nlin0302050/s1.tex
1: \documentclass[fleqn]{vch-book}
2: \usepackage{amsmath}
3: \usepackage{amssymb}
4: \usepackage{makeidx}
5:     \makeindex
6: \usepackage{times}
7: \usepackage{tabularx}
8: \usepackage{booktabs}
9:     \setlength{\heavyrulewidth}{0.4pt}
10:     \setlength{\lightrulewidth}{0.4pt}
11:     \setlength{\cmidrulewidth}{0.4pt}
12: 
13: \usepackage{epsfig}
14: 
15: \begin{document}
16: \chapter{ Nonlinear Vibration in Gear Systems
17: \tocauthor{G. Litak and M.I. 
18: Friswell}}\label{chap:editbooks}
19: \authorafterheading{Grzegorz Litak$^1$ and Michael I. Friswell$^2$} 
20: %Name\footnote{You may use the footnote for
21: %your current address, expression of thanks etc.}}
22: 
23: \affil{ $^1$ Department of Applied Mechanics, \\ Technical University of 
24: Lublin,
25: Nadbystrzycka 36, PL-20-618 Lublin, \\ Poland \\ \\
26: $^2$ Department of Areospace Engineering,  \\ 
27: University of Bristol,
28: Queens Building,  Bristol BS8 1TR,\\ United Kingdom}
29: 
30: 
31: Gear box dynamics is characterised by a periodically
32: changing
33: stiffness and a backlash which  can lead to a
34:  loss of the contact between the teeth. Due to backlash, the gear system
35: has piecewise linear stiffness characteristics and, in consequence,  
36: can vibrate
37: regularly or chaoticaly depending on the system parameters and the initial conditions.
38: We examine the possibility of a  nonfeedback system  control by introducing a weak 
39: resonant excitation term   and through 
40: adding an additional degree
41: of freedom to account for shaft flexibility on one side of the gearbox.
42: We shall show that by correctly choosing
43: the coupling values the system vibrations may be controlled.
44: 
45: 
46: 
47: 
48: \section{Introduction}
49: Gear box dynamics is based on  a periodically changing 
50: meshing stiffness complemented  by  a nonlinear  effect of  backlash 
51: between the teeth. 
52: Recently, their
53: regular and chaotic vibrations have been predicted theoretically and examined 
54: experimentally  
55: \cite{Kah90,Sat91,Bla95,Kah97,Rag99,War00,Lit03}. 
56: The theoretical description of this phenomenon
57: has been based mainly on single degree of freedom models 
58: \cite{Kah90,Sat91,Bla95,Kah97,Rag99,War00,Sza96} 
59: or multi degree models neglecting
60: backlash \cite{Sch86,War00b}. This paper examines the possibility of 
61: taming chaotic vibrations and reducing the amplitude by nonfeedback 
62: control methods. Firstly, we will introduce a weak resonant excitation 
63: through an additional small external excitation term (torque)  with a different 
64: phase \cite{Cha01,Cao03}. 
65: Secondly, we will examine  the effect of  an 
66: additional degree of freedom to account for
67: shaft flexibility on one side of the gearbox \cite{Lit03}.
68:  This may also be regarded as a single mode approximation to the
69: torsional system dynamics, or alternatively as a 
70: simple model for a vibration neutraliser or absorber
71: installed in one of the gears.
72: 
73: 
74: 
75: \section{One Stage Gear Model}
76: We start our analysis with modelling the relative vibrations of a single stage transmission
77: gear \cite{War00}. 
78: Figure \ref{GLf1}  shows a schematic picture of the physical system. 
79: The gear wheels are shown with
80: moments of inertia $I_1$ and $I_2$ and are coupled by the stiffness and damping of the teeth
81: mesh, represented by $k_Z$ and $c_Z$.
82: 
83: \begin{figure} 
84: \vspace{1cm} \begin{center}
85: \includegraphics[scale=0.5]{fig1.eps}  
86: \vspace{-1cm} 
87: \caption{ \label{GLf1}
88: The physical model of a one stage transmission gear system.}
89: %\vspace{1cm} 
90: \end{center}
91: \end{figure}
92: 
93: 
94: 
95: 
96: The equations of motion of the system may be written in terms of the
97: two degrees of freedom, $\psi_1$, $\psi_2$ that represent the rotational angles
98:  of the gear wheels. These angles are those that remain
99: after the steady rotation of the system is removed \cite{War00}.
