pobieranie: Simulink a Graphical Tool for Dynamic System Simulation (ang)
Pobierz dokument pdfPrzeglądaj wersję html pliku:
NCSU ASME Technical Sessions Spring 2001, Fall 2001, Spring 2002 Simulink: a Graphical Tool for Dynamic System Simulation
G. D. Buckner Department of Mechanical and Aerospace Engineering North Carolina State University
Simulink Tutorial G.D. Buckner
1
ASME Technical Sessions, Fall 2001 North Carolina State University
What is Simulink?
• a graphical, interactive software tool for modeling, simulating, and analyzing dynamic systems • enables rapid construction of "virtual prototypes" to explore design concepts at any level of detail with minimal effort • ideally suited to linear, nonlinear, continuous-time and discrete-time systems • commonly used in control system design, DSP design, communication system design, and other simulation applications • a graphical plug-in for MATLAB®, offering additional access to a range of non-graphical analysis and design tools • more information: www.mathworks.com Traditional approaches to system design typically include building a prototype followed by extensive testing and revision. This method can be both time-consuming and expensive. As an effective and widely accepted alternative, simulation is now the preferred approach to engineering design. Simulink is a powerful simulation software tool that enables you to quickly build and test virtual prototypes so that you can explore design concepts at any level of detail with minimal effort. By using Simulink to iterate and refine designs before building the first prototype, engineers can benefit from a faster, more efficient design process. Highlights Simulink provides an interactive, block-diagram environment for modeling and simulating dynamic systems. It includes an extensive library of predefined blocks that you can use to build graphical models of your systems using drag-and-drop operations. Supported model types include linear, nonlinear, continuous-time, discrete-time, multirate, conditionally executed, and hybrid systems. Models can be grouped into hierarchies to create a simplified view of components or subsystems. High-level information is presented clearly and concisely, while detailed information is easily hidden in subsystems within the model hierarchy.
Simulink Tutorial G.D. Buckner 2 ASME Technical Sessions, Fall 2001 North Carolina State University
How do you create a program in Simulink?
• Begin with !"#$Simulink Library Browser - a collection of tools for building dynamic models • Programming tools are organized into functional folders: Sources (function generators, etc), Sinks (oscilloscopes, data files, etc), Math functions, etc. • Click the New Model icon to create a blank project
Start with a very simple program:
• Open the Sources folder, and select the Sine Wave block • Drag the Sine Wave block onto your blank project, and double-click to setup its waveform parameters (Amplitude: 10, Frequency: 2*pi)
• Open the Sinks folder, and select the Scope block (an oscilloscope) • Drag the Scope block onto your blank project
Simulink Tutorial G.D. Buckner
3
ASME Technical Sessions, Fall 2001 North Carolina State University
• Click and drag your mouse cursor to create a connection between the Sine Wave output and the Scope input:
• Click the Start Simulation toolbar control: • Double-click the Scope block to view the simulation output: It doesn't look like a perfect sinewave because Simulink defaults to a variable-step solver, but in this trivial problem we have no states to solve • To change the simulation parameters, select the Simulation/Parameters menu bar item. • Change the Maximum Step Size to 0.01 sec, and the Stop Time to 5.0 sec. • Run the program again, and the Scope output now looks like: Looks better due to smaller step sizes
Now consider a more realistic simulation:
• A cart with mass m rolling on a frictionless surface.
F(t) x(t) m
• The equations of motion for this system can be found using Newton's 2nd Law:
Simulink Tutorial G.D. Buckner 4
ASME Technical Sessions, Fall 2001 North Carolina State University
m!! = ∑ Fx x !! = x F (t ) m
(1)
• Simulink integrates differential equations like (1) using Integrator blocks (in the Continuous folder). Since this system is second order, click and drag two Integrator blocks onto your project, along with one Gain block (in the Math folder), and connect them as shown:
• Assuming the cart mass is 10, run the program, and the Scope output now looks like this: This is a plot of x(t) vs. time Notice that the cart has net motion to the right due to the sinusoidal forcing • If you want to display the forcing function F(t) on the same plot, click and drag a Mux block (a multiplexer, in the Signals & Systems folder). Connect the Mux to the Scope as shown:
• Run the program, and the Scope output now looks like this: Both F(t) and x(t) vs. time
Simulink Tutorial G.D. Buckner
5
ASME Technical Sessions, Fall 2001 North Carolina State University
Let's add air drag (nonlinear damping):
• The equations of motion for this system are now:
m!! = ∑ Fx x
! F (t ) − C D x 2 !! = x m
• To add this nonlinearity, click and drag a Sum block, a Gain block, and a Math Function block, (all 3 in the Math folder) and connect them as shown:
• Notes: • To rotate a block, type Ctrl+R on the keyboard • To select the Square function on the Math Function block, doubleclick it and choose this function from the list • To change the +/- signs on the Sum block, double-click it θ
Consider a challenging control simulation:
2l
• A cart with mass m, rolling on a frictionless surface, with air drag, balancing a pole of mass mp.
F(t) m
x(t) FD(t)
Simulink Tutorial G.D. Buckner
6
ASME Technical Sessions, Fall 2001 North Carolina State University
• Control Objective: manipulate cart force F(t) to balance the pole (keep θ(t) near zero) • The equations of motion for this system can be found using Newton's 2nd Law, and making some simplifications:
(m + m )!! − C x
p
D
! ! x 2 + m p lθ! = F (t )
! m p l 2θ! + m p l!! = m p glθ x
• Assuming m=10, mp =1, l=1, J=1, and CD =1, the Simulink implementation of this complicated, nonlinear (and very unstable) system is:
• This might look like a mess, but imagine programming this system in Fortran or C! • One way to simplify the look of this system is to use a SubSystem block (in the Signals & Systems folder). Click and drag a SubSystem block onto your project, then select the entire cart/pole model and move it into the SubSystem (you can use Cut/Paste). Add an In1 and Out1 block (both in the Signals & Systems folder) as shown:
Simulink Tutorial G.D. Buckner
7
ASME Technical Sessions, Fall 2001 North Carolina State University
• Now, when you close this SubSystem, you will see only a block representing the entire cart/pole model. The input In1 is the control force F(t), the output Out1 is the angular position of the pole: • Designing a controller to stabilize this system is now much easier. One "low-tech" approach is to select a PID Controller block and combine it into your project as shown:
• By "tweaking" the proportional and derivative gains a bit (Kp=-1000, Kd=-100), this system can be stabilized from an initial deflection of 0.1 radians, as shown in the following figures.
Simulink Tutorial G.D. Buckner
8
ASME Technical Sessions, Fall 2001 North Carolina State University
Simulink Tutorial G.D. Buckner
9
ASME Technical Sessions, Fall 2001 North Carolina State University
Simulink a Graphical Tool for Dynamic System Simulation (ang)
NCSU ASME Technical Sessions Spring 2001, Fall 2001, Spring 2002 Simulink: a Graphical Tool for Dynamic System Simulation
G. D. Buckner Department of Mechanical and Aerospace Engineering North Carolina State University
Simulink Tutorial G.D. Buckner
1
ASME Technical Sessions, Fall 2001 North Carolina State University
What is Simulink?
• a graphical, interactive software tool for modeling, simulating, and analyzing dynamic systems • enables rapid construction of "virtual prototypes" to explore design concepts at any level of detail with minimal effort • ideally suited to linear, nonlinear, continuous-time and discrete-time systems • commonly used in control system design, DSP design, communication system design, and other simulation applications • a graphical plug-in for MATLAB®, offering additional access to a range of non-graphical analysis and design tools • more information: www.mathworks.com Traditional approaches to system design typically include building a prototype followed by extensive testing and revision. This method can be both time-consuming and expensive. As an effective and widely accepted alternative, simulation is now the preferred approach to engineering design. Simulink is a powerful simulation software tool that enables you to quickly build and test virtual prototypes so that you can explore design concepts at any level of detail with minimal effort. By using Simulink to iterate and refine designs before building the first prototype, engineers can benefit from a faster, more efficient design process. Highlights Simulink provides an interactive, block-diagram environment for modeling and simulating dynamic systems. It includes an extensive library of predefined blocks that you can use to build graphical models of your systems using drag-and-drop operations. Supported model types include linear, nonlinear, continuous-time, discrete-time, multirate, conditionally executed, and hybrid systems. Models can be grouped into hierarchies to create a simplified view of components or subsystems. High-level information is presented clearly and concisely, while detailed information is easily hidden in subsystems within the model hierarchy.
Simulink Tutorial G.D. Buckner 2 ASME Technical Sessions, Fall 2001 North Carolina State University
How do you create a program in Simulink?
• Begin with !"#$Simulink Library Browser - a collection of tools for building dynamic models • Programming tools are organized into functional folders: Sources (function generators, etc), Sinks (oscilloscopes, data files, etc), Math functions, etc. • Click the New Model icon to create a blank project
Start with a very simple program:
• Open the Sources folder, and select the Sine Wave block • Drag the Sine Wave block onto your blank project, and double-click to setup its waveform parameters (Amplitude: 10, Frequency: 2*pi)
• Open the Sinks folder, and select the Scope block (an oscilloscope) • Drag the Scope block onto your blank project
Simulink Tutorial G.D. Buckner
3
ASME Technical Sessions, Fall 2001 North Carolina State University
• Click and drag your mouse cursor to create a connection between the Sine Wave output and the Scope input:
• Click the Start Simulation toolbar control: • Double-click the Scope block to view the simulation output: It doesn't look like a perfect sinewave because Simulink defaults to a variable-step solver, but in this trivial problem we have no states to solve • To change the simulation parameters, select the Simulation/Parameters menu bar item. • Change the Maximum Step Size to 0.01 sec, and the Stop Time to 5.0 sec. • Run the program again, and the Scope output now looks like: Looks better due to smaller step sizes
Now consider a more realistic simulation:
• A cart with mass m rolling on a frictionless surface.
F(t) x(t) m
• The equations of motion for this system can be found using Newton's 2nd Law:
Simulink Tutorial G.D. Buckner 4
ASME Technical Sessions, Fall 2001 North Carolina State University
m!! = ∑ Fx x !! = x F (t ) m
(1)
• Simulink integrates differential equations like (1) using Integrator blocks (in the Continuous folder). Since this system is second order, click and drag two Integrator blocks onto your project, along with one Gain block (in the Math folder), and connect them as shown:
• Assuming the cart mass is 10, run the program, and the Scope output now looks like this: This is a plot of x(t) vs. time Notice that the cart has net motion to the right due to the sinusoidal forcing • If you want to display the forcing function F(t) on the same plot, click and drag a Mux block (a multiplexer, in the Signals & Systems folder). Connect the Mux to the Scope as shown:
• Run the program, and the Scope output now looks like this: Both F(t) and x(t) vs. time
Simulink Tutorial G.D. Buckner
5
ASME Technical Sessions, Fall 2001 North Carolina State University
Let's add air drag (nonlinear damping):
• The equations of motion for this system are now:
m!! = ∑ Fx x
! F (t ) − C D x 2 !! = x m
• To add this nonlinearity, click and drag a Sum block, a Gain block, and a Math Function block, (all 3 in the Math folder) and connect them as shown:
• Notes: • To rotate a block, type Ctrl+R on the keyboard • To select the Square function on the Math Function block, doubleclick it and choose this function from the list • To change the +/- signs on the Sum block, double-click it θ
Consider a challenging control simulation:
2l
• A cart with mass m, rolling on a frictionless surface, with air drag, balancing a pole of mass mp.
F(t) m
x(t) FD(t)
Simulink Tutorial G.D. Buckner
6
ASME Technical Sessions, Fall 2001 North Carolina State University
• Control Objective: manipulate cart force F(t) to balance the pole (keep θ(t) near zero) • The equations of motion for this system can be found using Newton's 2nd Law, and making some simplifications:
(m + m )!! − C x
p
D
! ! x 2 + m p lθ! = F (t )
! m p l 2θ! + m p l!! = m p glθ x
• Assuming m=10, mp =1, l=1, J=1, and CD =1, the Simulink implementation of this complicated, nonlinear (and very unstable) system is:
• This might look like a mess, but imagine programming this system in Fortran or C! • One way to simplify the look of this system is to use a SubSystem block (in the Signals & Systems folder). Click and drag a SubSystem block onto your project, then select the entire cart/pole model and move it into the SubSystem (you can use Cut/Paste). Add an In1 and Out1 block (both in the Signals & Systems folder) as shown:
Simulink Tutorial G.D. Buckner
7
ASME Technical Sessions, Fall 2001 North Carolina State University
• Now, when you close this SubSystem, you will see only a block representing the entire cart/pole model. The input In1 is the control force F(t), the output Out1 is the angular position of the pole: • Designing a controller to stabilize this system is now much easier. One "low-tech" approach is to select a PID Controller block and combine it into your project as shown:
• By "tweaking" the proportional and derivative gains a bit (Kp=-1000, Kd=-100), this system can be stabilized from an initial deflection of 0.1 radians, as shown in the following figures.
Simulink Tutorial G.D. Buckner
8
ASME Technical Sessions, Fall 2001 North Carolina State University
Simulink Tutorial G.D. Buckner
9
ASME Technical Sessions, Fall 2001 North Carolina State University
