CN100562823C - A kind of contour outline control method of complicated track - Google Patents

Abstract

The invention discloses a kind of contour outline control method of complicated track, this method is in conjunction with a kind of cross-coupling control framework with profile errors precompensation function, each that participates in servo motion set up the self-adapting data model, according to current goal location point and some historical position point values, determine servo controlled device parameter to be identified, by the RECURSIVE POLE PLACEMENT controlled variable of adjusting in real time.This method of current control output being adjusted according to historical controlled quentity controlled variable and following controlled quentity controlled variable suppressed the bounded process effectively and disturbed, and improved the precision that profile controls and the stability of process.

Description

A kind of contour outline control method of complicated track

Technical field

The present invention relates to the contour outline control method of a kind of contour outline control method, particularly a kind of complicated track.

Background technology

The existing method that improves the orbiting motion precision mainly contains two kinds: by improving each single shaft trace track tracking Control performance respectively, reduce each track profile errors after synthetic indirectly 1.; 2. servo dynamically by each single shaft of Coordination Treatment, be target to improve planar obit simulation profile control performance, take into account the servo dynamic perfromance of single shaft, directly reduce diaxon planar obit simulation profile errors.

Profile control on the ordinary meaning is on the basis that does not change single shaft servocontrol structure, by shared processing to information between each axis servomotor, promptly adopt the cross-coupling control strategy, provide additional profile errors information, realize the closed-loop control of track following to each.Do not adopt the former to be because: when one of them is disturbed when influence, other axles do not obtain corresponding feedback information, still think its be operate as normal and do not take the deterioration of effective indemnifying measure with minimizing profile performance.For different pursuit paths, as non-linear profile and deep camber corner, the influence that geometrical property is produced the track following process is not considered in single shaft servo tracking control yet.

The cross-coupling control based on attitude methods such as times that Koren proposed in 1980 (Cross-Coupled Control is called for short CCC) is the method for basic processing multiaxis Coupling Control.Its main thought is by obtaining each tracking error information, estimate profile errors in real time, and design Coupling Control rule is carried out feedback compensation to profile errors, with the compensation rate decoupling zero on each, obtain the linear computation model ε of profile errors=-C _xE _x+ C _yE _y, shown in Figure 10,11.Because that adopt is P type controller u _c=w _pε, it can not control non-linear profile.Then Koren etc. improves P (ratio) type controller, proposes the non-linear profile control of variable gain, as: circular arc, the mode that adopts straight line to approach, then according to different gain coefficient decoupling distribution profile errors, thus independent each single shaft of control.The profile errors of straight line or circular arc is followed the tracks of in this basic cross-coupling control device estimation under fixing coordinate system, the gained profile errors is the linear combination of each tracking error, compensator adopts the P type or the PID type controller of preset parameter, so only under the little situation of low speed or curved transition, be controlled at the contour accuracy aspect than uniaxiality tracking and improve a lot.

In addition, in order to reduce the computational complexity of contour motion Coupling Control, improve the stability of Coupling Control variable-gain, strengthen the robustness of Coupling Control Unit, people such as T.C.Chiu have designed in 1998 based on the profile errors transport function control mode of (ContourError Transfer Function is called for short CETF).It has been described by the dynamic relationship between the profile errors of the profile errors of Coupling Control system generation and not decoupled system generation, thereby CETF is regarded as the transport function of an equivalent system, the cross-coupling control device is converted into a single-input single-output system that becomes when of equal value designs.

Based on the cross-coupling control device design of profile errors transport function, still be the systematization method for designing of conventional cross Coupling Control in essence, by C that profile errors is gained _x, C _yBe similar in [1,1] scope, adopt Quantitative Feedback technology (QFT) design single-input single-output system of equal value, can be met the cross-coupling control compensator of the stability and the robustness of cross-couplings change in gain.

In addition, change the setting of general profile errors calculating in people such as Chin J.H. in 1999 based on Cartesian coordinates, the system dynamics equation is transformed to task coordinate system (Task Coordiante Frame, be called for short TCF) on, by make the dynamic perfromance coupling of each at the additional feedforward of each single shaft link precompensation device, the system transter matrix is diagonalizable under the task coordinate system thereby make, with normal direction error component approximate contours error, can be the design simplification of cross-coupling control device two separate single loop control unit designs, realize decoupling zero control normal error component and tangential error component.This Coupling Control Unit is based on the method design of frequency domain, keeps system performance constant by coordinate transform, and supposition track profile is little section straight line constitutes, and then this method is limited under the contour motion of low angular frequency and the piecewise linear reference contours and uses.Its essence is to come the approximate contours error with the error normal component on the task coordinate system, improve the bandwidth of normal error component emphatically, avoid the influence of system's latent instability factor of bringing because of each unidirectional bandwidth of raising of do not have selecting, considered the influence of speed and curvature direction simultaneously, as the controller feed-forward information system dynamics performance is compensated, thereby can be improved the profile performance of system under the high speed deep camber.Except straight line and circle, set up TCF for the complex geometry characteristic plane curve of any curved transition, on-line calculation is bigger; With the approximate profile errors that replaces of the normal error component under the TCF frame, fail reliably to realize profile errors decoupling zero control; Certainly, the processing power of controller and external disturbance uncertain for plant model also a little less than.

Other are applied to the method for profile control, as the passive coupling design of Controller, are that a kind of CCC control law based on the Lyapunov function is realized the multiaxis coordination control in the continuous path motion.By adjusting weights the profile performance is carried out different configurations with tracking performance, adopt integration Backstepping technical design control law, this control rate actual for the speed acceleration information that comprises reference locus the time become PD control, can realize local stability, be that cost improves the profile performance by increasing weights to sacrifice tracking performance.

Obviously, the essence that designs based on passivity CCC is that the notion of optimal energy control is incorporated in the controlled target design, sets the weights of energy function to adapt to different demands for control by adjustment.In actual application, for the complex outline curve, be difficult for finding its implied expression mode, and the discontinuous switching owing to control rate can reduce the profile performance when error approaches zero, has then limited the scope of its application to a great extent.

The present invention is directed to the profile control problem of plane or spatial complex track, complete solution is provided, and other advantages that exceed prior art are provided.