100: Thus, if the backlash is initially neglected, we have
101: 
102: \begin{eqnarray}
103: I_1 \ddot{\psi}_1 + \left[ k_Z(r_1\psi_1-r_2\psi_2)+
104: c_Z(r_1\dot{\psi}_1-r_2 \dot{\psi}_2) \right] r_1 & = & M_1,
105:  \nonumber \\
106: I_2 \ddot{\psi}_2 - \left[ k_Z(r_1\psi_1-r_2\psi_2)+
107: c_Z(r_1\dot{\psi}_1-r_2 \dot{\psi}_2) \right] r_2 & = & -M_2, \label{GLe1}
108: \end{eqnarray}
109: where $r_1$ and $r_2$ are the radii of the gear wheels and the overdot represents
110: differentiation with respect to time $t$.
111: 
112: 
113: 
114: \begin{figure}
115: \vspace{1cm} \begin{center}
116: \includegraphics[scale=0.5,angle=-90]{fig2.ps} 
117: \caption{\label{GLf2} The meshing stiffness $k(\tau)$.}
118: \end{center}
119: \vspace{0cm}
120: \end{figure}
121: 
122: 
123: \begin{figure}
124: \vspace{1cm} \begin{center}
125: \includegraphics[scale=0.5,angle=-90]{fig3a.ps}
126: \includegraphics[scale=0.5,angle=-90]{fig3b.ps}
127: \caption{ \label{GLf3}  The amplitude of vibration (a) and the maximal Lyapunov exponent $\lambda_1$ (b) versus frequency 
128: $\omega$ 
129: obtained by simulation.  The initial conditions were assumed to be   $x_0=-2.0$ and $v_0=\dot{x}_0=-0.5$ 
130: for small $\omega$ ($\omega_0=0.1$) and for each new  $\omega$ ($\omega_{i+1}$) calculations were 
131: performed for  
132: $400$ 
133: excitation cycles. Each time  
134: new initial conditions $(x_0,v_0)$ were chosen as the last pair of values of $(x,v)$ for previous 
135: $\omega_{i}$ ($(x_0,v_0)|_{new}=(x,v)|_{old}$). Curves '1' and '2' correspond to cases with $D=0$ and $D=2$, respectively.}
136: \end{center}
137: \vspace{0cm} 
138: \end{figure}
139: 
140: 
141: \begin{figure}
142: \vspace{1cm}
143: \hspace{-0.7cm}
144: \includegraphics[scale=0.4,angle=-90]{fig4a.ps} \hspace{-1.2cm} 
145: \includegraphics[scale=0.4,angle=-90]{fig4b.ps} 
146: \caption{\label{GLf4} The Poincare maps calculated for ten randomly chosen initial conditions for $\omega=1.5$ (a) and 
147: $\omega=1.6$  the gray colour corresponds to simulations with respect to (Eq. \ref{GLe2}) while the black 
148: one to (Eq. \ref{GLe6}), respectively.}
149: \vspace{0cm}
150: \end{figure}
151: 
152: 
153: 
154: 
155:  In fact, it is possible to reduce the above two equations  to one using the new relative
156: displacement coordinate  $x=r_1\psi_1 - r_2 \psi_2$. That coordinate
157: represents the relative displacement of the gear wheels at the teeth. 
158: The equation of motion for $x$ is obtained 
159: by subtracting $r_2/I_2$ times first equation from $r_1/I_1$ 
160: times the second one (Eqs. \ref{GLe1}). Thus, in non-dimensional form, the  
161: equation of motion can be rewritten as
162: \begin{equation}
163: \frac{{\rm d}^2}{{\rm d} \tau^2}x + \frac{2 \zeta}{\omega} \frac{{\rm d}}{{\rm d} \tau}x+ \frac{k(\tau) 
164: g(x,\eta)}{\omega^2}   =  \frac{\bar{B}(\tau)}{\omega^2} 
165: = \frac{B_0 + B_1 cos (\tau+ \Theta )}{\omega^2},
166: \label{GLe2}
167: \end{equation}
168: where the parameters are easily derived from Eqs. (\ref{GLe1}-\ref{GLe2}), as
169: \begin{eqnarray}
170:         \tau &=& \omega t, \nonumber \\
171: 	2 \zeta & = & c_Z \left[ r_1^2 /I_1 + r_2^2 / I_2 \right], 
172:  \label{GLe3}
173:  \\
174:   \bar{B}(\tau) & = & r_1M_1/I_1 + r_2M_2/I_2. \nonumber
175: \end{eqnarray}
176: $\zeta$, $k(\tau)$, $g(x,\eta)$ and $\bar{B}(\tau)$ are defined as  in reference \cite{War00}. 
