CN100589055C - Control method for changing structure of underwater hiding-machine space based on recursion fuzzy neural network - Google Patents

Abstract

The invention provides an underwater vehicle space variable structure control method based on a recurrent fuzzy neural network. A RFNN-based rudder control system, a RFNN-based envelope rudder controlsystem and a RFNN-based tail elevating rudder control system are designed, and combined together to form a united control system for underwater vehicle space motion. In this invention, the RFNN-basedrudder control system, the RFNN-based envelope rudder control system and the RFNN-based tail elevating rudder control system are designed, and further combined to compose a united control system forunderwater vehicle space motion, since the RFNN can make real-time adjustment of a controller gain epsilon, according to uncertain items of the system, the system not only has good dynamic property, but also effectively lowers buffeting and improves the robustness of an underwater vehicle automatic rudder control system.

Description

Underwater hiding-machine space variable structure control method based on the recurrence fuzzy neural network

(1) technical field

What the present invention relates to is a kind of control method, particularly a kind of control method of underwater hiding-machine auto-pilot control system.

(2) background technology

Underwater hiding-machine auto-pilot control technology is the important directions of underwater hiding-machine movement control technology development.Tradition underwater hiding-machine automatic pilot system normally is made up of surface level yaw rudder system and vertical plane elevator system, is called combined control system (or centralized control system).The core of this design philosophy is outstanding kinematic behavior of dividing plane motion, and controlling Design is simplified, and realizes easily, and adapt with the dirigibility of reality behaviour ship that therefore dividing graphic design method is the main mode of research underwater hiding-machine combined control system.The part but this design also comes with some shortcomings, major defect is: owing to do not consider the coupling effect of motion in controlling Design, last combined control system robustness is weakened.General compensation method is to utilize various corrections, compensation system to improve the anti-interference of system, and this makes system become complicated, and definite optimization work of controlled variable is cumbersome.

When in fact underwater hiding-machine is done spatial movement under water, the accurate movement equation is difficult to obtain usually, coupling influence between the non-linear and equation in the equation of motion, make the maneuvering motion of underwater hiding-machine become one and have stronger probabilistic system, the method that some are controlled based on mathematical models is difficult to the designing requirement that reaches satisfied as PID, decoupling zero, the most excellent control algolithm.Become the integrated approach of structure control as a kind of control, its major advantage is to adopt coarse mathematical model to carry out controlling Design, can estimate uncertain interference effect, stronger robustness is arranged, relatively be suitable for the design of underwater hiding-machine kinetic control system.Up to now, (Recurrent Fuzzy Neural Network, adaptive sliding moding structure control method RFNN) is not applied in the submersible space motion combined control system under water as yet based on the recurrence fuzzy neural network.

(3) summary of the invention

The object of the present invention is to provide and a kind ofly can estimate uncertain interference effect, have the underwater hiding-machine space variable structure control method based on the recurrence fuzzy neural network of stronger robustness.

The object of the present invention is achieved like this:

Design yaw rudder control system, casing rudder control system and tail elevating rudder control system respectively based on the recurrence fuzzy neural network, yaw rudder control system, tail elevating rudder control system, casing rudder control system are combined, constitute the combined control system of underwater submersible space motion: at first by system measurements device input course angle, trim angle, the degree of depth and conversion depth information, wherein, course angle inputs to the yaw rudder control system based on the recurrence fuzzy neural network, by adjusting gain parameter output actual direction rudder angle; Then actual direction rudder angle and speed of a ship or plane information are inputed to rectification building-out system, is the first rudder of rectification building-out system output that casing rudder rudder angle information and trim angle, the degree of depth and conversion depth information are input to the casing rudder control system based on the recurrence fuzzy neural network, export actual first rudder rudder angle by adjusting gain parameter, simultaneously the tail that the tail vane rudder angle information of rectification building-out system output and trim angle information are input to based on the recurrence fuzzy neural network rises the rudder control system, by adjusting gain parameter, export actual tail vane rudder angle

The control law of described course control is:

${\mathrm{\δ}}_r=\frac{(I_z-\frac{1}{2}\·\mathrm{\ρ}{L}^5{N}_{\stackrel{.}{r}}^{\′})[c_1{\stackrel{.}{e}}_1+{\stackrel{..}{\mathrm{\ψ}}}_d\left(t\right)+\widehat{\mathrm{\ϵ}}\·\mathrm{sat}\left(\frac{s}{\mathrm{\μ}}\right)]-\frac{1}{2}\mathrm{\ρ}{L}^4{N}_r^{\′}u\stackrel{.}{\mathrm{\ψ}}}{\frac{1}{2}\mathrm{\ρ}{L}^3{N}_{{\mathrm{\δ}}_r}^{\′}{u}^2}$

δ in the formula _rBe the yaw rudder rudder angle of underwater hiding-machine, ρ is a density of sea water, and L is a hull length, I _zBe the moment of inertia of underwater hiding-machine around the Z axle, u represents the linear velocity component of hull coordinate system on X-axis, i.e. longitudinal velocity, N ' _r,

Be the zero dimension hydrodynamic force coefficient of underwater hiding-machine, c ₁Be the coefficient of switching function, ride gain

Adjusted in real time by a recurrence fuzzy neural network, sat () is a saturation function, and s is the sliding mode of system's reality, and μ is a boundary layer thickness, is a little arithmetic number, e ₁Be the deviation of course angle, ψ _dBe given course angle.

