3}), Et _{C, 3}State variable when] ' be respectively switches to the multipolymer trade mark, input variable and output variable; T _sBe system's discrete time; MI _{C, k}And MI _{I, k}, Et _{C, k}And Et _{I, k}Be respectively accumulation melting index and instantaneous melting index, accumulation ethylene contents and the instantaneous ethylene contents of the individual reactor of k (k=1,2,3); T _k, C _Khm, C _Kmm, τ _kBe respectively temperature of reaction, hydrogen density of propylene ratio, ethylene, propylene concentration ratio, the reaction time of the individual reactor of k (k=1,2,3); θ _k(k=1,2,3) are the state-space model identified parameters;

With model (1a) and (1b) the unified formula (2) that is described as:

$\left\{\begin{array}{c}x(t+1)=\mathrm{Ax}\left(t\right)+\mathrm{Bv}\left(t\right),v\left(t\right)=g(u\left(t\right),t)\\ y\left(t\right)=\mathrm{Cx}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\end{array}\right.---\left(2\right)$

2), the steady-state value of calculating formula (2), referring to formula (3):

$v_s=\underset{t\&RightArrow;\&infin;}{\lim }g(u_s,t),{\mathrm{Ax}}_s=-{\mathrm{Bv}}_s---\left(3\right)$

X wherein _s, v _sAnd u _sThe steady-state value of representing system state, intermediate variable and input variable respectively;

3), according to state, intermediate variable and the input variable of formula (3) calibration system, referring to formula (4a):

S _x＝diag{s _x ⁽ⁱ⁾，i＝1，2，3，4}；S _v＝diag{s _v ⁽ⁱ⁾，i＝1，2，3，4}；

S _u1＝diag{s _u1 ⁽ⁱ⁾，i＝1，2，3}；S _u2＝diag{s _u2 ⁽ⁱ⁾，i＝1，2，3}； (4a)

S _u3＝diag{s _u3 ⁽ⁱ⁾，i＝1，2，3}

Wherein calibration coefficient is (4b):

s _x ⁽ⁱ⁾＝(x _s，m ⁽ⁱ⁾) ^-2，s _v ⁽ⁱ⁾＝(v _s，m ⁽ⁱ⁾) ^-2，i＝1，2，3，4；

${s_{u1}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}=0\mathrm{<figure-callout\; id="1"\; label="or\; u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">or</figure-callout>}{u_s}^{}{{1,o}_{}}^{\left(i\right)}={{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}\\ {({{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}-{{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)})}^{-2},\mathrm{else}\end{array}\right.,i=\mathrm{1,2,3};$

${s_{u2}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}=0\mathrm{<figure-callout\; id="2"\; label="or\; u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">or</figure-callout>}{u_s}^{}{{2,o}_{}}^{\left(i\right)}={{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}\\ {({{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}-{{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)})}^{-2},\mathrm{else}\end{array},\right.i=\mathrm{1,2,3};---\left(4b\right)$

${s_{u3}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}=0\mathrm{<figure-callout\; id="3"\; label="or\; u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">or</figure-callout>}{u_s}^{}{{3,o}_{}}^{\left(i\right)}={{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}\\ {({{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}-{{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)})}^{-2},\mathrm{else}\end{array}\right.,i=\mathrm{1,2,3};$

Wherein, " _{S, o} ⁽ⁱ⁾" the current production trade mark correspondence of expression i state, intermediate variable or import the steady-state value of component; " _{S, m} ⁽ⁱ⁾" steady-state value of expression target trade mark correspondence;

4), according to performance index of system, referring to formula (5):

${\mathrm{<figure-callout\; id="1"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_1={\mathrm{\&Sigma;}}_{t=0}^{\&infin;}\{(x\left(t\right)-x_s)\&prime;{\mathrm{QS}}_x(x\left(t\right)-x_s)+(v\left(t\right)-v_s)\&prime;{\mathrm{RS}}_v(v\left(t\right)-v_s)\}---\left(5\right)$

Set up formula (2) neutral line subsystem optimal controller, referring to formula (6):

v(t)＝-(RS _v+B′PB) ^-1B′PA(x(t)-x _s)+v _s (6)

Wherein matrix P is a matrix equation, referring to formula (7):

A′PA-P+A′PB(RS _v+B′PB) ^-1B′PA+QS _x＝0 (7)

The symmetric positive definite dematrix; Wherein Q 〉=0 and R〉0 be respectively the weighting matrix of state and intermediate variable;

5), according to the expectation limit of system state observer, utilization POLE PLACEMENT USING method design point observer, referring to formula (8):

$\hat{x}(t+1)=(A-\mathrm{LC})\hat{x}\left(t\right)+\mathrm{Bg}\left(t\right)+\mathrm{Ly}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;---\left(8\right)$

Wherein L is the observer gain matrix;

6), the secondary system performance index, with reference to formula (9):

${\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_2={\mathrm{\&Sigma;}}_{i=t}^{t+T_p-1}\{(g\left(i\right)-v\left(i\right))\&prime;(g\left(i\right)-v\left(i\right))+{\mathrm{\&Sigma;}}_{j=1}^3(u_j\left(i\right)-u_{s,j})\&prime;W_j{S}_{\mathrm{uj}}(u_j\left(i\right)-u_{s,j})\}---\left(9\right)$

(9)

Foundation is based on the model predictive controller of state observer, referring to formula (10):

$u{(t;\hat{x}\left(t\right),T_p)}^{*}=\underset{u}{\min }{\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_2$

s.t.x(i+1)＝Ax(i)+Bg(u(i)，i)，

v(i)＝-(RS _v+B′PB) ^-1B′PA(x(i)-x _s)+v _s (10)

x(i)∈Γ _x，v(i)∈Γ _v，u(i)∈Γ _u

$x\left(t\right)=\hat{x}\left(t\right),$ i＝t，…，t+T _p

Wherein, set Γ _x, Γ _vAnd Γ _uThe constraint conditions such as state, intermediate variable and input of expression handoff procedure;

(t) the current t of expression state observation value constantly; W _j0 (j=1,2,3) are the weighting matrix of input variable;

In the limited step-length optimal control in dynamic of line computation problem, i.e. formula (10), and, obtain Model Predictive Control amount based on state observer according to the rolling optimization principle, with reference to formula (11):

$u^{\mathrm{mpc}}\left(t\right)=u{(t;\hat{x}\left(t\right))}^{*},t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\left(11\right)$

Controller passes through state observer recording geometry state in each sampling instant, and upgrade the starting condition of optimal control problem with this observer state, rolling optimization calculates the PREDICTIVE CONTROL amount of current time then, go round and begin again, till trade mark handoff procedure carries out the transition to the target trade mark and steady production.

Technical conceive of the present invention is: the Spheripol propylene polymerization device with extensive employing at present is object (technological process of production as shown in Figure 1), bonding state observer technology and constraint Hammerstein model nonlinear prediction control technology, by the switching of control polypropylene melt index and these two Key Quality Indicator of ethylene contents, realize the polypropylene apparatus grade blocked operation.The advantage of method for designing of the present invention is easy understanding, easy to use, simultaneously can the various status informations of online forecasting handoff procedure.The controller of the present invention's design can realize that cost-effectively the trade mark switches the automatic control of (being that the non-similar trade mark switches especially) process according to the requirement of trade mark switching.The present invention and Seki, 2001; Bindlish ﹠amp; Rawlings, the difference of Model Predictive Control changing method is that these Model Predictive Control changing methods are based on the linearizing method for designing of the non-linear mechanism model pointwise of polymerization kinetics in 2003.