177: Note that the backlash and time dependent meshing stiffness have been included by 
178: rewriting the meshing force as \cite{War00}
179: \begin{equation}
180: 	k_Z \left[ r_1^2 /I_1 + r_2^2 / I_2 \right] x = k(\tau) g(x,\eta).
181: 	\label{GLe4}
182: \end{equation}
183: Equations (\ref{GLe3}) and (\ref{GLe4}) have assumed that the moments on the gear system are
184: composed of a sinusoidal moment at frequency $\omega$ with a constant offset.
185: 
186: 
187: The meshing stiffness is periodic and the backlash is described by a piecewise
188: linear function. Figure \ref{GLf2} shows a typical time dependent mesh stiffness 
189: variation $k(\tau)$ \cite{War00} and the backlash 
190: is modelled for a clearance $\eta$ as
191: 
192: \begin{equation}
193: g\left( {x,\eta } \right) = \left\{ {\begin{array}{*{20}c}
194:    x & \hspace{8mm} & {x \ge 0}  \\
195:    0 & & { - \eta  < x < 0}  \\
196:    {x + \eta } & & {x \le \eta }.  \\
197: \end{array}} \right.
198: \label{GLe5}
199: \end{equation}
200: 
201: 
202: 
203: 
204: 
205: 
206: \section{Vibrations of a Gear System in Presence of a Weak Resonance Term}
207: Now we are going to examine the gear system described by Eqs. (\ref{GLe1}-\ref{GLe5}) subjected to 
208: to an additional external periodic excitation with a relatively small value $D$  
209: \begin{equation}
210: \frac{{\rm d}^2}{{\rm d} \tau^2}x + \frac{2 \zeta}{\omega} \frac{{\rm d}}{{\rm d} \tau}x+ \frac{k(\tau)
211: g(x,\eta)}{\omega^2}   = 
212: \frac{B_0 + B_1 cos (\tau+ \Theta ) + D cos (\tau+ \Theta' )}{\omega^2},
213: \label{GLe6}
214: \end{equation}
215: where the phase angle $\Theta' \neq \Theta$. Such inclusion has been discussed earlier in the literature
216: as a suitable method of
217: chaos control for Josephson junction circuits \cite{Cha01} and Froude pendulum motion
218: \cite{Cao03}.  In those papers the authors claimed that the resonant term, if used adequately, can tame or 
219: include chaotic motion. 
220: We have performed numerical simulations of Eqs. (\ref{GLe2}) and (\ref{GLe6})  to highlight the effect of 
221: the addition of resonant term $D$ on the basic equation of motion (Eq. \ref{GLe2}). In particular we were 
222: interested in changes in vibration amplitude. The type of 
223: motion was analysed using Lyapunov exponents obtained by the algorithm of Wolf {\em et al.}  
224: \cite{Wol88}.  
225: System parameters have been used as in \cite{War00}:
226:  $\omega=1.5$, $B_0=1$, $B_1=4$, $\Theta=0$,  $\zeta = 0.08$, $\eta = 10$  
227: while the additional excitation term \ref{GLe6} was 
228: introduced as $D=2$,
229:  $\Theta'=0.1$. 
230: 
231: Figure \ref{GLf3}a shows 
232: the calculated vibration amplitude of the relative motions of the gear wheels, $A$,  defined as
233: \begin{equation}
234: A=\left| \frac{x_{max}-x_{min}}{2} \right|.
235: \label{GLe7}
236: \end{equation}
237: The maximal Lyapunov exponent 
238: $\lambda_1$ versus frequency
239: $\omega$,
240: obtained by simulation, is plotted in Fig. \ref{GLf3}b.  Note, the initial conditions were assumed to be   
241: $x_0=-2.0$ and 
242: $v_0=\dot{x}_0=-0.5$
243: for small starting $\omega$ ($\omega=0.1$) and for each new  $\omega_{i+1}$ calculations were performed 
244: for  
245: $400$
246: excitation cycles with
247: where new initial conditions were the last pair of values of $(x,v)$ for the previous frequency $\omega_{i}$. 
