CN102998973A - Multi-model self-adaptive controller of nonlinear system and control method - Google Patents

Abstract

The invention discloses a multi-model self-adaptive controller of a nonlinear system and a control method. A nonlinear robust indirect self-adaptive controller and a nonlinear neural network indirect self-adaptive controller are adopted to achieve control based on the fact that performance indexes are switched to an optimal controller at each sampling instant. Compared with a transitional nonlinear multi-model self-adaptive controller and a transitional control method, boundary of a nonlinear term of the nonlinear system is widened to zero-order close to boundedness, and adaptability of the multi-model self-adaptive controller can be effectively improved.

Description

A kind of multi-model Adaptive Control device and control method of nonlinear system

Technical field

The present invention relates to the controlled device of various reality in the every profession and trade production run, be specifically related to class multi-model Adaptive Control device and a control method thereof of the non-linear complax control system of a class zeroth order bounded.

Background technology

Because modern industry changes to technology-intensive type, occurred by miscellaneous subsystem and element forms and the Complex Industrial Systems process of internal relations complexity.This system often has the characteristics such as strong nonlinearity, fast time variant, model be uncertain.Although existing Nonlinear Processing method has some application, imperfection is high to the requirement of controlled device in theory, and does not have generality.

Because from characteristics such as a large amount of the unknowns of suitable parameter, method in classical control theory and the modern control theory be difficult to solve these characteristics, and adaptive control can be processed the control problem of uncertain system to a certain degree and be widely used in the control of nonlinear system.By to System Discrimination, adaptive control is the variation of compensation model order, parameter and input signal aspect automatically, can be controlled preferably effect.But for the time become very fast, parameter has the controlled device of saltus step, the poor effect of adaptive control system identification, thereby cause the dynamic property of system relatively poor, situation that transient error is excessive.Multi-model Adaptive Control develops on the basis of adaptive control, can effectively reduce the transient error of system.Therefore propose a kind of model structure of new nonlinear system, this model structure partly is comprised of linear segment and nonlinear terms and has more generality, for fast time variant, parameter saltus step object, adopts a plurality of identification models, obtains better control effect.But traditional multi-model Adaptive Control method requires the nonlinear terms part global bounded of controlled device, this condition restriction the application of multi-model Adaptive Control method in real system.

Summary of the invention

The present invention is directed to the technical matters that exists in the above-mentioned prior art, a kind of multi-model Adaptive Control device and control method of nonlinear system is provided.Relax the restrictive condition of nonlinear system, enlarge the usable range of multi-model Adaptive Control method.The method not only has been loosened to zeroth order near bounded with nonlinear terms global bounded condition, and has reduced the steady-state error of multi-model Adaptive Control system, has improved control accuracy.

For achieving the above object, the technical solution used in the present invention is:

A kind of multi-model Adaptive Control device of nonlinear system, this controller is by a non linear robust Indirect adaptive control device, nonlinear neural network adaptive controller and switching mechanism three parts form, wherein, a controller is non linear robust Indirect adaptive control device, another is nonlinear neural network Indirect adaptive control device, the input of controlled device is selected to produce between two controllers by switching mechanism, between the output of controlled device and two controllers a close loop negative feedback is arranged, be set to subtract each other relation between the model output of controlled device and two Indirect adaptive control devices, model error is used for the parameter of adjustment model and the weights of neural network.

Described non linear robust Indirect adaptive control device comprises a non linear robust adaptive model and a gamma controller, the non linear robust adaptive model is comprised of linear and nonlinear parts, linear segment after the linearization of linear segment correspondence system, non-linear partial is represented by the norm with the regression vector of coefficient, in order to the non-linear partial after the bucking-out system linearization.Regression vector is comprised of system's past input and output variable constantly.Adaptive parameter is the coefficient of linear regression vector, and adaptive law adopts the projection adaptive law.Gamma controller adopts leading controller of a step.

Described nonlinear neural network adaptive controller comprises a nonlinear neural network adaptive model and a nonlinear neural network adaptive controller.The nonlinear neural network adaptive model partly is comprised of linear segment and nonlinear neural network.The coefficient of linear segment can upgrade in any way as auto-adaptive parameter.Nonlinear neural network adopts the BP network, utilizes the error back propagation method to train.

In described switching mechanism, at first design performance index, these performance index comprise a cumulative errors part and a transient error part.In each control constantly, calculate the performance index of each controller, the controller that the selectivity index is less produces next control inputs constantly, can realize switching stably, and improve the transient performance of system.

The step that the control method of the multi-model Adaptive Control device of above-mentioned nonlinear system comprises is as follows:

S1: system initialization: the parameter of random initializtion non linear robust adaptive model, the parameter of random initializtion nonlinear neural network model and the weights of neural network, these parameters can be determined by certain priori;

In the S2:k=0 moment, object is output as 0; K ≠ 0 moment, object is output as the real output value of system, makes the poor departure e that obtains system with default value _cActual output is made the poor model error e that obtains with the output of non linear robust adaptive model ₁, make the poor model error e that obtains with the nonlinear neural network model ₂

S3: with departure e _cAs the input of non linear robust adaptive controller and nonlinear neural network adaptive controller, produce respectively controlled quentity controlled variable u by two controllers ₁And u ₂

S4: according to model error e ₁And e ₂Come calculation of performance indicators C ₁And C ₂Value, the input u that the less controller of selectivity desired value produces _i, as the control inputs u of controlled device and two models;

S5: utilize model error e ₁And e ₂Upgrade respectively parameter and the weights of non linear robust adaptive model and nonlinear neural network adaptive model;

S6: forward step S2 to.

Compared with prior art, beneficial effect of the present invention is as follows:

Can find out from multi-model Adaptive Control method of the present invention, comprise a non linear robust adaptive model in the non linear robust adaptive controller, this model increases an adaptive nonlinear compensation item on the basis of former linear adaption model, thereby the restrictive condition of the nonlinear terms of controlled device has been loosened to zeroth order near bounded by Bounded Linear, has widened greatly the scope of application of multi-model Adaptive Control device.By the non linear robust adaptive controller separately the nonlinear system of control can be proved to be and have stability and convergence.

A nonlinear neural network model that comprises system in the nonlinear neural network adaptive controller, omnipotent approximation theorem according to neural network, this model can approach with arbitrary accuracy the true output of system, and this has higher precision so that control method of the present invention is compared with traditional multi-model Adaptive Control method.Leading control thought of a step is adopted in the controller design, and calculated amount is little, can improve the computing velocity of system.

Design by switching mechanism, so that the controller of system switches between non linear robust adaptive controller and nonlinear neural network adaptive controller, can the selectivity desired value less controller can reduce the transient error of system like this as the control inputs of current system.The accumulation item that comprises an error in the performance index can prevent that locking system frequently switches between two controllers, and so that system's output is comparatively level and smooth.Because the non linear robust adaptive control system has stability, the present invention adopts non linear robust adaptive controller and nonlinear neural network adaptive controller to switch, and can prove that multi-model Adaptive Control utensil of the present invention has stability and convergence.

Description of drawings

Fig. 1 closes the feedback control system block scheme for what the present invention designed the multi-model neural network adaptive controller;

Fig. 2 is non linear robust Indirect adaptive control device structured flowchart;

Fig. 3 is nonlinear neural network Indirect adaptive control device structured flowchart;

Fig. 4 is the structured flowchart of neural network;

Fig. 5 is the process flow diagram of switching mechanism;

Fig. 6 (1) and Fig. 6 (2) are respectively curve of output and the input curve of controller of the present invention.

Concrete implementation method

Below in conjunction with accompanying drawing and example, further specify the present invention.

As shown in Figure 1, in the designed controller of multi-model Adaptive Control method of the present invention, formed by a non linear robust adaptive controller, a nonlinear neural network adaptive controller and a switching mechanism.Among the figure, r (t+1) is the tracking reference signal of system, and u (t) is the input of controlled device, and y (t+1) is the output of controlled device.Non linear robust Indirect adaptive control device comprises the non linear robust adaptive model With the non linear robust adaptive controller

It is controller

Output,

It is model

Output.Nonlinear neural network Indirect adaptive control device comprises the nonlinear neural network adaptive model

With the nonlinear neural network adaptive controller

It is controller

Output, It is model

Output.U (t) is existed by switching mechanism

With

Between select to produce.

The present invention is directed to the nonlinear discrete time system of following structure

$\mathrm{\Σ}:\left\{\begin{array}{c}x(t+1)=F(x\left(t\right),u\left(t\right))\\ y\left(t\right)=G\left(x\left(t\right)\right)\end{array}\right.---\left(1\right)$

In the formula, u (t), y (t) ∈ R is respectively the input and output of system, x (t) ∈ R ⁿBe n dimension state vector, F (), G () are smooth nonlinear functions.

System can be represented by following nonlinear model in a neighborhood of initial point:

$y(t+1)=\underset{i=0}{\overset{n_a-1}{\mathrm{\Σ}}}{a}_{i}y(t-i)+\underset{j=0}{\overset{n_b}{\mathrm{\Σ}}}{b}_{j}u(t-j)+f\left(w\left(t\right)\right)---\left(2\right)$

In the formula, a _i, i=0 ..., n _a-1; b _j, j=0 ..., n _bBe the parameter of system's the unknown, n _a, n _bBe the order of system, w (t)=[y (t) ..., y (t-n _a+ 1), u (t) ..., u (t-n _b)] ^TThe regression vector that is formed by system data.

Said system (2) is carried out following hypothesis:

A1. the order n of system _aAnd n _bKnown;

A2. parameter a _i, i=0 ..., n _a-1, b _j, j=0 ..., n _b, compact among the Ω at one and to change;

A3. system has overall uniform asympotically stable zero dynamic system, so that the growth rate of arbitrary list entries of system is no more than the growth rate of its corresponding output sequence;

A4. have known constant 0≤μ＜∞ so that function f (w (t)) for

Be zeroth order near bounded, namely satisfy

Wherein $\stackrel{\‾}{w}\left(t\right)={[y\left(t\right),...,y(t-n_a+1),u(t-1),...,u(t-n_b)]}^T,$ $g\left(\stackrel{\‾}{w}\left(t\right)\right)=\mathrm{\λ}\left|\right|\stackrel{\‾}{w}\left(t\right)\left|\right|$ Be nonlinear function, λ is unknown arbitrary constant.

Non linear robust Indirect adaptive control structural drawing shown in Figure 2.This controller comprises that non linear robust adaptive model and nonlinear adaptive controller two parts form.At first, the non linear robust adaptive model of design controlled device is designated as

$\underset{1}{y}(t+1)={\underset{1}{\mathrm{\θ}}}^T\left(t\right)w\left(t\right)+\underset{1}{\mathrm{\λ}}\left(t\right)\left|\right|\stackrel{\‾}{w}\left(t\right)\left|\right|$

(3)

$={\underset{1}{\mathrm{\δ}}}^T\left(t\right)\mathrm{\ψ}\left(t\right)$

In the formula $\underset{1}{\mathrm{\&theta;}}\left(t\right)={[\underset{1}{a_0}\left(t\right),...,\underset{1}{a_{n_a-1}}\left(t\right),\underset{1}{{\mathrm{<figure-callout\; id="0"\; label="b"\; filenames="HDA00002482983200031.png,HDA00002482983200032.png"\; state="\{\{state\}\}">b</figure-callout>}}_0}\left(t\right),...,\underset{1}{b_{n_b}}\left(t\right)]}^T$ With

It is model

In t parameter constantly, order

Can obtain (2) formula.At any system time t, by model

The estimated value that provides system's output is

Actual value by estimated value and system's output is poor, can obtain the model error of non linear robust adaptive model

Namely

According to model error

Adopt the following robust adaptive identification algorithm with the dead band to come model parameter is upgraded:

$\underset{1}{\overset{^}{\mathrm{\&delta;}}}\left(t\right)=\underset{1}{\overset{^}{\mathrm{\&delta;}}}(t-1)+\frac{\underset{1}{h}\left(t\right)\underset{1}{e}\left(t\right)\mathrm{\&psi;}(t-1)}{1+{\left|\right|w(t-1)\left|\right|}^2}---\left(4\right)$

In the formula, $\underset{1}{h}\left(t\right)=\left\{\begin{array}{cc}\frac{1}{2}& \left|\underset{1}{e}\left(t\right)\right|\&GreaterEqual;2\mathrm{\&mu;}\\ 0& \mathrm{otherwise}\end{array}\right..$

At each system time t, according to leading control thought of a step, by the parameter of non linear robust adaptive model

Design the nonlinear adaptive controller

$\underset{1}{u}\left(t\right)=\frac{1}{\underset{1}{{\hat{b}}_0\left(t\right)}}[r(t+1)-{\underset{1}{\overset{^}{\stackrel{\&OverBar;}{\mathrm{\&delta;}}}}}^T\left(t\right)\stackrel{\&OverBar;}{\mathrm{\&psi;}}\left(t\right)]---\left(5\right)$

In the formula, $\underset{1}{\overset{^}{\stackrel{\&OverBar;}{\mathrm{\&delta;}}}}\left(t\right)={[{\underset{1}{\overset{^}{\stackrel{\&OverBar;}{\mathrm{\&theta;}}}}}^T\left(t\right),\underset{1}{\overset{^}{\mathrm{\&lambda;}}}\left(t\right)]}^T,$ $\underset{1}{\overset{^}{\stackrel{\&OverBar;}{\mathrm{\&theta;}}}}\left(t\right)={[\underset{1}{{\hat{a}}_0}\left(t\right),...,\underset{1}{{\hat{a}}_{n_a-1}}\left(t\right),\underset{1}{{\hat{b}}_1}\left(t\right),...,\underset{1}{{\hat{b}}_{n_b}}\left(t\right)]}^T,$ $\stackrel{\&OverBar;}{\mathrm{\&psi;}}\left(t\right)={[{\stackrel{\&OverBar;}{w}}^T\left(t\right),|\left|\stackrel{\&OverBar;}{w}\left(t\right)\right|\left|\right]}^T.$

Nonlinear neural network Indirect adaptive control device structural drawing shown in Figure 3.This controller comprises nonlinear neural network adaptive model and nonlinear neural network controller two parts.

Give non-linear controlled device design a nonlinear neural network identification model

$\underset{2}{y}(t+1)={\underset{2}{\mathrm{\&theta;}}}^T\left(t\right)w\left(t\right)+\underset{2}{f}(W\left(t\right),\stackrel{\&OverBar;}{w}\left(t\right))---\left(6\right)$

In the formula, $\underset{2}{\mathrm{\&theta;}}\left(t\right)={[\underset{2}{a_0}\left(t\right),...,\underset{2}{a_{n_a-1}}\left(t\right),\underset{2}{{\mathrm{<figure-callout\; id="0"\; label="b"\; filenames="HDA00002482983200031.png,HDA00002482983200032.png"\; state="\{\{state\}\}">b</figure-callout>}}_0}\left(t\right),...,\underset{2}{b_{n_b}}\left(t\right)]}^T$ It is model

Parameter, Be that nonlinear function with the bounded of Neural Networks Representation approaches, W (t) is the weight coefficient of neural network, coefficient

And W (t) is predefined compacting among the S.

That t is constantly right

Debate knowledge,

Upgrade in the following manner:

$\underset{2}{\overset{^}{\mathrm{\&theta;}}}\left(t\right)=\underset{2}{\overset{^}{\mathrm{\&theta;}}}(t-1)+\frac{\underset{2}{h}\left(t\right)\underset{2}{e}\left(t\right)w(t-1)}{1+{\left|\right|w(t-1)\left|\right|}^2}---\left(7\right)$

In the formula, $\underset{2}{e}\left(t\right)=y\left(t\right)-\underset{2}{\overset{^}{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="HDA00002482983200011.png,HDA00002482983200022.png"\; state="\{\{state\}\}">y</figure-callout>}}}\left(t\right),$ $\underset{2}{h}\left(t\right)=\left\{\begin{array}{cc}\frac{1}{2}& \left|\underset{2}{e}\left(t\right)\right|\&GreaterEqual;2\mathrm{\&mu;}\\ 0& \mathrm{otherwise}\end{array}\right..$ If $\underset{2}{{\hat{b}}_0}\left(t\right)<b_{\min },$ Then order $\underset{2}{{\hat{b}}_0}\left(t\right)=b_{\min }.$

According to the nonlinear neural network model The Nonlinear control law that can obtain system is:

$\underset{2}{u}\left(t\right)=\frac{1}{\underset{2}{{\hat{b}}_0}\left(t\right)}[r(t+1)-{\underset{2}{\overset{^}{\stackrel{\&OverBar;}{\mathrm{\&theta;}}}}}^T\left(t\right)\stackrel{\&OverBar;}{w}\left(t\right)-\underset{2}{f}(W\left(t\right),\stackrel{\&OverBar;}{w}\left(t\right))]---\left(8\right)$

In the formula, $\underset{2}{\overset{^}{\stackrel{\&OverBar;}{\mathrm{\&theta;}}}}\left(t\right)={[\underset{2}{{\hat{a}}_0}\left(t\right),...,\underset{2}{{\hat{a}}_{n_a-1}}\left(t\right),\underset{2}{{\hat{b}}_1}\left(t\right),...,\underset{2}{{\hat{b}}_{n_b}}\left(t\right)]}^T.$

The structural drawing of neural network shown in Figure 4, this neural network are to have three layers of neuronic BP neural network, comprise input layer, hidden layer and output layer.Each neuron between the levels connects entirely, with not connecting between the layer neuron.Input layer is to the connection weight l in middle layer _Ij, i=1,2 ..., n _a+ n _b-2, j=1,2 ..., p; Hidden layer is to the connection weight v of output layer _J1, j=1,2 ..., p; The output threshold tau of each unit of hidden layer _j, j=1,2 ..., p; The output threshold value of output layer unit is γ ₁Parameter k=1,2 ..., m.Be input as $\stackrel{\&OverBar;}{w}\left(t\right)={[y\left(t\right),...,y(t-n_a+1),u(t-1),...,u(t-n_b)]}^T.$

Each neuronic input s of hidden layer _jFor: $s_j=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{l}_{\mathrm{ij}}{\stackrel{\&OverBar;}{w}}_i-{\mathrm{\&tau;}}_j,j=\mathrm{1,2},...,p.$ Use s _jCalculate each neuronic output b of hidden layer by transport function _jFor: b _j=g (s _j), j=1,2 ..., p. utilizes the output b of hidden layer _j, connection weight v _J1And threshold gamma ₁Calculate the neuronic output of output layer L _tFor:

Then calculate the neuronic response of output layer by transport function For:

Utilize connection weight v _J1, error

Output b with hidden layer _j, calculate each neuronic error d of hidden layer _j(t).

$d_j\left(t\right)=\left[\underset{2}{e}\left(t\right)v_{j1}\right]b_j(1-b_j)$

Utilize output error

With each neuronic output b of hidden layer _jRevise connection weight v _J1And threshold gamma ₁:

${\mathrm{<figure-callout\; id="1"\; label="v\; j"\; filenames="HDA00002482983200011.png,HDA00002482983200012.png"\; state="\{\{state\}\}">v</figure-callout>}}_j={\mathrm{<figure-callout\; id="1"\; label="v\; j"\; filenames="HDA00002482983200011.png,HDA00002482983200012.png"\; state="\{\{state\}\}">v</figure-callout>}}_j+\mathrm{\&kappa;}\underset{2}{e}\left(t\right)b_j$

${\mathrm{\&gamma;}}_1={\mathrm{\&gamma;}}_1+\mathrm{\&kappa;}\underset{2}{e}\left(t\right)$

j=1,2,…,p,0<κ<1

Utilize the error d of hidden neuron _j(t), the input of input layer $\stackrel{\&OverBar;}{w}\left(t\right)={[y\left(t\right),...,y(t-n_a+1),u(t-1),...,u(t-n_b)]}^T$ Revise connection weight l _IjAnd threshold tau _j:

τ _j=τ _j+σd _j(t) .

i＝1,2,…,n _a+n _b-2,j=1,2,…,p,0<σ<1

Shown in Figure 5 is the process flow diagram of switching mechanism, at first is respectively design performance index of non linear robust Indirect adaptive control and nonlinear neural network Indirect adaptive control

With

Its computing method are as follows:

$\underset{s}{J}\left(t\right)=\underset{i=1}{\overset{t}{\mathrm{\&Sigma;}}}\frac{\underset{s}{h}\left(i\right)[{\underset{s}{e}}^2\left(i\right)-{4\mathrm{\&mu;}}^2]}{2[1+{\left|\right|w(i-1)\left|\right|}^2]}+c\underset{j=t-1-N}{\overset{t}{\mathrm{\&Sigma;}}}[\frac{1}{2}-\underset{s}{h}\left(j\right)]{\underset{s}{e}}^2\left(j\right),s=\mathrm{1,2}---\left(10\right)$

In the formula $\underset{s}{e}\left(t\right)=y\left(t\right)-\underset{s}{\overset{^}{y}}\left(t\right),$ $\underset{s}{h}\left(t\right)=\left\{\begin{array}{cc}\frac{1}{2}& \left|\underset{s}{e}\left(t\right)\right|>2\mathrm{\&mu;}\\ 0& \mathrm{otherwise}\end{array}\right.,$ μ 〉=0, N are predefined integers, and c 〉=0 is a constant.Judge that two controller performances refer to the size of target value, the controller assignment that performance index value is less is to the control inputs u (t) of system, that is:

$u\left(t\right)=\left\{\begin{array}{cc}\underset{1}{u}\left(t\right)& \underset{1}{J}\left(t\right)\&le;\underset{2}{J}\left(t\right)\\ \underset{2}{u}\left(t\right)& J_1\left(t\right)>\underset{2}{J}\left(t\right)\end{array}\right.$

According to top discussion, the concrete real-time online control step of multi-model Adaptive Control method of the present invention is as follows:

S1: system initialization: random initializtion model With

Parameter and neural network, can determine according to priori; (neural network is set to single hidden layer, and the hidden neuron number is made as 6-10 usually);

In the S2:t=0 moment, system is output as zero, i.e. y (0)=0; When t ≠ 0 constantly, provide the true output valve y (t) of system by the controlled device of system, by model With

Provide respectively the estimation output of model With

The evaluated error of computation model is respectively

With $\underset{2}{e}\left(t\right)=y\left(t\right)-\underset{2}{y}\left(t\right);$

S3: by the reference input r (t) of system and the true departure e (t) that exports y (t) computing system of system;

S4: utilize model

With

Parameter come CONTROLLER DESIGN With

According to formula (5) (8), calculated respectively the output valve of Nonlinear Robust Controller and nonlinear neural network controller by the departure e (t) of system

With

S5: the performance index of being calculated each controller by the model evaluated error

With

Refer to the controller u that target value is less by switching mechanism (11) formula selectivity _i(t) as the control inputs u (t) of controlled device;

S6: by the model evaluated error With

According to adaptive law (4) and (7) separately, the difference Renewal model With

Parameter and the weights of neural network;

S7: get back to step S2.

Fig. 6 (1) and (2) are respectively the input and output curve of control system of the present invention.Can find out that the multi-model Adaptive Control device can well the track reference sinusoidal curve, control inputs is milder, is easy to realize.

Claims (7)

1. the multi-model Adaptive Control device of a nonlinear system, it is characterized in that, this controller is comprised of two Indirect adaptive control devices and a switching mechanism, wherein, a controller is non linear robust Indirect adaptive control device, another is nonlinear neural network Indirect adaptive control device, the input of controlled device is selected to produce between two controllers by switching mechanism, between the output of controlled device and two controllers a close loop negative feedback is arranged, be set to subtract each other relation between the model output of controlled device and two Indirect adaptive control devices, model error is used for the parameter of adjustment model and the weights of neural network.

2. the multi-model Adaptive Control device of nonlinear system according to claim 1, it is characterized in that, described non linear robust Indirect adaptive control device comprises a non linear robust adaptive model and a gamma controller, the non linear robust adaptive model increases the compensation term to system's nonlinear terms by the basis at linear model, guarantee that the Identification Errors of this model also can be asymptotic less than a normal number when the restrictive condition of system's nonlinear terms is loosened to zeroth order near bounded.

3. the multi-model Adaptive Control device of nonlinear system according to claim 1, it is characterized in that, described nonlinear neural network Indirect adaptive control device comprises a nonlinear neural network adaptive model and a nonlinear neural network controller, the nonlinear neural network adaptive model obtains the estimation output to controlled device by the online neural network weight of adjusting.

4. the multi-model Adaptive Control device of nonlinear system according to claim 2 is characterized in that, described neural network adaptive model contains an input layer, a hidden layer and an output layer.

5. the multi-model Adaptive Control device of nonlinear system according to claim 4 is characterized in that, contains 6-10 neuron in the hidden layer of described neural network adaptive model, and output layer has a neuron.

6. control method that is used for the arbitrary described multi-model Adaptive Control device of claim 1 to 5 is characterized in that the step of this control method is as follows:

S1: system initialization: the parameter of random initializtion non linear robust adaptive model, the parameter of random initializtion nonlinear neural network model and the weights of neural network, these parameters can be determined by certain priori;

In the S2:k=0 moment, object is output as 0; K ≠ 0 moment, object is output as the real output value of system, makes the poor departure e that obtains system with default value _cActual output is made the poor model error e that obtains with the output of non linear robust adaptive model ₁, make the poor model error e that obtains with the nonlinear neural network model ₂

S3: with departure e _cAs the input of non linear robust adaptive controller and nonlinear neural network adaptive controller, produce respectively controlled quentity controlled variable u by two controllers ₁And u ₂

S4: according to model error e ₁And e ₂Come calculation of performance indicators C ₁And C ₂Value, the input u that the less controller of selectivity desired value produces _i, as the control inputs u of controlled device and two models,

S5: utilize model error e ₁And e ₂Upgrade respectively parameter and the weights of non linear robust adaptive model and nonlinear neural network adaptive model;

S6: forward step S2 to.

7. the multi-model Adaptive Control method of nonlinear system according to claim 6, it is characterized in that, among the described step S7, in switching mechanism, at first design performance index, these performance index comprise a cumulative errors part and a transient error part, in each control constantly, calculate the performance index of each controller, the controller that the selectivity index is less produces next control inputs constantly.

Images