The control law of described casing rudder degree of depth control is:

${\mathrm{\δ}}_b=\frac{(m-\frac{1}{2}\mathrm{\ρ}{L}^3{Z}_{\stackrel{.}{w}}^{\′})[c_2{\stackrel{.}{e}}_2+{\stackrel{..}{\mathrm{\ζ}}}_d\left(t\right)+\widehat{\mathrm{\ϵ}}\·\mathrm{sat}\left(\frac{s}{\mathrm{\μ}}\right)]-\frac{1}{2}\mathrm{\ρ}{L}^2{Z}_w^{\′}u\stackrel{.}{\mathrm{\ζ}}}{\frac{1}{2}\mathrm{\ρ}{L}^2{Z}_{{\mathrm{\δ}}_b}^{\′}{u}^2}$

δ in the formula _bBe the casing rudder rudder angle of underwater hiding-machine, m is the quality of hull, and ρ is a density of sea water, and L is a hull length, and u represents the linear velocity component of hull coordinate system on X-axis, i.e. longitudinal velocity, Z ' _w,

Be the zero dimension hydrodynamic force coefficient of underwater hiding-machine, c ₂Be the coefficient of switching function, ride gain

Adjusted in real time by a recurrence fuzzy neural network, sat () is a saturation function, and s is the sliding mode of system's reality, and μ is a boundary layer thickness, is a little arithmetic number, e ₂Be the dark deviation of diving, ζ _dBe the instruction degree of depth.

The control law that described tail rises the control of the rudder degree of depth is:

${\mathrm{\δ}}_s=\frac{(I_y-\frac{1}{2}\mathrm{\ρ}{L}^5{M}_{\stackrel{.}{q}}^{\′})[c_3{\stackrel{.}{e}}_3+{\stackrel{..}{\mathrm{\θ}}}_d\left(t\right)+\widehat{\mathrm{\ϵ}}\·\mathrm{sat}\left(\frac{s}{\mathrm{\μ}}\right)]-\frac{1}{2}\mathrm{\ρ}{L}^4{M}_q^{\′}u\stackrel{..}{\mathrm{\θ}}+\mathrm{mgh\θ}}{\frac{1}{2}\mathrm{\ρ}{L}^3{M}_{{\mathrm{\δ}}_s}^{\′}}$

δ in the formula _sBe the tail vane rudder angle of underwater hiding-machine, m is the quality of hull, and ρ is a density of sea water, and h is that height of C.G., the L of hull is hull length, I _yBe the moment of inertia of underwater hiding-machine around Y-axis, u represents the linear velocity component of hull coordinate system on X-axis, i.e. longitudinal velocity, M ' _q,

Be the zero dimension hydrodynamic force coefficient of underwater hiding-machine, c ₃Be the coefficient of switching function, ride gain

Adjusted in real time by a recurrence fuzzy neural network, sat () is a saturation function, and s is the sliding mode of system's reality, and μ is a boundary layer thickness, is a little arithmetic number, e ₃Be the deviation of instruction trim angle and actual trim angle, θ _dBe the instruction trim angle.

The present invention utilizes the design philosophy on branch plane, yaw rudder control system, casing rudder control system and tail elevating rudder control system have been designed respectively based on RFNN, and yaw rudder control system, tail elevating rudder control system, casing rudder control system pressed the certain way group and be in the same place, constitute the combined control system of underwater submersible space motion.In the design of controller, other state variable relevant with controlled variable done to disturb processing, realized full decoupled control.

Underwater submersible space motion combined control system structured flowchart as shown in Figure 1, the major function of measurement system is the required state variable of Measurement and Control System among the figure.ψ _d, θ _d, ζ _d, H _dBe respectively that the instruction course angle is set, trim angle is set, and the degree of depth is set and the conversion degree of depth is set.Rolling corrector and side wash compensator all are housed, with the influence of compensating motion coupling in common underwater hiding-machine PID auto navigator.The underwater hiding-machine actual measurement

With

Deng signal is angular velocity under the inertial coordinates system, if bigger rolling motion appears in ship, under the inertial coordinates system

With

Not exclusively corresponding with the angular velocity q and the r of hull coordinate system, these signals will be changed before feeding back to control system, and this work is finished by the rolling corrector usually.In variable structure control system, this correction not necessarily because the strong robustness of variable structure control system, can be with uncorrected

With

Feed back in the control system,

With With the deviation of q and r, handle by the interference of control system.When the underwater hiding-machine underwater turning, owing to the normal navigation that change in depth influences underwater hiding-machine can appear in the influence of surveying wash.Therefore in the steering rudder, must be aided with suitable elevating rudder and change the depthkeeping cycle.Because becoming the design philosophy of structure controller is that coupling influence is counted in the interference, thereby can reach the coupling effect of control action.Therefore in variable structure control system, rectification building-out control can be cancelled, and has simplified the complicacy of system and device.The present invention rises the rudder control system by yaw rudder control system, casing rudder control system, the tail that designs based on RFNN, and then formation underwater submersible space motion combined control system, because RFNN can adjust controller gain in real time according to uncertain item size in the system The system that makes not only has good dynamic perfromance, can also reduce effectively to buffet, and improves the robustness of underwater hiding-machine auto-pilot control system.

(4) description of drawings

Fig. 1 is a underwater submersible space motion combined control system structured flowchart;

Fig. 2 is underwater submersible space motion combined control system implementing procedure figure;

Fig. 3 is the System with Sliding Mode Controller block diagram with the RFNN estimated gain;

Fig. 4-a to Fig. 4-d is underwater hiding-machine depthkeeping cycle simulation result figure, wherein speed of a ship or plane u=10kn, yaw rudder δ _r=10 °,----the motion simulation result when----expression vertical plane control system does not participate in controlling; Wherein: Fig. 4-a is the space motion path of underwater hiding-machine; Fig. 4-b, Fig. 4-c are the curves of output of trim angle and course angle; Fig. 4-d is the curve of output of yaw rudder and elevating rudder.

