CN104216396A - Method for evaluating performance of single-variable multi-time-variant disturbance systems in cracking furnaces - Google Patents

Abstract

The invention provides a method for evaluating the performance of single-variable multi-time-variant disturbance systems in cracking furnaces. The method includes steps of S100, acquiring SMTMV (single model time-variant minimum variance) control criteria for each single action of disturbance of the single-variable multi-time-variant disturbance systems at each time point according to preset time-variant disturbance models for different actions of disturbance; S200, acquiring MMTMV (multi-model mixing time-variant minimum variance) control criteria for all the actions of disturbance of the single-variable multi-time-variant disturbance systems at each time point by the aid of the SMTMV control criteria; S300, acquiring theoretical output variance according to the MMTMV control criteria; S400, comparing the theoretical output variance obtained in the step S300 by means of computation to actual output variance of the single-variable multi-time-variant disturbance systems under the control criteria to obtain quantitative performance indexes, and comparing the performance indexes to default index standards to obtain performance evaluation results of the single-variable multi-time-variant disturbance systems in the cracking furnaces. The actual output variance is acquired by means of actual detection and computation.

Description

The performance estimating method of perturbed system is become when single argument is many in pyrolysis furnace

Technical field

The present invention relates to petrochemical industry, particularly relate to the performance estimating method becoming perturbed system when single argument is many in a kind of pyrolysis furnace.

Background technology

Along with industrial process becomes increasingly complex, Performance Evaluation carries out to system just more and more important.System Performance Analysis method can evaluation control system performance effectively, judge whether control system operates in optimum Working according to assessment result, and improve control strategy on this basis, make technological process operate in optimum condition, and then improve productivity effect and the process safety of enterprise.

Control system Performance Evaluation is that Harris T proposed first in 1989, using the variance that obtains based on LMS control as assessment benchmark, is used for assessing the performance of single variable control system.Stanfej Nives etc. were application the method monitor and diagnosis single process control loop performance in 1991.The performance of constant disturbance control system when the method can assess list effectively, the performance of constant disturbance control system when but can not effectively assess many.Multi-model process is a kind of effective ways solving the uncertain complex industrial process control waiting system dynamic characteristic to change with operating mode.Luyuan L in 2002 etc. propose the shifting multi-mode type adaptive control based on on-line optimization.Multi-model switching method from all single model minimum variance principle, chooses the best control criterion of Evaluated effect to assess many perturbed systems performance, but in practical operation, multiple disturbance occurs that carrying out interval switches simultaneously, larger transient error can be produced, cause multi-model switching method to there is larger defect.

At present, performance estimating method be mostly for time constant perturbed system design LMS control criterion.Olaleye F in 2004 etc., based on becoming perturbed system temporary disturbance and most representative disturbance during list, propose a kind of control system assessment level assessing time-varying system performance.Mengfei Z in 2011 to by time become disturbance PID control system carried out Performance Evaluation.Although some researchists propose some control system performance estimating methods for becoming perturbed system time single, become disturbance control system for time many, correlative study is less.

Summary of the invention

The technical problem to be solved in the present invention how to carry out assessment to the performance becoming perturbed system when single argument is many in pyrolysis furnace to detect.

In order to solve this technical matters, the invention provides the performance estimating method becoming perturbed system when single argument is many in a kind of pyrolysis furnace, comprising the steps:

S100: according to different disturbance preset time become Disturbance Model and obtain becoming the SMTMV control criterion of single disturbance in perturbed system when the single argument of each time point is many;

S200: the MMTMV control criterion becoming perturbed system all disturbances when the single argument obtaining each time point by described SMTMV control criterion is many;

S300: obtain theoretical output variance according to described MMTMV control criterion;

S400: become compared with the actual output variance under perturbed system control criterion when the theoretical output variance calculated by step S300 detects with reality, the single argument that calculates is many, obtain the performance index quantized, and then by these performance index being compared with set criterion, obtain the performance evaluation result becoming perturbed system when single argument is many in pyrolysis furnace.

In described step S100, comprise the steps:

S101: become perturbed system disturbance transfer function according to becoming when one that presets when Disturbance Model calculates many with the single argument becoming disturbance corresponding to Disturbance Model time this in each moment;

S102: diophantus decomposition is carried out to described disturbance transfer function;

S103: substitute into described disturbance transfer function by decomposing the Diophantine equation that obtains, according to LMS control criterion, obtains becoming the SMTMV control criterion of this disturbance in perturbed system when single argument is many.

When becoming perturbed system disturbance into single disturbance when single argument, MMTMV control criterion is the SMTMV control criterion of this disturbance;

When single argument is many there is multiple disturbance in change perturbed system, based on the time length setting hybrid weight of each perturbation action, synthesizing the SMTMV control criterion of each disturbance according to this hybrid weight, finally obtaining becoming the MMTMV control criterion of perturbed system when single argument is many.

Become when single argument is many when perturbed system exists multiple and become disturbance N _{1, t}, N _{2, t}..., N _n,t, each perturbation action time period is respectively T ₁, T ₂..., T _n, T=∪ T _i, i=1,2 ..., n.The SMTMV controller designed for each disturbance is Q _i,t, then based on time each become disturbance SMTMV control criterion into

u _t＝-Q _1,ty _t,if t∈T ₁

u _t＝-Q _2,ty _t,if t∈T ₂

·

·

·

u _t＝-Q _n,ty _t,if t∈T _n

MMTMV control criterion is

u _t＝-Q _ty _t

Wherein, MMTMV controller Q _tmeet

Q _t＝α ₁(t)Q _1,t+α ₂(t)Q _2,t+...+α _n(t)Q _n,t

In formula, α ₁(t), α ₂(t) ..., α _nt () is respectively the hybrid weight of corresponding SMTMV controller; For any t ∈ T, α _it () meets:

${\mathrm{\α}}_i\left(t\right)=0,t\∉T_i,i=\mathrm{1,2},...,n$

α ₁(t),α ₂(t),...,α _n(t)≥0

α ₁(t)+α ₂(t)+...+α _n(t)＝1

Hybrid weight α _i(t) (i=1,2 ..., n) meet following formula

${\mathrm{\α}}_i\left(t\right)=\frac{{\widetilde{\mathrm{\α}}}_i\left(t\right)}{{\widetilde{\mathrm{\α}}}_1\left(t\right)+{\widetilde{\mathrm{\α}}}_2\left(t\right)+...+{\widetilde{\mathrm{\α}}}_n\left(t\right)},i=\mathrm{1,2},...,n$

for not normalized weight factor, it meets continuously differentiable condition;

get following formula

Wherein, U _i, L _ibe respectively T _iinterval bound, T _i={ t:L _i≤ t≤U _i;

choose and should make meet continuously differentiable condition, α _i(t)>=0, α _i(t)=0, i=1,2 ..., n, α ₁(t)+α ₂(t)+...+α _n(t)=1.

Described performance index definition is:

$\mathrm{\η}=\frac{{\mathrm{\δ}}_{\mathrm{MMTMV}}^2}{{\mathrm{\δ}}_y^2}$

In formula, be single argument many time become output variance under perturbed system MMTMV control criterion, for actual single argument many time become output variance under perturbed system control criterion, Performance Evaluation index 0≤η≤1.

The present invention becomes the appraisal procedure of minimum variance (Multi-model Mixing Time-variant Minimum Variance, MMTMV) control criterion when proposing multi-model mixing.Each moment according to become time each Disturbance Model formulate corresponding single argument many time become the single model of perturbed system time become minimum variance (Single Model Time-variant Minimum Variance, SMTMV) control criterion, then adopt the method for multi-model mixing each time point be mixed into single argument many time become the MMTMV control criterion of perturbed system.And then become when single argument is many output variance under the effect of perturbed system MMTMV control criterion many as single argument time become the benchmark of perturbed system Performance Evaluation and the actual output variance obtained compares, thus achieve check and evaluation.

Accompanying drawing explanation

Fig. 1 becomes disturbance feedback control system when single argument is many in one embodiment of the invention;

Fig. 2 be in one embodiment of the invention under working control device time become perturbation response curve figure;

Fig. 3 is hybrid parameter α in one embodiment of the invention ₁(t) valued curve;

Fig. 4 is hybrid parameter α in one embodiment of the invention ₂(t) valued curve;

Fig. 5 is hybrid parameter α in one embodiment of the invention ₃(t) valued curve.

Embodiment

The invention provides the performance estimating method becoming perturbed system when single argument is many in a kind of pyrolysis furnace, comprise the steps:

S100: according to different disturbance preset time become Disturbance Model and obtain becoming the SMTMV control criterion of single disturbance in perturbed system when the single argument of each time point is many;

S200: the MMTMV control criterion becoming perturbed system all disturbances when the single argument obtaining each time point by described SMTMV control criterion is many;

S300: obtain theoretical output variance according to described MMTMV control criterion;

S400: become compared with the actual output variance under perturbed system control criterion when the theoretical output variance calculated by step S300 detects with reality, the single argument that calculates is many, obtain the performance index quantized, and then by these performance index being compared with set criterion, obtain the performance evaluation result becoming perturbed system when single argument is many in pyrolysis furnace.

In described step S100, comprise the steps:

S101: become perturbed system disturbance transfer function according to becoming when one that presets when Disturbance Model calculates many with the single argument becoming disturbance corresponding to Disturbance Model time this in each moment;

S102: diophantus decomposition is carried out to described disturbance transfer function;

S103: substitute into described disturbance transfer function by decomposing the Diophantine equation that obtains, according to LMS control criterion, obtains becoming the SMTMV control criterion of this disturbance in perturbed system when single argument is many.

When becoming perturbed system disturbance into single disturbance when single argument, MMTMV control criterion is the SMTMV control criterion of this disturbance;

When single argument is many there is multiple disturbance in change perturbed system, based on the time length setting hybrid weight of each perturbation action, synthesizing the SMTMV control criterion of each disturbance according to this hybrid weight, finally obtaining becoming the MMTMV control criterion of perturbed system when single argument is many.

Specifically, time many for single argument, become disturbance control system, when the present invention proposes multi-model mixing, become the appraisal procedure of minimum variance (Multi-model Mixing Time-variant Minimum Variance, MMTMV) control criterion.Each moment according to become time each Disturbance Model formulate corresponding single argument many time become the single model of perturbed system time become minimum variance (Single Model Time-variant Minimum Variance, SMTMV) control criterion, then adopt the method for multi-model mixing each time point be mixed into single argument many time become the MMTMV control criterion of perturbed system.When becoming perturbed system disturbance when single argument into becoming disturbance time single, MMTMV Performance Evaluation criterion is the SMTMV control criterion for becoming time this designed by disturbance; When becoming change disturbance when perturbed system exists multiple when single argument, based on the time length setting hybrid weight of each perturbation action, hybrid weight determines the degree of participation of each SMTMV control criterion, synthesize final single argument many time become the MMTMV System Performance Analysis criterion of perturbed system.

The present invention is achieved through the following technical solutions:

According to each moment participate in single argument many time become perturbed system perturbation action time become Disturbance Model formulate single argument under this moment many time become the SMTMV control criterion of each disturbance of perturbed system;

Based on the thought of multi-model mixing, become when each time point utilizes single argument more the SMTMV control criterion of perturbed system be mixed to get single argument many time become the MMTMV control criterion of perturbed system;

Output variance when single argument is many under the effect of change perturbed system MMTMV control criterion becomes the benchmark of perturbed system Performance Evaluation time many as single argument;

The SMTMV control criterion step becoming perturbed system a certain disturbance when wherein acquisition single argument is many is as follows:

S1, according to some disturbances time become Disturbance Model calculate the single argument of each this disturbance of moment many time become perturbed system disturbance transfer function;

S2, diophantus decomposition is carried out to disturbance transfer function;

S3, the Diophantine equation that obtains will be decomposed substitute into transport function, and according to LMS control criterion, obtain becoming the SMTMV control criterion of this disturbance of perturbed system when single argument is many.

Become when single argument is many when perturbed system exists multiple and become disturbance N _{1, t}, N _{2, t}..., N _n,t, each perturbation action time period is respectively T ₁, T ₂..., T _n, T=∪ T _i, i=1,2 ..., n.The SMTMV controller designed for each disturbance is Q _i,t, then based on time each become disturbance SMTMV control criterion into

u _t＝-Q _1,ty _t,if t∈T ₁

u _t＝-Q _2,ty _t,if t∈T ₂

·

·

·

u _t＝-Q _n,ty _t,if t∈T _n

When then single argument is many become perturbed system MMTMV control criterion into

u _t＝-Q _ty _t

Wherein, MMTMV controller Qt meets

Q _t＝α ₁(t)Q _1,t+α ₂(t)Q _2,t+...+α _n(t)Q _n,t

In formula, α ₁(t), α ₂(t) ..., α _nt () is respectively the hybrid weight of corresponding SMTMV controller.For any t ∈ T, α _it () meets

${\mathrm{\α}}_i\left(t\right)=0,t\∉T_i,i=\mathrm{1,2},...,n$

α ₁(t),α ₂(t),...,α _n(t)≥0

α ₁(t)+α ₂(t)+...+α _n(t)＝1

Hybrid weight α _i(t) (i=1,2 ..., n) meet following formula

${\mathrm{\α}}_i\left(t\right)=\frac{{\widetilde{\mathrm{\α}}}_i\left(t\right)}{{\widetilde{\mathrm{\α}}}_1\left(t\right)+{\widetilde{\mathrm{\α}}}_2\left(t\right)+...+{\widetilde{\mathrm{\α}}}_n\left(t\right)},i=\mathrm{1,2},...,n$

for not normalized weight factor, it meets continuously differentiable condition.In the present invention get following formula

Wherein, U _i, L _ibe respectively T _iinterval bound, T _i={ t:L _i≤ t≤U _i.

choose and should make meet continuously differentiable condition, α _i(t)>=0, α _i(t)=0, i=1,2 ..., n, α ₁(t)+α ₂(t)+...+α _n(t)=1.

Definition performance index are

$\mathrm{\η}=\frac{{\mathrm{\δ}}_{\mathrm{MMTMV}}^2}{{\mathrm{\δ}}_y^2}$

In formula, be single argument many time become output variance under perturbed system MMTMV control criterion, for single argument many time become output variance under perturbed system working control criterion, Performance Evaluation index 0≤η≤1.

Below with reference to Fig. 1 to Fig. 5, the performance estimating method becoming perturbed system when single argument is many in pyrolysis furnace provided by the invention is described in detail, it is the present invention one optional embodiment, can think, those skilled in the art, in the scope not changing the present invention's spirit and content, can modify to it and polish.

Petrochemical industry is one of China's mainstay industry, and ethene, propylene are wherein most important products.The production scale of ethene, propylene, production technology indicate the development level of a national oil chemical engineering industry.Pyrolysis furnace is the faucet device of a whole set of petrochemical equipment, to even running and the energy and material consumption of whole chemical process, and even finally total economic benefit important.Along with pyrolysis furnace reaction is carried out, pyrolysis furnace tube wall is thickening because coking reaction, causes pyrolysis furnace Disturbance Model to present time-varying characteristics, therefore carries out Performance Evaluation to the pyrolysis furnace with time-varying characteristics significant.Pyrolysis furnace diene yield (ethene, propylene) and oil product attribute, pyrolysis furnace mean outlet temperature COT, pyrolysis furnace hydrocarbons gasoline ratio, feed rate, the factor such as are cross-over temperature relevant.

Disturbance feedback control system is become as shown in Figure 1 when single argument is many.In figure, T is the transport function not having time lag, and d is system time lags, and Q is controller transfer function, and N is disturbance transfer function, a _tfor white noise.The present invention is under the constant condition of oil product attribute, pyrolysis furnace hydrocarbons gasoline ratio, feed rate, the factor such as cross-over temperature, with pyrolysis furnace mean outlet temperature COT be performance variable, diene yield for controlled variable, carry out Performance Evaluation to becoming disturbance cracking furnace system during single-input single-output many.

First, become time many according to single argument disturbance cracking furnace system each time become Disturbance Model and formulate SMTMV controller.If time become Disturbance Model N _i,tact on single argument many time become disturbance cracking furnace system time, its export response be

$y_t=\frac{N_{i,t}}{1+q^{-d}T{Q}_{i,t}}{a}_t---\left(1\right)$

Adopt Diophantine equation decomposing N _i,t, have

(2) are substituted into (1),

$\begin{array}{c}{y}_t=\frac{F_{i,t}+q^{-d}{R}_{i,t}}{1+q^{-d}T{Q}_{i,t}}\\ =(F_{i,t}+\frac{R_{i,t}-F_{i,t}T{Q}_{i,t}}{1+q^{-d}T{Q}_{i,t}}{q}^{-d})a_t\\ =F_{i,t}{a}_t+L_{i,t}{a}_{t-d}\end{array}---\left(3\right)$

Wherein

$L_{i,t}=\frac{R_{i,t}-F_{i,t}T{Q}_{i,t}}{1+q^{-d}T{Q}_{i,t}}---\left(4\right)$

Then

Var(y _t)＝Var(F _i,ta _t)+Var(L _i,ta _t-d) (5)

Var(y _t)≥Var(F _i,ta _t) (6)

(6) formula and if only if L _i,tminimum when=0, namely

$L_{i,t}=\frac{R_{i,t}-F_{i,t}T{Q}_{i,t}}{1+q^{-d}T{Q}_{i,t}}=0---\left(4\right)$

So SMTMV controller under Disturbance Model effect can be become time this:

$Q_{i,t}=\frac{R_{i,t}}{T{F}_{i,t}}---\left(8\right)$

Multi-model switching method chooses the best control criterion of Evaluated effect to assess single argument many perturbed systems performance from the single model minimum variance principle participating in control action.Minimum variance (Multi-model Switching Time-variant Minimum Variance is become when becoming the multi-model switching of disturbance cracking furnace system when single argument is many, MSTMV) control criterion obtains as follows: in each moment, become time many from all participation single arguments the SMTMV control criterion of disturbance cracking furnace system control action, the SMTMV control criterion that when choosing many to single argument, change disturbance cracking furnace system Performance Evaluation effect is best becomes disturbance cracking furnace system MSTMV control criterion time many as single argument, with assess single argument many time become disturbance cracking furnace system performance.When adopting MSTMV control criterion many to single argument, change disturbance cracking furnace system carries out Performance Evaluation, effectively can assess the control performance of each time period of pyrolysis furnace control system, but in actual industrial process, may occur owing to becoming disturbance time multiple simultaneously, interval can be produced and switch the transient error making system generation larger, affect the accuracy of its assessment result.

This problem of transient error is there is for solving multi-model MSTMV control criterion, become Disturbance Model when can take into full account again each to the impact of Performance Evaluation, become disturbance cracking furnace system when the present invention's application MMTMV control criterion appraisal procedure is many to single argument and carry out Performance Evaluation.

Become when single argument is many when disturbance cracking furnace system exists multiple and become disturbance N _{1, t}, N _{2, t}..., N _n,t, each perturbation action pyrolysis furnace time period is respectively T ₁, T ₂..., T _n, T=∪ T _i, i=1,2 ..., n.The SMTMV controller designed for each disturbance is Q _i,t, then based on time each become disturbance SMTMV control criterion into

u _t＝-Q _1,ty _t,if t∈T ₁

u _t＝-Q _2,ty _t,if t∈T ₂

·

·

·

u _t＝-Q _n,ty _t,if t∈T _n (9)

Thus, at MMTMV control criterion middle controller Q _tmeet

u _t＝-Q _ty _t (10)

Q _t＝α ₁(t)Q _1,t+α ₂(t)Q _2,t+...+α _n(t)Q _n,t (11)

In formula, α ₁(t), α ₂(t) ..., α _nt () is respectively the hybrid weight of corresponding SMTMV controller.For any t ∈ T, α _it () meets

${\mathrm{\α}}_i\left(t\right)=0,t\∉T_i,i=\mathrm{1,2},...,n---\left(12\right)$

α ₁(t),α ₂(t),...,α _n(t)≥0 (13)

α ₁(t)+α ₂(t)+...+α _n(t)＝1 (14)

Hybrid weight α _i(t) (i=1,2 ..., n) meet following formula

${\mathrm{\α}}_i\left(t\right)=\frac{{\widetilde{\mathrm{\α}}}_i\left(t\right)}{{\widetilde{\mathrm{\α}}}_1\left(t\right)+{\widetilde{\mathrm{\α}}}_2\left(t\right)+...+{\widetilde{\mathrm{\α}}}_n\left(t\right)},i=\mathrm{1,2},...,n---\left(15\right)$

for not normalized weight factor, it meets continuously differentiable condition.In the present invention get following formula

Wherein, U _i, L _ibe respectively T _iinterval bound, T _i={ t:L _i≤ t≤U _i.

For making meet continuously differentiable condition, α _it () meets formula (12)-(14), get for such as minor function

Definition performance index are

$\mathrm{\η}=\frac{{\mathrm{\δ}}_{\mathrm{MMTMV}}^2}{{\mathrm{\δ}}_y^2}---\left(18\right)$

In formula, be single argument many time become disturbance cracking furnace system MMTMV control criterion control under output variance, for single argument many time become disturbance cracking furnace system working control criterion control under output variance, Performance Evaluation index 0< η <1.

The discrete transfer function getting pyrolysis furnace controlled device is

$q^{-d}T\left(q^{-1}\right)=q^{-4}\frac{0.3}{1-0.8q^{-1}}---\left(19\right)$

Have in pyrolysis furnace control action process when three and become disturbance, its Disturbance Model transport function form is as follows respectively

$N_1\left(q^{-1}\right)=\frac{1-b_1{q}^{-1}}{1-a_1{q}^{-1}},1\&le;t\&le;3000---\left(20\right)$

$N_2\left(q^{-1}\right)=\frac{1-b_2{q}^{-1}}{1-a_2{q}^{-1}},2000\&le;t\&le;3000---\left(21\right)$

$N_3\left(q^{-1}\right)=\frac{1-b_3{q}^{-1}}{1-a_3{q}^{-1}},1000\&le;t\&le;4000---\left(22\right)$

In formula, a ₁, b ₁, a ₂, b ₂, a ₃, b ₃changing Pattern is as follows

a ₁＝0.5+0.00006t

b ₁＝0.45+0.00001t

a ₂＝0.6+0.00005t

b ₂＝0.45-0.00001t

a ₃＝0.65+0.00005t

b ₃＝0.45-0.00002t (23)

N ₁, N ₂, N ₃acting on the pyrolysis furnace time is 1-3000,2000-3000,1000-4000 sampling respectively, overlapped between three.Noise source to be variance be 1 white noise.

The actual pyrolysis furnace controller form adopted in the present invention is as follows

$Q=\frac{0.55-0.45q^{-1}}{0.85-0.25q^{-1}-0.25q^{-2}}---\left(24\right)$

Under this controller action, the actual time becoming disturbance pyrolysis furnace control system when single argument is many becomes perturbation response curve as shown in Figure 2.Design MMTMV controller Q _ttime, hybrid weight α ₁(t), α ₂(t), α ₃(t) at the valued curve of each sampling instant as shown in Fig. 3,4,5.

Except employing MMTMV control criterion, the present invention also have employed respectively and becomes disturbance cracking furnace system time many to single argument carried out Performance Evaluation based on the SMTMV control criterion becoming Disturbance Model design when three, MSTMV control criterion.Become disturbance cracking furnace system when single argument is many under different control criterion controls, the output variance of different time sections and performance index are as shown in table 1 Yu table 2.

Pyrolysis furnace diene yield variance under the different appraisal procedure control action of table 1

Cracking furnace system performance index under the different appraisal procedure of table 2

Can see from table 1 and table 2, utilize based on N ₁sMTMV control criterion many to single argument time become disturbance pyrolysis furnace control system when carrying out Performance Evaluation, the single argument of energy Efficient Evaluation 0-1000,1000-2000,2000-3000 sampling time section becomes disturbance pyrolysis furnace control system performance time many, but 3000-4000 sampling time section, its assessment variance ratio realized variance is large, therefore its can not the single argument of Efficient Evaluation 3000-4000 sampling time section many time change disturbance cracking furnace system performance.In like manner, based on N ₂sMTMV control criterion can not the single argument of Efficient Evaluation 3000-4000 sampling time section many time become disturbance cracking furnace system performance.Adopt based on N ₃although SMTMV control criterion can effectively many to single argument time become disturbance pyrolysis furnace control system performance and assess, but it is not only large in the assessment variance of corresponding sampling time section Performance Evaluation variance ratio MMTMV control criterion, and in whole process 0-4000 sampling time section, its assessment variance is larger than MMTMV control criterion assessment variance equally, therefore based on N ₃sMTMV control criterion Evaluated effect be worse than MMTMV control criterion Evaluated effect.

When adopting MSTMV control criterion many to single argument, change disturbance cracking furnace system carries out Performance Evaluation, can effectively assess single argument many time become the control performance of each sampling time section of disturbance cracking furnace system, be only worse than MMTMV control criterion Evaluated effect at 1000-2000 its Evaluated effect of sampling time section.But in actual industrial process, may occur owing to becoming disturbance time multiple simultaneously, interval can be produced and switch the transient error making system generation larger, affect the accuracy of its assessment result.

The MMTMV control criterion appraisal procedure that the present invention proposes becomes the assessment result that disturbance cracking furnace system carries out Performance Evaluation more and is better than MSTMV control criterion appraisal procedure when 1000-2000 sampling time section is to single argument, and substantially identical with MSTMV control criterion assessment result in other sampling time section MMTMV control criterion assessment result.Because MMTMV control criterion avoids interval to switch by the method for hybrid weight, larger transient error impact evaluation result can not be produced because interval switches, solve MSTMV method Problems existing well.Therefore MMTMV appraisal procedure entirety is better than MSTMV method.

In order to accurate online evaluation single argument many time become disturbance control system performance, the present invention proposes a kind of single argument becoming LMS control when mix based on multi-model many time change perturbed system performance estimating method.Compared with the appraisal procedure of SMTMV control criterion, the method can fully take into account single argument many time become perturbed system each time become disturbance on the impact of System Performance Analysis, its accuracy and confidence level all improve a lot; Compared with MSTMV appraisal procedure, the method assessment result is substantially identical with MSTMV method assessment result, but the impact that the method can avoid the transient error switching generation because of interval to bring system evaluation result accuracy.Become disturbance cracking furnace system Performance Evaluation time many from single argument can obtain, appraisal procedure proposed by the invention can accurately, effectively assess single argument many time become the performance of disturbance control system, for the performance improvement of subsequent control system provides theoretical foundation.

Claims (5)

1. become a performance estimating method for perturbed system when single argument is many in pyrolysis furnace, it is characterized in that: comprise the steps:

S100: according to different disturbance preset time become Disturbance Model and obtain becoming the SMTMV control criterion of single disturbance in perturbed system when the single argument of each time point is many;

S200: the MMTMV control criterion becoming perturbed system all disturbances when the single argument obtaining each time point by described SMTMV control criterion is many;

S300: obtain theoretical output variance according to described MMTMV control criterion;

S400: become compared with the actual output variance under perturbed system control criterion when the theoretical output variance calculated by step S300 detects with reality, the single argument that calculates is many, obtain the performance index quantized, and then by these performance index being compared with set criterion, obtain the performance evaluation result becoming perturbed system when single argument is many in pyrolysis furnace.