Summary of the invention

The present invention promptly is for adapting to commercial production midplane/space profiles control field, wide type accuracy requirement that improves constantly and reliable control stabilization sexual demand, and the high-performance servo control algorithm of designing at complicated track profile control procedure.Purpose is: consider from profile sum of errors profile control stiffness two aspects, based on complex geometry characteristic locus tracking Control, propose the contour outline control method of a kind of high precision, high stability.

Realize technical scheme of the present invention, comprise following concrete steps:

(1) to the positional servosystem modeling, promptly carry out the Model Distinguish of position control system, determine model order, obtain the position control system transport function, linear regression data model and model parameter J to be identified thereof _Eq, B _Eq

(2) design according to the controlled device servo characteristic of identification gained, is determined stable control limit based on the positioner of PD type, and the real-time regulated controller parameter is to obtain the stability controller of little overshoot, no phase lag;

(3) the servo delay that interpolation feed rate and increment type position output valve are relatively obtained is input in the positioner, obtains the output of real-time controlled quentity controlled variable;

(4) as shown in Figure 3,, adopt motion command formation output buffering,, produce prediction feedforward input based on the data model that step (1) identification obtains according to the interpolation data of storing in the controller hardware fifo queue; Self-adaptation prediction controller APC by the present invention sets handles input quantity 301, servo delay 302, controlled quentity controlled variable 304, exists under the situation of external disturbance 306, obtains stable position output 308;

(5) as shown in figure 13, each axis servomotor outgoing position value is carried out cross-couplings handle, calculate coupling profile errors 1330 in real time,, speed ring inner control amount is once revised by compensating for coupling controller 1350,1350 '; Carry out decoupling zero through the predictive compensation gain process of setting 1320 and calculate, result of calculation is through v _Kx, v _KyHandle, that realizes feed rate repaiies accent in real time;

(6) on the basis of step (5), the self-adaptation prediction controller that integrating step (4) is realized, as Figure 14, the controlled quentity controlled variable U of 1410 pairs of each axis servomotors of each self-adaptation prediction controller _x, U _yFurther adjustment is done in output.Improve the stable output of real-time controlled quentity controlled variable according to historical position data and external disturbance information; Between centers is coordinated control and is adopted cross-couplings precompensation mechanism 1330, simultaneously to front end input feed rate 1315 and control output quantity U _x, U _yRepair and mediate reason.

The described specific implementation of above-mentioned steps (1) is: as shown in Figure 1, set up the linear servo controlling models of plane motion XY axle.Ignore electric time delay, determine that motor servo system 102,106 identification exponent numbers are 2, derive the linear system Differential Equation Model of simplification: $J\stackrel{\&CenterDot;\&CenterDot;}{y}+B\stackrel{\&CenterDot;}{y}+F\left(\stackrel{\&CenterDot;}{y}\right)+{\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">T</figure-callout>}}_1=u,$

Wherein: $J=\frac{J_1{R}_a}{K_g{K}_T}\&CenterDot;\frac{2\mathrm{\&pi;}}{P},$ $B=\frac{K_T{K}_g+R_{a}D}{K_g{K}_T}\&CenterDot;\frac{2\mathrm{\&pi;}}{P},$ $F_{\mathrm{fn}}\left(\stackrel{\&CenterDot;}{y}\right)=\frac{R_a}{K_g{K}_T}{F}_{\mathrm{fn}}^{\&prime;}\left(\frac{2\mathrm{\&pi;y}}{P}\right),$ $T_l=\frac{R_a}{K_g{K}_T}{T}_l^{\&prime;},$ P is a leading screw pitch, y,

Be respectively displacement, speed and the acceleration of lead screw transmission slide block.If do not consider earlier the non-linear friction item And external load T ' _l, obtain from servo input reference voltage u to displacement the transport function of output y, with J _Eq, B _EqBe model parameter to be identified, $G\left(s\right)=\frac{Y\left(s\right)}{U\left(s\right)}=\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="J\; eq\; s"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">J</figure-callout>}}_{\mathrm{eq}}{s}^{}{}^2+B_{\mathrm{eq}}s},$ Wherein: $J_{\mathrm{eq}}=\frac{J}{K_t{K}_{\mathrm{\&omega;P}}{K}_{\mathrm{eq}}R},$ $B_{\mathrm{eq}}=\frac{B+K_{\mathrm{\&omega;P}}{K}_{\mathrm{eq}}}{K_t{K}_{\mathrm{\&omega;P}}{K}_{\mathrm{eq}}R}.$ The frequency sweep identification experiment of being carried out is that the positional information of gathering output terminal y can obtain the linear Identification model from reference voltage input terminal to the position output terminal to the u of system end input continuing excitation sine sweep signal.Figure 2 shows that the physical model of vertical axes Z axle, the motion of Z axle is the dynamic process that workpiece and counterweight are moved simultaneously, transmits displacement by wire rope by two pulleys between the two, and according to the analysis to the XY axle, the dynamic model that can get the Z axle is: $\left\{\begin{array}{c}{m}_1{\stackrel{\&CenterDot;\&CenterDot;}{x}}_1+B_1{\stackrel{\&CenterDot;}{x}}_1+B_3({\stackrel{\&CenterDot;}{x}}_1-{\stackrel{\&CenterDot;}{x}}_2)+K(x_1-x_2)=\frac{\mathrm{\&tau;}}{R}\\ m_2{\stackrel{\&CenterDot;\&CenterDot;}{x}}_2+B_2{\stackrel{\&CenterDot;}{x}}_2+B_3({\stackrel{\&CenterDot;}{x}}_2-{\stackrel{\&CenterDot;}{x}}_1)+K(x_2-x_1)=0\end{array}\right.,$ After Laplace transformation, can get:

$\frac{x_1\left(s\right)}{\mathrm{\&tau;}\left(s\right)}=\frac{p_2{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">s</figure-callout>}}^2+p_{1}s+p_0}{R^{2}s(s^3+q_2{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">s</figure-callout>}}^2+q_{1}s+q_0)},$ Wherein: ${\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">p</figure-callout>}}_2=\frac{1}{m},$ ${\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">p</figure-callout>}}_1=\frac{B_2+B_3}{m_1{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">m</figure-callout>}}_2},$ ${\mathrm{<figure-callout\; id="0"\; label="p"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">p</figure-callout>}}_0=\frac{K}{m_1{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">m</figure-callout>}}_2},$ $q_2=\frac{m_1(B_2+B_3)+m_2(B_1+B_3)}{R{m}_1{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">m</figure-callout>}}_2},$ $q_1=\frac{K(m_1+m_2)+B_1{\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">B</figure-callout>}}_2+B_1{B}_3+B_2{B}_3}{R{m}_1{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">m</figure-callout>}}_2},$ ${\mathrm{<figure-callout\; id="0"\; label="q"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">q</figure-callout>}}_0=\frac{K(B_1+B_2)}{R{m}_1{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">m</figure-callout>}}_2}$ Be model parameter to be identified.