Fig. 5-a to Fig. 5-d is underwater hiding-machine space maneuver simulation result figure, and wherein extremely 10m, trim angle are set at θ under water from 50m floats under water for speed of a ship or plane u=12kn, underwater hiding-machine _d=5 °, the conversion degree of depth are 5m, require course angle ψ to change to 120 ° and the motion simulation curves that keep from 0 ° simultaneously; Wherein Fig. 5-a is the space motion path of underwater hiding-machine; Fig. 5-b, Fig. 5-c are the curves of output of trim angle and course angle; Fig. 5-d is the curve of output of yaw rudder and elevating rudder.

(5) embodiment

For example the present invention is done description in more detail below in conjunction with accompanying drawing:

1) RFNN of yaw rudder course control becomes the structure controller design

The horizontal plane motion of underwater hiding-machine comprises axially-movable, transverse movement and rolling motion, and its equation of motion is defined as follows respectively:

Axial equation: $u=U_0(1-e^{-0.52/\left|\stackrel{.}{\mathrm{\ψ}}\right|L})---\left(1\right)$

$m\left[\stackrel{.}{v}+\mathrm{ur}\right]=\frac{1}{2}\mathrm{\ρ}{L}^4\left[Y_{\stackrel{.}{r}}^{\′}\stackrel{.}{r}+Y_{\stackrel{.}{p}}^{\′}\stackrel{.}{p}r\right]+\frac{1}{2}\mathrm{\ρ}{L}^3[Y_r^{\′}\mathrm{ur}+Y_p^{\′}\mathrm{up}+Y_{\stackrel{.}{v}}^{\′}\stackrel{.}{v}]$

Horizontal equation: $+\frac{1}{2}\mathrm{\ρ}{L}^2[Y_v^{\′}\mathrm{uv}+Y_{v\left|v\right|}^{\′}v|{(v^2+w^2)}^{\frac{1}{2}}\left|\right]+\frac{1}{2}\mathrm{\ρ}{L}^2\left[Y_{{\mathrm{\δ}}_r}^{\′}{u}^2{\mathrm{\δ}}_r\right]---\left(2\right)$

The driftage equation:

$I_z\stackrel{.}{r}+(I_y-I_x)\mathrm{pq}=\frac{1}{2}\mathrm{\ρ}{L}^5\left[N_{\stackrel{.}{r}}^{\′}\stackrel{.}{r}+N_{\stackrel{.}{p}}^{\′}\stackrel{.}{p}\right]+\frac{1}{2}\mathrm{\ρ}{L}^4[N_{\stackrel{.}{v}}^{\′}\stackrel{.}{v}+N_p^{\′}\mathrm{up}+N_r^{\′}\mathrm{ur}]---\left(3\right)$

$+\frac{1}{2}\mathrm{\ρ}{L}^3\left[N_v^{\′}\mathrm{uv}\right]+\frac{1}{2}\mathrm{\ρ}{L}^3\left[N_{{\mathrm{\δ}}_r}^{\′}{u}^2{\mathrm{\δ}}_r\right]$

U, v, w represent three linear velocity components on X, Y, Z axle of hull coordinate system, i.e. longitudinal velocity, transverse velocity and vertical velocity in the formula; P, q, r are angular velocity in roll, angular velocity in pitch and course angle speed;

θ, ψ are heeling angle, trim angle and the course angle of underwater hiding-machine; M, L, h are respectively quality, length and the heights of C.G. of hull; ρ, g are density of sea water and acceleration of gravity; ξ, η, ζ are the geographic coordinate position of hull initial point; I _x, I _y, I _zBe the moment of inertia of underwater hiding-machine around X, Y, Z axle; x _G, y _G, z _GBarycentric coordinates position for underwater hiding-machine; δ _r, δ _b, δ _sBe the yaw rudder rudder angle of underwater hiding-machine, first rudder (casing rudder) rudder angle, tail vane rudder angle; a _T, b _T, c _TThe zero dimension propulsive coefficient; u _cThe speed of benchmark navigation attitude; X ' _Qq, Y ' _Pq, Z ' _Pp, K ' _Qr, M ' _Pp, N ' _PqBe respectively the zero dimension hydrodynamic force coefficient of underwater hiding-machine.

Consider following relation:

$\stackrel{.}{\mathrm{\ψ}}=r---\left(4\right)$

Under the speed control good premise, can think that the speed u of turning motion remains unchanged; If speed is not controlled, under the situation that keeps the main propulsion motor invariablenes turning speed, by the actual change of axial equation decision speed.General surface level course controlling Design only uses equation (3) and (4) to design, and measures because transverse velocity v is difficult, therefore often ignores the coupling influence of equation (2) to course angle speed r, and this influence is finally born by the robustness of system, can obtain:

$I_z\stackrel{..}{\mathrm{\ψ}}+(I_y-I_x)\mathrm{pq}=\frac{1}{2}\mathrm{\ρ}{L}^5[N_{\stackrel{.}{r}}^{\′}\stackrel{..}{\mathrm{\ψ}}+N_{\stackrel{.}{p}}^{\′}\stackrel{.}{p}]+\frac{1}{2}\mathrm{\ρ}{L}^4[N_{\stackrel{.}{v}}^{\′}\stackrel{.}{v}+N_p^{\′}\mathrm{up}+N_r^{\′}u\stackrel{.}{\mathrm{\ψ}}]---\left(6\right)$

