platoon of vehicles initially traveling at uniform speed and spacing when the lead vehicle po, 35 equation des 343¢01/5 SEL The Laplace Transform of The Linear Car-following Model Mostafa Ghandehari (ghandeha@uta.edu) and Siamak Ardekani (Ardekani@uta.edu) Department of Civil Engineering SEL ‘The University of Texas at Arlington Arlington, Texas 76019-0308 QA I" Laplace transform is used to analyze the differential-difference equation for the linear car~ Al following model. The car-following model has been developed to describe the dynamics of aT 43, i introduces a perturbation by decelerating or accelerating. Shifting theorem and the convolution theorem for Laplace transforms are used to model fluctuations in velocities and spacing of following vehicles as a function of the deceleration or acceleration of the lead car when the initial velocity and spacing are given. Introduction: ‘The ear-following model has been developed to describe the dynamics of a platoon of vehicles initially traveling at uniform speed and spacing when the lead vehicle introduces a perturbation by braking or accelerating. The linear car-following model was developed by Chandler, Herman and Montroll [1]. Assume that X,,(F) gives position of the n™ car in the platoon at time f, and v,,(t)is the corresponding velocity function. Thus the position and velocity functions for the Jeading car are given by X(t) and V(t), respectively. Chandler, Herman and Montroll assume that the acceleration of the 7!" car with time delay Tis proportional to the relative velocities of the n car and the (n-/)" car. In a subsequent paper, Herman et al. [2] have used a treatment of Laplace transform techniques on the car following model. The following differential-difference the motion of the cars in the platoon: Page 1mi +T)=-2 [2,0 - x40] o The constant A is called the sensitivity constant. If the delay time Tis small, as should be the case in driving, we obt the following second order differential-difference equation: xO =A [x,0-x,,0] @ From equation (2), the corresponding equation for velocities is v,Q=-A fp, -¥,.0] % In section 2, we use the Laplace transforms with given initial velocities to analyze equation (3). ‘The direct integration and differential operator techniques are used in section 3. In section 4 we use Taylor series expansion to analyze the differential difference equation (1). For a more detailed discussion of the car-following models see chapter 3 of Haberman [3]. A table of Laplace transforms is given in Abramowitz and Stegun [4]. 2. The Laplace Transform: ‘We use the notation ¥,, for the Laplace transform of V,,(t). Recall that the Laplace transform of derivative is given by =s¥,-v,(0) ) Where V,,(0) is the initial velocity of the n" car in platoon at time ¢. We now take the Laplace transform on both sides of equation (3) to obtain nV (O)= AV, + AV, 4 6) Equation (5) reduces to the following difference equation for ¥, Page 2(s+), =A V4 +¥, (0) © ‘The solution to the above difference equation is given by Gersting [5 ] x aye Ee] o The convolution theorem and the fact that a i Y aue* ee | fier vp(r)dt +2” Da 9) ‘We then integrate V,,(f) to obtain the position function x,(0 = fv, Oat +x, 0) (0) i In the next section we use direct integration to analyze equation (3). 3. Direct integration: Equation (3) is a standard example of a first order linear differential equation for ¥,, Let v, + 4v, = Ay, ay Multiplying both sides of equation (11) by the integrating factor €~' we obtain S(e,)= jev,,. (12) Page 3a , Lau, (4) =e" ¥, then ui) a3) Note that %,, (0) = V,,(0) . Assume that the leading car initially slows down to speed C at the same time that all other cars in the platoon are moving with initial velocities v,(O)=d, n21 and Vo(0)=C where o 00 (last vehicles in a long platoon) where as in equation (18) we allow f — © (asymptotic behavior in time). Equation (18) indicates the obvious fact that ifthe leading car insists on maintaining a constant slow speed of , then eventually all the cars in the platoon have to adjust to speed limv,@® The differential operator approach also gives results consistent with results shown above, By repeated application of the differential operator D + A J on both sides of equation (3) we obtain (D+A1)'v, =2"v% a9) ‘The homogenous solution to the linear differential equation (19) is given by v(t) 1 nae i Diet’ | where ¢;sareconsans. 