Signals and Systems 1 Lecture 8 Dr. Ali. A. Jalali September 6, 2002 Signals and Systems 1 Lecture # 8 Differential Equation Model EE 327 fall 2002 Differential Equation Model 1. Many physical systems are described by linear 2. 3. 4. 5. 6. differential equations. Reducing differential equations to algebraic equations Homogeneous solution, exponential solution and natural frequencies. Particular solution, system function and poleszeros Total solution, initial condition and steadystate Conclusion The Nth-order Differential Equation Model 1. 2. One important way of modeling LTI systems is by means of linear constant-coefficient equations. Example for DE model: Analog Filter 2nd-order ED 2 1 1 vt vt 2 v t 2 et . RC1 R C1C2 R C1C2 The Nth-order Differential Equation Model 1. Example for DE model: Instrument servo 3rd-order ED kk p A kk p A RJ BL RB t t t t r t . LJ LJ LJ LJ The Nth-order Differential Equation Model 1. Example for DE model: Electric Network 2nd-order ED The Nth-order Differential Equation Model 1. Example for DE model: Electric Network 2nd-order ED The Nth-order Differential Equation Model 1. The general linear constant-coefficient Nth-Order DE for SISO systems are: dyt d N y t a0 yt a1 a N b0 xt N dt dt dxt d L x t b1 a L . L dt dt OR: d y t d x t ak bk . k k dt dt k 0 k 0 N k L k X(t) system input, Y(t) system output and our practical restriction order NL Initial Condition Solution of Differential Equation 1. The character equation of DE of system is found by substituting a trial solution N into the k d y(t ) ak 0 k dt k 0 homogeneous DE: The result is: y(t ) Cest a0 s 0 Ce st a1s1 Ce st a N s N Ce st 0 . Since the Ce st cannot be zero it can be factored out. The remaining term must satisfy the algebraic equation. a0 s a1s a N s 0 0 1 This is Characteristic Equation. N Initial Condition Solution of Differential Equation This is Characteristic Equation Can be written as: Or as factored form: a0 s 0 a1s1 a N s N 0 N k a S k 0. k 0 N k a s k aN s r1 s r2 s rN 0 k 0 Characteristic roots are: s1 r1 , s2 r2 ,, sN rN Where r1 , r2 , , rN may be real or complex (conjugate pairs). Initial Condition Solution of Differential Equation The solution of the homogeneous DE of: k N d y(t ) ak 0 k dt k 0 For a given set of initial conditions: yt t 0 , dyt dt t 0 ,, d N 1 yt dt N 1 Is called the initial condition (IC) solution and for simple (non repeating) roots is of the form: y IC t C1e r1t C2 e C N e r2 t N or: y IC t ck e rk t k 1 rN t . t 0 Initial Condition Solution of Differential Equation N y IC t ck e rk t In IC solution: k 1 k 1, 2,, N are coefficients that must be determined in order to satisfy the given set of initial conditions and k 1, 2, , N rk are the characteristic roots. Multiple roots: If the CE contains multiple roots indicated by factor terms of the form s sk q C1t q 1e sk t C2 t q 2 e sk t Cq e sk t will appear in IC solution. System Stability in term of IC response A casual system is stable if the IC response decays to zero as: t . This happens if and only if all the rk ‘s are negative real, or if there are any complex roots, their real parts must be negative. Img 2 r1 1, r2 2, r3, 4 2 2 j Real -3 -2 -1 -2 Natural Frequency The roots of the characteristic polynomial are called natural frequencies. These are frequencies which there is an output in the absence of an input. Also the roots of the characteristic polynomial are called eigenvalues. Example: The flexible shaft subjected to an applied torque (t ) with the shaft angle (t ) described by the DE: t 3 t 2t 1 3t 1- Find CE? 2- Catachrestic roots? 3- Is this system stable? 4- Give the algebraic form of the initial condition response? 5- Find the constants in the IC solution if (t ) 1 and [d (t ) / dt ] t 0 0. Example Solution 1- Find CE? The homogeneous differential equation is: s 2 3s 2 0. Yielding the CE as: 2- CR? Using the quadratic formula, CR are: t 3t 2 t 0 r1, 2 1, 2. 3- Is this system stable? Both roots are negative real so the system is stable. 4- Algebraic form of IC response? IC (t ) C1e t C2 e 2t 5- Constant IC solution? IC (0) 1 C1 C2 and [d IC (t ) / dt ] t 0 0 C1 2C2 . Solving gives: C1 2 and C2 1. Thus (t ) 2e t IC 2 t e , t 0. MATLAB Comment Function roots in MATLAB: If the CE is 5th order as: 5s5 4s 4 3s 3 2s 2 s 0 >>p=[5 4 3 2 1 0]; >>r=roots(p) The results are: >> p=[5 4 3 2 1 0];roots(p) ans = 0 0.1378 + 0.6782i 0.1378 - 0.6782i -0.5378 + 0.3583i -0.5378 - 0.3583i r1 0 r2 ,3 01378 . j0.6782 r4 ,5 05378 . j0.3583 The Unit Impulse Response Model 1. The unit impale function is defined implicitly by its sifting property: (t t0 ) f (t )dt f (t0 ) where f(t) is assumed to be continuous at t t0 . pt . Approximation to impulse function: t t 0 lim 0 EE 327 fall 2002 Signals and Systems 1 The Unit Impulse Response Model 1. Using the sifting property lead the following product function. Approximation of the sifting property The value of integral is: f t 0 . f t 0 EE 327 fall 2002 Signals and Systems 1 Unit Impulse Response The response of an LTI system to an input of unit impulse function is called the unit impulse response. x(t)=(t) LTI y(t)=h(t) Important: When determining the unit impulse response h(t) of an LTI system, it is necessary to make all initial conditions zero. (output due to input not energy stored in system) EE 327 fall 2002 Signals and Systems 1 Convolution If the unit impulse response h(t) of a linear continuous system is known, the system output y(t) can be found for any input x(t). Approximation by pulse Sifting property of pulse EE 327 fall 2002 x(t ) x( ) f (t )d Signals and Systems 1 Convolution Integral 1. The convolution integral is one of the most important results used in the study of the response of linear systems. 2. If we know the unit impulse response h(t) for a linear system, by using the convolution integral we can compute the system output for any known input x(t). 3. In the following integration integral h(t) is the system’s unit impulse response. x(t ) x( )h(t )d EE 327 fall 2002 Signals and Systems 1 Convolution Evaluation 1. The convolution integral can be evaluated in three distinct ways. a) Analytical method, b) Graphical method, c) Numerical convolution We will discuss about these and about convolution properties in class. (see class notes) EE 327 fall 2002 Signals and Systems 1