CN101846976B - Polymerization process grade switch-over track optimizing method based on shooting technique - Google Patents

Abstract

The invention relates to a polymerization process grade switch-over track optimizing method based on a shooting technique. The track optimization calculation of the conventional grade switch-over process is very difficult. In the method, a general track optimizing model for the polymerization process grade switchover is firstly established and merged into a dynamic track optimizing model, then the track of the dynamic track optimizing model is optimized by a self-adaptation shooting technique, and an on-site production device is controlled and regulated. The self-adaptation shooting technique firstly calls a result worked out by a single shooting technique, provides the result for a multipoint shooting technique as an initial value thereof, and then adopts the multipoint shooting technique to work out a feasible solution on the basis of the given initial value, and if an optimal solution is found, then iteration is ended, and otherwise recirculation is carried out. The method adopts two different shooting strategies, greatly enhances the feasibility and high efficiency of a track optimizing method, can effectively reduce the production of substandard products in the polymerization process grade switch-over process, and reduces the switch-over time.

Description

A kind of method for optimization of grade switching track of polymerization process based on the target practice technology

Technical field

The invention belongs to areas of information technology, relate to a kind of optimisation technique, particularly relate to a kind of method for optimization of grade switching track of polymerization process based on self-adaptation target practice technology.

Background technology

The trade mark switching of polymerization process is meant the polymer production device is switched to the process that another organizes production technology from one group of production technology, the polymeric articles of the corresponding a kind of specification of a kind of trade mark.Because modern society is very vigorous to the demand of polymeric articles, product diversification there is higher requirement, for meeting the need of market and the maximization of economic interests, enterprise all can carry out the trade mark continually and switch in the process of producing polymkeric substance.And the process that the trade mark switches can produce underproof transitional product inevitably.By seeking optimum switching track, make handoff procedure transitional product minimum number, transit time the shortest, the track optimizing problem of Here it is trade mark handoff procedure.And the optimization of grade switching track problem is very complicated, relates to differential and finds the solution with mixing of algebraic equation, relates to the dynamic optimization technology, and classic method is found the solution very difficult.

Summary of the invention

Target of the present invention is the weak point at existing technology, and a kind of track optimizing method of polymerization process grade handoff procedure is provided, and specifically is based on self-adaptation target practice technology, improves the feasibility and the high efficiency of trade mark handoff procedure track optimizing.

The inventive method has adopted means such as adaptive technique, dynamic optimization technology, numerical analysis technology, numerical technology, computer technology, and concrete steps are:

Step (1) is set up the general track optimizing model that polymerization process grade switches, and comprises the shortest model and substandard product minimum model switching time;

No matter be which kind of polymerization process, the objective function of its trade mark handover optimization model all comprises the shortest of switching time, or the output minimum of substandard product.Select control variable as the case may be, and provide constraint condition, the specification of the initial trade mark and the target trade mark is retrained.

Described switching time, the shortest model representation was:

$\underset{u}{\mathrm{Min}}\left\{t_{\mathrm{gt}}\right\}$

s.t.

$\frac{\mathrm{dx}}{\mathrm{dt}}=f(x\left(t\right),u\left(t\right),t)$

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

Described substandard product minimum model representation is:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{{\∫}_0^{t_{\mathrm{gt}}}{[p(x\left(t\right),u\left(t\right),t)-p_{t\arg \mathrm{et}}]}^2\mathrm{dt}\right\}$

s.t.

$\frac{\mathrm{dx}}{\mathrm{dt}}=f(x\left(t\right),u\left(t\right),t)$

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

Wherein, t _GtRepresent that switching time, p represent product index, the p in arbitrary moment in the handoff procedure _TargerThe product index of the expression target trade mark, x represents that state variable, u represent control variable, t express time constant, and Min represents minimum, and f represents function of state, and g represents constraint condition, promptly the product index to the initial trade mark or the target trade mark claims;

Step (2) with set up switching time the shortest model and substandard product minimum model merge into following dynamic trajectory Optimization Model:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{J(x\left(t\right),u\left(t\right),t)\right\}$

s.t.

$\frac{\mathrm{dx}}{\mathrm{dt}}=f(x\left(t\right),u\left(t\right),t)$

g(x(t _gt)，u(t _gt)，t)＜＝0

Wherein, J represents objective function.The dynamic trajectory Optimization Model is the differential algebraic equations problem in essence.

Step (3) adopts self-adaptation target practice technology to carry out track optimizing the dynamic trajectory Optimization Model.Self-adaptation target practice technology is that singles' target technology is combined with multiple spot target practice technology, and concrete steps comprise:

A. call singles' target technology and try to achieve the result, offer multiple spot target practice technology as its initial value;

B. on given initial value basis, adopt multiple spot target practice technology to try to achieve feasible solution;

C. on the basis of feasible solution, utilize singles' target technology to calculate.If find optimum solution, iteration finishes; Otherwise turn back to a. step, again circulation.

The step of described singles' target technology comprises:

1) with t switching time _GtAverage mark is slit into n segment, and n represents total interval number;

2) state variable x is integrated to the t finish time by existing Euler's integral method constantly from initial 0 of handoff procedure _Gt, obtain following steady-state model:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{J(x\left(t\right),u\left(t\right),t)\right\}$

s.t.

$x={\∫}_0^{t_{\mathrm{gt}}}f(x\left(t\right),u\left(t\right),t)\mathrm{dt}$

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

3) steady-state model adopts existing nonlinear programming approach to find the solution, and obtains feasible solution.

The step of described multiple spot target practice technology comprises:

1) with t switching time _GtLast average mark is slit into n segment;

2) with state variable x in each segment, by existing Euler's integral method, be integrated to end from the top of segment;

3) add one group of algebraic equation constraint y: state variable x will be equated in the interval top value of each extremity of an interval value and next;

4) thus obtain steady-state model:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{J(x\left(t\right),u\left(t\right),t)\right\}$

s.t.

$x={\∫}_{t_k}^{t_{k+1}}f(x\left(t\right),u\left(t\right),t)\mathrm{dt}$

y(x(t)，u(t)，t)＝0

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

T wherein _kRepresent k interval zero hour in n the segment, t _K+1Represent k+1 the interval zero hour, k represents interval sequence number;

5) steady-state model adopts existing nonlinear programming approach to find the solution, and obtains feasible solution.

Self-adaptation target practice technology combines the advantage of singles' target technology and multiple spot target practice technology, has abandoned shortcoming separately.The advantage of singles' target technology is to restrain than being easier to, but the speed of finding the solution is slow relatively; And that the advantage of multiple spot target practice technology is a speed of convergence is very fast relatively, but simultaneously because the whole discretizes of variable, the expansion of the dimension that throws into question, convergence ratio is difficulty.

And the initial value that self-adaptation target practice technology requires singles' target technology to provide is always circularly found the solution to many target practices technology, has both utilized the convergence of singles' target technology to be easy to characteristics, has also utilized multiple spot target practice technology to find the solution characteristics fast.Therefore, this track optimizing algorithm based on self-adaptation target practice technology has greatly improved feasibility and the high efficiency of finding the solution.

Step (4) offers the polymer production on-site control device with the control variable result of calculation in the handoff procedure that obtains, and control device carries out regulating and controlling according to this control strategy to field production device.

The inventive method can greatly shorten the transit time of trade mark handoff procedure by Field adjustment and enforcement, reduces the generation of defective intermediate product in the handoff procedure, solves the trouble and worry of trade mark handoff procedure.Like this, manufacturing enterprise can carry out the trade mark and switch in continuous production run, and does not need first parking, the new trade mark of driving again, has reduced the waste of the material and the energy.Thereby be beneficial on the maximized basis in the economy and the energy, satisfied the demand of market the different product trade mark.

Embodiment

A kind of method for optimization of grade switching track of polymerization process based on the target practice technology has adopted means such as adaptive technique, dynamic optimization technology, numerical analysis technology, numerical technology, computer technology, and concrete steps are:

Step (1) is set up the general track optimizing model that polymerization process grade switches, and comprises the shortest model and substandard product minimum model switching time.

Described switching time, the shortest model representation was:

$\underset{u}{\mathrm{Min}}\left\{t_{\mathrm{gt}}\right\}$

s.t.

$\frac{\mathrm{dx}}{\mathrm{dt}}=f(x\left(t\right),u\left(t\right),t)$

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

Described substandard product minimum model representation is:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{{\∫}_0^{t_{\mathrm{gt}}}{[p(x\left(t\right),u\left(t\right),t)-p_{t\arg \mathrm{et}}]}^2\mathrm{dt}\right\}$

s.t.

$\frac{\mathrm{dx}}{\mathrm{dt}}=f(x\left(t\right),u\left(t\right),t)$

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

Wherein, t _GtRepresent that switching time, p represent product index, the p in arbitrary moment in the handoff procedure _TargerThe product index of the expression target trade mark, x represents that state variable, u represent control variable, t express time constant, and Min represents minimum, and f represents function of state, and g represents constraint condition, promptly the product index to the initial trade mark or the target trade mark claims.

Step (2) with set up switching time the shortest model and substandard product minimum model merge into following dynamic trajectory Optimization Model:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{J(x\left(t\right),u\left(t\right),t)\right\}$

s.t.

$\frac{\mathrm{dx}}{\mathrm{dt}}=f(x\left(t\right),u\left(t\right),t)$

g(x(t _gt)，u(t _gt)，t)＜＝0

Wherein, J represents objective function.

