Sunday, July 21, 2019
Stability and Control of Nonlinear Systems
Stability and Control of Nonlinear Systems The following system was provided to study about passivity, asymptotic stability, and input to state stability properties at conditions. The given system of differential equation for analysis is given below Also, The state space representation of the system is as follows. Let, Similarly Hence Based on the system equation is given by Similarly, based on the state space representation The state of system , where n=4, where m=2 and where p=2. Hence p=m. A dissipative system with respect to supply rate is said to be passive, if The Lyapunov function for the system is given as Hence with respect to definition of state variables, it can be rewritten as Hence, Also, based on the definition of S, For the given system , hence the lossless system, is passive from u to y. The PD feedback controller of the system with and is represented as Hence the state space representation of the system, is given by The modified Lyapunov function with potential energy is given by It can be observed that V(x) is differentiable (V: R4 → R and a C1 function). Based on the equation of V(x) it can be observed that, The term and all other terms are quadratic in nature. Hence where Hence V(x) is positive definite. Let V(x) is bounded by V(x)≠¤M, where M à  µ R, then it implies that ⇒ ⇒ and ⇒ ⇒ and ⇒ ⇒ and ⇒ ⇒ and ⇒ ⇒ Hence V(x) is radially unbounded. The derivative of V(x) can be obtained as follows Upon substitution and solving the equations, At It can be observed that Based on the above equation, it can be observed that It can be observed that , has only quadratic terms with a negative sign prefixed hence where Hence is negative definite. The equilibrium point of the system at is given by Hence origin is the only equilibrium point of the system. Based on the above observations, it can be concluded that the system is globally asymptotically stable at the origin. The given systems were simulated for different values of and , modified one at a time with other disturbance set to zero and the initial condition set at origin. The following observations can be found from subplots of and Hence the disturbance in both the coordinates of the system are additive in nature. It can be observed that, whenever initial state of the system set to origin and disturbance is induced in one of the coordinate ( or ), the other coordinate of the system is not disturbed. Figure 1 State of System with disturbance at origin with rd=0 Figure 2 State of System with disturbance at origin with thd=0 The Figure 3 indicates the state of the system, when is modified from -5 to 5 with , with the initial condition as x = {7,3,5,1}. The settling time of the system varies with the magnitude of disturbance and the initial condition. Also, it can be observed from the plot of that the system settles to a point which is offset from the origin (equilibrium) by the value of disturbance. Also, the settling time of the system is more for d=-c, when compared to d=c. Also, disturbance in one of the coordinate (), has its effect in another coordinate. Figure 3 State of System with rd=0 at x = [7,3,5,1] The observations of disturbance induced in when , is applicable for the disturbance induced in with  Also, it can be observed from Figure 4 that the settling time of the system is higher when a disturbance is induced in r-coordinate, when compared to -coordinate. Figure 4 State of System with thd = 0 at x = [7,3,5,1] The effect of having both and was observed by simulating the system response for and . Also, it can be observed that settling time of the system is similar to disturbance induced only in the r-coordinate. Figure 5 State of System with thd = -5, rd=5 at x = [7,3,5,1] In all the above plots, it can be observed from the subplot of that the settling point of state as t→, and , indicating that the state of the system tracks the input in the respective coordinate. It can also be observed from the previous plots for d=0, system exhibits the property of global asymptotic stability to the origin (equilibrium point). Also, , the state implies the Bounded Input Bounded State property of the system. The input to state stability of the closed loop system with respect to and for the system was validated by adding a destabilizing feedback with and . The function k(x) of the disturbance is selected, such that the power transferred to the system is maximized, which can be performed when . From the above equation, it can be observed that the power transferred to the system can be maximized by choosing same sign of with c≠¥0. The nature of system response for different range of c is listed in the Table 1 below. Table 1 System Response for Variation in c at initial condition of [7,3,5,1]] Value of c Observation c ≠¤ 1.99 The energy of the system decreases initially, indicated