$+\frac{1}{2}\mathrm{\ρ}{L}^3\left[N_v^{\′}\mathrm{uv}\right]+\frac{1}{2}\mathrm{\ρ}{L}^3\left[N_{{\mathrm{\δ}}_r}^{\′}{u}^2{\mathrm{\δ}}_r\right]$

Equation (6) is changed into following form:

$\stackrel{..}{\mathrm{\ψ}}=\frac{1}{I_z-\frac{1}{2}\·\mathrm{\ρ}{L}^5{N}_{\stackrel{.}{r}}^{\′}}[\frac{1}{2}\mathrm{\ρ}{L}^4{N}_r^{\′}u\stackrel{.}{\mathrm{\ψ}}+\frac{1}{2}\mathrm{\ρ}{L}^3{N}_{{\mathrm{\δ}}_r}^{\′}{u}^2{\mathrm{\δ}}_r+d_1\left(t\right)]---\left(7\right)$

Wherein

$d_1\left(t\right)=\frac{1}{2}\mathrm{\ρ}{L}^5{N}_{\stackrel{.}{p}}^{\′}\stackrel{.}{p}+\frac{1}{2}\mathrm{\ρ}{L}^4{N}_{\stackrel{.}{v}}^{\′}\stackrel{.}{v}+\frac{1}{2}\mathrm{\ρ}{L}^4{N}_p^{\′}\mathrm{up}+\frac{1}{2}\mathrm{\ρ}{L}^3{N}_v^{\′}\mathrm{uv}-(I_y-I_x)\mathrm{pq}---\left(8\right)$

Following formula shows that we move other degree of freedom to the coupling influence of yawing rotation, the nonlinear characteristic of yawing rotation itself, and the model mismatch of horizontal plane motion model (designing a model of being adopted itself is exactly the motion model of simplifying) all disturbs d by surveying ₁(t) handle.Choose:

e ₁(t)＝ψ _d(t)-ψ(t)

${\stackrel{.}{e}}_1\left(t\right)={\stackrel{.}{\mathrm{\ψ}}}_d\left(t\right)-\stackrel{.}{\mathrm{\ψ}}\left(t\right)$

${\stackrel{..}{e}}_1\left(t\right)={\stackrel{..}{\mathrm{\ψ}}}_d\left(t\right)-\stackrel{..}{\mathrm{\ψ}}\left(t\right)$

In the formula, ψ _d(t) be given course angle,

Be respectively single order, second derivative; e ₁(t) be the deviation of course angle.

The selection switching function is:

${\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="C2008100642560013H1.png,C2008100642560013H5.png"\; state="\{\{state\}\}">s</figure-callout>}}_1=c_1{\mathrm{<figure-callout\; id="1"\; label="e"\; filenames="C2008100642560013H1.png,C2008100642560013H5.png"\; state="\{\{state\}\}">e</figure-callout>}}_1+{\stackrel{.}{e}}_1$

Can in the hope of:

${\stackrel{.}{s}}_1=c_1{\stackrel{.}{e}}_1+{\stackrel{..}{e}}_1=c_1{\stackrel{.}{e}}_1+({\stackrel{..}{\mathrm{\&psi;}}}_d\left(t\right)-\stackrel{..}{\mathrm{\&psi;}}\left(t\right))$

$=c_1{\stackrel{.}{e}}_1+{\stackrel{..}{\mathrm{\&psi;}}}_d\left(t\right)-\frac{1}{I_z-\frac{1}{2}\&CenterDot;\mathrm{\&rho;}{L}^5{N}_{\stackrel{.}{r}}^{\&prime;}}[\frac{1}{2}\mathrm{\&rho;}{L}^4{N}_r^{\&prime;}u\stackrel{..}{\mathrm{\&psi;}}+\frac{1}{2}\mathrm{\&rho;}{L}^3{N}_{{\mathrm{\&delta;}}_r}^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2{\mathrm{\&delta;}}_r+d_1\left(t\right)]---\left(9\right)$

Use and become the control law that structure control can get course control:

${\mathrm{\&delta;}}_r=\frac{(I_z-\frac{1}{2}\&CenterDot;\mathrm{\&rho;}{L}^5{N}_{\stackrel{.}{r}}^{\&prime;})[{\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="C2008100642560013H1.png,C2008100642560013H5.png"\; state="\{\{state\}\}">c</figure-callout>}}_1+{\stackrel{.}{e}}_1+{\stackrel{..}{\mathrm{\&psi;}}}_d\left(t\right)+\widehat{\mathrm{\&epsiv;}}\&CenterDot;\mathrm{sat}\left(\frac{s}{\mathrm{\&mu;}}\right)]-\frac{1}{2}\mathrm{\&rho;}{L}^4{N}_r^{\&prime;}u\stackrel{.}{\mathrm{\&psi;}}}{\frac{1}{2}\mathrm{\&rho;}{L}^3{N}_{{\mathrm{\&delta;}}_r}^{\&prime;}{u}^2}---\left(10\right)$

Ride gain in the formula

Adjust by a RFNN network real-time.Sat () is a saturation function, and μ is a boundary layer thickness, is a little arithmetic number.

$\mathrm{sat}(s/\mathrm{\&mu;})=\left\{\begin{array}{cc}1& s>\mathrm{\&mu;}\\ s/\mathrm{\&mu;}& \left|s\right|\&le;\mathrm{\&mu;}\\ -1& s<-\mathrm{\&mu;}\end{array}\right.$

2) RFNN of casing rudder degree of depth control becomes the structure controller design

The vertical plane motion of underwater hiding-machine comprises catenary motion and pitching, and its equation of motion is defined as follows respectively:

Vertical equation:

$m[\stackrel{.}{w}+\mathrm{vp}-\mathrm{uq}]=\frac{1}{2}\mathrm{\&rho;}{L}^4\left[Z_{\stackrel{.}{q}}^{\&prime;}\stackrel{.}{q}\right]+\frac{1}{2}\mathrm{\&rho;}{L}^3[Z_{\stackrel{.}{w}}^{\&prime;}\stackrel{.}{w}+Z_q^{\&prime;}\mathrm{uq}+Z_{\mathrm{vp}}^{\&prime;}\mathrm{vp}]$

$+\frac{1}{2}\mathrm{\&rho;}{L}^2[{\mathrm{<figure-callout\; id="0"\; label="Z"\; filenames="C2008100642560013H2.png,C2008100642560013H5.png"\; state="\{\{state\}\}">Z</figure-callout>}}_0^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2+Z_w^{\&prime;}\mathrm{uw}+Z_{\mathrm{ww}}^{\&prime;}|w{({\mathrm{<figure-callout\; id="2"\; label="v"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">v</figure-callout>}}^2+w^2)}^{\frac{1}{2}}|+Z_{\mathrm{vv}}^{\&prime;}{v}^2]---\left(11\right)$

$+\frac{1}{2}\mathrm{\&rho;}{L}^2[Z_{{\mathrm{\&delta;}}_s}^{\&prime;}{u}^2{\mathrm{\&delta;}}_s+Z_{{\mathrm{\&delta;}}_b}^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2{\mathrm{\&delta;}}_b]$

The trim equation:

$I_y\stackrel{.}{q}+(I_x-I_z)\mathrm{rp}=\frac{1}{2}\mathrm{\&rho;}{L}^5[M_{\stackrel{.}{q}}^{\&prime;}\stackrel{.}{q}+M_{\mathrm{rp}}^{\&prime;}\mathrm{rp}]+\frac{1}{2}\mathrm{\&rho;}{L}^4[M_{\stackrel{.}{w}}^{\&prime;}\stackrel{.}{w}+M_q^{\&prime;}\mathrm{uq}]$

$+\frac{1}{2}\mathrm{\&rho;}{L}^3[{\mathrm{<figure-callout\; id="0"\; label="M"\; filenames="C2008100642560013H2.png,C2008100642560013H5.png"\; state="\{\{state\}\}">M</figure-callout>}}_0^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2+M_w^{\&prime;}\mathrm{uw}+M_{\mathrm{ww}}^{\&prime;}|w{({\mathrm{<figure-callout\; id="2"\; label="v"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">v</figure-callout>}}^2+w^2)}^{\frac{1}{2}}|+M_{\mathrm{vv}}^{\&prime;}{v}^2]---\left(12\right)$

$+\frac{1}{2}\mathrm{\&rho;}{L}^3[M_{{\mathrm{\&delta;}}_s}^{\&prime;}{u}^2{\mathrm{\&delta;}}_s+M_{{\mathrm{\&delta;}}_b}^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2{\mathrm{\&delta;}}_b]-\mathrm{mgh}\sin \mathrm{\&theta;}$

Consider relational expression:

$\stackrel{.}{\mathrm{\&zeta;}}\&ap;-u\sin \mathrm{\&theta;}+w\cos \mathrm{\&theta;}---\left(13\right)$

$\stackrel{.}{\mathrm{\&theta;}}=q---\left(14\right)$

The manipulation of physical of underwater hiding-machine shows, deepen motor-driven on, generally requiring trim is 3 °～5 °, deepening rapidly and requiring trim is 5 °～7 °, and underwater hiding-machine is limited by the speed of a ship or plane and the sea area degree of depth, and from safety perspective, the trim angle maximum is in 7 °～10 °.Therefore, trim angle changes less generally speaking, and formula (13) can be reduced to

$\stackrel{.}{\mathrm{\&zeta;}}=w---\left(15\right)$

(15) formula is updated to (11) formula, and the form that changes the Linear Control equation into can get:

$\stackrel{..}{\mathrm{\&zeta;}}=\frac{1}{m-\frac{1}{2}\mathrm{\&rho;}{L}^3{Z}_{\stackrel{.}{w}}^{\&prime;}}[\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_w^{\&prime;}u\stackrel{.}{\mathrm{\&zeta;}}+\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_{{\mathrm{\&delta;}}_b}^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2{\mathrm{\&delta;}}_b+d_2\left(t\right)]---\left(16\right)$

In the formula

$d_2\left(t\right)=\frac{1}{2}\mathrm{\&rho;}{L}^4{Z}_{\stackrel{.}{q}}^{\&prime;}\stackrel{.}{q}-m[\mathrm{vp}-\mathrm{uq}]+\frac{1}{2}\mathrm{\&rho;}{L}^3[Z_q^{\&prime;}\mathrm{uq}+Z_{\mathrm{vp}}^{\&prime;}\mathrm{vp}]$

$+\frac{1}{2}\mathrm{\&rho;}{L}^2[{\mathrm{<figure-callout\; id="0"\; label="Z"\; filenames="C2008100642560013H2.png,C2008100642560013H5.png"\; state="\{\{state\}\}">Z</figure-callout>}}_0^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2+Z_{\mathrm{ww}}^{\&prime;}|w{({\mathrm{<figure-callout\; id="2"\; label="v"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">v</figure-callout>}}^2+w^2)}^{\frac{1}{2}}|+Z_{\mathrm{vv}}^{\&prime;}{v}^2]+\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_{{\mathrm{\&delta;}}_s}^{\&prime;}{u}^2{\mathrm{\&delta;}}_s---\left(17\right)$

As can be seen, we are distracter d to the influence of the influence of trim angle and manipulation tail vane from formula (17) ₂(t) handled, embodied the automatic synchronization of change structure controller the head and the tail rudder.Choose

e ₂(t)＝ζ _d(t)-ζ(t)

${\stackrel{.}{e}}_2\left(t\right)={\stackrel{.}{\mathrm{\&zeta;}}}_d\left(t\right)-\stackrel{.}{\mathrm{\&zeta;}}\left(t\right)$

${\stackrel{..}{e}}_2\left(t\right)={\stackrel{..}{\mathrm{\&zeta;}}}_d\left(t\right)-\stackrel{..}{\mathrm{\&zeta;}}\left(t\right)$

ζ in the formula _d(t) be the instruction degree of depth, e ₂(t) be the dark deviation of diving.

Choosing switching function is

${\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">s</figure-callout>}}_2=c_2{\mathrm{<figure-callout\; id="2"\; label="e"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">e</figure-callout>}}_2+{\stackrel{.}{e}}_2---\left(18\right)$

Can get this switching function differentiate

${\stackrel{.}{s}}_2=c_2{\stackrel{.}{e}}_2+{\stackrel{..}{e}}_2=c_2{\stackrel{.}{e}}_2+({\stackrel{..}{\mathrm{\&zeta;}}}_d\left(t\right)-\stackrel{..}{\mathrm{\&zeta;}\left(t\right)})$

$=c_2{\stackrel{.}{e}}_2+{\stackrel{..}{\mathrm{\&zeta;}}}_d\left(t\right)-\frac{1}{m-\frac{1}{2}\mathrm{\&rho;}{L}^3{Z}_{\stackrel{.}{w}}^{\&prime;}}[\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_w^{\&prime;}u\stackrel{.}{\mathrm{\&zeta;}}+\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_{{\mathrm{\&delta;}}_b}^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2{\mathrm{\&delta;}}_b+d_2\left(t\right)]---\left(19\right)$

Use and become the control law that structure control can get degree of depth control:

${\mathrm{\&delta;}}_b=\frac{(m-\frac{1}{2}\mathrm{\&rho;}{L}^3{Z}_{\stackrel{.}{w}}^{\&prime;})[{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">c</figure-callout>}}_2+{\stackrel{.}{e}}_2+{\stackrel{..}{\mathrm{\&zeta;}}}_d\left(t\right)+\widehat{\mathrm{\&epsiv;}}\&CenterDot;\mathrm{sat}\left(\frac{s}{\mathrm{\&mu;}}\right)]-\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_w^{\&prime;}u\stackrel{.}{\mathrm{\&zeta;}}}{\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_{{\mathrm{\&delta;}}_b}^{\&prime;}{u}^2}---\left(20\right)$

Ride gain in the formula

Adjust by a RFNN network real-time.

3) RFNN of tail elevating rudder trim control becomes the structure controller design

Wushu (14) substitution formula (12), and the form that changes into the Linear Control equation can get:

$\stackrel{..}{\mathrm{\&theta;}}=\frac{1}{I_y-\frac{1}{2}\mathrm{\&rho;}{L}^5{M}_{\stackrel{.}{q}}^{\&prime;}}[\frac{1}{2}\mathrm{\&rho;}{L}^4{M}_q^{\&prime;}u\stackrel{..}{\mathrm{\&theta;}}-\mathrm{mgh}\sin \mathrm{\&theta;}+\frac{1}{2}\mathrm{\&rho;}{L}^3{M}_{{\mathrm{\&delta;}}_s}^{\&prime;}{\mathrm{\&delta;}}_s+d_3\left(t\right)]---\left(21\right)$

In the formula

$d_3\left(t\right)=\frac{1}{2}\mathrm{\&rho;}{L}^5{M}_{\mathrm{rp}}^{\&prime;}\mathrm{rp}+\frac{1}{2}\mathrm{\&rho;}{L}^3[{\mathrm{<figure-callout\; id="0"\; label="M"\; filenames="C2008100642560013H2.png,C2008100642560013H5.png"\; state="\{\{state\}\}">M</figure-callout>}}_0^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2+M_w^{\&prime;}\mathrm{uw}+M_{\mathrm{ww}}^{\&prime;}|w{({\mathrm{<figure-callout\; id="2"\; label="v"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">v</figure-callout>}}^2+w^2)}^{\frac{1}{2}}|+M_{\mathrm{vv}}^{\&prime;}{v}^2]---\left(22\right)$

$+\frac{1}{2}\mathrm{\&rho;}{L}^4{M}_q^{\&prime;}\mathrm{uq}+\frac{1}{2}\mathrm{\&rho;}{L}^3{M}_{{\mathrm{\&delta;}}_b}^{\&prime;}{\mathrm{<figure-callout\; id="2"\; label="u"\; filenames="C2008100642560013H1.png"\; state="\{\{state\}\}">u</figure-callout>}}^2{\mathrm{\&delta;}}_b-(I_x-I_z)\mathrm{rp}$

Equally, in the trim controller, the influence of the influence of vertical velocity w and casing rudder is all contributed to distracter d ₃(t) in.Because trim angle is all smaller usually, thus sin θ ≈ θ, substitution formula (21)

$\stackrel{..}{\mathrm{\&theta;}}=\frac{1}{I_y-\frac{1}{2}\mathrm{\&rho;}{L}^5{M}_{\stackrel{.}{q}}^{\&prime;}}[\frac{1}{2}\mathrm{\&rho;}{L}^4{M}_q^{\&prime;}u\stackrel{..}{\mathrm{\&theta;}}-\mathrm{mgh}\mathrm{\&theta;}+\frac{1}{2}\mathrm{\&rho;}{L}^3{M}_{{\mathrm{\&delta;}}_s}^{\&prime;}{\mathrm{\&delta;}}_s+d_3\left(t\right)]---\left(22\right)$

Choose

e ₃＝θ _d-θ

${\stackrel{.}{e}}_3={\stackrel{.}{\mathrm{\&theta;}}}_d-\stackrel{.}{\mathrm{\&theta;}}$

${\stackrel{..}{e}}_3={\stackrel{..}{\mathrm{\&theta;}}}_d-\stackrel{..}{\mathrm{\&theta;}}$

In the formula, θ _dBe instruction trim angle, e ₃Deviation for instruction trim angle and actual trim angle.