Several sections shown in the accompanying drawing 2: trade mark handoff procedure, expected performance index, state observer, loop control unit, fallout predictor, switch controller are the important module of forming the technical solution used in the present invention.Wherein trade mark handoff procedure is obtained by the identification of polypropylene process units, and the requirement according to on-the-spot blocked operation simultaneously obtains multinomial control expected performance indexs such as switching controls accuracy, rapidity, stability and economy.Parameter of these identification objects self and performance demands are taken all factors into consideration in switch controller design link.

The present invention can directly move enforcement on existing industrial control computer, specifically see accompanying drawing 3.This method application process can roughly be divided into four-stage:

1, produces trade mark management, promptly in the configuration interface, confirm the operating conditions and the Key Quality Indicator such as melting index and ethylene contents of the current production trade mark and the target production trade mark.The upper ledge of surface chart is the operating conditions and the physical index of the current production trade mark, marks with redness; The lower frame of surface chart is the operating conditions and the physical index of the target trade mark, can stir the target trade mark that inquiry will be switched by clicking " last one " and " next one ".Behind the target trade mark that affirmation will be switched, click the operation that " switching to the target trade mark " beginning trade mark switches, promptly switch to the configuration of the target trade mark, by industrial computer data are sent in the dynamic data base simultaneously and preserved from the trade mark configuration (with red mark) of current production.

2, the generation of trade mark switching model and constraint condition.In the configuration interface, system generates handoff procedure mathematical model and constraint condition automatically according to the current trade mark and the target trade mark, also can revise input by hand.Wherein the handoff procedure mathematical model is following multivariate Hammerstein nonlinear model:

$\left\{\begin{array}{c}\frac{d\ln \left({\mathrm{MI}}_{c,1}\left(s\right)\right)}{\mathrm{ds}}=-\frac{1}{{\mathrm{\&tau;}}_1}\ln \left({\mathrm{MI}}_{c,1}\left(s\right)\right)+\frac{1}{{\mathrm{\&tau;}}_1}\ln \left({\mathrm{MI}}_{i,1}\left(s\right)\right)\\ \frac{d\ln \left({\mathrm{MI}}_{c,2}\left(s\right)\right)}{\mathrm{ds}}=\frac{1}{{\mathrm{\&tau;}}_1}\ln \left({\mathrm{MI}}_{c,1}\left(s\right)\right)-(\frac{1}{{\mathrm{\&tau;}}_1}+\frac{1}{{\mathrm{\&tau;}}_2})\ln \left({\mathrm{MI}}_{c,2}\left(s\right)\right)+\frac{1}{{\mathrm{\&tau;}}_2}\ln \left({\mathrm{MI}}_{i,2}\left(s\right)\right)\\ \frac{{\mathrm{dEt}}_{c,1}\left(s\right)}{\mathrm{ds}}=-\frac{1}{{\mathrm{\&tau;}}_1}{\mathrm{Et}}_{c,1}\left(s\right)+\frac{1}{{\mathrm{\&tau;}}_1}{\mathrm{Et}}_{i,1}\left(s\right)\\ \frac{{\mathrm{dEt}}_{c,2}\left(s\right)}{\mathrm{ds}}=\frac{1}{{\mathrm{\&tau;}}_1}{\mathrm{Et}}_{c,1}\left(s\right)-(\frac{1}{{\mathrm{\&tau;}}_1}+\frac{1}{{\mathrm{\&tau;}}_2}){\mathrm{Et}}_{c,2}\left(s\right)+\frac{1}{{\mathrm{\&tau;}}_2}{\mathrm{Et}}_{i,2}\left(s\right)\\ \ln \left({\mathrm{MI}}_{i,1}\left(s\right)\right)={\mathrm{\&theta;}}_{11}\left(s\right)+{\mathrm{\&theta;}}_{12}\left(s\right)/T_1\left(s\right)+{\mathrm{\&theta;}}_{13}\left(s\right)\ln ({\mathrm{\&theta;}}_{14}\left(s\right)+{\mathrm{\&theta;}}_{15}\left(s\right){\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{16}\left(s\right){\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{mm}}\left(s\right))\\ \ln \left({\mathrm{MI}}_{i,2}\left(s\right)\right)={\mathrm{\&theta;}}_{21}\left(s\right)+{\mathrm{\&theta;}}_{22}\left(s\right)/T_2\left(s\right)+{\mathrm{\&theta;}}_{23}\left(s\right)\ln ({\mathrm{\&theta;}}_{24}\left(s\right)+{\mathrm{\&theta;}}_{25}\left(s\right){\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{26}\left(s\right){\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{mm}}\left(s\right))\\ {\mathrm{Et}}_{i,1}\left(s\right)=\frac{2({\mathrm{\&theta;}}_{18}\left(s\right){\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{mm}}\left(s\right)+1)}{3{\mathrm{\&theta;}}_{17}\left(s\right)/{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{18}\left(s\right){\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{mm}}\left(s\right)+4}\\ {\mathrm{Et}}_{i,2}\left(s\right)=\frac{2({\mathrm{\&theta;}}_{28}\left(s\right){\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{mm}}\left(s\right)+1)}{3{\mathrm{\&theta;}}_{27}\left(s\right)/{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{28}\left(s\right){\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{mm}}\left(s\right)+4}\end{array}\right.$

$\left\{\begin{array}{c}\frac{d\ln \left({\mathrm{MI}}_{c,1}\left(s\right)\right)}{\mathrm{ds}}=-\frac{1}{{\mathrm{\&tau;}}_1}\ln \left({\mathrm{MI}}_{c,1}\left(s\right)\right)+\frac{1}{{\mathrm{\&tau;}}_1}\ln \left({\mathrm{MI}}_{i,1}\left(s\right)\right)\\ \frac{d\ln \left({\mathrm{MI}}_{c,2}\left(s\right)\right)}{\mathrm{ds}}=\frac{1}{{\mathrm{\&tau;}}_1}\ln \left({\mathrm{MI}}_{c,1}\left(s\right)\right)-(\frac{1}{{\mathrm{\&tau;}}_1}+\frac{1}{{\mathrm{\&tau;}}_2})\ln \left({\mathrm{MI}}_{c,2}\left(s\right)\right)+\frac{1}{{\mathrm{\&tau;}}_2}\ln \left({\mathrm{MI}}_{i,2}\left(s\right)\right)\\ \frac{d\ln \left({\mathrm{MI}}_{c,3}\left(s\right)\right)}{\mathrm{ds}}=\frac{1}{{\mathrm{\&tau;}}_2}\ln \left({\mathrm{MI}}_{c,2}\left(s\right)\right)-(\frac{1}{{\mathrm{\&tau;}}_2}+\frac{1}{{\mathrm{\&tau;}}_3})\ln \left({\mathrm{MI}}_{c,3}\left(s\right)\right)+\frac{1}{{\mathrm{\&tau;}}_3}\ln \left({\mathrm{MI}}_{i,3}\left(s\right)\right)\\ \frac{{\mathrm{dEt}}_{c,3}\left(s\right)}{\mathrm{ds}}=-\frac{1}{{\mathrm{\&tau;}}_3}{\mathrm{Et}}_{c,3}\left(s\right)+\frac{1}{{\mathrm{\&tau;}}_3}{\mathrm{Et}}_{i,3}\left(s\right)\\ \ln \left({\mathrm{MI}}_{i,1}\left(s\right)\right)={\mathrm{\&theta;}}_{11}\left(s\right)+{\mathrm{\&theta;}}_{12}\left(s\right)/T_1\left(s\right)+{\mathrm{\&theta;}}_{13}\left(s\right)\ln ({\mathrm{\&theta;}}_{14}\left(s\right)+{\mathrm{\&theta;}}_{15}\left(s\right){\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{16}\left(s\right){\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{mm}}\left(s\right))\\ \ln \left({\mathrm{MI}}_{i,2}\left(s\right)\right)={\mathrm{\&theta;}}_{21}\left(s\right)+{\mathrm{\&theta;}}_{22}\left(s\right)/T_2\left(s\right)+{\mathrm{\&theta;}}_{23}\left(s\right)\ln ({\mathrm{\&theta;}}_{24}\left(s\right)+{\mathrm{\&theta;}}_{25}\left(s\right){\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{26}\left(s\right){\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{mm}}\left(s\right))\\ \ln \left({\mathrm{MI}}_{i,3}\left(s\right)\right)={\mathrm{\&theta;}}_{31}\left(s\right)+{\mathrm{\&theta;}}_{32}\left(s\right)/T_3\left(s\right)+{\mathrm{\&theta;}}_{33}\left(s\right)\ln ({\mathrm{\&theta;}}_{34}\left(s\right)+{\mathrm{\&theta;}}_{35}\left(s\right){\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{hm}}\left(s\right)+{\mathrm{\&theta;}}_{36}\left(s\right){\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{mm}}\left(s\right))-\ln (1+{\mathrm{\&theta;}}_{37}\left(s\right){\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{mn}}\left(s\right)))\\ {\mathrm{Et}}_{i,3}\left(s\right)=\frac{2({\mathrm{\&theta;}}_{29}\left(s\right){\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{mm}}\left(s\right)+1)}{15{\mathrm{\&theta;}}_{28}\left(s\right)/{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{hm}}\left(s\right)+{14\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{mm}}\left(s\right)+29}\end{array}\right.$

Wherein, the last same form is switched mathematical model for switching to the homopolymerization or the random copolymerization trade mark, and the following same form is switched mathematical model for switching to the crushing-resistant copolymerization trade mark; MI _{C, k}And MI _{I, k}, Et _{C, k}And Et _{I, k}Be respectively accumulation melting index and instantaneous melting index, accumulation ethylene contents and the instantaneous ethylene contents of the individual reactor of k (k=1,2,3); T _k, C _Khm, C _Kmm, τ _kBe respectively temperature of reaction, hydrogen density of propylene ratio, ethylene, propylene concentration ratio, the reaction time of the individual reactor of k (k=1,2,3); θ _k={ θ _Kj, (k=1,2,3; J=1 ..., 9) be the system model identified parameters.The production data that can call the dynamic data library storage carries out identification, and discrimination method has a variety of, as Subspace Identification method, nonlinear least square method etc., can be with reference to the correlation technique document, and here no longer describe in detail and suppose that this identification process finishes.With T _sFor discrete time is set up trade mark handoff procedure discrete time state-space model, and (1b) with reference to formula (1a).

3, click " configurable controller " button on the configuration interface, the CPU that starts industrial computer calls " switching control program " software package that weaves in advance and designs the rolling optimization switch controller.This algorithm is innovated on the basis of the new constraint Hammerstein system model PREDICTIVE CONTROL that proposes just and is obtained.

Click " controller simulation " button in the configuration interface, enter the model predictive controller off-line simulation debug phase.Adjust the adjustable parameters such as performance index, quadratic performance index weighting matrix and prediction step in the configuration interface, and observe the closed-loop system response curve, determine the good design of Controller parameters of integrated performance index such as dynamic property, steady-state behaviour and economic performance thus.Adjust matrix Q, R and W _jThe rule of adjusting: turn parameter q down _i, r _iAnd w _{I, j}Can accelerate the response speed of corresponding states, improve the dynamic property of control system, but the output energy of corresponding required i controller will increase, and its required output energy that provides of pairing topworks also will increase, can tend to exceed its range of capacity, when facing the not modeling dynamic perfromance of controlled process, be easy to show aggressive behavior simultaneously, be unfavorable for the robust stability of control system; On the contrary, increase parameter q _i, r _iAnd w _{I, j}To slow down the response speed of corresponding states, but the output energy of i required controller reduces, and its required output energy that provides of pairing topworks can reduce also, thereby can help the robust stability of control system.Regulate parameter T _pThe rule of adjusting: turn parameter T down _pCan reduce the on-line calculation of optimal control problem (10), accelerate the response speed of system, but will worsen the dynamic property and the robust performance of system; On the contrary, increase parameter T _pCan improve the dynamic property and the robust performance of system, but increase the on-line calculation of optimal control problem (10), the real-time of influence control.Therefore actual adjusting when regulating parameter should be weighed between the integrated performance indexs such as dynamic property, steady-state behaviour and economic performance of control system.

4, " trade mark switchover operation " button in the click configuration interface, the CPU that starts industrial computer reads the optimizing controller parameter, and to carry out the controlled quentity controlled variable that " trade mark switching automatic control program " obtain the current time optimum be polymeric reaction condition, should be sent to loop control unit and upgrade its setting value by the optimum polymeric reaction condition then.Loop control unit is the telo merization device in view of the above, and polymerization process is operated in the scope of setting.Show on the configuration interface this moment is paradigmatic system closed-loop response curve under the online situation.When next sampling instant arrived, the starting condition of the new state update controller by the observation paradigmatic system repeated whole implementation afterwards, so goes round and begins again and realizes the automatic control that the trade mark switches.

A complete set of trade mark switching controls process can be finished on industry control level configuration interface, the industrial exemplary application that this process can hereinafter provide with reference to this instructions.Compare with traditional trade mark changing method, the maximum characteristics of the polypropylene apparatus grade switch controller method for designing that the present invention provides are to realize the automatic switchover control of the trade mark according to the virtual condition of system's operation, and the economy that guarantees trade mark handoff procedure is with energy-saving and cost-reducing.Hereinafter specific implementation method illustrates actual effect of the present invention to switch to example in the polypropylene trade mark handoff procedure between modal homopolymer and homopolymer, homopolymer and multipolymer, multipolymer and the multipolymer, but range of application of the present invention is not exceeded with the switching trade mark among these embodiment.As previously mentioned, the present invention can also be used for the trade mark handoff procedure of various types of olefins polyplants such as tygon, Polyvinylchloride except can being used for the polypropylene apparatus grade handoff procedure.

Beneficial effect of the present invention mainly shows: 1, simplicity of design, understand easily, realize online put into operation easy, practical; 2, realize that in quite wide scope the polypropylene trade mark is the continuous automatic switchover of the non-similar trade mark especially, can greatly shorten trade mark switching time, reduce the discharging of substandard product, improve the production economic benefit and the market competitiveness.

Description of drawings

Fig. 1 is a polypropylene plant Spheripol process chart,

Wherein, Catalyst is a catalyzer; Cocatalyst is a promotor; RW is each reactor cycles chilled water; TIC represents the temperature of reaction controller; RAC1 represents hydrogen alkene concentration ratio controller; RAC2 represents ethylene, propylene concentration ratio controller.

The design proposal synoptic diagram of the trade mark switch controller that Fig. 2 provides for the present invention.

The structural representation that Fig. 3 adopts during for actual motion of the present invention.

Fig. 4 switches to the handoff procedure closed loop curve of output and the input curve of the crushing-resistant copolymerization trade mark for the homopolymerization trade mark,

Wherein, dotted line and solid line are respectively 1# and 2# annular-pipe reactor accumulation melt index forecast output trajectory among Fig. 4 (a); Fig. 4 (b) is fluidized-bed reactor accumulation melting index output trajectory (wherein solid line is control result of the present invention, and dotted line is the actual blocked operation result of factory, down together); Fig. 4 (c) is fluidized bed accumulation ethylene contents output trajectory; Fig. 4 (d) is a 1# annular-pipe reactor temperature of reaction input curve; Fig. 4 (e) is a 1# annular-pipe reactor hydrogen alkene concentration ratio input curve; Fig. 4 (f) is a 2# annular-pipe reactor temperature of reaction input curve; Fig. 4 (g) is a 2# annular-pipe reactor hydrogen alkene concentration ratio input curve; Fig. 4 (h) is a fluidized-bed reactor temperature of reaction input curve; Fig. 4 (i) is a fluidized-bed reactor hydrogen alkene concentration ratio input curve; Fig. 4 (j) is a fluidized-bed reactor ethylene, propylene concentration ratio input curve; Times express time scale in each subgraph.

Embodiment

Below in conjunction with accompanying drawing the present invention is further described.

With reference to Fig. 1～Fig. 4, a kind of polypropylene apparatus grade switching and controlling method, described control method comprises the steps:

1), set up polypropylene trade mark handoff procedure discrete time state-space model, referring to formula (1a) and formula (1b):

$\left\{\begin{array}{c}x(t+1)=A_{1}x\left(t\right)+B_{1}v\left(t\right),v\left(t\right)=g_1(u\left(t\right),t),y\left(t\right)=C_{1}x\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A_1=\left[\begin{array}{cccc}1-T_s/{\mathrm{\&tau;}}_1& 0& 0& 0\\ T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2& 0& 0\\ 0& 0& 1-T_s/{\mathrm{\&tau;}}_1& 0\\ 0& 0& T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2\end{array}\right],\\ {\mathrm{<figure-callout\; id="1"\; label="B"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">B</figure-callout>}}_1=\left[\begin{array}{cccc}{T}_s/{\mathrm{\&tau;}}_1& & & 0\\ & T_s/{\mathrm{\&tau;}}_2& & \\ & & T_s/{\mathrm{\&tau;}}_1& \\ 0& & & T_s/{\mathrm{\&tau;}}_2\end{array}\right],g_1(u,t)=\left[\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ {\mathrm{Et}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\\ {\mathrm{Et}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\end{array}\right]\\ {\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_1=\left[\begin{array}{cccc}0& 1& 0& 0\\ 0& 0& 0& 1\end{array}\right]\end{array}\right.---\left(1a\right)$

(1a)

$\left\{\begin{array}{c}x(t+1)=A_{2}x\left(t\right)+B_{2}v\left(t\right),v\left(t\right)=g_2(u\left(t\right),t),y\left(t\right)=C_{2}x\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A_2=\left[\begin{array}{cccc}1-T_s/{\mathrm{\&tau;}}_1& 0& 0& 0\\ T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2& 0& 0\\ 0& T_s/{\mathrm{\&tau;}}_2& 1-T_s/{\mathrm{\&tau;}}_2-T_s/{\mathrm{\&tau;}}_3& 0\\ 0& 0& 0& 1-T_s/{\mathrm{\&tau;}}_3\end{array}\right],\\ {\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">B</figure-callout>}}_2=\left[\begin{array}{cccc}{T}_s/{\mathrm{\&tau;}}_1& & & 0\\ & T_s/{\mathrm{\&tau;}}_2& & \\ & & T_s/{\mathrm{\&tau;}}_3& \\ 0& & & T_s/{\mathrm{\&tau;}}_3\end{array}\right],g_2(u,t)=\left[\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ \ln ({\mathrm{MI}}_{i,3}({\mathrm{\&theta;}}_3,u_3))\\ {\mathrm{Et}}_{i,3}({\mathrm{\&theta;}}_3,u_3)\end{array}\right]\\ {\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_2=\left[\begin{array}{cccc}0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]\end{array}\right.---\left(1b\right)$

(1b)

Wherein, in the formula (1a), x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), Et _{C, 1}, Et _{C, 2}] ', u=[u ₁, u ₂]=[T ₁, C _1hm, C _1mmT ₂, C _2hm, C _2mm] ' and y=[y ₁, y ₂[ln (the MI of] '= _{C, 2}), Et _{C, 2}State variable, input variable and output variable when] ' be respectively switches to the homopolymer or the random copolymers trade mark; In the formula (1b), x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), ln (MI _{C, 3}), Et _{C, 3}] ', u=[u ₁, u ₂, u ₃]=[T ₁, C _1hm, C _1mmT ₂, C _2hm, C _2mmT ₃, C _3hm, C _3mm] ' and y=[y ₁, y ₂[ln (the MI of] '= _{C, 3}), Et _{C, 3}State variable when] ' be respectively switches to the multipolymer trade mark, input variable and output variable; T _sBe system's discrete time; MI _{C, k}And MI _{I, k}, Et _{C, k}And Et _{I, k}Be respectively accumulation melting index and instantaneous melting index, accumulation ethylene contents and the instantaneous ethylene contents of the individual reactor of k (k=1,2,3); T _k, C _Khm, C _Kmm, τ _kBe respectively temperature of reaction, hydrogen density of propylene ratio, ethylene, propylene concentration ratio, the reaction time of the individual reactor of k (k=1,2,3); θ _k(k=1,2,3) are the state-space model identified parameters.

With model (1a) and (1b) the unified formula (2) that is described as:

$\left\{\begin{array}{c}x(t+1)=\mathrm{Ax}\left(t\right)+\mathrm{Bv}\left(t\right),v\left(t\right)=g(u\left(t\right),t)\\ y\left(t\right)=\mathrm{Cx}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\end{array}\right.---\left(2\right)$

2), the steady-state value of calculating formula (2), referring to formula (3):

$v_s=\underset{t\&RightArrow;\&infin;}{\lim }g(u_s,t),{\mathrm{Ax}}_s=-{\mathrm{Bv}}_s---\left(3\right)$

X wherein _s, v _sAnd u _sThe steady-state value of representing system state, intermediate variable and input variable respectively;

3), according to state, intermediate variable and the input variable of formula (3) calibration system, referring to formula (4a):

S _x＝diag{s _x ⁽ⁱ⁾，i＝1，2，3，4}；S _v＝diag{s _v ⁽ⁱ⁾，i＝1，2，3，4}；

S _u1＝diag{s _u1 ⁽ⁱ⁾，i＝1，2，3}；S _u2＝diag{s _u2 ⁽ⁱ⁾，i＝1，2，3}； (4a)

S _u3＝diag{s _u3 ⁽ⁱ⁾，i＝1，2，3}

Wherein calibration coefficient is (4b):

s _x ⁽ⁱ⁾＝(x _s，m ⁽ⁱ⁾) ^-2，s _v ⁽ⁱ⁾＝(v _s，m ⁽ⁱ⁾) ^-2，i＝1，2，3，4；

${s_{u1}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{u_{s1,o}}^{\left(i\right)}=0\mathrm{or}{{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}={{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}\\ {({{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}-{{\mathrm{<figure-callout\; id="1"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)})}^{-2},\mathrm{else}\end{array}\right.,i=\mathrm{1,2,3};$

${s_{u2}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{u_{s2,o}}^{\left(i\right)}=0\mathrm{or}{{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}={{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}\\ {({{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}-{{\mathrm{<figure-callout\; id="2"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)})}^{-2},\mathrm{else}\end{array},\right.i=\mathrm{1,2,3};---\left(4b\right)$

${s_{u3}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{u_{s3,o}}^{\left(i\right)}=0\mathrm{or}{{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}={{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}\\ {({{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)}-{{\mathrm{<figure-callout\; id="3"\; label="u\; s"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">u</figure-callout>}}_s}^{\left(i\right)})}^{-2},\mathrm{else}\end{array}\right.,i=\mathrm{1,2,3};$

Wherein, " _{S, o} ⁽ⁱ⁾" the current production trade mark correspondence of expression i state, intermediate variable or import the steady-state value of component; " _{S, m} ⁽ⁱ⁾" steady-state value of expression target trade mark correspondence;

4), according to performance index of system, referring to formula (5):

${\mathrm{<figure-callout\; id="1"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_1={\mathrm{\&Sigma;}}_{t=0}^{\&infin;}\{(x\left(t\right)-x_s)\&prime;{\mathrm{QS}}_x(x\left(t\right)-x_s)+(v\left(t\right)-v_s)\&prime;{\mathrm{RS}}_v(v\left(t\right)-v_s)\}---\left(5\right)$

Set up formula (2) neutral line subsystem optimal controller, referring to formula (6):

v(t)＝-(RS _v+B′PB) ^-1B′PA(x(t)-x _s)+v _s (6)

Wherein matrix P is a matrix equation, referring to formula (7):

A′PA-P+A′PB(RS _v+B′PB) ^-1B′PA+QS _x＝0 (7)

The symmetric positive definite dematrix; Wherein Q 〉=0 and R〉0 be respectively the weighting matrix of state and intermediate variable;

5), according to the expectation limit of system state observer, utilization POLE PLACEMENT USING method design point observer, referring to formula (8):

$\hat{x}(t+1)=(A-\mathrm{LC})\hat{x}\left(t\right)+\mathrm{Bg}\left(t\right)+\mathrm{Ly}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;---\left(8\right)$

Wherein L is the observer gain matrix;

6), the secondary system performance index, with reference to formula (9):

${\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_2={\mathrm{\&Sigma;}}_{i=t}^{t+T_p-1}\{(g\left(i\right)-v\left(i\right))\&prime;(g\left(i\right)-v\left(i\right))+{\mathrm{\&Sigma;}}_{j=1}^3(u_j\left(i\right)-u_{s,j})\&prime;W_j{S}_{\mathrm{uj}}(u_j\left(i\right)-u_{s,j})\}---\left(9\right)$

(9)

Foundation is based on the model predictive controller of state observer, referring to formula (10):

$u{(t;\hat{x}\left(t\right),T_p)}^{*}=\underset{u}{\mathrm{<figure-callout\; id="2"\; label="min\; u\; J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">min</figure-callout>}}{J}_{}{}_2$

s.t.x(i+1)＝Ax(i)+Bg(u(i)，i)，

v(i)＝-(RS _v+B′PB) ^-1B′PA(x(i)-x _s)+v _s (10)

x(i)∈Γ _x，v(i)∈Γ _v，u(i)∈Γ _u

$x\left(t\right)=\hat{x}\left(t\right),$ i＝t，…，t+T _p

Wherein, set Γ _x, Γ _vAnd Γ _uThe constraint conditions such as state, intermediate variable and input of expression handoff procedure;

(t) the current t of expression state observation value constantly; W _j0 (j=1,2,3) are the weighting matrix of input variable.

In the limited step-length optimal control in dynamic of line computation problem, i.e. formula (10), and, obtain Model Predictive Control amount based on state observer according to the rolling optimization principle, with reference to formula (11):

$u^{\mathrm{mpc}}\left(t\right)=u{(t;\hat{x}\left(t\right))}^{*},t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\left(11\right)$

Controller passes through state observer recording geometry state in each sampling instant, and upgrade the starting condition of optimal control problem with this observer state, rolling optimization calculates the PREDICTIVE CONTROL amount of current time then, go round and begin again, till trade mark handoff procedure carries out the transition to the target trade mark and steady production.

Present embodiment specifically has for the homopolymerization trade mark switches to the crushing-resistant copolymerization trade mark:

Suppose that the current production trade mark is homopolymer J_PP_10 in producing trade mark administration interface.1# and 2# annular-pipe reactor still are the steady state process value of trade mark J_PP_10 at this moment, and density of hydrogen and ethylene concentration are zero in the 3# fluidized-bed reactor, simultaneously according to actual trade mark blocked operation, suppose that temperature has been stabilized in 340K in the preceding fluidized bed of trade mark switching.Set forth the specific operation process that the homopolymerization trade mark switches to the crushing-resistant copolymerization trade mark below.

1, in producing trade mark administration interface, be crushing-resistant copolymerization trade mark K_PP_02 by the target trade mark of clicking " last one " or " next one " key selection switching.

2, according to the switching target trade mark, (modification) generation trade mark switches mathematical model and switches constraint condition as follows in the configuration interface:

Switching model:

$\left\{\begin{array}{c}\frac{d\ln \left({\mathrm{MI}}_{c,1}\left(t\right)\right)}{\mathrm{ds}}=-0.8696\ln \left({\mathrm{MI}}_{c,1}\left(t\right)\right)+0.8696\ln \left({\mathrm{MI}}_{i,1}\left(t\right)\right)\\ \frac{d\ln \left({\mathrm{MI}}_{c,2}\left(t\right)\right)}{\mathrm{ds}}=0.8696\ln \left({\mathrm{MI}}_{c,1}\left(t\right)\right)-2.0460\ln \left({\mathrm{MI}}_{c,2}\left(t\right)\right)+1.1765\ln \left({\mathrm{MI}}_{i,2}\left(t\right)\right)\\ \frac{d\ln \left({\mathrm{MI}}_{c,3}\left(t\right)\right)}{\mathrm{ds}}=1.1765\ln \left({\mathrm{MI}}_{c,2}\left(t\right)\right)-3.1725\ln \left({\mathrm{MI}}_{c,3}\left(t\right)\right)+1.9960\ln \left({\mathrm{MI}}_{i,3}\left(t\right)\right)\\ \frac{{\mathrm{dEt}}_{c,3}\left(t\right)}{\mathrm{ds}}=-{1.9960\mathrm{Et}}_{c,3}\left(t\right)+{1.9960\mathrm{Et}}_{i,3}\left(t\right)\\ \ln \left({\mathrm{MI}}_{i,1}\left(t\right)\right)=0.322-46.175/T_1\left(t\right)+5.379\ln \left(1.085{+1.259C}_{1\mathrm{hm}}\left(t\right)\right)\\ \ln \left({\mathrm{MI}}_{i,2}\left(t\right)\right)=0.116-46.175/T_2\left(t\right)+4.379\ln \left(1.065{+1.259C}_{2\mathrm{hm}}\left(t\right)\right)\\ \ln \left({\mathrm{MI}}_{i,3}\left(t\right)\right)=12.457-19.491/T_3\left(t\right)+21.447\ln (0.429+{0.988C}_{3\mathrm{hm}}\left(t\right)+{0.209C}_{3\mathrm{hm}})-\ln (1+0.439C_{3\mathrm{hm}}))\\ {\mathrm{Et}}_{i,3}\left(s\right)=\frac{{0.1048C}_{3\mathrm{mm}}\left(s\right)+2}{0.826/C_{3\mathrm{hm}}\left(t\right)+{14C}_{3\mathrm{mm}}\left(t\right)+29}\end{array}\right.$

Switch constraint:

$\left\{\begin{array}{c}{\mathrm{\&Gamma;}}_x=\{0.5\&le;\ln \left({\mathrm{MI}}_{c,1}\left(t\right)\right)\&le;\mathrm{0.9,0.3}\&le;\ln \left({\mathrm{MI}}_{c,2}\left(t\right)\right)\&le;\mathrm{1.0,0}\&le;\ln \left({\mathrm{MI}}_{c,3}\left(t\right)\right)\&le;\mathrm{0.8,0}\&le;{\mathrm{Et}}_{c,3}\left(t\right)\&le;10\},\\ {\mathrm{\&Gamma;}}_y=\{0.5\&le;\ln \left({\mathrm{MI}}_{i,1}\left(t\right)\right)\&le;\mathrm{1.0,0}.2\&le;\ln \left({\mathrm{MI}}_{i,2}\left(t\right)\right)\&le;1.3,0\&le;\ln \left({\mathrm{MI}}_{i,3}\left(t\right)\right)\&le;\mathrm{0.9,0}\&le;{\mathrm{Et}}_{i,3}\left(t\right)\&le;9\},\\ {\mathrm{\&Gamma;}}_u=\{341.15K\&le;T_1\left(t\right)\&le;345.15K,0.007\%\&le;{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{1\mathrm{hm}}\left(t\right)\&le;0.060\%,\\ 341.15K\&le;T_2\left(t\right)\&le;345.15K,0.007\%\&le;{\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{2\mathrm{hm}}\left(t\right)\&le;0.060\%,\\ 341.15K\&le;T_3\left(t\right)\&le;350.17K,0\&le;{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{hm}}\left(t\right)\&le;6.50\%,0\&le;{\mathrm{<figure-callout\; id="3"\; label="C"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">C</figure-callout>}}_{3\mathrm{mm}}\left(t\right)\&le;100\%\}\end{array}\right.$

3, click " configurable controller " button on the configuration interface and enter next configuration interface, the CPU that starts industrial computer calls " switching control program " software package that weaves in advance and designs the rolling optimization switch controller.Concrete computation process is as follows:

(1) makes x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), ln (MI _{C, 3}), Et _{C, 3}] ', u=[u ₁, u ₂, u ₃]=[T ₁, C _1hm, 0; T ₂, C _2hm, 0; T ₃, C _3hm, C _3mm] ', y=[ln (MI _{C, 3}), Et _{C, 3}] ' and the discrete time T of system _s=0.5h then can get trade mark handoff procedure discrete time state-space model by formula (2) and is

$\left\{\begin{array}{c}x(t+1)=\mathrm{Ax}\left(t\right)+\mathrm{Bv}\left(t\right),v\left(t\right)=g(u\left(t\right),t),y\left(t\right)=\mathrm{Cx}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A=\left[\begin{array}{cccc}0.5652& 0& 0& 0\\ 0.4348& -0.023& 0& 0\\ 0& 0.5882& -0.5862& 0\\ 0& 0& 0& 0.002\end{array}\right],B=\left[\begin{array}{cccc}0.4348& & & 0\\ & 0.5882& & \\ & & 0.988& \\ 0& & & 0.998\end{array}\right],\\ C=\left[\begin{array}{cccc}0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]\\ g(u,t)=\left[\left(,\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ \ln \left({\mathrm{MI}}_{i,3}({\mathrm{\&theta;}}_3,u_3)\right)\\ {\mathrm{Et}}_{i,3}({\mathrm{\&theta;}}_3,u_3)\end{array},\right)\right]\end{array}\right.$

(2) according to the steady-state value of the switching target trade mark and formula (3) calculating switching model, get

$u_s=\left[\begin{array}{ccc}{u}_{s,1}& u_{s,2}& u_{s,3}\end{array}\right]=\left[\begin{array}{ccc}343.15& 343.15& 348.17\\ 0.01& 0.01& 4.0\\ 0& 0& 82.0\end{array}\right],$ $v_s=\left[\begin{array}{c}0.6886\\ 0.3087\\ 0.7428\\ 7.4997\end{array}\right],$ $x_s=\left[\begin{array}{c}0.6886\\ 0.4702\\ 0.6419\\ 7.5\end{array}\right]$

Simultaneously computational data being sent into industrial computer dynamic data base system preserves.

(3) according to the switching target trade mark and formula (4) calibration system state, intermediate variable and input variable,

$S_x=\left[\begin{array}{cccc}2.1089& & & 0\\ & 4.5231& & \\ & & 2.427& \\ 0& & & 0.0178\end{array}\right],$ $S_v=\left[\begin{array}{cccc}2.1089& & & 0\\ & 10.4952& & \\ & & 1.8123& \\ 0& & & 0.0178\end{array}\right],$

${\mathrm{<figure-callout\; id="1"\; label="S\; u"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">S</figure-callout>}}_u=\left[\begin{array}{ccc}1& 0& 0\\ 0& 625& 0\\ 0& 0& 1\end{array}\right],$ ${\mathrm{<figure-callout\; id="2"\; label="S\; u"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">S</figure-callout>}}_u=\left[\begin{array}{ccc}1& 0& 0\\ 0& 816& 0\\ 0& 0& 1\end{array}\right],$ ${\mathrm{<figure-callout\; id="3"\; label="S\; u"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">S</figure-callout>}}_u=\left[\begin{array}{ccc}1& 0& 0\\ 0& 1& 0\\ 0& 0& 1\end{array}\right],$

And these nominal datas are sent into industrial computer dynamic data base system preserve.

(4) performance index of setting trade mark handoff procedure are as follows

${\mathrm{<figure-callout\; id="1"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_1={\mathrm{\&Sigma;}}_{t=0}^{\&infin;}\{(x\left(t\right)-x_s)\&prime;{\mathrm{QS}}_x(x\left(t\right)-x_s)+(v\left(t\right)-v_s)\&prime;{\mathrm{RS}}_v(v\left(t\right)-v_s)\}$

Q=diag{1 wherein, 1,1,1}, R=diag{5,5,5,5} then according to the optimal controller that linear subsystem is calculated in formula (6) and (7) is

$\left\{\begin{array}{c}v\left(t\right)=-K\&CenterDot;(x\left(t\right)-x_s)+v_s\\ K=\left[\begin{array}{cccc}0.0888& 0.0047& -0.0044& 0\\ 0.0606& -0.023& 0.0196& 0\\ -0.0188& 0.1267& -0.1251& 0\\ 0& 0& 0& 00.0003\end{array}\right]\end{array}\right.$

(5) the expectation limit P_ob={0.09 of initialization system state observer, 0.095,0.1,0.098} then according to Lyapunov method and formula (8) must system state observer equation be

$\left\{\begin{array}{c}\hat{x}(t+1)=(A-\mathrm{LC})\hat{x}\left(t\right)+\mathrm{Bg}\left(t\right)+\mathrm{Ly}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ L\&prime;=\left[\begin{array}{cccc}0.3996& 0.3005& -0.3370& 0\\ 0& 0& 0& -0.0880\end{array}\right]\end{array}\right.$

(6) setting trade mark handoff procedure quadratic performance index is as follows

${\mathrm{<figure-callout\; id="2"\; label="J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">J</figure-callout>}}_2={\mathrm{\&Sigma;}}_{i=t}^{t+T_p-1}\{(g\left(i\right)-v\left(i\right))\&prime;(g\left(i\right)-v\left(i\right))+{\mathrm{\&Sigma;}}_{j=1}^3(u_j\left(i\right)-u_{s,j})\&prime;W_j{S}_{\mathrm{uj}}(u_j\left(i\right)-u_{s,j})\}$

T wherein _p=4, W _j=diag{0.1,0.1,0.1}, j=1,2,3, then according to the limited step-length optimal control in dynamic problem of formula (10) definition

$u{(t;\hat{x}\left(t\right),T_p)}^{*}=\underset{u}{\mathrm{<figure-callout\; id="2"\; label="min\; u\; J"\; filenames="A200810163417E00231.png,A200810163417E00241.png"\; state="\{\{state\}\}">min</figure-callout>}}{J}_{}{}_2$

s.t.x(i+1)＝Ax(i)+Bg(u(i)，i)，

v(i)＝-K·(x(i)-x _s)+v _s

x(i+1)∈Γ _x，v(i)∈Γ _v，u(i)∈Γ _u

$x\left(t\right)=\hat{x}\left(t\right),$ i＝t，…，t+3

The utilization genetic algorithm gets optimal control solution in the above-mentioned optimal control problem of line computation

$u{(t;\hat{x}\left(t\right),T_p)}^{*}=\{u{(t;\hat{x}\left(t\right))}^{*},u{(t+1;\hat{x}\left(t\right))}^{*},u{(t+2;\hat{x}\left(t\right))}^{*},u{(t+3;\hat{x}\left(t\right))}^{*}\}$

According to formula (12) must be based on the trade mark switching model PREDICTIVE CONTROL amount of state observer

$u^{\mathrm{mpc}}\left(t\right)=u{(t;\hat{x}\left(t\right))}^{*},t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;$

Controller upgrades the starting condition of limited step-length optimal control in dynamic problem in view of the above at the state of each sampling instant by the state observer recording geometry, and rolling optimization calculates the PREDICTIVE CONTROL amount of current time then.

(7) design parameter Q, R, the W of artificial debugging model predictive controller ₁, W ₂And W ₃,

Q＝diag{1，1，1，1}，R＝diag{5，5，5，5}，W ₁＝W ₂＝W ₃＝diag{0.1，0.1，0.1}

4, " trade mark switchover operation " button in the click configuration interface, the CPU that starts industrial computer reads the optimizing controller parameter, and to carry out the controlled quentity controlled variable that " trade mark switching automatic control program " obtain the current time optimum be polymeric reaction condition, should be sent to loop control unit and upgrade its setting value by the optimum polymeric reaction condition then.Loop control unit is the telo merization device in view of the above, and polymerization process is operated in the scope of setting.Show on the configuration interface this moment is paradigmatic system closed-loop response curve under the online situation.When next sampling instant arrived, the starting condition of the new state update controller by the observation paradigmatic system repeated whole implementation afterwards, so goes round and begins again and realizes the automatic control that the trade mark switches.

What more than set forth is the good trade mark switching controls effect that embodiment showed that the present invention provides.It may be noted that, the present invention is not only limited to the foregoing description, for switching between the similar trades mark such as the homopolymerization trade mark, the random copolymerization trade mark and the crushing-resistant copolymerization trade mark, be blocked operation between the non-similar trade mark especially, the method design trade mark switch controller that adopts the present invention to provide can be realized good trade mark switching controls effect equally.The method for designing of the polypropylene apparatus grade switch controller that the present invention provides can be widely used in the trade mark switching controls of various types of olefins polyplants such as tygon, Polyvinylchloride.

Claims (1)

1, a kind of polypropylene apparatus grade switching and controlling method is characterized in that: described control method comprises the steps:

1), set up polypropylene board bow handoff procedure discrete time state-space model, referring to formula (1a) and formula (1b):

$\left\{\begin{array}{c}x(t+1)=A_{1}x\left(t\right)+B_{1}v\left(t\right),v\left(t\right)=g_1(u\left(t\right),t),y\left(t\right)=C_{1}x\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A_1=\left[\begin{array}{cccc}1-T_s/{\mathrm{\&tau;}}_1& 0& 0& 0\\ T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2& 0& 0\\ 0& 0& 1-T_s/{\mathrm{\&tau;}}_1& 0\\ 0& 0& T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2\end{array}\right]\\ B_1=\left[\begin{array}{cccc}{T}_s/{\mathrm{\&tau;}}_1& & & 0\\ & T_s/{\mathrm{\&tau;}}_2& & \\ & & T_s/{\mathrm{\&tau;}}_1& \\ 0& & & T_s/{\mathrm{\&tau;}}_2\end{array}\right],g_1(u,t)=\left[\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ {\mathrm{Et}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\\ {\mathrm{Et}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\end{array}\right]\\ C_1=\left[\begin{array}{cccc}0& 1& 0& 0\\ 0& 0& 0& 1\end{array}\right]\end{array}\right.---\left(1a\right)$

$\left\{\begin{array}{c}x(t+1)=A_{2}x\left(t\right)+B_{2}v\left(t\right),v\left(t\right)=g_2(u\left(t\right),t),y\left(t\right)=C_{2}x\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\\ A_2=\left[\begin{array}{cccc}1-T_s/{\mathrm{\&tau;}}_1& 0& 0& 0\\ T_s/{\mathrm{\&tau;}}_1& 1-T_s/{\mathrm{\&tau;}}_1-T_s/{\mathrm{\&tau;}}_2& 0& 0\\ 0& T_s/{\mathrm{\&tau;}}_2& 1-T_s/{\mathrm{\&tau;}}_2-T_s/{\mathrm{\&tau;}}_3& 0\\ 0& 0& 0& 1-T_s/{\mathrm{\&tau;}}_3\end{array}\right]\\ B_2=\left[\begin{array}{cccc}{T}_s/{\mathrm{\&tau;}}_1& & & 0\\ & T_s/{\mathrm{\&tau;}}_2& & \\ & & T_s/{\mathrm{\&tau;}}_3& \\ 0& & & T_s/{\mathrm{\&tau;}}_3\end{array}\right],g_2(u,t)=\left[\begin{array}{c}\ln \left({\mathrm{MI}}_{i,1}({\mathrm{\&theta;}}_1,u_1)\right)\\ \ln \left({\mathrm{MI}}_{i,2}({\mathrm{\&theta;}}_2,u_2)\right)\\ \ln ({\mathrm{MI}}_{i,3}({\mathrm{\&theta;}}_3,u_3))\\ {\mathrm{Et}}_{i,3}({\mathrm{\&theta;}}_3,u_3)\end{array}\right]\\ C_2=\left[\begin{array}{cccc}0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]\end{array}\right.---\left(1b\right)$

Wherein, in the formula (1a), x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), Et _{C, 1}, Et _{C, 2}] ', u=[u ₁, u ₂]=[T ₁, C _1hm, C _1mmT ₂, C _2hm, C _2mm] ' and y=[y ₁, y ₂[ln (the MI of] '= _{C, 2}), Et _{C, 2})] ' be respectively state variable, input variable and output variable when switching to the homopolymer or the random copolymers trade mark; In the formula (1b), x=[x ₁, x ₂, x ₃, x ₄[ln (the MI of] '= _{C, 1}), ln (MI _{C, 2}), ln (MI _{C, 3}), Et _{C, 3}] ', u=[u ₁, u ₂, u ₃]=[T ₁, C _1hm, C _1mmT ₂, C _2hm, C _2mmT ₃, C _3hm, C _3mm] ' and y=[y ₁, y ₂[ln (the MI of] '= _{C, 3}), Et _{C, 3}State variable when] ' be respectively switches to the multipolymer trade mark, input variable and output variable; T _sBe system's discrete time; MI _{C, k}And MI _{I, k}, Et _{C, k}And Et _{I, k}Be respectively accumulation melting index and instantaneous melting index, accumulation ethylene contents and the instantaneous ethylene contents of k reactor, wherein, k=1,2,3; T _k, C _Khm, C _Kmm, τ _kBe respectively temperature of reaction, hydrogen density of propylene ratio, ethylene, propylene concentration ratio, the reaction time of k reactor; θ _kBe the state-space model identified parameters;

With model (1a) and (1b) the unified formula (2) that is described as:

$\left\{\begin{array}{c}x(t+1)=\mathrm{Ax}\left(t\right)+\mathrm{Bv}\left(t\right),v\left(t\right)=g(u\left(t\right),t)\\ y\left(t\right)=\mathrm{Cx}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\end{array}\right.---\left(2\right)$

2), the steady-state value of calculating formula (2), referring to formula (3):

$v_s=\underset{t\&RightArrow;\&infin;}{\lim }g(u_s,t),{\mathrm{Ax}}_s=-{\mathrm{Bv}}_s---\left(3\right)$

X wherein _s, v _sAnd u _sThe steady-state value of representing system state, intermediate variable and input variable respectively;

3), according to state, intermediate variable and the input variable of formula (3) calibration system, referring to formula (4a):

S _x＝diag{s _x ⁽ⁱ⁾，i＝1，2，3，4}；S _v＝diag{s _v ⁽ⁱ⁾，i＝1，2，3，4}；

S _u1＝diag{s _u1 ⁽ⁱ⁾，i＝1，2，3}；S _u2＝diag{s _u2 ⁽ⁱ⁾，i＝1，2，3}；(4a)

S _u3＝diag{s _u3 ⁽ⁱ⁾，i＝1，2，3}

Wherein calibration coefficient is (4b):

s _x ⁽ⁱ⁾＝(x _s， _m ⁽ⁱ⁾) ^-2，s _v ⁽ⁱ⁾＝(v _s， _m ⁽ⁱ⁾) ^-2，i＝1，2，3，4；

${s_{u1}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{u_{s1,o}}^{\left(i\right)}=0\mathrm{or}{u_{s1,o}}^{\left(i\right)}={u_{s1,m}}^{\left(i\right)}\\ {({u_{s1,o}}^{\left(i\right)}-{u_{s1,m}}^{\left(i\right)})}^{-2},\mathrm{else}\end{array}\right.i=\mathrm{1,2,3};$

${s_{u2}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{u_{s2,o}}^{\left(i\right)}=0\mathrm{or}{u_{s2,o}}^{\left(i\right)}={u_{s2,m}}^{\left(i\right)}\\ {({u_{s2,o}}^{\left(i\right)}-{u_{s2,m}}^{\left(i\right)})}^{-2},\mathrm{else}\end{array},\right.i=\mathrm{1,2,3};---\left(4b\right)$

${s_{u3}}^{\left(i\right)}=\left\{\begin{array}{c}1,\mathrm{if}{u_{s3,o}}^{\left(i\right)}=0\mathrm{or}{u_{s3,o}}^{\left(i\right)}={u_{s3,m}}^{\left(i\right)}\\ {({u_{s3,o}}^{\left(i\right)}-{u_{s3,m}}^{\left(i\right)})}^{-2},\mathrm{else}\end{array}\right.i=\mathrm{1,2,3};$

Wherein, " _{S, o} ⁽ⁱ⁾" the current production trade mark correspondence of expression i state, intermediate variable or import the steady-state value of component; " _{S, m} ⁽ⁱ⁾" steady-state value of expression target trade mark correspondence;

4), according to performance index of system, referring to formula (5):

$J_1={\mathrm{\&Sigma;}}_{t=0}^{\&infin;}\{(x\left(t\right)-x_s)\&prime;{\mathrm{QS}}_x(x\left(t\right)-x_s)+(v\left(t\right)-v_s)\&prime;{\mathrm{RS}}_v(v\left(t\right)-v_s)\}---\left(5\right)$

Set up formula (2) neutral line subsystem optimal controller, referring to formula (6):

v(t)＝-(RS _v+B′PB) ^-1B′PA(x(t)-x _s)+v _s (6)

Wherein matrix P is a matrix equation, referring to formula (7):

A′PA-P+A′PB(RS _v+B′PB) ^-1B′PA+QS _x＝0 (7)

The symmetric positive definite dematrix, wherein Q 〉=0 and R〉0 be respectively the weighting matrix of state and intermediate variable;

5), according to the expectation limit of system state observer, utilization POLE PLACEMENT USING method design point observer, referring to formula (8):

$\hat{x}(t+1)=(A-\mathrm{LC})\hat{x}\left(t\right)+\mathrm{Bg}\left(t\right)+\mathrm{Ly}\left(t\right),t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;---\left(8\right)$

Wherein L is the observer gain matrix;

6), according to the secondary system performance index, with reference to formula (9):

$J_2={\mathrm{\&Sigma;}}_{i=t}^{t+T_p-1}\{(g\left(i\right)-v\left(i\right))\&prime;(g\left(i\right)-v\left(i\right))+{\mathrm{\&Sigma;}}_{j=1}^3(u_j\left(i\right)-u_{s,j})\&prime;W_j{S}_{\mathrm{uj}}(u_j\left(i\right)-u_{s,j})\}---\left(9\right)$

Foundation is based on the model predictive controller of state observer, referring to formula (10):

$u{(t;\hat{x}\left(t\right),T_p)}^{*}=\underset{u}{\min }{J}_2$

s.t.x(i+1)＝Ax(i)+Bg(u(i)，i)，

v(i)＝-(RS _v+B′PB) ^-1B′PA(x(i)-x _s)+v _s (10)

x(i)∈Γ _x，v(i)∈Γ _v，u(i)∈Γ _u

$x\left(t\right)=\hat{x}\left(t\right),$ i＝t，…，t+T _p

Wherein, set Γ _x, Γ _vAnd Γ _uThe constraint conditions such as state, intermediate variable and input of expression handoff procedure; (t) the current t of expression state observation value constantly; W _j0 (j=1,2,3) are the weighting matrix of input variable.

In the limited step-length optimal control in dynamic of line computation problem, i.e. formula (10), and, obtain Model Predictive Control amount based on state observer according to the rolling optimization principle, with reference to formula (11):

$u^{\mathrm{mpc}}\left(t\right)=u{(t;\hat{x}\left(t\right))}^{*},t=\mathrm{0,1},\&CenterDot;\&CenterDot;\&CenterDot;\left(11\right)$

Controller passes through state observer recording geometry state in each sampling instant, and upgrade the starting condition of optimal control problem with this observer state, rolling optimization calculates the PREDICTIVE CONTROL amount of current time then, go round and begin again, till trade mark handoff procedure carries out the transition to the target trade mark and steady production.

Images