The described specific implementation of above-mentioned steps (2) is as shown in Figure 3, to set second order controlled device 305:

$\left\{\begin{array}{c}A\left(z^{-1}\right)y\left(k\right)=z^{-1}B\left(z^{-1}\right)u\left(k\right)+y_d\\ A\left(z^{-1}\right)=1+a_1{z}^{-1}+a_2{z}^{-1}\\ B\left(z^{-1}\right)=b_0+b_1{z}^{-1}\end{array}\right.,$ Wherein: y _dBe the disturbance 306 of normal value.Adopt PD controller 303, establish the closed loop proper polynomial of carrying out POLE PLACEMENT USING and be: A _m(z ^-1), comprise integral element in the controller, get:

$\left\{\begin{array}{c}F\left(z^{-1}\right)=(1-z^{-1})(1+f_2{z}^{-1})=1+(f_2-1)z^{-1}-f_2{z}^{-2}(-1<{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">f</figure-callout>}}_2<0)\\ G\left(z^{-1}\right)={\mathrm{<figure-callout\; id="0"\; label="g"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_0+g_1{z}^{-1}+g_2{z}^{-2}\end{array}\right.,$ If

Have by Fig. 4 transport function: F (z ^-1) u (k)=H (z ^-1) r (k)-G (z ^-1) y (k), 405 for calculating Fu (k), therefore can get controlled quentity controlled variable 410 expression formula u (k) and be:

$u\left(k\right)=\frac{K_D(1-z^{-1})}{1+f_2{z}^{-1}}\&CenterDot;y\left(k\right)+[K_P+\frac{K_I}{(1-z^{-1})(1-f_2{z}^{-1})}]\&CenterDot;[r\left(k\right)-y\left(k\right)],$ See Fig. 5, obtain PD controller 501 behind the 303 process transforms, it is output as 505.Amount wherein to be asked is: f ₂, g ₀, g ₁, g ₂Or f, K _P, K _I, K _DThe POLE PLACEMENT USING condition that need to satisfy: i.e. closed loop proper polynomial A _m(z ^-1), thereby have: (AF+z ^-1BG) y (k)=z ^-1BHr (k), obviously characteristic equation is:

(AF+z ^-1BG)=A _m(z ^-1), wherein: AF and z ^-1BG is 4 order polynomials, and relatively homogeneous power coefficient can get 4 independent equations, thereby solves unknown quantity f ₂, g ₀, g ₁, g ₂, counter separating obtains K _P, K _i, K _dWith F (z ^-1) be set as quadratic polynomial, select A _m(z ^-1) time get deg A _m=2, A _m=1+a _M1z ^-1+ a _M2z ^-2, corresponding continuous second order polynomial is:

${\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">s</figure-callout>}}^2+2\mathrm{\&xi;}{\mathrm{\&omega;}}_{n}s+{\mathrm{\&omega;}}_{\mathrm{<figure-callout\; id="2"\; label="n"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">n</figure-callout>}}^2=(s+s_1)\left({s+s}_2\right).$

By servo controlled device characteristic ξ, ω _nValue and the servo sample period T of setting ₀, can get a of this quadratic polynomial _M1, a _M2Value satisfy following equation: $\left\{\begin{array}{c}{a}_{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">m</figure-callout>}1}=-2\exp (-\mathrm{\&xi;}{\mathrm{\&omega;}}_n{T}_0)\cos \left({\mathrm{\&omega;}}_{\mathrm{<figure-callout\; id="0"\; label="n\; T"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">n</figure-callout>}}{T}_{}{}_0\sqrt{1-{\mathrm{\&xi;}}^2}\right)\\ a_{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">m</figure-callout>}2}=\exp (-2\mathrm{\&xi;}{\mathrm{\&omega;}}_n{T}_0)\end{array}\right..$ At this moment, when adaptive characteristic promptly was embodied in A, the B variation of reaction plant model characteristic, F, G be servo-actuated simultaneously also; On the basis of object parameters on-line identification, dispose stable control limit, the self-adaptation PD controlled variable of adjusting.

The described specific implementation of above-mentioned steps (3) is: realize motion command fifo queue transmission mechanism, as shown in Figure 6, deposit (601) successively in default motion command buffer zone (that is the default fifo queue in hardware store district) by the motion command that backdrop procedure produces.Each command order is specified offset position by pointer, and a base address is set, and complies with the setting of biasing address in the servo period of setting from formation, takes out (602) interpolation position data, calls the movement function storehouse and carries out corresponding servo motion task.

The described specific implementation of above-mentioned steps (4) is: by identification gained I type servo-drive system, provide following quadratic form weighting objective function, $J=\underset{j=-M_R}{\overset{\&infin;}{\mathrm{\&Sigma;}}}[e^T\left(j\right)\mathrm{Qe}\left(j\right)+{\mathrm{\&Delta;u}}^T\left(j\right)\mathrm{H\&Delta;u}\left(j\right)],$ Wherein e (j) is respectively the real-time controlled quentity controlled variable of servo tracking sum of errors with u (j), and Q, H are weighting coefficient matrix to be designed, and the controlled quentity controlled variable computing formula that z then adds interference value is: u (k)=F _e∑ e (i)+F _xX (k)+F _Pr(z) P (k)+F _Pd(z) d (k), wherein:

$F_{\mathrm{pr}}\left(z\right)=F_r\left(1\right)z+F_r\left(2\right){\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">z</figure-callout>}}^2+\&CenterDot;\&CenterDot;\&CenterDot;+F_r\left(M_r\right)z^{M_r},$ $F_{\mathrm{pd}}\left(z\right)=F_d\left(0\right)z+F_d\left(1\right){\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">z</figure-callout>}}^2+\&CenterDot;\&CenterDot;\&CenterDot;+F_d\left(M_d\right)z^{M_d},$

Then optimum prediction servo-control system constitutes as shown in Figure 7.715,725 are and prediction feedforward desired value and prediction feedforward interference value.Based on univariate SISO control system, FEEDBACK CONTROL 710 invariant vector during with feedforward control matrix 715 one dimensions, then in u (k) formula one, binomial is same as typical servo-control system, three, four are embodied prediction servo-control system characteristics.F _Pr(z) and F _Pd(z) procedure parameter to be identified is input, output, the M according to the last step _R(M _d) the following desired value in rank and interference value make optimum and predict control action.It is rationally effectively to use the exponent number of following information that this prediction method of servo-controlling is applied in the greatest problem that solves in the real-time servo-control system, the big more then servo-controlled precision of quantity of information is high more, but it is also many more that DSP handles resource consumption, and the data of emulation are in theory: prediction surpasses exponent number can not influence servo-controlled performance 30 times yet.Fig. 8 promptly shows concrete algorithm and realizes, wherein 301,308 is the I/O of system, 302 is servo delay, export 840 actings in conjunction in self-adaptation prediction controller 830 with prediction feedforward interference value 720, FEEDBACK CONTROL, handle through feed-forward module 820, act on feedback controller 810, whole control is output as 850, acts on controlled device 710 and obtains final prediction position output valve.Fig. 9 shows the concrete observing and controlling flow process under Fig. 1 servocontrol structure, and wherein 910 is semiclosed loop speed control control model by motor band encoder feedback, and 920 are adorned the position control mode of grating chi feedback by linear servo mechanism.

The described specific implementation of above-mentioned steps (5) is: at first calculate real-time contour motion error, as shown in figure 10,1010 is the tracking target track, y _d(k) and y _d(k-1) be adjacent 2 points on the tracking target track, y (k) and y (k-1) they are real-time follow-up Place object point, then among the figure in the geometric relationship, and ε _kWith ε _K-11020 is the target trajectory profile errors, e _kWith e _K-11030 is the target trajectory tracking error.Accordingly, transformed coordinate system in Figure 11, under pursuit path 1010 ', 1102 and 1105 are the tangential and normal tracing error under the real-time task coordinate system, and similar, 1130 ' is tracking error, and 1102 can to admit be 1020 profile errors at this moment.Can get geometric relationship thus:

$\left\{\begin{array}{c}{\stackrel{\&RightArrow;}{e}}_k=|y\left(k\right)-y_d\left(k\right)|={\stackrel{\&RightArrow;}{e}}_{\mathrm{kx}}+{\stackrel{\&RightArrow;}{e}}_{\mathrm{ky}}\\ {\stackrel{\&RightArrow;}{\mathrm{\&epsiv;}}}_k=\underset{\mathrm{\&tau;}\&Element;[{\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">k</figure-callout>}}_0,k_f]}{\min }\left|\right|y\left(k\right)-y_d\left(\mathrm{\&tau;}\right)\left|\right|\end{array}\right.,$ If 1102 equivalences of normal direction deviation are handled for profile errors, are then had:

$\left\{\begin{array}{c}{\stackrel{\&RightArrow;}{e}}_k={\stackrel{\&RightArrow;}{e}}_{\mathrm{kx}}+{\stackrel{\&RightArrow;}{e}}_{\mathrm{ky}}\\ {\stackrel{\&RightArrow;}{\mathrm{\&epsiv;}}}_k={\stackrel{\&RightArrow;}{\mathrm{\&epsiv;}}}_{\mathrm{kx}}+{\stackrel{\&RightArrow;}{\mathrm{\&epsiv;}}}_{\mathrm{ky}}\end{array}\right.\&DoubleRightArrow;\left\{\begin{array}{c}{e}_{\mathrm{kx}}=x_r-x_m\\ e_{\mathrm{ky}}=y_r-y_m\\ {\mathrm{\&epsiv;}}_{\mathrm{kx}}=-\mathrm{\&epsiv;}\sin \mathrm{\&theta;}\\ {\mathrm{\&epsiv;}}_{\mathrm{ky}}=\mathrm{\&epsiv;}\cos \mathrm{\&theta;}\end{array}\right.,$ Y (x wherein _m, y _m), y _d(x _r, y _r), wherein:

$\mathrm{\&theta;}\left(k\right)={\tan }^{-1}\frac{f_x\left(k\right)}{f_y\left(k\right)}={\tan }^{-1}\frac{x_r\left(k\right)-x_r(k-1)}{y_r\left(k\right)-y_r(k-1)},$ I.e. this time instructions location track tangent line and X-axis angle, f _x(k), f _y(k) be the feed rate of XY axle in each servo period τ, profile errors ε can be with the following formula approximate representation:

ε _k=e _KyCos θ-e _KxSin θ has then wherein promptly embodied the relation of tracking error and profile errors.By position control system speed control model, I type system speed tracking error $e(\&infin;)=\frac{1}{K_v},$ That is: $K_v=\frac{v_{\max }}{e},$ Can get speed ring tracking error computing formula: $E=\frac{v}{K},$ Then can rewrite the general computing formula of profile errors is:

$\mathrm{\&epsiv;}=\frac{K_x-K_y}{K_x{K}_y}\&CenterDot;[y_r\left(t\right)-y_r(t-1)]\&CenterDot;\cos \mathrm{\&theta;}\&DoubleLeftRightArrow;\mathrm{\&epsiv;}=\frac{K_x-K_y}{K_x{K}_y}\&CenterDot;[x_r\left(t\right)-x_r(t-1)]\&CenterDot;\sin \mathrm{\&theta;}.$

In addition, as Figure 12, describe by the parameter vector function by how much offset point target geometric locuses: r (u)={ x (u), y (u) }, order $x^{\&prime;}=\frac{\mathrm{dx}}{\mathrm{du}},$ $y^{\&prime;}=\frac{\mathrm{dy}}{\mathrm{du}},$ Then: k ₀Be that curve is at r ₀The relative curvature of point: ${\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">k</figure-callout>}}_0=[\frac{x^{\&prime;}{y}^{\&prime;\&prime;}+x^{\&prime;\&prime;}{y}^{\&prime;}}{{(x^{\&prime;2}+y^{\&prime;2})}^{3/2}}{]}_{u=u_0},$ Be to follow the tracks of trace arbitrarily wherein: 1010 ", follow the tracks of data point P _iCoordinate be:

Distance is: D _i(s)=[(X _i-X (s)) ²+ (Y _i-Y (s)) ²] ^1/2, the track following error is 1230, thus can push away profile errors 1220 calculating formulas of arbitrary plane curve:

On the basis of path predictive compensation algorithm (PM) and cross-couplings (CCC), servo feed rate (pulsatile once rough interpolation value) and controlled quentity controlled variable output are repaiied accent.As shown in figure 13,1305 the expression be the rough interpolation feed rate that the front end interpolator provides, 1310 the expression be X-axis input instruction position, obtain the initial feed rate 1315,1315 ' of rough interpolation through differentiation element, obtain profile errors calculated value 1330 according to servo output 1380,1380 ', after speed ring adjustment factor 1360, by feed rate 1315, the 1315 ' input end of 1320 decoupling zeros in the XY axle, it is carried out 1325 repair accent, obtain reference instruction position input 1335 through an integral element.Wherein, 1340 is each controlled device, and 1370 is each axis Position Control interference components; In addition, profile errors calculating 1330 is also carried out a controlled quentity controlled variable adjustment to ring 1350 in the speed.

The described specific implementation of above-mentioned steps (6) is: establishing p is current sampling point, and e (k)=R (k)-y (k), y are system's output, and R is a desired value, and e is a tracking error, and u (k) is the input of controlled device, establishes F _R=[F _R1, F _R2..., F _Rj..., F _RM], F _d=[F _D0, F _D1..., F _Dj..., F _DM], controlled quentity controlled variable u (k) can be expressed as: $u\left(k\right)=\underset{j=0}{\overset{M}{\mathrm{\&Sigma;}}}{F}_{\mathrm{Rj}}R(k+j)+F_{\mathrm{dj}}d(k+j),$ Be expressed as with incremental form:

$\mathrm{\&Delta;u}\left(k\right)=\underset{j=0}{\overset{M}{\mathrm{\&Sigma;}}}{F}_{\mathrm{Rj}}\mathrm{\&Delta;R}(k+j)+F_{\mathrm{dj}}\mathrm{\&Delta;d}(k+j).$ By the prediction control APC module of Fig. 8, adjustable parameter F _Rj, F _DjThe adaptive control rate, can pass through J _pAsk local derviation that it is descended along the negative gradient direction to it, obtain:

$F_{\mathrm{Rj}}={2B}_{\mathrm{Rj}}T\underset{k=-M+1}{\overset{p}{\mathrm{\&Sigma;}}}\frac{y\left(k\right)}{u\left(k\right)}R(k+j)\mathrm{Qe}\left(k\right)+\mathrm{\&Delta;R}(k+j)\mathrm{H\&Delta;u}\left(k\right),$ $F_{\mathrm{dj}}=2B_{\mathrm{bj}}T\underset{k=-M+1}{\overset{p}{\mathrm{\&Sigma;}}}\frac{y\left(k\right)}{u\left(k\right)}d(k+j)\mathrm{Qe}\left(k\right)+\mathrm{\&Delta;d}(k+j)\mathrm{H\&Delta;u}\left(k\right),$ In the use, controlled device G _p(k)=y (k)/u (k) and linear Identification model are stable convergence, otherwise may cause the output vibration, shown in Figure 14 being on the basis of Figure 13 cross-couplings precompensation control, the prediction control module 1410 that adds each, utilize cross-couplings precompensation control to rough interpolation primary feed rate 1315,1315 ' and controller 1345,1345 ' output repair accent, be aided with of the correction of each APC module, make that the cross-coupling control process is stable more, the result is more accurate in control controlled quentity controlled variable.

Compare with existing track following, profile control technology, the remarkable result that the present invention embodied is:

(1) by the servo-drive system linear Identification, obtains the position control data model, determine the parameter and the adaptive law of on-line identification model, thereby can obtain accurate, stable prediction control output.This method is the servo-control system of no plant model commonly used, and controlled quentity controlled variable output is more stable, tracking position of object output is more accurate.

(2) method of prediction adaptive control laws used in the present invention and cross-couplings precompensation control is particularly suitable for complex geometry feature contour control, than typical case's PD Position Tracking algorithm that feedovers, is reducing aspect the profile errors good performance is arranged.

(3) the adaptive real-time control method of prediction that is adopted among the present invention, make full use of following and historical control information, adopt the Quadratic Optimal Control method, compare with some track profile control algolithms that propose in the early time, by setting suitable self-adaptation prediction control rate, can obtain more stable controlled quentity controlled variable output and more accurate contour accuracy.

By the detailed description of following relevant example embodiment and with reference to relevant drawings, can characterize the above feature of the present invention and make other advantages also apparent.

Description of drawings

Fig. 1 shows the servomechanism installation that carries out track profile control experiment.

Fig. 2 is the analysis chart about Z axle motion principle.

Fig. 3 shows the control block diagram that the method that adopts POLE PLACEMENT USING is carried out PD control.

Fig. 4 is the discrete control system block diagram of band disturbance.

Fig. 5 shows the discrete control system block diagram that carries out the PID POLE PLACEMENT USING.

Fig. 6 shows the formation mode of front end interpolation data transmission.

Fig. 7 is the schematic diagram of prediction feedforward control.

Fig. 8 shows the self-adaptation prediction control block diagram that has external interference.

Fig. 9 shows the observing and controlling configuration flow figure based on Fig. 1.

Figure 10 shows profile errors and the tracking error geometric relationship under the Cartesian coordinates.

Figure 11 shows wide error and the tracking error geometric relationship under the real-time task coordinate system.

Figure 12 shows wide error of the arbitrary curve with how much data points and tracking error geometric relationship.

Figure 13 is the control structure figure of a typical application cross-couplings profile errors precompensation mechanism.

Figure 14 is the cross-couplings profile errors precompensation control structure figure based on prediction control.

Embodiment

A concrete example embodiment is, with reference to the topworks of figure 1, and plane complicated track following that carries out and profile control examples.

The servo-control mechanism figure of Fig. 1 promptly shows the working control platform of carrying out this example.102 and 106 are respectively XY axle linear motion guide rail, link together in orthogonal mode, and are fixed on 101 mechanism's stage support; XY axle travel position information is surveyed by the grating dipstick metering, 106 moving component carrying real work task; Limit switch 104 reaches back origin switch before and after all being equipped with on each kinematic axis.Parts moving linearly drives the ball-screw transmission by AC servo motor 105 by servoamplifier, and velocity information is returned by motor encoder, and positional information then obtains (all joining for three) by 103 (X-axis) grating chi.The servo motion structure of vertical axes Z axle 109 is same as the XY axle, carry moving component on the linear motion axis, owing to be the vertical axes structure, then the Z axle stops by the motor band-type brake or counterweight before and after installing additional, 108 promptly is the counterweight that this experimental provision installed additional, 108 are connected by the coaxial cable pulley 110,110 ' that is fixed in Z axle top with rigid cord 203 with Z axle moving component 207, counterweight 108 is by straight line optical axis guide rod constrained motion direction, 206 is vertical axes transmission ball-screw, 107 take advantage of for fixing of Z axle, and 208,209 are respectively Z axis drive motor and scrambler.Motion is by servo controller (DSP-Based) drive controlling, and as shown in Figure 9, the control system configuration all can be according to basic servocontrol pattern, and wherein 910 loops are speed ring encoder information feedbacks, and 920 loops are position ring grating chi information feedback.

Realize high-performance track profile control algolithm of the present invention, adopt position control mode.Can obtain preferable control effect and profile control accuracy according to following control flow.Twin shaft cross-coupling control block diagram shown in Figure 14 has promptly embodied essence spirit of the present invention, and its core has mainly embodied a kind of real-time servo control algorithm based on plant model.

Concrete identification flow process to each controlled device is:

At first, construct the nominal model of each, it has mainly been explained the structural dynamic characteristics of profile Control work platform by one group of linear differential equation.Identification controlled device nominal model need have the electromechanical properties of servo control mechanism accurately and understand, and can utilize empirical data/theoretical formula/measurement data to obtain the properest understanding of this respect, and this category information is used for determining the parameter of structure of models and definition model;

Then, the control performance index of design servo-drive system, the feasible design that runs on the control algolithm of controller has feasibility, reaches optimization;

At last, model being verified, is the control system model with robustness with the model of guaranteeing to pick out.It not only comprises the theoretical model of system dynamics, and comprises uncertainty description and noise description.For the purpose of the present invention, the modelling verification problem is explained clearly becomes a kind of linear time invariant system, it has the uncertainty and the experimental data of norm bounded structure, carries out modelling verification in frequency domain, determine model whether with inputoutput data coupling and controller whether with Model Matching.

With the X-axis is example, and the mode of employing sine pulse frequency sweep picks out the identification model from the input reference voltage to the outgoing position, is the sine sweep signal of 1.5V in the lasting input of controlled device input end 0.1～500Hz, amplitude, is described as: $u\left({\mathrm{kT}}_s\right)=\mathrm{\&alpha;}\sin \left(\mathrm{\&omega;k}{T}_s\right)+{\hat{F}}_{\mathrm{fu}}\left(v\right),$ K=0,1,2 ..., N, wherein: Ts is the sampling period, and k is the identification sampling number, and ω is the input angle frequency, and α is the input amplitude,

Be the moment of friction compensation term, the positional information of gathering output terminal is in real time used the method for least square identification objects is carried out off-line identification in the System Discrimination tool box of Matlab/Simulink.Obtain each is input to position output from reference voltage transport function.For the identification of vertical axes Z axle, because the Z axle has bigger phase lag than the XY axle, need add a pure lag system when determining Z shaft model structure, set suitable time constant, can obtain its simplified model in the low-frequency range scope with same method.

Obtain after the linear Identification model of each kinematic axis, then need set up the real-time controlling models of contact with it.The present invention adopts the mode of POLE PLACEMENT USING that the real time position controlled variable is adjusted.

Set the second order controlled device $\left\{\begin{array}{c}A\left(z^{-1}\right)y\left(k\right)=z^{-1}B\left(z^{-1}\right)u\left(k\right)+y_d\\ A\left(z^{-1}\right)=1+a_1{z}^{-1}+a_2{z}^{-1}\\ B\left(z^{-1}\right)={\mathrm{<figure-callout\; id="0"\; label="b"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">b</figure-callout>}}_0+b_1{z}^{-1}\end{array}\right.,$ Wherein, y _dBe the disturbance of normal value, A _m(z ^-1) promptly be the proper polynomial of carrying out POLE PLACEMENT USING, get:

$\left\{\begin{array}{c}F\left(z^{-1}\right)=(1-z^{-1})(1+f_2{z}^{-1})=1+(f_2-1)z^{-1}-f_2{z}^{-2}(-1<{\mathrm{<figure-callout\; id="2"\; label="f"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">f</figure-callout>}}_2<0)\\ G\left(z^{-1}\right)={\mathrm{<figure-callout\; id="0"\; label="g"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_0+g_1{z}^{-1}+g_2{z}^{-2}\end{array}\right.,$ Then have:

Because F (z ^-1) u (k)=H (z ^-1) r (k)-G (z ^-1) y (k), can get the controlled quentity controlled variable expression formula and be:

$u\left(k\right)=\frac{K_D(1-z^{-1})}{1+f_2{z}^{-1}}\&CenterDot;y\left(k\right)+[K_P+\frac{K_I}{(1-z^{-1})(1-f_2{z}^{-1})}]\&CenterDot;[r\left(k\right)-y\left(k\right)],$ As seen ratio and integral control action are in deviation signal, and the differential action is used for output signal y (k).Then control corresponding parameter real-time update amount is: $\left\{\begin{array}{c}{K}_P=-\frac{{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_1+{2\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">g</figure-callout>}}_2}{1+f_2}\\ T_d=\frac{({\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_1+g_2)f_2-g_2}{{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_1+{2\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">g</figure-callout>}}_2}\&CenterDot;T_0\\ T_i=-\frac{({\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_1+{2g}_2)\&CenterDot;T_0}{(1+f_2)({\mathrm{<figure-callout\; id="0"\; label="g"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_0+{\mathrm{<figure-callout\; id="1"\; label="g"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">g</figure-callout>}}_1+g_2)}\end{array}\right.,$ When adaptive characteristic was embodied in A, the B variation of reaction plant model, F, G be servo-actuated simultaneously also.On the basis that the POLE PLACEMENT USING pid parameter self-adaptation of above definite system is adjusted, to object parameters on-line identification, thereby the step of asking for controlled quentity controlled variable u (k) is as follows:

(1) sets the controlled quentity controlled variable initial value, the input of reading system, output numerical value;

(2) input controlled quentity controlled variable u (k), the outgoing position value y (k) according to controlled device estimates controlled device data model translation operator coefficient

(3) solve F, G and obtain H according to POLE PLACEMENT USING Adaptive PID Control algorithm;

(4) solve the value of real-time u (k).

Wherein,

The on-line identification method adopt discrimination method to LTI system linearity parameter model.

At last, use forgetting factor method (RFF) but real-time identification goes out model parameter:

$\left\{\begin{array}{c}\mathrm{\&Theta;}(k+1)=\frac{P\left(k\right)\mathrm{\&psi;}(k+1)}{\mathrm{\&lambda;}+{\mathrm{\&psi;}}^T(k+1)P\left(k\right)\mathrm{\&psi;}(k+1)}\\ P(k+1)=\frac{1}{\mathrm{\&lambda;}}(P\left(k\right)-\mathrm{\&Theta;}(k+1)\mathrm{\&psi;}(k+1)P\left(k\right))\\ \widehat{\mathrm{\&theta;}}(k+1)=\widehat{\mathrm{\&theta;}}\left(k\right)+\mathrm{\&Theta;}(k+1)(y(k+1)-{\mathrm{\&psi;}}^T(k+1)\widehat{\mathrm{\&theta;}}\left(k\right))\end{array}\right.,$ Wherein Θ (k) is an intermediate variable with P (k), and λ is that forgetting factor is selected to approach 1 decimal, if data are in the past handled in λ=1 expression on an equal basis.

Fig. 6 shows the function flow graph of motion control commands queue operation.Use the intention of queue operation to be in the present invention in order to solve and the unbalanced interpolation iterative computation time, the interpolation calculation task is positioned in the background process program of DSP, thereby reduce real-time servocontrol task, guarantee the hard real-time feature of position servo control task.Specific implementation is, in the background process program, whenever calculates a Place object point, all is sent in the servo instruction, deposits (function call) pop down among the FIFO as a motion control instruction, i.e. 601 operations.In each servo period arrives, all need to obtain a movement instruction from FIFO and carry out servo operation, promptly 602 operations play stack.Owing to need to manage some job tasks in the backdrop procedure, then the hardware resource outside the job task management is all allocated the usefulness of the interpolation calculating of giving complex curve.Be the continuity of guaranteeing to move, then need guarantee to have in the fifo queue stack enough motion control instruction for servo calling, RAM opens up one section space in the DRAM of motion controller or outside the sheet, realizes the dynamic memory process of movement instruction.Set the memory space upper and lower limit, guarantee at first that in the motion incipient stage instruction number that is not less than the lower limit capacity is arranged in the fifo queue, can start whole servo motion, in operational process, regularly monitor the stack contents amount, stop the interpolation action in case memory capacity surpasses the upper limit, only offer the real time kinematics steering order at each servo period.The length setting that reaches the fifo queue stack of determining of fifo queue stack upper and lower limit need be decided on interpolation calculation task complicacy, but its cardinal rule then is to interrupt the supply of movement instruction in real-time servocontrol process.In the present invention, the FIFO stack capability is set at 1K * 16 bytes, and the upper limit promptly is the stack top of formation, and lower limit set is 25% of a capacity of queue.If the motion control instruction capacity is less than minimum tolerance limit in the formation, then need increase interpolation cycle to improve the motion command capacity.

Carry out the control of track profile and adopt cross-coupling control method shown in Figure 13, calculate profile errors 1330 in real time, it is adjusted controlled quentity controlled variable as value of feedback.The compensation of profile errors value of feedback, generalized case is to act on loop feedback, controlled quentity controlled variable to controlled device is proofreaied and correct, be about to 1330 and act on positioner 1345 and 1345 ' output terminal by compensating for coupling controller (Figure 13 only comprises proportional component) 1350 and 1350 ', 1345 and 1345 ' has embodied the servo dynamic of kinematic axis, and simplifying becomes the proportional component processing.Coupling Control Unit 1350 and 1350 ' value are determined according to 1345 and 1345 ' matching properties.The calculating of profile errors 1330 of wherein being coupled is asked for by Figure 10 and plane geometry shown in Figure 11 relation, has than detailed theoretical derivation in invention step (2), and specific embodiment adopts formula:

$\mathrm{\&epsiv;}=\frac{K_x-K_y}{K_x{K}_y}\&CenterDot;[y_r\left(t\right)-y_r(t-1)]\&CenterDot;\cos \mathrm{\&theta;}\&DoubleLeftRightArrow;\mathrm{\&epsiv;}=\frac{K_x-K_y}{K_x{K}_y}\&CenterDot;[x_r\left(t\right)-x_r(t-1)]\&CenterDot;\sin \mathrm{\&theta;},$ Or formula:

Parameter declaration can contrast Figure 13.Speed ring proportional gain K _PxAnd K _PyThe aforementioned self-adaptation proportional gain factor of definite reference K _p, T _i, T _dMode obtain, adopt the trapezoidal integration method of approximation by general incremental PID control law, obtain $\left\{\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="C2007100302280017H1.png,C2007100302280017H2.png"\; state="\{\{state\}\}">p</figure-callout>}}_1=K_p\&CenterDot;(1+\frac{T_d}{{\mathrm{<figure-callout\; id="0"\; label="T"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">T</figure-callout>}}_0}+\frac{{\mathrm{<figure-callout\; id="0"\; label="T"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">T</figure-callout>}}_0}{2\&CenterDot;T_i})\\ {\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="C2007100302280019H1.png"\; state="\{\{state\}\}">p</figure-callout>}}_2=-K_p\&CenterDot;(1+\frac{{2T}_d}{T_0}-\frac{{\mathrm{<figure-callout\; id="0"\; label="T"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">T</figure-callout>}}_0}{2\&CenterDot;T_i})\\ p_3=K_p\&CenterDot;\frac{T_d}{{\mathrm{<figure-callout\; id="0"\; label="T"\; filenames="C2007100302280017H2.png"\; state="\{\{state\}\}">T</figure-callout>}}_0}\end{array}\right.,$ Thereby obtain PID adaptive control laws u (k)=u (the k-1)+p of each kinematic axis ₁E (k)+p ₂E (k-1)+p ₃E (k-2).K _{ε x}And K _{ε y}Parameter needs to set according to the servo characteristic and the track profile geometric properties of different motion axle.