20) In the case that Yo is a constant, a particular solution is, v, =e et We add homogenous and a particular solution to obtain wet v,(Q=e*'| Yet’ |+C S Using initial condition for velocities, equation (22) is equivalent to the equation (14). Page 5‘A case of particular interest is when the leading car decelerates, that is, assume that the velocity of leading car is given by 23) ‘The operator theory approach will yield < na nat i hr ~at b+ a 4, Taylor Series: ‘The first order Taylor series approximation is given by v(t T) =v, (D+ 7 vi (0) 3) ‘Substitute equation (25) into equation (1) and simplify to obtain the following second order differential-difference equation Tv) +, (0) + Ay, (0) = 2,40 026) Also recall hat, L(y") = s*L(y) — sy(0) — y'(0) Using ¥,, for the Laplace transforms of V, and taking transform of both sides of equation (26) we obtain Thr, — sv, (0) —vi(0)]+ sv, —v, (0) + Av, = AV, 27) Simplification of equation (27) will result in As Tsv, (0), _Tv,0) Tada! + Fades THAts Tt+ats T+At+s TtAr+s (28) Page 6Tsv, (0), _T¥,(0) T+Ats T+a+s Letting , equation (28) will yield 9) ‘The solution to differential equation (29) will in turn yield = 5, = AG, + BL YA 630) ‘a Using the formula for the sum of geometric progression, equation (30) simplifies to A’-1 +3IT re A-1 ) Substituting the values of A and B , and simplifying the algebra yields ee , (T+Ats) -2" (T+Aassy slr, +0) A | 2) With various assumptions for initial velocities, we can take the inverse Laplace transform of equation (32) to find an expression for, . For example, assuming V,(0) = ¢, v;,(0) =, 21 and assuming that the lead car has a constant deceleration, ie. Vp = —At + b,, equation (32) is further simplified as - X_ [Psv,(0)+ 7,0] Ws %, =m ht IF oe (T+A+sy T+s (T+A+s) ‘After using a shifting theorem and the convolution theorem fo PPvo(a)de+ a or F 64 cler+ a= |,-_4 T+s (T+A+s) Page 7We use the notation * for convolution to obtain x ¥ AT +A\ t=) “=D! Je MOV (Ede + Ler) + (Td eT?) e! ~eT 2 et" _ + (fd cP?) eT x Ae (n-1)! Using the fact that the inverse of 1 (in the second bracket of equation 34) can be considered as a delta distribution, we obtain fe (PMD Cdr + (n-1!3 (Id -~cT*)e7' ~cT ATOM + CTE(1) + (@-) (5) r, zu (Td -cT?) fem Ferg, a (n-1)! ‘Simplification of equation (35) and extra assumption on initial conditions could yield simpler expressions for V, (2). Integration of V, (¢) will in tum yield the position function. In summary, the Laplace transform has been used to discuss the linear car-following model. For example, assuming that the leading car has constant velocity d < ¢ , we derived equation (14). Assuming that the leading car has constant deceleration we obtained equation (24), First order Taylor series approximation was used to obtain equation (35) for ‘velocity of the 1” car in the platoon. Page 8References: [1] Chandler, Robert E., Herman Robert, Montroll Elliott W. “Trafie dynamics: studies car following” Operation Res. 6 1958, 165-184. {2} Herman, Robert, Montroll, Elliott W.; Potts, Renfrey B: Rothery, Richard W. “Traffic dynamics: analysis of stability in car-following” Opertions Res. 7,1959, 86-106. [3] Haberman, Richard, Mathematical Models: Mechanical Vibrations, Population Dynamics, and Traffic Flow”, SIAM, 2007 [4] Milton Abramowitz, Irene A. Stegun; “Handbook of Mathematical Functions”, Dover Publications, Inc New York, 1956 [5] Judith L. Gersting, Mathematical Structures for Computer Science, 5" ed, Freeman, 2003 Acknowledgement: ‘The authors wish to thank Dr. Feraydune Kashefi for useful discussions related to this work.. Page 9