CN1945470A - Two freedom decoupling smith pre-evaluating control system of industrial multiple variable time lag process - Google Patents

Abstract

A kind of two degrees of freedom decouple Smith estimate control system with industrial multivariable process of multi-skewing, is composed of the n-dimensional set-point tracking controller, n-dimensional perturbation estimator, controlled process identification module and two multi-route signals mixer, in which, n is the dimension of the charged multivariable process. Through setting internal control loop between the process of input and output, and perturbation estimator, it effectively suppresses the loaded jamming signals in the process, to obtain a smooth process output. The given value response of system is adjusted in the way of open-loop control by the set-point tracking controller, so that the given value response and process loading jamming respond can be adjusted respectively and independently. Meanwhile, the impact of the system skewing to performance can be effectively compensated. The control system of the invention can achieve the notable decoupling between the output responses, maintain a good stability, and adapt the modeling errors of the charged process in a wide range and process parameter perturbation.

Description

Two-degree-of-freedom decoupling Smith prediction control system for industrial multivariable time-lag process

Technical Field

The invention relates to a system, in particular to a two-degree-of-freedom decoupling Smith pre-estimation control system of an industrial multivariable process with multiple time lags, and belongs to the technical field of industrial process control.

Background

With the development of industry, the production scale is more and more complex, and the requirement on control is higher and higher. To achieve efficient production of high quality products, many processes are configured as high-dimensional multivariable processes, which are one of the most common types of processes in industrial production. Due to the transmission and detection time lag of each output of the multivariable process and the cross-linking coupling effect between output channels, most of the developed univariate control methods are difficult to be used in the MIMO process, especially for the process with obvious time lag, the coupling effect between system outputs is very outstanding, and the system output response performance is seriously deteriorated. Therefore, how to implement multi-skew compensation and decoupling control is a current research and application problem.

For coupling between multivariable systems, the preferred method is to overcome by using appropriate matching between the controlled and manipulated variables, with the most representative method being the relative gain pairing method. However, for a system with a relatively serious relationship, a satisfactory decoupling effect cannot be achieved even by adopting the best variable pairing relationship, and at this time, a decoupling network (or referred to as a compensation network) must be added into the system for decoupling, so that a strongly coupled object becomes a non-coupled or weakly coupled control object. In current industrial practice, a static decoupler is usually adopted to reduce the coupling effect between each output of a multivariable control system, namely, a constant matrix decoupler is firstly arranged at the multi-path input end of a controlled process, the transmission matrix of the constant matrix decoupler is the inverse matrix of the steady-state gain transmission matrix of the controlled process, and then the control system is constructed and set by utilizing a developed and mature univariate control design method for the transmission matrix of the controlled process obtained by the augmentation. The main disadvantage is that the coupling effect of the dynamic response stage of the control system is not considered, so that the dynamic coupling of the system outputs of each path is still serious, and the control quality is not high. In terms of skew compensation, the estimated structure proposed by o.j.smith is representative. The structure has the greatest advantage that a time lag link is moved out of a closed loop, so that the control quality is greatly improved, but the structure has the greatest defect that the structure depends too much on an accurate mathematical model, when an estimation model and an actual object have errors, the control quality can be remarkably deteriorated and even dispersed, and the structure is very sensitive to external disturbance and poor in robustness. The conventional Smith predictive control structure is difficult to be truly applied in practice. To solve this problem, many scholars have proposed an improved method based on the conventional Smith predictive control structure.

The search of the prior art documents shows that the representative Smith prediction control scheme aiming at the industrial multivariable process with multiple time lags is international famous Wang Q-G.professor in the document Decoupling Smith predictor Design for multivariable systems with multiple time lags (decoupling Smith predictor Design of multivariable system with multiple time lags, published in chemical engineering Research and Design, Transactions of the Institute of chemical Engineers, chemical engineering society, 2000, 78, 565 and 572.) proposes a method of firstly decoupling a control object by a decoupler, then designing a plurality of single variable Smith prediction control systems aiming at the decoupled process, although the remarkably improved control effect is obtained, the decoupling process is usually processed by approximate control of model reduction, and the process is not completely decoupled by a diagonal matrix, the decoupling process cannot be matched with the proposed desired decoupling model accurately, and therefore ideal performance cannot be obtained. Moreover, the control structure cannot independently optimize the given value response of each system and the load disturbance response thereof, but at present, the industrial production practice has strong expectations for solving the problem. Recently, professor s.p. hung has proposed a Two-degree-of-Freedom Decoupling control system in document "Decoupling Multivariable control with Two Degrees of Freedom" published in industrial engineering Research publication 2006, 45, 3161 and 3173 ", and although Decoupling of the given value response and the load disturbance response of each road system is realized, the controller matrix design adopts a numerical method, and the required data computation amount is quite large, which is not convenient for practical popularization and application and on-line adjustment. It should be noted that other Smith predictive control schemes for industrial multivariable processes with multiple time lags are proposed under the assumption that the process model satisfies certain constraints, and thus cannot be applied to actual industrial production.

Disclosure of Invention

The invention aims to overcome the defects in the prior art, provides a two-degree-of-freedom decoupling Smith pre-estimation control system of an industrial multivariable time-lag process, so as to effectively compensate the influence of the time-lag of a system on the performance of the system, simultaneously realize the obvious decoupling among all paths of output responses of a nominal system, realize the mutually independent adjustment of the set value response and the load interference response of all paths of systems, and realize an online single-parameterization adjustment controller, so as to ensure the simple and convenient operation, and can be widely applied to various different industrial multiple-input multiple-output production process.