If profile errors value of feedback compensating action in position ring, can be used profile errors precompensation mechanism, be provided with and estimate decoupling zero yield value 1360, calculate 1320 by each axis error decoupling zero, the rough interpolation feed rate of each kinematic axis is repaiied accent, K _vThe setting of value can reference:

Wherein ρ is a curvature.Can embody the present invention's basic compensating for coupling spirit like this.

Figure 14 shows a kind of cross-couplings precompensation motion control strategy that comprises each single kinematic axis prediction feedforward control.Based on the position control model that the identification of step (1) institute is come out, be translated into the state-space model of servo-drive system, $\left\{\begin{array}{c}x(k+1)=\mathrm{Ax}\left(k\right)+\mathrm{Bu}\left(k\right)+\mathrm{Ed}\left(k\right)\\ y\left(k\right)=\mathrm{Cx}\left(k\right)\end{array}\right.,$ A wherein _x, B _x, C _x, A _y, B _y, C _yValue can determine according to the system linear transport function.In the formula: d (k) is a known interfering signal, and when satisfying controllability ornamental condition, the target setting value signal is R (k), and then error signal is: e (k)=R (k)-y (k), and then derive following error system equation:

$\left[\begin{array}{c}e[k+1]\\ \mathrm{\&Delta;x}[k+1]\end{array}\right]=\left[\begin{array}{cc}{I}_M& -\mathrm{CA}\\ 0& A\end{array}\right]\left[\begin{array}{c}e\left[k\right]\\ \mathrm{\&Delta;x}\left[k\right]\end{array}\right]+\left[\begin{array}{c}-\mathrm{CB}\\ B\end{array}\right]\mathrm{\&Delta;u}\left(k\right)+\left[\begin{array}{c}{I}_M\\ 0\end{array}\right]\mathrm{\&Delta;R}(k+1)+\left[\begin{array}{c}-\mathrm{CE}\\ E\end{array}\right]\mathrm{\&Delta;d}\left(k\right),$ Or

X ₀(k+1)=φ X ₀(k)+G Δ u (k)+G _RΔ R (k+1)+G _dΔ d (k), wherein φ is the coefficient { a to be identified of dynamic data model ₁, a ₂..., a _mb ₁, b ₂..., b _n, discrimination method is with reference to aforementioned forgetting factor method (RFF).Fig. 8 promptly shows application self-adapting prediction controller action in the synoptic diagram of feedback controller and controlled device, under the input of reference instruction position 301 and external interference 725, self-adaptation prediction controller 830, prediction control module 820 and feedback control module 810 actings in conjunction and controlled device 710 obtain prediction control output 308.Corresponding Figure 14, each single shaft control amount U is further adjusted in self-adaptive control module 1410 outputs _x, U _y, jointly single shaft is done in order to reach better control effect with cross-coupling control device 1350,1350 ' regulating and controlling amount.Another aspect that each single shaft is regulated is the feedrate override to 1315 and 1315 ', repaiies the accent amount by decoupling controller 1320 decisions, and the feed rate 1325,1325 ' after the adjusting passes through an integral element output order position.

Thus, a specific embodiment of the present invention is described, concrete implementing procedure needs in conjunction with different execution units.Determined controlled device nominal model is the data model through linearization process, and the uncertainty description of model should be contained the attribute information of model structure and parameter to greatest extent, is provided at nominal model envelope on every side.In the present embodiment, each kinematic axis servo loop all comprises input, the controlled quentity controlled variable adjustment of a plurality of feedforwards, feedback, and therefore the control system that is constituted is described and should be adopted state-space method.

Should understand, even in above-mentioned example embodiment, proposed a plurality of implementation steps and implementation method involved in the present invention, comprise control structure and concrete function detail, but this announcement only is illustrative, and many variations of its details aspect, the particularly adjustment of part-structure in principle of the invention scope and layout, the maximum magnitude of answering continuation to express to the upperseat concept of the expressed term of appended claims.Its special Application Example will change according to special type of driver and execution unit, keep the identical functions type basically simultaneously, and not depart from scope of the present invention and design.

Claims (1)

1, a kind of contour outline control method of complicated track may further comprise the steps:

(1) to the positional servosystem modeling, promptly carry out the Model Distinguish of position control system, determine model order, obtain the position control system transport function, linear regression data model and model parameter J to be identified thereof _Eq, B _Eq

(2) design according to the controlled device servo characteristic of identification gained, is determined stable control limit based on the positioner of PD type, and the real-time regulated controller parameter obtains the stability controller of little overshoot, no phase lag;

(3) the servo delay that interpolation feed rate and increment type position output valve are relatively obtained is input in the positioner, obtains the output of real-time controlled quentity controlled variable;

(4) according to the interpolation data of storing in the controller hardware fifo queue, adopt motion command formation output buffering, based on the data model that step (1) identification obtains, produce prediction feedforward input; By self-adaptation prediction controller input quantity (301), servo delay (302), controlled quentity controlled variable (304) are handled, obtained stable position output (308);

(5) each axis servomotor outgoing position value is carried out cross-couplings and handle, calculate coupling profile errors (1330) in real time,, speed ring inner control amount is revised by compensating for coupling controller (1350,1350 '); Through the precompensation gain process, to carry out decoupling zero and calculate, result of calculation is through XY axial feed rate correction component v _Kx, v _KyHandle, realize feed rate repaiied accent in real time;

(6) on the basis of step (5), the self-adaptation of integrating step (4) prediction controller, each self-adaptation prediction controller

(1410) to the controlled quentity controlled variable U of each axis servomotor _x, U _yFurther adjustment is done in output; Improve the stable output of real-time controlled quentity controlled variable according to historical position data and external disturbance information; Between centers is coordinated control and is adopted cross-couplings precompensation mechanism, simultaneously to front end input feed rate (1315) and control output quantity U _x, U _yRepair and mediate reason.

Images