Get switching function:

$s_3=c_3{e}_3+{\stackrel{.}{e}}_3---\left(23\right)$

This switching function differentiate is got:

${\stackrel{.}{s}}_3=c_3{\stackrel{.}{e}}_3+{\stackrel{..}{e}}_3=c_3{\stackrel{.}{e}}_3+({\stackrel{..}{\mathrm{\&theta;}}}_d\left(t\right)-\stackrel{..}{\mathrm{\&theta;}}\left(t\right))$

$=c_3{\stackrel{.}{e}}_3+{\stackrel{..}{\mathrm{\&theta;}}}_d\left(t\right)-\frac{1}{I_y-\frac{1}{2}\mathrm{\&rho;}{L}^5{M}_{\stackrel{.}{q}}^{\&prime;}}[\frac{1}{2}\mathrm{\&rho;}{L}^4{M}_q^{\&prime;}u\stackrel{..}{\mathrm{\&theta;}}-\mathrm{mgh}\mathrm{\&theta;}+\frac{1}{2}\mathrm{\&rho;}{L}^3{M}_{{\mathrm{\&delta;}}_s}^{\&prime;}{\mathrm{\&delta;}}_s+d_3\left(t\right)]---\left(24\right)$

Use and become the control law that structure control can get degree of depth control:

${\mathrm{\&delta;}}_s=\frac{(I_y-\frac{1}{2}\mathrm{\&rho;}{L}^5{M}_{\stackrel{.}{q}}^{\&prime;})[c_3+{\stackrel{.}{e}}_3+{\stackrel{..}{\mathrm{\&theta;}}}_d\left(t\right)+\widehat{\mathrm{\&epsiv;}}\&CenterDot;\mathrm{sat}\left(\frac{s}{\mathrm{\&mu;}}\right)]-\frac{1}{2}\mathrm{\&rho;}{L}^4{M}_q^{\&prime;}u\stackrel{..}{\mathrm{\&theta;}}+\mathrm{mgh\&theta;}}{\frac{1}{2}\mathrm{\&rho;}{L}^3{M}_{{\mathrm{\&delta;}}_s}^{\&prime;}}---\left(25\right)$

Ride gain in the formula

Adjust by a RFNN network real-time.

Can mediate to yaw rudder control system, casing rudder control system and tail elevating rudder control system by adaptive sliding mode controller as shown in Figure 3, realize jointly controlling of underwater hiding-machine based on RFNN.Below by emulation mode the main forms of motion of underwater hiding-machine space maneuver is verified.Considered the motor-driven of rudder in the emulation, the yaw rudder speed setting is | δ _r| _Max=3 °/s, the elevating rudder rotating speed is made as | δ _{B, s}| _Max=5 °/s.In direction controller, RFNN (1) is 2 input nodes, 12 regular nodes, 1 output node; RFNN (2) is 3 input nodes, 18 regular nodes, and 1 output node, 3 inputs are respectively: underwater hiding-machine speed of a ship or plane u, yaw rudder rudder angle δ _rWith course angle ψ.In depth controller and trim angle controller, RFNN (1) is 2 input nodes, 12 regular nodes, 1 output node; RFNN (2) is 5 input nodes, 33 regular nodes, and 1 output node, 5 inputs are respectively: underwater hiding-machine speed of a ship or plane u, yaw rudder rudder angle δ _r, head and the tail elevating rudder rudder angle δ _bAnd δ _s, and system's output feedback, for first rudder control system degree of depth ξ, be trim angle θ for the tail vane control system.

Fig. 4 provides depthkeeping turning motion simulation result, as can be seen, if the vertical plane control system does not participate in control, when circling round, underwater hiding-machine will do to change the spatially spiral motion of the degree of depth, and under the control of combined control system, can guarantee depthkeeping break-in campaign, the steady-state error of the degree of depth is less than 0.2m, and this explanation control system offside wash has good inhibitory effect.Fig. 5 provides the space maneuver simulation result.Since RFNN can be online the adjustment controller gain, make control system that good dynamic perfromance not only be arranged, also have higher steady precision.To the control requirement of the degree of depth, course and trim etc., dynamic quality was good when designed combined control system can be finished underwater submersible space motion preferably, and control accuracy is higher, has stronger robustness.

Claims (4)

1, a kind of underwater hiding-machine space variable structure control method based on the recurrence fuzzy neural network, it is characterized in that: design yaw rudder control system respectively based on the recurrence fuzzy neural network, casing rudder control system and tail elevating rudder control system, the yaw rudder control system, tail elevating rudder control system, casing rudder control system is combined, constitute the combined control system of underwater submersible space motion: at first by system measurements device input course angle, trim angle, the degree of depth and conversion depth information, wherein, course angle inputs to the yaw rudder control system based on the recurrence fuzzy neural network, by adjusting gain parameter output actual direction rudder angle; Then actual direction rudder angle and speed of a ship or plane information are inputed to rectification building-out system, the output information of rectification building-out system is input to the casing rudder control system based on the recurrence fuzzy neural network, export actual first rudder rudder angle by adjusting gain parameter, the output information of rectification building-out system is input to the tail elevating rudder control system based on the recurrence fuzzy neural network simultaneously, by adjusting gain parameter, export actual tail vane rudder angle.

2, the underwater hiding-machine space variable structure control method based on the recurrence fuzzy neural network according to claim 1 is characterized in that: the control law of described course control is:

${\mathrm{\&delta;}}_r=\frac{(I_z-\frac{1}{2}\&CenterDot;\mathrm{\&rho;}{L}^5{N}_{\stackrel{\&CenterDot;}{r}}^{\&prime;})[c_1{\stackrel{\&CenterDot;}{e}}_1+{\stackrel{\&CenterDot;\&CenterDot;}{\mathrm{\&psi;}}}_d\left(t\right)+\widehat{\mathrm{\&epsiv;}}\&CenterDot;\mathrm{sat}\left(\frac{s}{\mathrm{\&mu;}}\right)]-\frac{1}{2}\mathrm{\&rho;}{L}^4{N}_r^{\&prime;}u\stackrel{\&CenterDot;}{\mathrm{\&psi;}}}{\frac{1}{2}\mathrm{\&rho;}{L}^3{N}_{{\mathrm{\&delta;}}_r}^{\&prime;}{u}^2}$

δ in the formula _rBe the yaw rudder rudder angle of underwater hiding-machine, ρ is a density of sea water, and L is a hull length, I _zBe the moment of inertia of underwater hiding-machine around the Z axle, u represents the linear velocity component of hull coordinate system on X-axis, i.e. longitudinal velocity, N ' _r,

Be the zero dimension hydrodynamic force coefficient of underwater hiding-machine, c ₁Be the coefficient of switching function, ride gain

Adjusted in real time by a recurrence fuzzy neural network, sat () is a saturation function, and s is the sliding mode of system's reality, and μ is a boundary layer thickness, is a little arithmetic number, e ₁Be the deviation of course angle, ψ _dBe given course angle.

3, the underwater hiding-machine space variable structure control method based on the recurrence fuzzy neural network according to claim 2 is characterized in that: the control law of described casing rudder degree of depth control is:

${\mathrm{\&delta;}}_b=\frac{(m-\frac{1}{2}\mathrm{\&rho;}{L}^3{Z}_{\stackrel{\&CenterDot;}{w}}^{\&prime;})[c_2{\stackrel{\&CenterDot;}{e}}_2{+\stackrel{\&CenterDot;\&CenterDot;}{\mathrm{\&zeta;}}}_d\left(t\right)+\widehat{\mathrm{\&epsiv;}}\&CenterDot;\mathrm{sat}\left(\frac{s}{\mathrm{\&mu;}}\right)]-\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_w^{\&prime;}u\stackrel{\&CenterDot;}{\mathrm{\&zeta;}}}{\frac{1}{2}\mathrm{\&rho;}{L}^2{Z}_{{\mathrm{\&delta;}}_b}^{\&prime;}{u}^2}$

δ in the formula _bBe the casing rudder rudder angle of underwater hiding-machine, m is the quality of hull, and ρ is a density of sea water, and L is a hull length, and u represents the linear velocity component of hull coordinate system on X-axis, i.e. longitudinal velocity, Z ' _w,

Be the zero dimension hydrodynamic force coefficient of underwater hiding-machine, c ₂Be the coefficient of switching function, ride gain

Adjusted in real time by a recurrence fuzzy neural network, sat () is a saturation function, and s is the sliding mode of system's reality, and μ is a boundary layer thickness, is a little arithmetic number, e ₂Be the dark deviation of diving, ζ _dBe the instruction degree of depth.

4, the underwater hiding-machine space variable structure control method based on the recurrence fuzzy neural network according to claim 3 is characterized in that: the control law of described tail elevating rudder degree of depth control is:

${\mathrm{\&delta;}}_s=\frac{(I_y-\frac{1}{2}\mathrm{\&rho;}{L}^5{M}_{\stackrel{\&CenterDot;}{q}}^{\&prime;})[c_3{\stackrel{\&CenterDot;}{e}}_3+{\stackrel{\&CenterDot;\&CenterDot;}{\mathrm{\&theta;}}}_d\left(t\right)+\widehat{\mathrm{\&epsiv;}}\&CenterDot;\mathrm{sat}\left(\frac{s}{\mathrm{\&mu;}}\right)]-\frac{1}{2}\mathrm{\&rho;}{L}^4{M}_q^{\&prime;}u\stackrel{\&CenterDot;\&CenterDot;}{\mathrm{\&theta;}}+\mathrm{mgh\&theta;}}{\frac{1}{2}\mathrm{\&rho;}{L}^3{M}_{{\mathrm{\&delta;}}_s}^{\&prime;}}$

δ in the formula _sBe the tail vane rudder angle of underwater hiding-machine, m is the quality of hull, and ρ is a density of sea water, and h is that height of C.G., the L of hull is hull length, I _yBe the moment of inertia of underwater hiding-machine around Y-axis, u represents the linear velocity component of hull coordinate system on X-axis, i.e. longitudinal velocity, M ' _q,

Be the zero dimension hydrodynamic force coefficient of underwater hiding-machine, c ₃Be the coefficient of switching function, ride gain

Adjusted in real time by a recurrence fuzzy neural network, sat () is a saturation function, and s is the sliding mode of system's reality, and μ is a boundary layer thickness, is a little arithmetic number, e ₃Be the deviation of instruction trim angle and actual trim angle, θ _dBe the instruction trim angle.

Images