CN105223808A - Based on the mechanical arm system saturation compensation control method that neural network dynamic face sliding formwork controls - Google Patents

Abstract

Based on the mechanical arm system saturation compensation control method that neural network dynamic face sliding formwork controls, comprising: the dynamic model setting up mechanical arm servo-drive system, initialization system state, sampling time and controling parameters; According to Order Derivatives in Differential Mid-Value Theorem, by the saturated linearization process of non-linear input in system, derive with the saturated mechanical arm servo system models of the unknown; Based on dynamic surface sliding-mode control, calculating control system tracking error, sliding-mode surface and differential.The invention provides one can effective compensation the unknown saturated, the neural network dynamic face sliding-mode control of the complexity explosion issues avoiding the method for inversion to bring, realizes the stable of system and follows the tracks of fast.

Description

Based on the mechanical arm system saturation compensation control method that neural network dynamic face sliding formwork controls

Technical field

The present invention relates to a kind of mechanical arm system saturation compensation control method controlled based on neural network dynamic face sliding formwork, particularly with the control method of the mechanical arm servo-drive system of input saturation constraints.

Background technology

Mechanical arm servo-drive system is widely used in robot, the contour performance system of aviation aircraft, and the accurate fast control how realizing mechanical arm servo-drive system has become a hot issue.Trajectory Tracking Control System and the flexible mechanical arm problem of robotic arm are subject to increasing attention.But unknown saturation nonlinearity link is extensively present in mechanical arm servo-drive system, the efficiency of control system is often caused to reduce or even lost efficacy.Therefore, input saturated constraint must consider in Controller gain variations process.For the control problem of mechanical arm servo-drive system, there is a lot of control method, such as PID controls, adaptive control, sliding formwork control etc.

Sliding formwork controls to be considered to an effective robust control method in and external disturbance uncertain at resolution system.The advantages such as sliding-mode control has that algorithm is simple, fast response time, to external world noise and Parameter Perturbation strong robustness.But sliding formwork controls demand fulfillment matching condition in the design process, the uncertainty of real system matching condition becomes the obstacle of sliding formwork control design case.The method of inversion has improves sliding mode controller performance, loosens the advantage of matching condition.Sliding formwork is controlled to combine with the method for inversion, in each step design of controller, introduces virtual controlling variable.But the method for inversion will introduce the problem of complexity blast.Therefore, adopt dynamic surface sliding formwork to control, make the input of the controller of system become simple, become an important research direction.

Saturation nonlinearity link is extensively present in mechanical arm servo-drive system, Hydrauservo System and other Industrial Engineering fields.Saturated existence often causes the efficiency of control system reduce or even lost efficacy.Therefore, be improve control performance, for saturated compensation and control method essential.Traditional saturation compensation method is generally set up saturated inversion model or approximate inverse model, and by estimating saturated bound parameter designing adaptive controller, to compensate saturated impact.But in the nonlinear system such as mechanical arm servo-drive system, saturated inversion model often not easily accurately obtains.For the unknown input saturation existed in system, based on Order Derivatives in Differential Mid-Value Theorem through line linearity, become a simple time-varying system, avoid ancillary relief.Neural network is widely used in the non-linear and uncertain of disposal system, and achieves good control effects.Thus neural network can be utilized to approach the unknown parameter of unknown function and system model, the complexity explosion issues simultaneously avoiding the method for inversion to bring improves the tracing control performance of system.

Summary of the invention

Cannot saturation compensation effectively in order to what overcome existing mechanical arm servo-drive system, model parameter uncertainty, and the deficiency of the complexity blast that brings of the method for inversion etc., the invention provides a kind of mechanical arm system saturation compensation control method controlled based on neural network dynamic face sliding formwork, simplify the project organization of controller, achieve the mechanical arm system Position Tracking Control of band input saturation, ensure that system stability is followed the tracks of fast.

In order to the technical scheme solving the problems of the technologies described above proposition is as follows:

Based on the mechanical arm system saturation compensation control method that dynamic surface sliding formwork controls, comprise the following steps:

Step 1, sets up the dynamic model of mechanical arm servo-drive system, initialization system state, sampling time and controling parameters, and process is as follows:

The dynamic model expression-form of 1.1 mechanical arm servo-drive systems is

$\left\{\begin{array}{c}I\stackrel{\·\·}{q}+K(q-\mathrm{\θ})+MgLsin\left(q\right)=0\\ J\stackrel{\·\·}{\mathrm{\θ}}-K(q-\mathrm{\θ})=v\left(u\right)\end{array}\right.---\left(1\right)$

Wherein, q and θ is respectively the angle of robot linkage and motor; G is acceleration of gravity; I is the inertia of connecting rod; J is the inertia of motor; K is spring rate; M and L is quality and the length of connecting rod respectively; U is control signal; V (u), for saturated, is expressed as:

$v\left(u\right)=sat\left(u\right)=\left\{\begin{array}{cc}{v}_{max}\mathrm{sgn}\left(u\right),& \left|u\right|\≥v_{max}\\ u,& \left|u\right|\≤v_{max}\end{array}\right.---\left(2\right)$

Wherein sgn (u) is unknown nonlinear function; v _maxfor unknown parameter of saturation, meet v _max> 0;