248: Curves marked by '1' correspond to solutions of Eq. \ref{GLe2} while those marked by '2' correspond to
249: Eq. \ref{GLe6}. 
250: One can easily note that the amplitude of the modified system is slightly larger but simultaneously the 
251: system behaves
252: more regularly.  In most of cases where the original system was in a chaotic state ($\lambda_1 > 1$ for the 
253: curve '1', Fig. \ref{GLf3}b) it vibrates  regularly  in the presence resonant term  ($\lambda_1 < 1$ for the 
254: curve 
255: '2' in Fig. \ref{GLf3}b). Unfortunately the taming of the chaotic motion was accompanied by 
256: additional features 
257: visible in Fig. \ref{GLf3}a. Clearly, the curve '2' describing the amplitude of motion shows a number of 
258: jumps, i.e. for $\omega \approx 1.36$ and 1.70. These jumps may be associated with transitions to other 
259: solutions (attractors)
260:  with changing $\omega$. This effect is typical for many nonlinear systems but it seems to be more 
261: transparent for $D \neq 0$.  
262: To explore further this effect we have also calculated Poincare 
263: maps for many initial conditions chosen randomly.  The results for frequencies $\omega=1.5$ and 
264: 1.6 are shown in Fig.  \ref{GLf4}. In this figure, one can see that for $\omega=1.5$ attractors 
265: (in gray colour) for $D=0$ and (in black colour) for $D \neq 0$ are similar. Here, numbers '1' and '2' 
266: denote the regular and 
267: chaotic attractors, respectively. 
268: But if we move $\omega$ to a larger value  ($\omega$= 1.6) the regular attractor for $D=0$ 
269: disappears. The attractor remains for $D \neq 0$, but interestingly another chaotic
270:  attractor emerges. For some other 
271: frequencies additional attractors also exist, which supports the proposed explanation of the jumping 
272: phenomenon.
273:  
274: 
275: 
276: 
277: 
278: 
279: 
280: 
281: \section{Vibrations of a Gear System with a Flexible Shaft}
282: 
283: 
284: \begin{figure}
285: \vspace{0.5cm} \begin{center}
286: \includegraphics[scale=0.5]{fig5.eps}
287: \vspace{1cm}
288: \caption{\label{GLf5}
289:  The physical model of the gear model with an additional degree of freedom
290: introduced by  a  disk fixed by a flexible shaft to one of gear wheels.
291: }
292: \end{center}
293: \end{figure}  
294: 
295: 
296: Here we examine the effect of adding an
297: additional degree of freedom to the gear system (Fig. \ref{GLf5}) by using
298: a flexible shaft on one side of the gearbox \cite{Lit03}. 
299: 
300: Now the equations of motion of the system may be written in terms of the
301: three degrees of freedom, $\psi_1$, $\psi_2$ and $\psi_3$, related to the rotational angles 
302: of the gear wheels and disk, and have the following form
303: 
304: \begin{eqnarray}
305: I_1 \ddot{\psi}_1 &+& \left[ k_Z(r_1\psi_1-r_2\psi_2)+
306: c_Z(r_1\dot{\psi}_1-r_2 \dot{\psi}_2) \right] r_1  =  M_1,
307: \nonumber \\
308: I_2 \ddot{\psi}_2 &-& \left[ k_Z(r_1\psi_1-r_2\psi_2)+
309: c_Z(r_1\dot{\psi}_1-r_2 \dot{\psi}_2) \right] r_2 - k_S ( \psi_3-\psi_2 )  =  -M_2, \nonumber
310: \\
311: I_3 \ddot{\psi}_3 &+& k_S ( \psi_3-\psi_2 )  =  -M_3.
312: \label{GLe8}
313: \end{eqnarray}
314: 
315: 
316: \begin{figure}
317: \vspace{1cm} \begin{center}
318: \includegraphics[scale=0.5,angle=-90]{fig6a.ps} 
319: \includegraphics[scale=0.5,angle=-90]{fig6b.ps}
320: \includegraphics[scale=0.5,angle=-90]{fig6c.ps}
321: \includegraphics[scale=0.5,angle=-90]{fig6d.ps}
322: \caption{\label{GLf6}
323:  The amplitude of vibration (a) and the maximal Lyapunov exponent $\lambda_1$ (b) versus frequency
324: $\omega$
325: obtained by simulations.  The initial conditions were assumed to be   $x_{10}=-2.0$, 
326: $v_{10}=\dot{x}_{10}=-0.5$ and $x_{20}=0$ $v_{20}=\dot{x}_{20}=0$
327: for small $\omega$ ($\omega=0.1$) and for each new  $\omega_{i+1}$ calculations were performed for  $400$
328: excitation cycles 
329: where the new initial conditions were the last pair of values of $(x_1,v_1,x_2,v_2)$ for the previous 
330: $\omega_{i}$.