by the plot of Lyapunov function shown in Figure 6 and the same result can be observed on the plot of r and ÃŽ ¸, where the magnitude decreases initially and oscillates with the bounded magnitude, for the bounded input indicated in plot of theta-d. c>1.99 For c=5, the energy of the system increases, indicated by the plot of Lyapunov function shown in Figure 7 and the plot of r and ÃŽ ¸ indicates that the magnitude continues to increases resulting in unbounded state for the bounded input indicated in plot of theta-d. Also, it can be observed that the rate of increase in energy of the system, decreases with time. Figure 6 State of System at c=1.75 The system is not Input to state stable (ISS) for c>1.99 and Figure 7 indicates a system which is not ISS for c=5. The value of transition from bounded state to unbounded state was observed at c=1.93 for an initial point of [1,2,1,2]. Based on the above observation, the transition value of c is dependent of initial condition (energy) of the system. Figure 7 State of System at c=5 The PD control used in the r-coordinate is modified as The simulations were carried out, to identify the properties of ISS satisfied by the system, with respect to and as inputs. All the simulations were carried out with respect to the initial condition x0 = (7,3,5,1) Condition 1: The system is evaluated with zero disturbance and , the result is indicated in Figure 8. Figure 8 System with Zero Disturbance For the no disturbance conditions, it can be observed that the system is asymptotically stable about the origin (equilibrium), indicating the Global asymptotic stability of the system about the origin. Also from the plot of Lyapunov function, it can be observed that the energy of the system settles down to zero. Condition 2: The destabilizing feedback input used in question 5 for the system was fed to the system and it its response is indicated in figure Figure 9 State of System at c=5 with modified PD Control The following observations can be made with respect to figure For an input , the state , indicating bounded input bounded state property of the system. It can be observed that, though the energy of the system increases initially, but upper bounded over a period. The energy and the state of the system gets bounded over period of simulation. Hence for the bounded input, state of the system is bounded. Also, the system exhibits property of asymptotic gain, since the state of the system is upper bounded by disturbance with gain of the system. Also, it was observed that though the system is ISS for the c=5, as the value of c increases energy of the system increases (example for c=10, v(x) is upper bounded to 10,000). Hence modifying the PD control, makes the system ISS for a larger range of disturbances, when compared to earlier control. Condition 3: The system was fed with the input Figure 10 State of system rd=0 and theta d=5*exp(t) It can be observed from the plot that d(t)→ 0 as , also aysmptotically. Hence the system indicates the property of converging input, converging state. The response of the system was evaluated with different possible inputs for , such as , the state of the system x1, x3 was chosen based on observations made in earlier simulations (q5) where predominantly these states grew out of bound Similarly, the above input conditions were simulated with =0 and defined as one of the input, few combinations of the above input disturbances and few possible system interconnections such as positive feedback interconnection, negative feedback interconnection, series interconnection. System response for various types of disturbance Constant Disturbance The disturbance of the system is set to constant values, as indicated in Figure 5 Figure 11 State of system at theta d=-5 rd=5 It can be observed from the plot of Figure 11 and Figure 5 that the settling time of system in r-coordinate has reduced almost by half, when compared to previous control. Positive Feedback Interconnection The disturbance input condition is mentioned below and the system response is shown in Figure 12 Figure 12 System Response for Positive Feedback Interconnection The state of the system indicates the converging nature, also it can be observed that after the transient period system follows the input. Series Interconnection The system is connected in series, with the following disturbance input configuration for each of the subsystem and the plot for the same is shown in Figure 13. Figure 13 System Response for Series Interconnection It can be observed that the behavior of the system is similar with respect to condition 2, but the energy of the system settles down at a higher level when compared to the similar condition with System with different disturbances acting simultaneously The type of disturbance added to the system is given below and the response of the system is shown in figure Figure 14 System response of simultaneous time varying disturbance It can be observed that the system exhibit the property of bounded input bounded state, even if the disturbance is of time varying. In all the above simulation conditions, it was observed that the system exhibits bounded state nature for a wider range of inputs with higher magnitude, when compared to the PD control implemented earlier. This phenomenon can be attributed to the cubic terms with the negative sign, as it can reduce the rate at which energy of the system increases, before it goes out of bound. APPENDIX Code Used for Generation of Plots Contents Q4 Constant Value of Theta-d and r-d Q5 for ISS Q6 for ISS with new u2 Q4 Constant Value of Theta-d and r-d clc clear all close all global x1d; global x3d; ts=500; %Duration for solving ip=[7,3,5,1]; options=odeset(AbsTol,1e-7,RelTol,1e-5); thd=[-5]; rd=[5]; for i=1:size(thd,2)    for j=1:size(rd,2)%-29:30:31        x1d=thd(i);  %x1d is Theta-d        x3d=rd(j);  %x3d is r-d        [t,x]=ode23(@deeqn,[0 ts],ip,options);        figure(1)        subplot(2,2,1)        hold on        plot(t,x(:,1))        title(Plot of Theta)        xlabel(Time)        ylabel(Theta)        grid on        grid minor        subplot(2,2,2)        hold on        plot(t,x(:,2))        title(Plot of Theta-dot)        xlabel(Time)        ylabel(Theta-dot)        grid on        grid minor        subplot(2,2,3)        hold on        plot(t,x(:,3))        title(Plot of r)        xlabel(Time)        ylabel(r)        grid on        grid minor        subplot(2,2,4)        hold on        plot(t,x(:,4))        title(Plot of r-dot)        xlabel(Time)        ylabel(r-dot)        grid on        grid minor    end end Q5 for ISS clc close all global x1d; global x3d; ts=10000; %Duration for solving ip=[7,3,5,1]; options=odeset(AbsTol,1e-7,RelTol,1e-5); x1=ip;  global c;  cval=[1.92] %1.993 is transition point  for i=1:size(cval,2)  c=cval(i); %4.0125        x1d=0;  %x1d is Theta-d        x3d=0;  %x3d is r-d        [t,x]=ode23(@deeqnvx,[0 ts],ip,options);        figure(2)        subplot(2,3,1)        hold on        plot(t,x(:,1))        title(Plot of Theta)        xlabel(Time)        ylabel(Theta)        grid on        grid minor        subplot(2,3,2)        hold on        plot(t,x(:,2))        title(Plot of Theta-dot)        xlabel(Time)        ylabel(Theta-dot)        grid on        grid minor        subplot(2,3,4)        hold on        plot(t,x(:,3))        title(Plot of r)        xlabel(Time)        ylabel(r)        grid on        grid minor        subplot(2,3,5)        hold on        plot(t,x(:,4))        title(Plot of r-dot)        xlabel(Time)        ylabel(r-dot)        grid on        grid minor        subplot(2,3,3)        hold on        thdin=c.*sign(x(:,2));        plot(t,thdin)        title(Plot of theta-d)        xlabel(Time)        ylabel(theta-dot)        grid on        grid minor        subplot(2,3,6)        hold on       vxfn=(1/2).*(((x(:,3).^2)+1).*(x(:,2).^2)+(x(:,4).^2)+(x(:,1).^2)+(x(:,3).^2));        plot(t,vxfn)        title(Plot Lyapunov Function)        xlabel(Time)        ylabel(v(x))        grid on        grid minor  end Q6 for ISS with new u2 clc close all global x1d; global x3d; ts=100; %Duration for solving ip=[7,3,5,1]; options=odeset(AbsTol,1e-7,RelTol,1e-5); x1=ip;  global c;  cval=[5] %1.993 is transition point  for i=1:size(cval,2)  c=cval(i); %4.0125        x1d=0;  %x1d is Theta-d        x3d=0;  %x3d is r-d        [t,x]=ode23(@deeqnr,[0 ts],ip,options);        figure(3)        subplot(2,3,1)        hold on        plot(t,x(:,1))        title(Plot of Theta)        xlabel(Time)        ylabel(Theta)        grid on        grid minor        subplot(2,3,2)        hold on        plot(t,x(:,2))        title(Plot of Theta-dot)        xlabel(Time)        ylabel(Theta-dot)        grid on        grid minor        subplot(2,3,4)        hold on        plot(t,x(:,3))        title(Plot of r) Function for constant disturbance function dx = deeqn(t,x) % Function for system model %  Argument function for ODE Solver global x1d; global x3d; dx=[x(2); (-2*x(3)*x(4)*x(2)-x(2)-x(1)+x1d)/((x(3).^2)+1);x(4);x(3)*(x(2).^2)-x(4)-x(3)+x3d]; end System with Destabilizing Feedback function dx = deeqnvx(t,x) % Function for system model %  Argument function for ODE Solver global x1d; global x3d; global c; x1d=c.*sign((+1).*x(2));; dx=[x(2); (-2*x(3)*x(4)*x(2)-x(2)-x(1)+x1d)/((x(3).^2)+1); x(4); x(3)*(x(2).^2)-x(4)-x(3)+x3d]; end Function with new u2 and old u1 function dx = deeqnr(t,x) % Function for system model %  Argument function for ODE Solver global x1d; global x3d; global c; x1d=x(4);%c.*sign((+1).*x(2));; x3d=x(2); dx=[x(2); (-2*x(3)*x(4)*x(2)-x(2)-x(1)+x1d)/((x(3).^2)+1);x(4);x(3)*(x(2).^2)-x(4)-x(3)+x3d-(x(3).^3)+(x3d.^3)]; end Published with MATLAB ® R2016b
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.