Definition x ₁=q, x ₃=θ, formula (1) is rewritten as

$\left\{\begin{array}{c}{\stackrel{\·}{x}}_1=x_2\\ {\stackrel{\·}{x}}_2=-\frac{MgL}{I}sin\left(x_1\right)-\frac{K}{I}(x_1-x_3)\\ {\stackrel{\·}{x}}_3=x_4\\ {\stackrel{\·}{x}}_4=\frac{1}{J}v\left(u\right)+\frac{K}{J}(x_1-x_3)\\ y=x_1.\end{array}\right.---\left(3\right)$

Wherein, y is system output trajectory;

1.2 defining variable z ₁=x ₁, z ₂=x ₂, $z_4=-x_2\frac{MgL}{I}cos\left(x_1\right)-\frac{K}{I}(x_2-x_4),$ Then formula (3) is rewritten into

$\left\{\begin{array}{c}{\stackrel{\·}{z}}_1=z_2\\ {\stackrel{\·}{z}}_2=z_3\\ {\stackrel{\·}{z}}_3=z_4\\ {\stackrel{\·}{z}}_4=f_1\left(\stackrel{\‾}{z}\right)+b_{1}v\left(u\right)\\ y=z_1\end{array}\right.---\left(4\right)$

Wherein, $\stackrel{\‾}{z}={\[z_1,z_2,z_3\]}^T,f_1\left(\stackrel{\‾}{z}\right)=\frac{MgL}{I}sin\left(z_1\right)(z_2^2-\frac{K}{J})-\left(\frac{MgL}{I}cos\right(z_1)+\frac{K}{J}+\frac{K}{I})z_3,b_1=\frac{K}{IJ};$

Step 2, according to Order Derivatives in Differential Mid-Value Theorem, carry out linearization process by saturated for the non-linear input in system, derive with the saturated mechanical arm servo system models of the unknown, process is as follows:

2.1 pairs of saturated models carry out smooth treatment

$\begin{array}{c}g\left(u\right)=v_{max}\×\tanh \left(\frac{u}{v_{\max }}\right)\\ =v_{\max }\×\frac{e^{u/v_{\max }}-e^{-u/v_{\max }}}{e^{u/v_{ma\×}}+e^{-u/v_{max}}}\end{array}---\left(5\right)$

Then

v(u)＝sat(u)＝g(u)+d(u)(6)

Wherein, d (u) represents the error existed between smooth function and saturated model;

2.2 according to Order Derivatives in Differential Mid-Value Theorem, there is δ ∈ (0,1) and makes

$g\left(u\right)=g\left(u_0\right)+g_{u_{\mathrm{\ξ}}}(u-u_0)---\left(7\right)$

Wherein $g_{u_{\mathrm{\ξ}}}=\frac{\∂g\left(u\right)}{\∂u}{|}_{u=u_{\mathrm{\ξ}}}>0,u_{\mathrm{\ξ}}=\mathrm{\ξ}u+(1-\mathrm{\ξ})u_0,$ u ₀∈(0,u)；

Select u ₀=0, formula (7) is rewritten as

$g\left(u\right)=g_{u_{\mathrm{\ξ}}}u---\left(8\right)$

Formula (4), by formula (6) and formula (8), is rewritten as following equivalents by 2.2:

$\left\{\begin{array}{c}{\stackrel{\·}{z}}_1=z_2\\ {\stackrel{\·}{z}}_2=z_3\\ {\stackrel{\·}{z}}_3=z_4\\ {\stackrel{\·}{z}}_4=f_2\left(\stackrel{\‾}{z}\right)+b_{2}u\\ y=z_1\end{array}\right.---\left(9\right)$

Wherein, $f_2\left(\stackrel{\‾}{z}\right)=f_1\left(\stackrel{\‾}{z}\right)+d\left(u\right),b_2=b_1\×g_{u_{\mathrm{\ξ}}};$

Step 3, calculating control system tracking error, sliding-mode surface and differential, process is as follows:

The tracking error of 3.1 definition control system, sliding-mode surface is

$\left\{\begin{array}{c}e=y-y_d\\ s_1=e+\mathrm{\λ}\∫edt\end{array}\right.---\left(10\right)$

Wherein, y _dfor second order can lead desired trajectory, λ is constant, and λ > 0;

The differentiate of 3.2 pairs of formulas (10) obtains:

$\{\begin{array}{c}\stackrel{\·}{e}=\stackrel{\·}{y}-{\stackrel{\·}{y}}_d=z_2-{\stackrel{\·}{y}}_d\\ {\stackrel{\·}{s}}_1=\stackrel{\·}{e}+\mathrm{\λ}e=z_2-{\stackrel{\·}{y}}_d+\mathrm{\λ}e\end{array}---\left(11\right)$

3.3 design virtual controlling amounts

${\stackrel{\‾}{z}}_2=-k_1{s}_1+{\stackrel{\·}{y}}_d-\mathrm{\λ}e---\left(12\right)$

Wherein, k ₁for constant, and k ₁> 0;

The variable β that 3.4 definition one are new ₂, allow virtual controlling amount be τ by time constant ₂firstorder filter:

${\mathrm{\τ}}_2{\stackrel{\·}{\mathrm{\β}}}_2+{\mathrm{\β}}_2={\stackrel{\‾}{z}}_2,{\mathrm{\β}}_2\left(0\right)={\stackrel{\‾}{z}}_2\left(0\right)---\left(13\right)$

3.5 definition then

$\stackrel{\·}{\mathrm{\β}}2=\frac{{\stackrel{\‾}{z}}_2-{\mathrm{\β}}_2}{{\mathrm{\τ}}_2}=-\frac{y_2}{{\mathrm{\τ}}_2}---\left(14\right)$

Step 4, for formula (4), design virtual controlling amount, process is as follows:

4.1 definition error variances

s _i＝z _i-β _i,i＝2,3(15)

The first differential of formula (15) is

${\stackrel{\·}{s}}_i=z_{i+1}-{\stackrel{\·}{\mathrm{\β}}}_i,i=2,3---\left(16\right)$

4.2 design virtual controlling amounts

${\stackrel{\‾}{z}}_{i+1}=-k_i{s}_i-s_{i-1}+{\stackrel{\·}{\mathrm{\β}}}_i---\left(17\right)$

Wherein, k _ifor constant, and k _i> 0;

The variable β that 4.3 definition one are new _i+1, allow virtual controlling amount be τ by time constant ₂firstorder filter:

${\mathrm{\τ}}_{i+1}{\stackrel{\·}{\mathrm{\β}}}_{i+1}+{\mathrm{\β}}_{i+1}={\stackrel{\‾}{z}}_{i+1},{\mathrm{\β}}_{i+1}\left(0\right)={\stackrel{\‾}{z}}_{i+1}\left(0\right)---\left(18\right)$

4.4 definition $y_{i+1}={\mathrm{\β}}_{i+1}-{\stackrel{\‾}{z}}_{i+1},$ Then

${\stackrel{\·}{\mathrm{\β}}}_{i+1}=\frac{{\stackrel{\‾}{z}}_{i+1}-{\mathrm{\β}}_{i+1}}{{\mathrm{\τ}}_{i+1}}=-\frac{y_{i+1}}{{\mathrm{\τ}}_{i+1}}---\left(19\right)$

Step 5, CONTROLLER DESIGN inputs, and process is as follows:

5.1 definition error variances

s ₄＝z ₄-β ₄(20)

The first differential of calculating formula (20) is

${\stackrel{\·}{s}}_4=f_2\left(\stackrel{\‾}{z}\right)+b_{2}u-{\stackrel{\·}{\mathrm{\β}}}_4---\left(21\right)$

5.2 in order to approach the Nonlinear uncertainty that can not directly obtain and b ₂, define following neural network

Wherein, W ^*for ideal weight, ε ^*for neural network perfect error value, meet | ε ^*|≤ε _n, expression formula is:

Wherein, a, b, c, d are constant;

5.3 CONTROLLER DESIGN input u:

Wherein, for the estimated value of ideal weight W, for perfect error upper bound ε ^*estimated value;

5.4 design adaptive rates:

Wherein, Γ=Γ ^t> 0, Γ ₃adaptive gain matrix, σ, v _{ε N}all constant, and σ > 0, v _{ε N}> 0;

Step 6, design Lyapunov function

$V=\frac{1}{2}\underset{i=1}{\overset{3}{\Σ}}(s_i^2+y_{i+1}^2)+\frac{1}{2b_2}{s}_4^2+\frac{1}{2}{\tilde{W}}^T{\mathrm{\Γ}}^{-1}\tilde{W}+\frac{1}{2v_{\mathrm{\ϵ}N}}{\widetilde{\mathrm{\ϵ}}}_N^2---\left(26\right)$

Carry out differentiate to formula (26) to obtain:

$\stackrel{\·}{V}=\underset{i=1}{\overset{3}{\Σ}}(s_i{\stackrel{\·}{s}}_i+y_{i+1}{\stackrel{\·}{y}}_{i+1})+\frac{1}{b_2}{s}_4{\stackrel{\·}{s}}_4+{\tilde{W}}^T{\mathrm{\Γ}}^{-1}\stackrel{\·}{\hat{W}}+\frac{1}{v_{\mathrm{\ϵ}N}}{\widetilde{\mathrm{\ϵ}}}_N{\stackrel{\·}{\widehat{\mathrm{\ϵ}}}}_N---\left(27\right)$

If then decision-making system is stable.

The present invention is based on neural network, dynamic surface sliding-mode control, under considering unknown input saturated conditions, the control method of the saturation compensation of design mechanical arm servo-drive system, realizes the tenacious tracking of system, ensures that tracking error is at finite time convergence control.

Technical conceive of the present invention is: can not survey for state, and with the mechanical arm servo-drive system of unknown input saturation, utilizes Order Derivatives in Differential Mid-Value Theorem to optimize saturated structures, propose the mechanical arm servo-drive system based on saturated model.Control in conjunction with neural network, adaptive control and dynamic surface sliding formwork again, the saturation compensation control method of the arm servo-drive system that designs a mechanism.By Order Derivatives in Differential Mid-Value Theorem, make saturated continuously differentiable, then approach unknown function by neural network, eliminate the ancillary relief that tradition is saturated.And utilize the virtual error variance of dynamic surface sliding mode design, avoid the complexity explosion issues that the method for inversion is brought, realize the Position Tracking Control of system.The invention provides one can effective compensation the unknown saturated, the neural network dynamic face sliding-mode control of the complexity explosion issues avoiding the method for inversion to bring, realizes the stable of system and follows the tracks of fast.

Advantage of the present invention is: avoid unknown input saturation on the impact of alliance tracing control performance, and the complexity explosion issues that the method for inversion is brought, bucking-out system Unknown Model indeterminate, and the position realizing system is followed the tracks of.

Accompanying drawing explanation

Fig. 1 is non-linear saturated schematic diagram of the present invention;

Fig. 2 is the schematic diagram of tracking effect of the present invention;

Fig. 3 is the schematic diagram of tracking error of the present invention;

Fig. 4 is the schematic diagram of controller of the present invention input;

Fig. 5 is control flow chart of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the present invention will be further described.

With reference to Fig. 1-Fig. 5, a kind of mechanical arm system saturation compensation control method controlled based on neural network dynamic face sliding formwork, comprises the following steps:

Step 1, sets up the dynamic model of mechanical arm servo-drive system, initialization system state, sampling time and controling parameters, and process is as follows:

The dynamic model expression-form of 1.1 mechanical arm servo-drive systems is

$\left\{\begin{array}{c}I\stackrel{\·\·}{q}+K(q-\mathrm{\θ})+MgLsin\left(q\right)=0\\ J\stackrel{\·\·}{\mathrm{\θ}}-K(q-\mathrm{\θ})=v\left(u\right)\end{array}\right.---\left(1\right)$

Wherein, q and θ is respectively the angle of robot linkage and motor; G is acceleration of gravity; I is the inertia of connecting rod; J is the inertia of motor; K is spring rate; M and L is quality and the length of connecting rod respectively; U is control signal; V (u), for saturated, is expressed as:

$v\left(u\right)=sat\left(u\right)=\left\{\begin{array}{cc}{v}_{max}\mathrm{sgn}\left(u\right),& \left|u\right|\≥v_{max}\\ u,& \left|u\right|\≤v_{max}\end{array}\right.---\left(2\right)$

Wherein sgn (u) is unknown nonlinear function; v _maxfor unknown parameter of saturation, meet v _max> 0;

Definition x ₁=q, x ₃=θ, formula (1) is rewritten as

$\left\{\begin{array}{c}{\stackrel{\·}{x}}_1=x_2\\ {\stackrel{\·}{x}}_2=-\frac{MgL}{I}sin\left(x_1\right)-\frac{K}{I}(x_1-x_3)\\ {\stackrel{\·}{x}}_3=x_4\\ {\stackrel{\·}{x}}_4=\frac{1}{J}v\left(u\right)+\frac{K}{J}(x_1-x_3)\\ y=x_1.\end{array}\right.---\left(3\right)$

Wherein, y is system output trajectory;

1.2 defining variable z ₁=x ₁, z ₂=x ₂, $z_4=-x_2\frac{MgL}{I}cos\left(x_1\right)-\frac{K}{I}(x_2-x_4),$ Then formula (3) is rewritten into

$\left\{\begin{array}{c}{\stackrel{\·}{z}}_1=z_2\\ {\stackrel{\·}{z}}_2=z_3\\ {\stackrel{\·}{z}}_3=z_4\\ {\stackrel{\·}{z}}_4=f_1\left(\stackrel{\‾}{z}\right)+b_{1}v\left(u\right)\\ y=z_1\end{array}\right.---\left(4\right)$

Wherein, $\stackrel{\‾}{z}={\[z_1,z_2,z_3\]}^T,f_1\left(\stackrel{\‾}{z}\right)=\frac{MgL}{I}sin\left(z_1\right)(z_2^2-\frac{K}{J})-\left(\frac{MgL}{I}cos\right(z_1)+\frac{K}{J}+\frac{K}{I})z_3,b_1=\frac{K}{IJ};$

Step 2, according to Order Derivatives in Differential Mid-Value Theorem, carry out linearization process by saturated for the non-linear input in system, derive with the saturated mechanical arm servo system models of the unknown, process is as follows:

2.1 pairs of saturated models carry out smooth treatment

$\begin{array}{c}g\left(u\right)=v_{max}\×\tanh \left(\frac{u}{v_{\max }}\right)\\ =v_{\max }\×\frac{e^{u/v_{\max }}-e^{-u/v_{\max }}}{e^{u/v_{ma\×}}+e^{-u/v_{max}}}\end{array}---\left(5\right)$

Then

v(u)＝sat(u)＝g(u)+d(u)(6)

Wherein, d (u) represents the error existed between smooth function and saturated model;

2.2 according to Order Derivatives in Differential Mid-Value Theorem, there is δ ∈ (0,1) and makes

$g\left(u\right)=g\left(u_0\right)+g_{u_{\mathrm{\ξ}}}(u-u_0)---\left(7\right)$

Wherein $g_{u_{\mathrm{\ξ}}}=\frac{\∂g\left(u\right)}{\∂u}{|}_{u=u_{\mathrm{\ξ}}}>0,u_{\mathrm{\ξ}}=\mathrm{\ξ}u+(1-\mathrm{\ξ})u_0,$ u ₀∈(0,u)；

Select u ₀=0, formula (7) is rewritten as

$g\left(u\right)=g_{u_{\mathrm{\ξ}}}u---\left(8\right)$

Formula (4), by formula (6) and formula (8), is rewritten as following equivalents by 2.2:

$\left\{\begin{array}{c}{\stackrel{\·}{z}}_1=z_2\\ {\stackrel{\·}{z}}_2=z_3\\ {\stackrel{\·}{z}}_3=z_4\\ {\stackrel{\·}{z}}_4=f_2\left(\stackrel{\‾}{z}\right)+b_{2}u\\ y=z_1\end{array}\right.---\left(9\right)$

Wherein, $f_2\left(\stackrel{\‾}{z}\right)=f_1\left(\stackrel{\‾}{z}\right)+d\left(u\right),b_2=b_1\×g_{u_{\mathrm{\ξ}}};$

Step 3, calculating control system tracking error, sliding-mode surface and differential, process is as follows:

The tracking error of 3.1 definition control system, sliding-mode surface is

$\left\{\begin{array}{c}e=y-y_d\\ s_1=e+\mathrm{\λ}\∫edt\end{array}\right.---\left(10\right)$

Wherein, y _dfor second order can lead desired trajectory, λ is constant, and λ > 0;

The differentiate of 3.2 pairs of formulas (10) obtains:

$\{\begin{array}{c}\stackrel{\·}{e}=\stackrel{\·}{y}-{\stackrel{\·}{y}}_d=z_2-{\stackrel{\·}{y}}_d\\ {\stackrel{\·}{s}}_1=\stackrel{\·}{e}+\mathrm{\λ}e=z_2-{\stackrel{\·}{y}}_d+\mathrm{\λ}e\end{array}---\left(11\right)$

3.3 design virtual controlling amounts

${\stackrel{\‾}{z}}_2=-k_1{s}_1+{\stackrel{\·}{y}}_d-\mathrm{\λ}e---\left(12\right)$

Wherein, k ₁for constant, and k ₁> 0;

The variable β that 3.4 definition one are new ₂, allow virtual controlling amount be τ by time constant ₂firstorder filter:

${\mathrm{\τ}}_2{\stackrel{\·}{\mathrm{\β}}}_2+{\mathrm{\β}}_2={\stackrel{\‾}{z}}_2,{\mathrm{\β}}_2\left(0\right)={\stackrel{\‾}{z}}_2\left(0\right)---\left(13\right)$

3.5 definition $y_2={\mathrm{\β}}_2-{\stackrel{\‾}{z}}_2,$ Then

$\stackrel{\·}{\mathrm{\β}}2=\frac{{\stackrel{\‾}{z}}_2-{\mathrm{\β}}_2}{{\mathrm{\τ}}_2}=-\frac{y_2}{{\mathrm{\τ}}_2}---\left(14\right)$

Step 4, for formula (4), design virtual controlling amount, process is as follows:

4.1 definition error variances

s _i＝z _i-β _i,i＝2,3(15)

The first differential of formula (15) is

${\stackrel{\·}{s}}_i=z_{i+1}-{\stackrel{\·}{\mathrm{\β}}}_i,i=2,3---\left(16\right)$

4.2 design virtual controlling amounts

${\stackrel{\‾}{z}}_{i+1}=-k_i{s}_i-s_{i-1}+{\stackrel{\·}{\mathrm{\β}}}_i---\left(17\right)$

Wherein, k _ifor constant, and k _i> 0;

The variable β that 4.3 definition one are new _i+1, allow virtual controlling amount be τ by time constant ₂firstorder filter:

${\mathrm{\τ}}_{i+1}{\stackrel{\·}{\mathrm{\β}}}_{i+1}+{\mathrm{\β}}_{i+1}={\stackrel{\‾}{z}}_{i+1},{\mathrm{\β}}_{i+1}\left(0\right)={\stackrel{\‾}{z}}_{i+1}\left(0\right)---\left(18\right)$

4.4 definition $y_{i+1}={\mathrm{\β}}_{i+1}-{\stackrel{\‾}{z}}_{i+1},$ Then

${\stackrel{\·}{\mathrm{\β}}}_{i+1}=\frac{{\stackrel{\‾}{z}}_{i+1}-{\mathrm{\β}}_{i+1}}{{\mathrm{\τ}}_{i+1}}=-\frac{y_{i+1}}{{\mathrm{\τ}}_{i+1}}---\left(19\right)$

Step 5, CONTROLLER DESIGN inputs, and process is as follows:

5.1 definition error variances

s ₄＝z ₄-β ₄(20)

The first differential of calculating formula (20) is

${\stackrel{\·}{s}}_4=f_2\left(\stackrel{\‾}{z}\right)+b_{2}u-{\stackrel{\·}{\mathrm{\β}}}_4---\left(21\right)$

5.2 in order to approach the Nonlinear uncertainty that can not directly obtain and b ₂, define following neural network

Wherein, W ^*for ideal weight, ε ^*for neural network perfect error value, meet | ε ^*|≤ε _n, expression formula is:

Wherein, a, b, c, d are constant;

5.3 CONTROLLER DESIGN input u:

Wherein, for the estimated value of ideal weight W, for perfect error upper bound ε ^*estimated value;

5.4 design adaptive rates:

Wherein, Γ=Γ ^t> 0, Γ ₃adaptive gain matrix, σ, v _{ε N}all constant, and σ > 0, v _{ε N}> 0;

Step 6, design Lyapunov function

$V=\frac{1}{2}\underset{i=1}{\overset{3}{\Σ}}(s_i^2+y_{i+1}^2)+\frac{1}{2b_2}{s}_4^2+\frac{1}{2}{\tilde{W}}^T{\mathrm{\Γ}}^{-1}\tilde{W}+\frac{1}{2v_{\mathrm{\ϵ}N}}{\widetilde{\mathrm{\ϵ}}}_N^2---\left(26\right)$

Carry out differentiate to formula (26) to obtain:

$\stackrel{\·}{V}=\underset{i=1}{\overset{3}{\Σ}}(s_i{\stackrel{\·}{s}}_i+y_{i+1}{\stackrel{\·}{y}}_{i+1})+\frac{1}{b_2}{s}_4{\stackrel{\·}{s}}_4+{\tilde{W}}^T{\mathrm{\Γ}}^{-1}\stackrel{\·}{\hat{W}}+\frac{1}{v_{\mathrm{\ϵ}N}}{\widetilde{\mathrm{\ϵ}}}_N{\stackrel{\·}{\widehat{\mathrm{\ϵ}}}}_N---\left(27\right)$

If then decision-making system is stable.

For the validity of checking institute extracting method, The present invention gives the contrast of three kinds of control methods: be with the dynamic surface sliding-mode control (S1) of saturation compensation, not with the dynamic surface sliding-mode control (S2) of saturation compensation and the dynamic surface control method not with saturation compensation (S3).

Contrast in order to more effective, all optimum configurations are all consistent system initialization parameters is [x ₁, x ₂, x ₃, x ₄] ^t=[0,0,0,0] ^t, [β ₂, β ₃, β ₄] ^t=[0,0,0] ^t; Neural network parameter is Γ=diag{5}, a=10, b=10, c=1, d=-1; Adaptive control rate parameter is v _{ε N}=0.1, σ=0.01, δ=0.1; The time constant parameter of firstorder filter is t ₂=t ₃=t ₄=0.02; System model parameter is Mgl=5, I=1, J=1, K=40, I=1; Controller parameter is k ₁=0.5, k ₂=8, k ₃=8, k ₄=2, λ=7.

Tracking trapezoidal wave inputs, and its expression formula is such as formula (28).The controller saturation input v of S1 _max=75, S2, S3 controller are input as v _max=295.As seen from Figure 2, the tracking effect of S1 is better than S2, S3, and S2 finally can not realize the stable convergence of system; As can be seen from Figure 3, the tracking steady-state error of S1 method is minimum.As can be seen from Figure 4, with in input saturation controller situation, even if saturated restriction is comparatively large, the tenacious tracking of system can still be realized.Therefore, the invention provides one can effective compensation the unknown saturated, the neural network dynamic face sliding-mode control of the complexity explosion issues avoiding the method for inversion to bring, realizes the stable of system and follows the tracks of fast.

$y_d=\{\begin{array}{cc}0,& 0\&le;t<2\\ 5(t-2),& 2\&le;t<4\\ 10,& 4\&le;t<6\\ -5(t-8),& 6\&le;t<10\\ -10,& 10\&le;t<12\\ -5(t-14),& 12\&le;t<16\\ 10,& 16\&le;t<18\\ -5(t-20),& 18\&le;t\&le;20\end{array}---\left(28\right)$

What more than set forth is the excellent effect of optimization that an embodiment that the present invention provides shows, obvious the present invention is not just limited to above-described embodiment, do not depart from essence spirit of the present invention and do not exceed scope involved by flesh and blood of the present invention prerequisite under can do all distortion to it and implemented.

Claims (1)

1., based on the mechanical arm system saturation compensation control method that neural network dynamic face sliding formwork controls, it is characterized in that: described control method comprises the following steps:

Step 1, sets up the dynamic model of mechanical arm servo-drive system, initialization system state, sampling time and controling parameters, and process is as follows:

The dynamic model expression-form of 1.1 mechanical arm servo-drive systems is

$\{\begin{array}{c}I\stackrel{\&CenterDot;\&CenterDot;}{q}+K\left(q-\mathrm{\&theta;}\right)+MgLsin\left(q\right)=0\\ J\stackrel{\&CenterDot;\&CenterDot;}{\mathrm{\&theta;}}-K\left(q-\mathrm{\&theta;}\right)=v\left(u\right)\end{array}---\left(1\right)$

Wherein, q and θ is respectively the angle of robot linkage and motor; G is acceleration of gravity; I is the inertia of connecting rod; J is the inertia of motor; K is spring rate; M and L is quality and the length of connecting rod respectively; U is control signal; V (u), for saturated, is expressed as:

$v\left(u\right)=sat\left(u\right)=\left\{\begin{array}{cc}{v}_{max}\mathrm{sgn}\left(u\right),& \left|u\right|\&GreaterEqual;v_{max}\\ u,& \left|u\right|\&le;v_{max}\end{array}\right.---\left(2\right)$

Wherein sgn (u) is unknown nonlinear function; v _maxfor unknown parameter of saturation, meet v _max> 0;

Definition x ₁=q, x ₃=θ, formula (1) is rewritten as

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{x}}_1=x_2\\ {\stackrel{\&CenterDot;}{x}}_2=-\frac{MgL}{I}sin\left(x_1\right)-\frac{K}{I}\left(x_1-x_3\right)\\ {\stackrel{\&CenterDot;}{x}}_3=x_4\\ {\stackrel{\&CenterDot;}{x}}_4=\frac{1}{J}v\left(u\right)+\frac{K}{J}\left(x_1-x_3\right)\\ y=x_1.\end{array}\right.---\left(3\right)$

Wherein, y is system output trajectory;

1.2 defining variable z ₁=x ₁, z ₂=x ₂, $z_4=-x_2\frac{MgL}{I}cos\left(x_1\right)-\frac{K}{I}\left(x_2-x_4\right),$ Then formula (3) is rewritten into

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{z}}_1=z_2\\ {\stackrel{\&CenterDot;}{z}}_2=z_3\\ {\stackrel{\&CenterDot;}{z}}_3=z_4\\ {\stackrel{\&CenterDot;}{z}}_4=f_1\left(\stackrel{-}{z}\right)+b_{1}v\left(u\right)\\ y=z_1\end{array}\right.---\left(4\right)$

Wherein, $\stackrel{-}{z}={\&lsqb;z_1,z_2,z_3\&rsqb;}^T,f_1\left(\stackrel{-}{z}\right)=\frac{MgL}{I}sin\left(z_1\right)\left(z_2^2-\frac{K}{J}\right)-\left(\frac{MgL}{I}cos\left(z_1\right)+\frac{K}{J}+\frac{K}{I}\right)z_3,b_1=\frac{K}{IJ};$

Step 2, according to Order Derivatives in Differential Mid-Value Theorem, carry out linearization process by saturated for the non-linear input in system, derive with the saturated mechanical arm servo system models of the unknown, process is as follows:

2.1 pairs of saturated models carry out smooth treatment

$\begin{array}{c}g\left(u\right)=v_{\max }\&times;tan\left(\frac{u}{v_{\max }}\right)\\ =v_{\max }\&times;\left(\frac{e^{u/v_{\max }}-e^{-u/v_{\max }}}{e^{u/v_{\max }}+e^{-u/v_{\max }}}\right)\end{array}---\left(5\right)$

Then

v(u)＝sat(u)＝g(u)+d(u)(6)

Wherein, d (u) represents the error existed between smooth function and saturated model;

2.2 according to Order Derivatives in Differential Mid-Value Theorem, there is δ ∈ (0,1) and makes

$g\left(u\right)=g\left(u_0\right)+g_{u\mathrm{\&xi;}}\left(u-u_0\right)---\left(7\right)$

Wherein $g_{u_{\mathrm{\&xi;}}}=\frac{\&part;g\left(u\right)}{\&part;u}{|}_{u=u_{\mathrm{\&xi;}}}>0,u_{\mathrm{\&xi;}}=\mathrm{\&xi;}u+(1-\mathrm{\&xi;})u_0,$ u ₀∈(0,u)；

Select u ₀=0, formula (7) is rewritten as

$g\left(\text{u}\right)=g_{u\mathrm{\&xi;}}u\text{---}\left(8\right)$

Formula (4), by formula (6) and formula (8), is rewritten as following equivalents by 2.2:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{z}}_1=z_2\\ {\stackrel{\&CenterDot;}{z}}_2=z_3\\ {\stackrel{\&CenterDot;}{z}}_3=z_4\\ {\stackrel{\&CenterDot;}{z}}_4=f_2\left(\stackrel{-}{z}\right)+b_{2}u\\ y=z_1\end{array}\right.---\left(9\right)$

Wherein, $f_2\left(\stackrel{-}{z}\right)=f_1\left(\stackrel{-}{z}\right)+d\left(u\right),b_2=b_1\&times;g_{u\mathrm{\&xi;}};$

Step 3, calculating control system tracking error, sliding-mode surface and differential, process is as follows:

The tracking error of 3.1 definition control system, sliding-mode surface is

$\left\{\begin{array}{c}e=y-y_d\\ s_1=e+\mathrm{\&lambda;}\&Integral;edt\end{array}\right.---\left(10\right)$

Wherein, y _dfor second order can lead desired trajectory, λ is constant, and λ > 0;

The differentiate of 3.2 pairs of formulas (10) obtains:

$\left\{\begin{array}{c}\stackrel{\&CenterDot;}{e}=\stackrel{\&CenterDot;}{y}-{\stackrel{\&CenterDot;}{y}}_d=z_2-{\stackrel{\&CenterDot;}{y}}_d\\ {\stackrel{\&CenterDot;}{s}}_1=\stackrel{\&CenterDot;}{e}+\mathrm{\&lambda;}e=z_2-{\stackrel{\&CenterDot;}{y}}_d+\mathrm{\&lambda;}e\end{array}\right.---\left(11\right)$

3.3 design virtual controlling amounts

${\stackrel{-}{z}}_2=-k_1{s}_1+{\stackrel{\&CenterDot;}{y}}_d-\mathrm{\&lambda;}e---\left(12\right)$

Wherein, k ₁for constant, and k ₁> 0;

The variable β that 3.4 definition one are new ₂, allow virtual controlling amount be τ by time constant ₂firstorder filter: ${\mathrm{\&tau;}}_2{\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_2+{\mathrm{\&beta;}}_2={\stackrel{-}{z}}_2,{\mathrm{\&beta;}}_2\left(0\right)={\stackrel{-}{z}}_2\left(0\right)---\left(13\right)$

3.5 definition $y_2={\mathrm{\&beta;}}_2-{\stackrel{-}{z}}_2,$ Then

${\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_2=\frac{{\stackrel{-}{z}}_2-{\mathrm{\&beta;}}_2}{{\mathrm{\&tau;}}_2}=-\frac{y_2}{{\mathrm{\&tau;}}_2}---\left(14\right)$

Step 4, for formula (4), design virtual controlling amount, process is as follows:

4.1 definition error variances

s _i＝z _i-β _i,i＝2,3(15)

The first differential of formula (15) is

${\stackrel{\&CenterDot;}{s}}_i=z_{i+1}-{\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_{i,j}=2,3---\left(16\right)$

4.2 design virtual controlling amounts

${\stackrel{\&OverBar;}{z}}_{i+1}=-k_i{s}_i-s_{i-1}+{\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_i---\left(17\right)$

Wherein, k _ifor constant, and k _i> 0;

The variable β that 4.3 definition one are new _i+1, allow virtual controlling amount be τ by time constant ₂firstorder filter:

${\mathrm{\&tau;}}_{i+1}{\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_{i+1}+{{\mathrm{\&beta;}}_i}_{+1}={\stackrel{-}{z}}_{i+1},{{\mathrm{\&beta;}}_i}_{+1}\left(0\right)={\stackrel{-}{z}}_{i+1}\left(0\right)---\left(\text{18}\right)$

4.4 definition $y_{i+1}={{\mathrm{\&beta;}}_i}_{+1}\text{-}{\stackrel{\&CenterDot;}{z}}_{i+1},$ Then

${\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_{i+1}=\frac{{\stackrel{-}{z}}_{i+1}-{{\mathrm{\&beta;}}_i}_{+1}}{{\mathrm{\&tau;}}_{i+1}}=-\frac{y_{i+1}}{{\mathrm{\&tau;}}_{i+1}}---\left(19\right)$

Step 5, CONTROLLER DESIGN inputs, and process is as follows:

5.1 definition error variances

s ₄＝z ₄-β ₄(20)

The first differential of calculating formula (20) is

${\stackrel{\&CenterDot;}{s}}_4=f_2\left(\stackrel{-}{z}\right)+b_{2}u-{\stackrel{\&CenterDot;}{\mathrm{\&beta;}}}_4\text{---}\left(21\right)$

5.2 in order to approach the Nonlinear uncertainty that can not directly obtain and b ₂, define following neural network

Wherein, W ^*for ideal weight, ε ^*for neural network perfect error value, meet | ε ^*|≤ε _n, expression formula is:

Wherein, a, b, c, d are constant;

5.3 CONTROLLER DESIGN input u:

Wherein, for the estimated value of ideal weight W, for perfect error upper bound ε ^*estimated value;

5.4 design adaptive rates:

Wherein, Γ=Γ ^t> 0, Γ ₃adaptive gain matrix, σ, v _{ε N}all constant, and σ > 0, v _{ε N}> 0;

Step 6, design Lyapunov function

$V=\frac{1}{2}\underset{i=1}{\overset{3}{\&Sigma;}}\left(s_i^2+y_{i+1}^2\right)+\frac{1}{2b_2}{s}_4^2+\frac{1}{2}{\tilde{W}}^T{\mathrm{\&Gamma;}}^{-1}\tilde{W}+\frac{1}{2v_{\mathrm{\&epsiv;}N}}{\widetilde{\mathrm{\&epsiv;}}}_N^2---\left(26\right)$

Carry out differentiate to formula (26) to obtain:

$\stackrel{\&CenterDot;}{V}=\underset{i=1}{\overset{3}{\&Sigma;}}\left(s_i{\stackrel{\&CenterDot;}{s}}_i+y_{i+1}{\stackrel{\&CenterDot;}{y}}_{i+1}\right)+\frac{1}{b_2}{s}_4{\stackrel{\&CenterDot;}{s}}_4+{\tilde{W}}^T{\mathrm{\&Gamma;}}^{-1}\stackrel{\&CenterDot;}{\hat{W}}+\frac{1}{v_{\mathrm{\&epsiv;}N}}{\widetilde{\mathrm{\&epsiv;}}}_N{\stackrel{\&CenterDot;}{\widehat{\mathrm{\&epsiv;}}}}_N---\left(27\right)$

If then decision-making system is stable.