331: Curves '1','2','3' were obtained for $k_S=0$, 0.1 and 1.0, respectively.
332: Four nonzero Lyapunov exponents $\lambda_i$ ($i=1,2,3,4$) versus frequency $\omega$ are shown
333: for $k_S=0.1$ (c) and $k_S=1.0$ (d).
334: }
335: \end{center}
336: \vspace{0cm}
337: \end{figure}
338: 
339: 
340: 
341: 
342: As in the previous case (Eqs. \ref{GLe1}, \ref{GLe2}) the above set of equations may be decoupled
343: by using the new relative 
344: displacement coordinates  $x_1=r_1\psi_1 - r_2 \psi_2$ and $x_2=r_2(\psi_3 -
345: \psi_2)$. Thus we 
346: reduce the number of equations from three to two
347: 
348: 
349: \begin{eqnarray}
350: \frac{{\rm d}^2}{{\rm d} \tau^2}x_1  &+& \frac{2 \zeta}{\omega} \frac{{\rm d}}{{\rm d} \tau}x_1 + 
351: \frac{k(\tau) 
352: g(x_1,\eta)}{\omega^2} + \frac{\beta_1 k_S x_2}{\omega^2} \nonumber \\ &=&  \frac{B_0 + B_1 cos ( \omega t+ \Theta 
353: )}{\omega^2},
354: \label{GLe9} \\
355: \frac{{\rm d}^2}{{\rm d} \tau^2}x_2   &+& \frac{\beta_2 k_S x_2}{\omega^2} + \frac{2 \beta_3 
356: \zeta}{\omega} 
357: \frac{{\rm d}}{{\rm d} \tau}x_1+ \frac{\beta_3 k(\tau) g(x_1,\eta)}{\omega^2}  
358: =  0, \nonumber
359: \end{eqnarray}
360: where most  of  parameters have already been  defined in Eqs. \ref{GLe3} and \ref{GLe4}.
361: The rest of them  are easily 
362: derived from Eqs. 
363: (\ref{GLe8}-\ref{GLe9}), as
364: \begin{eqnarray}
365:         \beta_1 & = & 1 / I_2, \nonumber \\
366:         \beta_2 & = & 1/I_2 + 1/I_3, \nonumber  \\
367:         \beta_3 & = & (r_2^2/I_2) / \left[ r_1^2/I_1 + r_2^2/I_2 \right], \label{GLe10} \\
368:              0 & = & r_2M_2/I_2 - r_2M_3/I_3. \nonumber
369: \end{eqnarray}
370: 
371: \begin{figure}
372: \vspace{1cm}
373: \hspace{-0.7cm}
374: \includegraphics[scale=0.4,angle=-90]{fig7a.ps} \hspace{-0.8cm}
375: \includegraphics[scale=0.4,angle=-90]{fig7b.ps}
376: 
377: \hspace{-0.7cm}
378: \includegraphics[scale=0.4,angle=-90]{fig7c.ps} \hspace{-0.8cm}
379: \includegraphics[scale=0.4,angle=-90]{fig7d.ps}
380: \caption{\label{GLf7}
381:  Phase portraits (gray lines) and Poincare maps (black points) for $\omega=1.5$ and $k_S=0$ (a), 
382: $k_S=1.0$ (b) and $k_S=0.1$ (c-d). The initial conditions:
383:  $x_{10}=-2.0$,
384: $v_{10}=\dot{x}_{10}=-0.5$ and $x_{20}=0$ $v_{20}=\dot{x}_{20}=0$.}
385: \vspace{0cm}
386: \end{figure}
387: 
388: The simulation results of the above system (Eq. \ref{GLe9}, Fig. \ref{GLf5}) are presented 
389: in Fig. \ref{GLf6} and \ref{GLf7}. First of all we have repeated the calculations of the amplitude and 
390: Lyapunov exponents. The results for the amplitude $A$ with various couplings ($k_S$)  are 
391: summarized  in Fig. \ref{GLf6}a. Curves '1','2','3' correspond to for $k_S=0$, 0.1 and 1.0, respectively.