The invention is realized by the following technical scheme, and the invention consists of the following parts: the system comprises an n-dimensional set point tracking controller, an n-dimensional disturbance estimator, a controlled process identification module and two multipath signal mixers. Where n is the dimension of the controlled multivariable process. A first multi-path signal mixer is provided at the n-dimensional input of the process to be controlled and has a set of n-dimensional positive input terminals, a set of n-dimensional negative input terminals and a set of n-dimensional output terminals, with the positive input terminals of the first multi-path signal mixer being connected to the n-dimensional output signal of the setpoint tracking controller, the negative input terminals of the first multi-path signal mixer being connected to the n-dimensional output signal of the disturbance estimator, and the negative input terminals of the first multi-path signal mixer being connected to the n-dimensional input. The second multi-path signal mixer is arranged at the n-dimensional output end of the controlled process and is provided with a group of n-dimensional positive polarity input ends, a group of n-dimensional negative polarity input ends and a group of n-dimensional output ends, wherein the group of positive polarity input ends of the multi-path signal mixer is connected with the n-dimensional output measuring signal of the controlled process, the group of negative polarity input ends of the multi-path signal mixer is connected with the n-dimensional output end of the controlled process identification module, and the other group of output ends of the multi-path signal mixer. The output signal of the set point tracking controller is divided into two paths, one path is sent to the positive polarity input end of the first multi-path signal mixer, and the other path is sent to the input end of the controlled process identification module.

The function of the set point tracking controller is to process and operate the system set value input signal and provide n-dimensional input energy required by the operation of the controlled process, so that the n-dimensional output of the controlled process meets the requirements of each set value. The function of the multi-channel signal mixer is to mix two sets of n-dimensional input signals into a set of n-dimensional output signals in the order of the input channels. The disturbance estimator has the function of processing and operating each path of output deviation signals of the detected controlled process, so that the n-dimensional input quantity of the controlled process is adjusted, and the purposes of eliminating system output deviation and inhibiting load interference signals are achieved. The executable structure of the disturbance estimator comprises: the device comprises an intermediate-stage controlled process identification module, a correction disturbance filter and an intermediate-stage multipath signal mixer. The middle-stage multipath signal mixer is arranged at the n-dimensional input end of the correction disturbance filter and is provided with two groups of n-dimensional positive input ends and one group of n-dimensional output ends, one group of positive input ends of the middle-stage multipath signal mixer is connected with the n-dimensional input signal of the disturbance estimator, the other group of positive input ends of the middle-stage multipath signal mixer is connected with the n-dimensional output signal of the middle-stage controlled process identification module, and the other group of output ends of the middle-stage multipath signal mixer is connected with the; the n-dimensional output of the modified disturbance filter is connected to the n-dimensional input of the process reference model.

During actual operation, n-dimensional multi-channel set value input signals of the control system are sequentially sent to the set point tracking controller according to process requirements, and are subjected to operation processing and amplification to provide n-dimensional multi-channel input energy required by the operation of a controlled n-dimensional multivariable process, so that the output of the n-dimensional control system meets the requirements of the n-dimensional set value input signals. When a load interference signal is mixed into the controlled process, the system output changes, so that the deviation between the actual system output and the output signal of the controlled process identification module is generated. The deviation signal is sent to the n-dimensional input end of the disturbance estimator, the disturbance estimator processes the deviation signal and generates an n-dimensional control output signal, and the output signal is sent to the n-dimensional input end of the controlled process to be adjusted, so that the system output change and fluctuation caused by the load interference signal are counteracted and balanced, and the purpose of gradually eliminating the system output deviation is achieved.

The set point tracking controller is designed based on the robust H2 optimal performance index, the optimal decoupling of the output response of the nominal system can be realized, the output response transfer function of the nominal system meets the diagonal form, and meanwhile, the corresponding diagonal elements in each row of controllers of the set point tracking controller and the output response transfer function of the nominal system are adjusted by the same adjusting parameter lambda_cjjThe setting can be adjusted in a monotonous way on line, so that the time domain response index of the corresponding j system output is represented by lambda_cjjMonotonously and quantitatively setting, and meanwhile, each row of controllers of the disturbance estimator is provided with the same regulating parameter lambda_fjjSetting, which can be adjusted monotonously on line, so that the load interference response output by the corresponding jth system is determined by lambda_fjjAnd monotonously and quantitatively adjusting.

The basic idea of the invention is: the system set value response adopts an open-loop control mode, and an industrial multivariable time-lag process is stabilized by adopting a low-order rational regular controller matrix on a forward input channel, so that the coupling effect between the controller matrix and a control closed loop for inhibiting load interference is avoided, namely, the complete decoupling between the system set value response and the load interference response is realized, and meanwhile, a set point tracking controller is designed by utilizing an internal model controller design method aiming at a 'no-time-lag' part in a controlled process model, so that the influence of the system time-lag on the system performance is effectively eliminated, and the set value response of a control system can be ensured to reach the optimal decoupling. The control inner ring for inhibiting the process load interference signal is arranged between the input end and the output end of the process load interference signal, the deviation between the output measurement signal of the actual process and the output signal of the controlled process identification module is used as the feedback regulation information quantity of the process load interference signal and is sent to a disturbance estimator arranged on a feedback channel of the control inner ring, and after the calculation processing of the disturbance estimator, the obtained estimation signal is sent to an input regulation device of the controlled process for regulation, thereby realizing the purposes of eliminating the output deviation of the system and inhibiting the load interference signal.

The two-degree-of-freedom decoupling Smith prediction control system provided by the invention has the following outstanding advantages: 1. the influence of system time lag on system output can be effectively compensated, so that the system performance is obviously improved; 2. the method can realize the obvious decoupling between the output responses of the nominal system and respectively optimize the given value response and the load interference response of each output of the control system; 3. each column of sub-controllers in the set point tracking controller matrix and the disturbance estimator matrix are set by a single parameter and are set by the same parameter, so that the on-line monotonous quantitative adjustment of the tracking performance of the set point of the system, the nominal performance of the system and the disturbance resistance performance of the system can be realized; 4. the control system can ensure good robust stability, is insensitive to process parameter change, and can adapt to the modeling error of the controlled process and the perturbation of the process parameter in a larger range. Therefore, the two-degree-of-freedom decoupling Smith estimation control system provided by the invention has obvious superiority and practicability, and can show an advanced control effect in actual industrial application.

Drawings

FIG. 1 is a block schematic diagram of a two-degree-of-freedom decoupling Smith prediction control system according to the invention.

In FIG. 1, G(s) is an n-dimensional controlled multivariable multi-lag process, C(s) is an n-dimensional setpoint tracking controller, F(s) is an n-dimensional disturbance estimator, G (G) is a time-lag estimator, and_m(s) is a controlled process identification module, the circle node in the figure is a multipath signal mixer, r is an n-dimensional system given value input signal, y is an n-dimensional system output, u is an n-dimensional output signal of C(s),

the n-dimensional output signal of F(s), d is a load interference signal, and v is a deviation signal between the n-dimensional output measurement signal of the actual controlled process and the n-dimensional output of the controlled process identification module.

Fig. 2 is a schematic diagram of an executable structure of the disturbance estimator proposed in the present invention.

In FIG. 2, v is the difference between the actual measured n-dimensional output signal of the controlled process and the n-dimensional output of the controlled process identification moduleThe signal of the deviation is used to determine,is referred to as a disturbance estimation signal, G_m(s) means intermediate stage controlled process identification module, G_mo(s)F_fo(s) is a modified disturbance filter, F_fo(s) refers to the perturbation filter.

Fig. 3 is a schematic diagram of the output closed loop response of the rectifying tower object under the nominal condition for two unit step given value input signals and an inverse step load interference signal with the amplitude of 0.1.

Where fig. 3(a) shows a response curve for a first process output and fig. 3(b) shows a response curve for a second process output. The solid line represents the invention and the dotted line represents the Hung process.

Fig. 4 is a schematic diagram of output closed loop response of a rectifying tower object to two unit step given value input signals and an inverse step load interference signal with an amplitude of 0.1 under a multiplicative uncertainty condition.

Where fig. 4(a) shows a response curve for a first process output and fig. 4(b) shows a response curve for a second process output. The solid line represents the output response curve of the system without the controller parameter being adjusted in the presence of multiplicative input uncertainty, and the dotted line represents the output response curve of the system with the controller parameter being adjusted in the presence of multiplicative input uncertainty.

Detailed Description

The embodiments of the present invention will be described in detail below with reference to the accompanying drawings: the present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiments.

The decoupling Smith estimation control system shown in fig. 1 in this embodiment is composed of the following parts: controlled multivariable multi-lag process G(s), n-dimensional set point withA tracking controller C(s), an n-dimensional disturbance estimator F(s), a controlled process identification module G_mAnd two multipath signal mixers (circle nodes in the figure), wherein the first multipath signal mixer is arranged at the n-dimensional input end of the controlled process G(s) and comprises a group of n-dimensional positive polarity input ends, a group of n-dimensional negative polarity input ends and a group of n-dimensional output ends, the positive polarity input ends of the multipath signal mixers are connected with the n-dimensional output signals of the set point tracking controller C(s), the negative polarity input ends of the multipath signal mixers are connected with the n-dimensional output signals of the disturbance estimator F(s), and the other group of output ends of the multipath signal mixers are connected with the n-dimensional input end of the controlled process G(s). The second multi-path signal mixer is arranged at the n-dimensional output end of the controlled process G(s) and is provided with a group of n-dimensional positive polarity input ends, a group of n-dimensional negative polarity input ends and a group of n-dimensional output ends, wherein one group of positive polarity input ends is connected with the n-dimensional output measuring signals of the controlled process G(s), and one group of negative polarity input ends is connected with the controlled process identification module G_mAn n-dimensional output and a further set of outputs are connected to an n-dimensional input of a disturbance estimator matrix F(s). The output signal of the set point tracking controller C(s) is divided into two paths, one path is sent to the positive polarity input end of the first multi-path signal mixer, and the other path is sent to the controlled process identification module G_m(s) input terminal. The executable structure of the disturbance estimator comprises: middle-stage controlled process identification module G_m(s) modifying the perturbation filter G_mo(s)F_fo(s) and an intermediate stage multi-channel signal mixer. The intermediate-stage multi-path signal mixer is arranged in the correction disturbance filter G_mo(s)F_fo(s) having two sets of n-dimensional positive input terminals and one set of n-dimensional output terminals, one set of positive input terminals being connected to the n-dimensional input signals of the disturbance estimator F(s), the other set of positive input terminals being connected to the intermediate stage controlled process identification module G_m(s) n-dimensional output signal, another group of output terminals connected to the modified disturbance filter G_mo(s)F_fo(s) an n-dimensional input; modified disturbance filter G_mo(s)F_foThe n-dimensional output of(s) is connected to the intermediate stage process reference model G_mAn n-dimensional input of(s).

Actually carrying out the embodimentWhen the two-degree-of-freedom decoupling Smith estimation control system is used, firstly, a given value input signal r of the control system is sent to a set point tracking controller C(s), the set point tracking controller C(s) amplifies and smoothes the given value input signal r, and input energy u required by the operation of a controlled multivariable multi-time-lag process G(s) is provided, so that the output of the controlled multivariable multi-time-lag process G(s) meets the requirement of the given value signal r. The output signal u of the set point tracking controller C(s) is divided into two paths, one path is sent to the actual controlled process G(s), and the other path is sent to the controlled process identification module G_m(s) input terminal. At the same time, the detection signal of output y of actual controlled process G(s) and the controlled process identification module G_mAnd(s) sending the output signal into a second multi-path signal mixer for difference operation, sending the deviation value signal v into a disturbance estimator F(s) of a control inner ring, processing and amplifying the deviation value signal v by the disturbance estimator F(s) of the control inner ring, and sending the deviation value signal v into the input end of the actual controlled process G(s) in a negative feedback mode, thereby adjusting the input control quantity u of the actual controlled process G(s) and achieving the purpose of eliminating the mixed load interference signal d to output the controlled process P1.

In practice, the controlled industrial multivariable time-lag process is generally described by the following frequency domain mathematical expression form

Wherein <math> <mrow> <msub> <mi>g</mi> <mi>ij</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <msub> <mi>g</mi> <mi>oij</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <msup> <mi>e</mi> <mrow> <msub> <mrow> <mo>-</mo> <mi>&theta;</mi> </mrow> <mi>ij</mi> </msub> <mi>s</mi> </mrow> </msup> <mo>,</mo> </mrow> </math> It refers to the transfer function from the ith input to the jth output of the controlled process, g_0ij(s) is the rational transfer function part of its stability regularity, θ_ijIs its corresponding process transmission time lag, set point tracking controller is given belowC(s), the design formula of the disturbance estimator F(s):

(1) firstly, decomposing a transfer function matrix identification model of an industrial multivariable process with multiple time lags, wherein the decomposition form is as follows:

G(s)＝G_D(s)G_mo(s)＝G_A(s)G_N(s)G_mo(s) (2)

wherein,

<math> <mrow> <msub> <mi>G</mi> <mi>A</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>diag</mi> <mo>{</mo> <msup> <mi>e</mi> <mrow> <mo>-</mo> <msub> <mi>&theta;</mi> <mi>jj</mi> </msub> <mi>s</mi> </mrow> </msup> <mo>}</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein theta is_jjIs taken as G^-1The largest estimate in column j of(s).

<math> <mrow> <mrow> <msub> <mi>G</mi> <mi>N</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>diag</mi> <mo>{</mo> <munderover> <mrow> <mi>&Pi;</mi> <mo></mo> </mrow> <mrow> <mi>r</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>r</mi> <mn>1</mn> </msub> </munderover> <msup> <mrow> <mo>(</mo> <mfrac> <mrow> <mo>-</mo> <mi>s</mi> <mo>+</mo> <msub> <mi>z</mi> <mi>rj</mi> </msub> </mrow> <mrow> <mi>s</mi> <mo>+</mo> <msub> <mi>z</mi> <mi>rj</mi> </msub> </mrow> </mfrac> <mo>)</mo> </mrow> <msub> <mi>k</mi> <mi>rj</mi> </msub> </msup> <mo>}</mo> </mrow> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> </mrow> </math>

Wherein z is_rjIs G^-1Unstable pole, k, in column j of(s)_ijIs G_O ^-1Unstable pole z in j column of(s)_rjThe maximum number of (2).

G_D(s)＝G_A(s)G_N(s) (5)

<math> <mrow> <msub> <mi>G</mi> <mi>mo</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <msubsup> <mi>G</mi> <mi>D</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mi>G</mi> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>6</mn> <mo>)</mo> </mrow> </mrow> </math>

(2) The perturbation filter is designed to be of the form:

<math> <mrow> <msub> <mi>F</mi> <mi>fo</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>diag</mi> <mo>{</mo> <mfrac> <mn>1</mn> <msup> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mi>fjj</mi> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <msub> <mi>n</mi> <mi>j</mi> </msub> </msup> </mfrac> <mo>}</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>7</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein λ is_fjjThe system is an adjustable parameter and is used for adjusting the output of the jth system to achieve the actual required disturbance resistance.

(3) From steps (1) and (2), the disturbance estimator f(s) is designed in the form:

$F\left(s\right)={(I-G\left(s\right)G_{\mathrm{mo}}^{-1}\left(s\right)F_o\left(s\right))}^{-1}{G}_{\mathrm{mo}}^{-1}\left(s\right)F_o\left(s\right)$

<math> <mrow> <mo>=</mo> <msup> <mrow> <mo>(</mo> <mi>I</mi> <mo>-</mo> <msub> <mi>G</mi> <mi>D</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <msub> <mi>F</mi> <mi>o</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msup> <msubsup> <mi>G</mi> <mi>mo</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <msub> <mi>F</mi> <mi>o</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>8</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein n is_jRepresents G_mo ^-1(s) maximum relative order. As can be seen from expression (8), the proposed disturbance estimator can be conveniently implemented using the structure shown in fig. 2. It is to be noted that if G is_mo ^-1(s)F_o(s) cannot be physically implemented and can be approximated using rational approximation techniques, see the following analysis.

(4) The setpoint tracking filter is designed to be of the form:

<math> <mrow> <msub> <mi>F</mi> <mi>co</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>diag</mi> <mo>{</mo> <mfrac> <mn>1</mn> <msup> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mi>cjj</mi> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <msub> <mi>n</mi> <mi>j</mi> </msub> </msup> </mfrac> <mo>}</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>9</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein λ is_cjjThe system is an adjustable parameter and is used for adjusting the output of the jth system to reach the actual required set point tracking performance.

(5) From steps (1) and (5), the setpoint tracking controller c(s) is designed in the form of:

<math> <mrow> <mi>C</mi> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <msubsup> <mi>G</mi> <mi>mo</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <msub> <mi>F</mi> <mi>co</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>10</mn> <mo>)</mo> </mrow> </mrow> </math>

in the above design process, there is a special case if G is after the object transfer function matrix decomposition_moThe(s) still contains time lag terms, which result in that the designed controller is in an irrational form and cannot be physically realized, and the controller can be reasonably approximated by using an approximation technology, and the specific steps are as follows: first, for G_mo ^-1(s) decomposing in the form:

<math> <mrow> <msubsup> <mi>G</mi> <mi>mo</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <msubsup> <mover> <mi>G</mi> <mo>~</mo> </mover> <mi>mo</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <msub> <mi>G</mi> <mi>r</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>11</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein G is_r(s) is G_mo ^-1(s) decomposable carbon moieties which may be directly according to G_mo ^-1The expression for(s) is obtained. Then, toA rational approximation is made, the form of the approximation being:

<math> <mrow> <msubsup> <mover> <mi>G</mi> <mo>~</mo> </mover> <mi>mo</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>0</mn> </mrow> <mi>U</mi> </munderover> <msub> <mi>&alpha;</mi> <mi>i</mi> </msub> <msup> <mi>s</mi> <mi>i</mi> </msup> </mrow> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>0</mn> </mrow> <mi>V</mi> </munderover> <msub> <mi>&beta;</mi> <mi>j</mi> </msub> <msup> <mi>s</mi> <mi>j</mi> </msup> </mrow> </mfrac> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>12</mn> <mo>)</mo> </mrow> </mrow> </math>

where U and V are parameters specified by the user according to the design specifications. The larger the values of U and V are, the more the expressionThe smaller the error of this rational approximation, the better the system performance. Typically U and V can be taken to be 1 or 2. Parameter alpha_iAnd beta_jThis can be conveniently determined according to the following two equations,

<math> <mrow> <mfenced open='[' close=']'> <mtable> <mtr> <mtd> <msub> <mi>&alpha;</mi> <mn>0</mn> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>&alpha;</mi> <mn>1</mn> </msub> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>&alpha;</mi> <mi>U</mi> </msub> </mtd> </mtr> </mtable> </mfenced> <mo>=</mo> <mfenced open='[' close=']'> <mtable> <mtr> <mtd> <msub> <mi>d</mi> <mn>0</mn> </msub> </mtd> <mtd> <mn>0</mn> </mtd> <mtd> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> </mtd> <mtd> <mn>0</mn> </mtd> </mtr> <mtr> <mtd> <msub> <mi>d</mi> <mn>1</mn> </msub> </mtd> <mtd> <msub> <mi>d</mi> <mn>0</mn> </msub> </mtd> <mtd> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> </mtd> <mtd> <mn>0</mn> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> <mtd> <mo>&CenterDot;</mo> </mtd> <mtd> </mtd> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> <mtd> <mo>&CenterDot;</mo> </mtd> <mtd> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> </mtd> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> <mtd> <mo>&CenterDot;</mo> </mtd> <mtd> </mtd> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>d</mi> <mi>U</mi> </msub> </mtd> <mtd> <msub> <mi>d</mi> <mrow> <mi>U</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> </mtd> <mtd> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> </mtd> <mtd> <msub> <mi>d</mi> <mrow> <mi>U</mi> <mo>-</mo> <mi>V</mi> </mrow> </msub> </mtd> </mtr> </mtable> </mfenced> <mfenced open='[' close=']'> <mtable> <mtr> <mtd> <msub> <mi>&beta;</mi> <mn>0</mn> </msub> </mtd> </mtr> <mtr> <mtd> <msub> <mi>&beta;</mi> <mn>1</mn> </msub> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <mo>&CenterDot;</mo> </mtd> </mtr> <mtr> <mtd> <msub> <mi>&beta;</mi> <mi>V</mi> </msub> </mtd> </mtr> </mtable> </mfenced> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>13</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein

<math> <mrow> <msub> <mi>d</mi> <mi>k</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mi>k</mi> <mo>!</mo> </mrow> </mfrac> <munder> <mi>lim</mi> <mrow> <mi>s</mi> <mo>&RightArrow;</mo> <mn>0</mn> </mrow> </munder> <mfrac> <mrow> <msup> <mi>d</mi> <mi>k</mi> </msup> <msubsup> <mover> <mi>G</mi> <mo>~</mo> </mover> <mi>mo</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> </mrow> <msup> <mi>ds</mi> <mi>k</mi> </msup> </mfrac> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>15</mn> <mo>)</mo> </mrow> </mrow> </math>

<math> <mrow> <msub> <mi>&beta;</mi> <mi>o</mi> </msub> <mo>=</mo> <mfenced open='{' close=''> <mtable> <mtr> <mtd> <mn>1</mn> <mo>,</mo> <mrow> <mo>(</mo> <msub> <mi>&beta;</mi> <mi>j</mi> </msub> <mo>&GreaterEqual;</mo> <mn>0</mn> <mo>)</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mo>-</mo> <mn>1</mn> <mo>,</mo> <mrow> <mo>(</mo> <msub> <mi>&beta;</mi> <mi>j</mi> </msub> <mo>&lt;</mo> <mn>0</mn> <mo>)</mo> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mo>&CenterDot;</mo> <mrow> <mo>(</mo> <mn>16</mn> <mo>)</mo> </mrow> </mrow> </math>

It should be noted that the design formulas of the set point tracking controller and the disturbance estimator can be conveniently realized on an industrial personal computer, a single chip microcomputer and the like in a digital discretization mode, and the sampling time can be generally between 0.01 and 0.1 second. Each column of controllers of the setpoint tracking controller matrix designed in this embodiment is set by the same adjusting parameter, and each column of controllers of the disturbance estimator matrix is also set by the same adjusting parameter. On-line setting of adjustable parameters lambda of C(s) and F(s)_cjjAnd λ_fjjThe rule of (1) is: can initially set lambda_cjjAnd λ_fjjAt (5-10) theta_jjWithin the range. Setting parameter lambda of C(s) is reduced_cjjThe corresponding system output response speed can be accelerated, and the nominal response performance of the control system is improved, but the correspondingly required output energy of the jth row of controllers is increased, and the energy required to be provided by the corresponding actuators is also increased, which tends to exceed the capacity range of the controllers, and when the unmodeled dynamic characteristics of the controlled process are encountered, the controllers are prone to show an over-excitation behavior, which is not favorable for the robust stability of the control system; conversely, increasing the tuning parameter λ_fjjThe corresponding system output response is slowed down, but the required output energy of the jth column controller is reduced, and the energy required by the corresponding actuator is also reduced, thereby being beneficial to improving the robust stability of the control system. Thus, the setting parameter lambda of the actual setpoint tracking controller is adjusted_cjjThere is a trade-off between the nominal performance of the system output response and the output capacity of each column of controllers and their actuators. Similarly, the tuning parameter λ of the actual tuning disturbance estimator_fjjThe output capacity of each row controller and its actuator should be between the load disturbance rejection performance and the robust stability of the control systemA trade-off is made.

Investigating a widely investigated chemical hydrocarbon fractionator process

$G\left(s\right)=\left[\begin{array}{cc}\frac{{12.8e}^{-s}}{16.7s+1}& \frac{{-18.9}^{-3s}}{21s+1}\\ \frac{{6.6e}^{-7s}}{10.9s+1}& \frac{{-19.4e}^{-3s}}{14.4s+1}\end{array}\right]$

By applying the decoupling Smith pre-estimation control structure, a control system is constructed according to a structural block diagram shown in the attached figure 1; and then designing and setting a controller: the first step, applying formulas (2) - (6) to obtain the decomposition factor, which is as follows:

$G_D\left(s\right)=\left[\begin{array}{cc}{e}^{-\mathrm{<figure-callout\; id="0"\; label="s"\; filenames="A20061011788100173.png"\; state="\{\{state\}\}">s</figure-callout>}}& 0\\ 0& e^{-3s}\end{array}\right]$

$G_{\mathrm{mo}}\left(s\right)=\left[\begin{array}{cc}\frac{12.8}{\mathrm{16,7}s+1}& \frac{{-18.9e}^{-2s}}{21s+1}\\ \frac{{6.6e}^{-3s}}{10.9s+1}& \frac{-19.4}{14.4s+1}\end{array}\right]$

secondly, applying a formula (7) to obtain a disturbance estimator:

$F\left(s\right)=\frac{F_o\left(s\right)}{1-G\left(s\right)F_o\left(s\right)}$

wherein

<math> <mrow> <msub> <mi>F</mi> <mi>o</mi> </msub> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mo>=</mo> <msup> <mi>D</mi> <mrow> <mo>-</mo> <mn>1</mn> </mrow> </msup> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> <mfenced open='[' close=']'> <mtable> <mtr> <mtd> <mfrac> <mrow> <mo>(</mo> <mn>16.7</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mn>12.8</mn> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>f</mi> <mn>11</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> <mtd> <mfrac> <mrow> <mn>18.9</mn> <mrow> <mo>(</mo> <mn>16.7</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mn>14.4</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> <mrow> <mrow> <mo>(</mo> <mn>21</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>f</mi> <mn>22</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> <msup> <mi>e</mi> <mrow> <mo>-</mo> <mn>2</mn> <mi>s</mi> </mrow> </msup> </mtd> </mtr> <mtr> <mtd> <mfrac> <mrow> <mo>-</mo> <mn>6.6</mn> <mrow> <mo>(</mo> <mn>16.7</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mn>14.4</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> <mrow> <mrow> <mo>(</mo> <mn>10.9</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>f</mi> <mn>11</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> <msup> <mi>e</mi> <mrow> <mo>-</mo> <mn>4</mn> <mi>s</mi> </mrow> </msup> </mtd> <mtd> <mfrac> <mrow> <mo>(</mo> <mn>14.4</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>19.4</mn> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>f</mi> <mn>22</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> </mtr> </mtable> </mfenced> </mrow> </math>

$D\left(s\right)=1-\frac{0.5023(14.4s+1)(16.7s+1)}{(21s+1)(10.9s+1)}{e}^{-6s}$

Applying equation (8) to obtain a setpoint tracking controller:

$C\left(s\right)=G^{-1}\left(s\right)T_s\left(s\right)$

<math> <mrow> <mo>=</mo> <mfrac> <mfenced open='[' close=']'> <mtable> <mtr> <mtd> <mfrac> <mrow> <mo>(</mo> <mn>16.7</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mn>12.8</mn> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>c</mi> <mn>1</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> <mtd> <mfrac> <mrow> <mn>18.9</mn> <mrow> <mo>(</mo> <mn>16.7</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mn>14.4</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> <mrow> <mrow> <mo>(</mo> <mn>21</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>c</mi> <mn>2</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> <msup> <mi>e</mi> <mrow> <mo>-</mo> <mn>2</mn> <mi>s</mi> </mrow> </msup> </mtd> </mtr> <mtr> <mtd> <mfrac> <mrow> <mo>-</mo> <mn>6.6</mn> <mrow> <mo>(</mo> <mn>16.7</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mn>14.4</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> <mrow> <mrow> <mo>(</mo> <mn>10.9</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>c</mi> <mn>1</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> <msup> <mi>e</mi> <mrow> <mo>-</mo> <mn>4</mn> <mi>s</mi> </mrow> </msup> </mtd> <mtd> <mfrac> <mrow> <mo>(</mo> <mn>14.4</mn> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> <mrow> <mo>-</mo> <mn>19.4</mn> <mrow> <mo>(</mo> <msub> <mi>&lambda;</mi> <mrow> <mi>c</mi> <mn>2</mn> </mrow> </msub> <mi>s</mi> <mo>+</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> </mtr> </mtable> </mfenced> <mrow> <mi>D</mi> <mrow> <mo>(</mo> <mi>s</mi> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow> </math>

from the above controller form, it can be seen that to implement the controller, a very complicated control structure must be adopted, so that a linear approximation technique is used to simplify the calculation and maintain the design requirements, and the irrational portion d(s) in the controller can be approximated as follows according to the approximation equations (9) - (14):

$D\left(s\right)=\frac{{74.9789\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="A20061011788100173.png,A20061011788100181.png"\; state="\{\{state\}\}">s</figure-callout>}}^2+16.066s+0.4977}{36.6518{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="A20061011788100173.png,A20061011788100181.png"\; state="\{\{state\}\}">s</figure-callout>}}^2+25.419s+1}$

the above process is reduced by an order of 2, and if the order selection is larger, the design accuracy is higher, but the complexity of the controller is larger.

And thirdly, constructing a closed-loop control system according to the closed-loop control structure diagram shown in the attached figure 1, wherein the disturbance estimator is constructed according to the figure 2. When the decoupling Smith prediction control system is actually operated, only n-dimensional multi-channel given value input signals are required to be sequentially sent to the set point tracking controller according to working requirements, the set point tracking controller can carry out operation processing and amplification on the n-dimensional multi-channel given value input signals and provide the n-dimensional input energy required by the working of the controlled multivariable time lag process G(s), and therefore the output of the n-dimensional multi-channel control system can respectively reach the n-dimensional given value tracking requirements. When a load interference signal is mixed in the controlled process G(s), the system output changes, the deviation generated by the change and the reference output signal provided by the controlled process identification module is sent to the n-dimensional input end of the disturbance estimator F(s), F(s) generates corresponding change and sends the n-dimensional disturbance estimation signal to the n-dimensional input end of the controlled process G(s) for regulation, so that the change and fluctuation of the system output caused by the load interference signal are gradually counteracted and balanced, and the aim of gradually eliminating the system output deviation is fulfilled.

And fourthly, online adjusting the set point tracking controller and the disturbance estimator, and observing the closed loop response of the system, thereby determining the optimal controller parameters. During simulation experiments, unit step input signals are added to the first path of input quantity at the moment when t is 0 seconds: r is₁1/s and the second input signal is r₂And (3) adding a unit-order input signal to the second input at the time when t is 100 seconds, and adding an inverse step input with the amplitude of 0.1 to the two controlled process input ends at the time when t is 200 seconds. To obtain the same response speed as that of the response curve obtained in the Hung method, λ is set here_c11＝3.8，λ_c22＝3.5，λ_f11＝2.0 and λ_f22The resulting closed loop response curve of the system is shown in fig. 3, 2.0. As can be seen from fig. 3, the two-degree-of-freedom decoupled Smith prediction control system (solid line) provided by the present embodiment achieves near complete decoupling between the output responses of the nominal systems. It can be seen that the given value response of the system output and the suppression of the load disturbance signalThe manufacturing performance is obviously superior to that of a two-degree-of-freedom prediction control system (point line) of the Hung method.

Now assume that there is a multiplicative input uncertainty Δ for the actual controlled process G(s)_IBiag { (s +0.3)/(s +1), (s +0.3)/(s +1) }. Where Δ_IIt can be approximately physically interpreted that two input regulator valves of a controlled process have up to 100% uncertainty in the high frequency band and nearly 30% uncertainty in the low frequency band operating range. The simulation experiment described above is performed under such severe process input uncertainty, and the computer simulation result of the process output response obtained by the tuning method of the controller according to the present embodiment is shown in fig. 4. As can be seen from fig. 4, the tuning method (solid line) of the controller provided in this embodiment can well ensure the robust stability of the given value response and the load disturbance response of the system. In addition, it can be seen that the monotonically increasing set point tracks the control parameter (λ) in the controller_c11＝6.8，λ_c226.5), the oscillation of the system setpoint response decreases, as shown by the dotted line in fig. 4, while the tuning parameter (λ) in the disturbance estimator controller is monotonically increased_f11＝6.2 and λ_f226.9), the anti-disturbance response speed of the process output becomes slower, as shown by the dotted line in fig. 4, but the robust stability margin of the system becomes larger.

It should be noted that various modifications can be made without departing from the scope of the present invention. The invention provides a design method of a set point tracking controller and a disturbance estimator aiming at a general industrial multivariable process identification model with multiple time lags, so the method is suitable for various multiple-input multiple-output production processes with time lags. The two-degree-of-freedom decoupling Smith estimation control system provided by the invention can be widely applied to the production process of the industries such as petrifaction, metallurgy, medicine, building materials, textile and the like.

Claims (3)

1. A two-degree-of-freedom decoupling Smith estimation control system for an industrial multivariable time-lag process comprises the following steps: the system comprises an n-dimensional set point tracking controller, an n-dimensional disturbance estimator, a controlled process identification module and two multipath signal mixers, and is characterized in that: wherein n is the dimension of the controlled multivariable process, the first multipath signal mixer is arranged at the n-dimensional input end of the controlled process and comprises a group of n-dimensional positive polarity input ends, a group of n-dimensional negative polarity input ends and a group of n-dimensional output ends, one group of positive polarity input ends is connected with the n-dimensional output signal of the setpoint tracking controller, one group of negative polarity input ends is connected with the n-dimensional output signal of the disturbance estimator, and the other group of output ends is connected with the n-dimensional input end of the controlled process; the second multi-channel signal mixer is arranged at the n-dimensional output end of the controlled process and is provided with a group of n-dimensional positive polarity input ends, a group of n-dimensional negative polarity input ends and a group of n-dimensional output ends, wherein the group of positive polarity input ends of the second multi-channel signal mixer is connected with the n-dimensional output end of the controlled process identification module, the group of negative polarity input ends of the second multi-channel signal mixer is connected with the n-dimensional output end of the controlled process identification module, the other group of output ends of the second multi-channel signal mixer is connected with the n-dimensional input end of the disturbance estimator matrix, the output signals of the set point tracking controller are divided into two paths; the set point tracking controller processes and calculates the system set value input signals and provides n-dimensional input energy required by the work of the controlled process, so that n-dimensional output of the controlled process meets the requirement of each set value, the disturbance estimator processes and calculates detected output deviation signals of each path of the controlled process, so that the n-dimensional input quantity of the controlled process is adjusted, the purposes of eliminating system output deviation and inhibiting load interference signals are achieved, and the multi-path signal mixer sequentially mixes two groups of n-dimensional input signals into one group of n-dimensional output signals according to input channels.

2. The two-degree-of-freedom decoupled Smith prediction control system for industrial multivariable time-lag process of claim 1, wherein: the executable structure of the disturbance estimator comprises: the middle-stage controlled process identification module, a correction disturbance filter and a middle-stage multipath signal mixer, wherein the middle-stage multipath signal mixer is arranged at an n-dimensional input end of the correction disturbance filter and is provided with two groups of n-dimensional positive polarity input ends and a group of n-dimensional output ends, one group of positive polarity input ends is connected with an n-dimensional input signal of a disturbance estimator, the other group of positive polarity input ends is connected with an n-dimensional output signal of the middle-stage controlled process identification module, and the other group of output ends is connected with an n-dimensional input end of the correction disturbance filter; the n-dimensional output of the modified disturbance filter is connected to the n-dimensional input of the process reference model.

3. The two-degree-of-freedom decoupled Smith prediction control system for industrial multivariable time-lag process of claim 1, wherein: the set point tracking controller is designed based on the optimal performance index of robust H2,

each column of controllers of the setpoint tracking controller contains the same regulating parameter lambda_cjjThe set point tracking performance for adjusting the output of the jth system to meet the actual requirement is achieved by adjusting lambda monotonically on line_cjjThe time domain response index output by the jth system can be quantitatively adjusted; each column controller of the disturbance estimator comprises the same regulating parameter lambda_fjjThe method is used for adjusting the output of the jth system to achieve the actual required disturbance resistance performance by adjusting lambda monotonically on line_fjjCan realize the quantitative regulation of the load interference response of the jth system output and the on-line setting of the adjustable parameter lambda_cjjAnd λ_fjjThe rule of (1) is: initial setting of lambda_cjjAnd λ_fjjAt (5-10) theta_jjWithin the range, the regulating parameter lambda is reduced_cjjThe corresponding system output response speed can be accelerated, the nominal response performance of the control system is improved, but the robustness and the stability of the control system are not facilitated; increasing the tuning parameter lambda_fjjThe corresponding system output response is slowed down but the robust stability of the control system is advantageously improved.

Images