EEL325: Control Engineering II
Department of Electrical Engineering, IIT Delhi Final Project Total Marks: 100 Consider an inverted pendulum on a moving cart as shown below: Due Date: Nov 13 2013
Figure 1: Inverted pendulum on a moving cart system The equations of motion of the system can be derived using Newtons laws as
Where
(t)]T . It is required to stabilize the pendulum system in the vertically The state vector is dened asx(t) = [z (t), z (t), (t), upward position using the control F (t). Only the horizontal position of the pendulum can be measured at the output using some position sensor, i.e. y (t) = z (t) + Lsin(). Do the following: 1. Linearize the system about its equilibrium point when the pendulum is vertically upward and the cart is at z (t) = 0 position and write the perturbed linear system in state space matrix form. Dene the state space model in MATLAB using the command ss and convert to transfer function using ss2tf. Find out the open-loop poles of the system. 2. Implement the state space realization in Simulink MATLAB using only adders, ampliers and integrators (available in the Simulink browser). Simulate using dierent inputs (e.g., unit step, sinusoidal etc.) to the cart and dierent initial conditions. Compare the output response with the full nonlinear model (provided in pendulum_cart_dynamics.m). Deliverables: Snapshot of the Simulink model, Comparison plots of the output y (t) for the linearized and the nonlinear model with dierent inputs and initial conditions, and your comments on the validity of the linearized model. 3. Comment on the stability of the nonlinear system for the equilibrium point considered in (1). 4. Comment on the possibility of designing a PD control law to stabilize the nonlinear system.
5. Is it possible to design a state feedback controller to stabilize the pendulum-cart system? Justify. If yes, place the closed-loop poles such that the following design requirement are satised (a) Settling time for z (t) and (t) <5 sec. (b) Rise time for (t) < 0.5 sec. (c) Steady state error for both z (t) and (t) < 2% (d) Control eort F (t) < 0.5 N m Deliverables: poles chosen with justication, Hand calculations to nd the state feedback gain K and verication using MATLAB command place, snapshot of the Simulink block diagram used to implement the controller. Simulation plots of the states x(t) on the linearized model using small non-zero initial conditions. 6. Is it possible to design an observer to estimate the states x(t)? Justify. If yes, place the poles 5 times the controller poles found in the previous part. Deliverables: Hand calculations to nd the observer gain L and verication using MATLAB command place, snapshot of the Simulink block diagram used to implement the observer. Simulation plots of the state estimation error x (t). Note that the observer initial condition x (0) should be chosen dierent to the initial states x(0). 7. Combine the controller and observer found in (6) and (7). Deliverables: snapshot of the Simulink block diagram used to implement the controller+observer. Simulation plots of the states x(t) on the linearized and the nonlinear model using small non-zero initial conditions. Note: Although the output of the system is y (t) , you are given extra outputs in pendulum_cart_dynamics.m for plotting purpose. Do not use the extra outputs for control+observer design. You are expected to upload the pdf of your project on Sakai. Please do not give me hard copies of your report. The grade will be based on originality, clarity of presentation, correctness and completeness of results. Honesty policy should be strictly followed. Discussion is allowed but not copying of code, plots, report. Non adherence will lead to a penalty varying from reduction of nal grade to a grade of F. All project reports will be tested through the plagiarism checking software turnitin .