392: One can see that a suitable change to $k_S$ may lead to the lowering of the amplitude. 
393: This effect is more efficient 
394: in the case of $k_S=1.0$. Comparing the corresponding values of the maximal Lyapunov exponents one  can  
395: investigate the chaotic nature of the vibrations (Fig. \ref{GLf6}b). Namely, the  case 
396: $k_S=0.1$ (the curve '2') does not reduce the original chaoticity in region  
397: around $\omega \approx 1.0$ and  $\omega \in 
398: [1.24,1.62]$ (the curve '1') while $k_S=1.0$ is large enough to tame the chaos in these regions as well 
399: as to make the vibration amplitude smaller.   
400: 
401: It should be noted that for $k_S \neq 0$ there are four nonzero exponents to be examined. The 
402: positive value of the 
403: largest one $\lambda_1$, is presented in Fig.  
404: \ref{GLf6}b. It detects frequency regions of chaotic vibrations. But it is also interesting to see what has 
405: happened to the other 
406: exponents, especially for our multidimensional system. They were also calculated and are plotted for comparison in Figs.
407: \ref{GLf6}c-d. For a small coupling stiffness, $k_S=0.1$, we have the surprising result that
408:  most of the chaotic 
409: regions are characterized by two positive exponents (Fig. \ref{GLf6}c), which is a signal that the system 
410: is  truly hyperchaotic \cite{Kap94,War99,War00b}. In this case two initially close  trajectories escape 
411: exponentially in two different 
412: directions. The results of calculations with larger $k_S$ ($k_S=1.0$) are different (Fig. \ref{GLf6}), 
413: where regions of chaotic motion with only one positive Lyapunov exponent were found, which signals 
414: typical chaotic behaviour. To clarify this point we plot the phase portrait and Poincare maps for the chosen
415: frequency $\omega=1.5$. Figure \ref{GLf7}a shows the portrait plane and the Poincare map of chaotic 
416: attractor 
417: for $k_S=0$
418: while in Fig. \ref{GLf7}b, for $k_S=1.0$,
419:   the corresponding attractor is regular, synchronized with the excitation frequency $\omega$. Figures 
420: \ref{GLf7}c and d show the  phase portrait and the Poincare map of the hyperchaotic attractor. 
421: 
422: 
423: 
424: \section{Conclusions}
425: This paper has examined the effect of adding an additional small resonance excitation term or an 
426: additional degree of freedom to a simple
427: model of gear vibration (Figs. \ref{GLf1} and \ref{GLf2}). We have shown that the resonance term 
428: can successively reduce chaoticity in the dynamical system (Fig. \ref{GLf3}b), although  
429: the vibration 
430: amplitude increases to a larger value than that for the basic system. We also noted several 
431: sudden jumps in  
432: the vibration amplitude value (Fig. \ref{GLf3}a) which could destroy a real gearbox. The jumps 
433: are presumably caused by the 
434: creation of a larger number of attractors in the presence of the resonance 
435: term (Fig. \ref{GLf4}).  The extra 
436: degree of freedom  (Fig. \ref{GLf5}), which may represent a flexible shaft or 
437: a vibration neutraliser, also has a considerable effect on the dynamics. 
438: We noted that the suitable use of the coupling value, shaft stiffness $k_S$, can lead to regular
439: attractors ($k_S=1.0$) and, simultaneously, to a reduction in the vibration amplitude (Fig. 
440: \ref{GLf6}a-b). Interestingly we have found that for a small value of the coupling stiffness,
441: $k_S=0.1$, two
442: of the four Lyapunov exponents were positive, leading to hyperchaos phenomenon. The study of  
443: the hyperchaotic 
444: attactor will be reported in a separate article. 
445: Thus the coupling $k_S$ is a proper bifurcation parameter giving a wide 
446: range of system behaviour.  The correct choice of this stiffness value can also be used to control the system vibrations.
447:  
448: 
449: 
450: \renewcommand\bibname{References}
451: 
452: \begin{thebibliography}[1]{99}\label{sect:labib}
453: 
454: 
455: \bibitem{Kah90} A.~Kahraman and  R.~Singh,
456: \emph{ Non--linear dynamics of a spur gear pair},
457: Journal of Sound and Vibration \textbf{142} (1990), 49--75.