2. become the performance estimating method of perturbed system when single argument is many in pyrolysis furnace as claimed in claim 1, it is characterized in that: in described step S100, comprise the steps:

S101: become perturbed system disturbance transfer function according to becoming when one that presets when Disturbance Model calculates many with the single argument becoming disturbance corresponding to Disturbance Model time this in each moment;

S102: diophantus decomposition is carried out to described disturbance transfer function;

S103: substitute into described disturbance transfer function by decomposing the Diophantine equation that obtains, according to LMS control criterion, obtains becoming the SMTMV control criterion of this disturbance in perturbed system when single argument is many.

3. become the performance estimating method of perturbed system when single argument is many in pyrolysis furnace as claimed in claim 1, it is characterized in that: when becoming perturbed system disturbance into single disturbance when single argument, MMTMV control criterion is the SMTMV control criterion of this disturbance more;

When single argument is many there is multiple disturbance in change perturbed system, based on the time length setting hybrid weight of each perturbation action, synthesize the SMTMV control criterion of each disturbance again according to this hybrid weight, finally obtain becoming the MMTMV control criterion of perturbed system when single argument is many.

4. become the performance estimating method of perturbed system when single argument is many in the pyrolysis furnace as described in one of as any in claims 1 to 3, it is characterized in that: become when single argument is many when perturbed system exists multiple and become disturbance N _{1, t}, N _{2, t}..., N _n,t, each perturbation action time period is respectively T ₁, T ₂..., T _n, T=∪ T _i, i=1,2 ..., n.The SMTMV controller designed for each disturbance is Q _i,t, then based on time each become disturbance SMTMV control criterion into

u _t＝-Q _1,ty _t,if t∈T ₁

u _t＝-Q _2,ty _t,if t∈T ₂

·

·

·

u _t＝-Q _n,ty _t,if t∈T _n

MMTMV control criterion is

u _t＝-Q _ty _t

Wherein, MMTMV controller Q _tmeet

Q _t＝α ₁(t)Q _1,t+α ₂(t)Q _2,t+...+α _n(t)Q _n,t

In formula, α ₁(t), α ₂(t) ..., α _nt () is respectively the hybrid weight of corresponding SMTMV controller; For any t ∈ T, α _it () meets:

${\mathrm{\&alpha;}}_i\left(t\right)=0,t\&NotElement;T_i,i=\mathrm{1,2},...,n$

α ₁(t),α ₂(t),...,α _n(t)≥0

α ₁(t)+α ₂(t)+...+α _n(t)＝1

Hybrid weight α _i(t) (i=1,2 ..., n) meet following formula

${\mathrm{\&alpha;}}_i\left(t\right)=\frac{{\widetilde{\mathrm{\&alpha;}}}_i\left(t\right)}{{\widetilde{\mathrm{\&alpha;}}}_1\left(t\right)+{\widetilde{\mathrm{\&alpha;}}}_2\left(t\right)+...+{\widetilde{\mathrm{\&alpha;}}}_n\left(t\right)},i=\mathrm{1,2},...,n$

for not normalized weight factor, it meets continuously differentiable condition;

get following formula

Wherein, U _i, L _ibe respectively T _iinterval bound, T _i={ t:L _i≤ t≤U _i;

choose and should make meet continuously differentiable condition, α _i(t)>=0, α _i(t)=0, i=1,2 ..., n, α ₁(t)+α ₂(t)+...+α _n(t)=1.

5. become the performance estimating method of perturbed system when single argument is many in pyrolysis furnace as claimed in claim 1, it is characterized in that: described performance index definition is:

$\mathrm{\&eta;}=\frac{{\mathrm{\&delta;}}_{\mathrm{MMTMV}}^2}{{\mathrm{\&delta;}}_y^2}$

In formula, be single argument many time become output variance under perturbed system MMTMV control criterion, for actual single argument many time become output variance under perturbed system control criterion, Performance Evaluation index 0≤η≤1.