Step (3) adopts self-adaptation target practice technology to carry out track optimizing the dynamic trajectory Optimization Model.Self-adaptation target practice technology is that singles' target technology is combined with multiple spot target practice technology, and concrete steps comprise:

A. call singles' target technology and try to achieve the result, offer multiple spot target practice technology as its initial value;

B. on given initial value basis, adopt multiple spot target practice technology to try to achieve feasible solution;

C. on the basis of feasible solution, utilize singles' target technology to calculate.If find optimum solution, iteration finishes; Otherwise turn back to a. step, again circulation.

Wherein, the step of singles' target technology comprises:

1) with t switching time _GtAverage mark is slit into n segment, and n represents total interval number;

2) state variable x is integrated to the t finish time by existing Euler's integral method constantly from initial 0 of handoff procedure _Gt, obtain following steady-state model:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{J(x\left(t\right),u\left(t\right),t)\right\}$

s.t.

$x={\∫}_0^{t_{\mathrm{gt}}}f(x\left(t\right),u\left(t\right),t)\mathrm{dt}$

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

3) steady-state model adopts existing nonlinear programming approach to find the solution, and obtains feasible solution.

The step of multiple spot target practice technology comprises:

1) with t switching time _GtLast average mark is slit into n segment;

2) with state variable x in each segment, by existing Euler's integral method, be integrated to end from the top of segment;

3) add one group of algebraic equation constraint y: state variable x will be equated in the interval top value of each extremity of an interval value and next;

4) thus obtain steady-state model:

$\underset{u,t_{\mathrm{gt}}}{\mathrm{Min}}\left\{J(x\left(t\right),u\left(t\right),t)\right\}$

s.t.

$x={\∫}_{t_k}^{t_{k+1}}f(x\left(t\right),u\left(t\right),t)\mathrm{dt}$

y(x(t)，u(t)，t)＝0

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

T wherein _kRepresent k interval zero hour in n the segment, t _K+1Represent k+1 the interval zero hour, k represents interval sequence number;

5) steady-state model adopts existing nonlinear programming approach to find the solution, and obtains feasible solution.

Step (4) offers the polymer production on-site control device with the control variable result of calculation in the handoff procedure that obtains, and control device carries out regulating and controlling according to this control strategy to field production device.

Claims (3)

1. method for optimization of grade switching track of polymerization process based on the target practice technology is characterized in that the concrete steps of this method are:

Step (1) is set up the general track optimizing model that polymerization process grade switches, and comprises the shortest model and substandard product minimum model switching time;

Described switching time, the shortest model representation was:

s.t.

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

Described substandard product minimum model representation is:

s.t.

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

Wherein, t _GtRepresent that switching time, p represent product index, the p in arbitrary moment in the handoff procedure _TargetThe product index of the expression target trade mark, x represents that state variable, u represent control variable, t express time constant, and Min represents minimum, and f represents function of state, and g represents constraint condition;

Step (2) with set up switching time the shortest model and substandard product minimum model merge into following dynamic trajectory Optimization Model:

s.t.

g(x(t _gt)，u(t _gt)，t)＜＝0

Wherein, J represents objective function;

Step (3) adopts self-adaptation target practice technology to carry out track optimizing the dynamic trajectory Optimization Model, specifically:

A. at first call singles' target technology and try to achieve the result, offer multiple spot target practice technology as its initial value;

B. then on given initial value basis, adopt multiple spot target practice technology to try to achieve feasible solution;

C. on the basis of feasible solution, utilize singles' target technology to calculate: if find optimum solution, then iteration finishes; If do not find optimum solution, then turn back to step a, again circulation;

Step (4) offers the polymer production on-site control device with the control variable result of calculation in the handoff procedure that obtains, and control device carries out regulating and controlling according to this control strategy to field production device.

2. a kind of method for optimization of grade switching track of polymerization process based on the target practice technology as claimed in claim 1 is characterized in that the step of the singles' target technology described in the step (3) is:

1) with t switching time _GtAverage mark is slit into n segment, and n represents total interval number;

2) state variable x is integrated to the t finish time by existing Euler's integral method constantly from initial 0 of handoff procedure _Gt, obtain following steady-state model:

s.t.

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

3) steady-state model adopts existing nonlinear programming approach to find the solution, and obtains feasible solution.

3. a kind of method for optimization of grade switching track of polymerization process based on the target practice technology as claimed in claim 1 is characterized in that the step of the multiple spot target practice technology described in the step (3) is:

1) with t switching time _GtLast average mark is slit into n segment;

2) with state variable x in each segment, by existing Euler's integral method, be integrated to end from the top of segment;

3) add one group of algebraic equation constraint y: state variable x will be equated in the interval top value of each extremity of an interval value and next;

4) thus obtain steady-state model:

s.t.

y(x(t)，u(t)，t)＝0

g(x(t _gt)，u(t _gt)，t _gt)＜＝0

T wherein _kRepresent k interval zero hour in n the segment, t _K+1Represent k+1 the interval zero hour, k represents interval sequence number;

5) steady-state model adopts nonlinear programming approach to find the solution, and obtains feasible solution.