458: 
459: \bibitem{Sat91} K.~Sato, S.~Yammamoto, T.~Kawakami, \emph{Bifurcation sets and 
460: chaotic states of a gear system subjected to harmonic excitation}, 
461: Computational Mechanics \textbf{7} (1991), 171--182. 
462: 
463: \bibitem{Bla95} G.W.~Blankenship and A.~Kahraman,
464: \emph{Steady state forced response of a mechanical oscillator with combined
465: parametric excitation and clearance type non--linearity},
466: Journal of Sound and Vibration \textbf{185} (1995), 743--765.
467: 
468: \bibitem{Kah97} A.~Kahraman and G.W.~Blankenship, \emph{Experiments on nonlinear
469: dynamic behavior of an oscillator with clearance and periodically
470: time--varying parameters}, ASME Journal of Applied Mechanics \textbf{64} (1997), 217--226.
471: 
472: \bibitem{Rag99} A.~Raghothama and S.~Narayanan,
473: \emph{Bifurcation and chaos in geared rotor bearing system by incremental 
474: harmonic
475: balance method},
476: Journal of Sound and Vibration \textbf{226} (1999), 469--492.
477: 
478: \bibitem{War00}
479: J.~Warmi\'nski, G.~Litak, K.~Szabelski, \emph{Dynamic phenomena in gear boxes},
480:  in  M.~Wiercogroch and B.~De
481: Kraker, editors. \emph{Applied nonlinear dynamics and chaos of mechanical systems with
482: discontinuities}, Series on Nonlinear Science Series A vol. 28, World Scientific,
483: Singapore 2000, 177--205.
484: 
485: 
486: \bibitem{Lit03} G.~Litak and M.I.~Friswell, \emph{Vibrations in gear 
487: systems},  
488: Chaos, Solitons \&
489: Fractals  \textbf{16}  (2003), 145--150.
490: 
491: 
492: 
493: \bibitem{Sza96} K.~Szabelski, G.~Litak, J.~Warmi\'nski and
494: G.~Spuz--Szpos, \emph{Chaotic vibrations  of the parametric system with baclash and
495: non--linear elasticity}, in
496: Proc. of EUROMECH -- 2nd European Nonlinear Oscillation
497: Conference, vol. 1, Prague, September 1996, 431--435.
498: 
499: \bibitem{Sch86} G.~Schmidt, A.~Tondl, \emph{Non-linear vibrations}, 
500: Akademie-Verlag, Berlin 1986.
501: 
502: \bibitem{War00b} J.~Warmi\'nski, G.~Litak and  K.~Szabelski,
503: \emph{Synchronisation and chaos in a parametrically and
504: self--excited system with two degrees of freedom},
505: Nonlinear Dynamics \textbf{22} (2000), 135--153.
506: 
507: 
508: \bibitem{Cha01} R.~Chakon, F.~Palmero and F.~Balibrea, \emph{Taming chaos 
509: in a driven Josephson 
510: junction},
511: International Journal of Bifurcation and Chaos \textbf{11} (2001), 1897--1909.
512: 
513: 
514: \bibitem{Cao03} H.~Cao, X.~Chi and G.~Chen,  \emph{Suppresing or including chaos by weak resonant 
515: excitations in an externally-forced Froude pendulum}, International Journal of Bifurcation and Chaos 
516: (2003), in press. 
517: 
518: \bibitem{Wol88} A.~Wolf, J.B.~Swift, H.L.~Swinney and J.A.~Vastano, \emph{Determining Lyapunov exponents 
519: from a time series}, Physica D \textbf{16} (1988) 285--317. 
520: 
521: \bibitem{Kap94} T.~Kapitaniak and L.O.~Chua, \emph{Hyperchaotic attractors of undirectionally coupled
522: Chua's circuits}, International Journal of Bifurcation and Chaos \textbf{4} (1994) 477--482.
523: 
524: \bibitem{War99} J.~Warmi\'nski, G.~Litak and K.~Szabelski, \emph{Vibrations of a
525: parametrically  
526:  and self--excited system with two degrees of freedom}, in Proc. of Second
527:  International Conference on Identification in Engineering Systems, 
528:  Swansea, March 1999, (Eds. M.I.~Friswell, J.M.~Mottershead and A.W.~Lees,
529:  University of Wales, Swansea 1999), 285--294.
530: 
531:  
532: 
533: \end{thebibliography}
534: 
535: 
536: 
537: 
538: 
539: 
540: \end{document}
541: