Application of neural networks to identification of nonlinear characteristics in cushioning packaging

The structural neural network method is applied to the identification of nonlinear characteristics of cushioning liners in cushioning packaging. The simulated results on the two typical cushioning liner models show that the nonlinear characteristics can be identified perfectly.     

Global stability in switched recurrent neural networks with time-varying delay via nonlinear measure

In this paper, based on switched systems and recurrent neural networks (RNNs) with time-varying delay, the model of switched RNNs is formulated. Global asymptotical stability (GAS) and global robust stability (GRS) for such switched neural networks are studied by employing nonlinear measure and linear matrix inequality (LMI) techniques. Some new sufficient conditions are obtained to ensure GAS or GRS of the unique equilibrium of the proposed switched system. Furthermore, the proposed LMI results are computationally efficient as it can be solved numerically with standard commercial software. Finally, three examples are provided to illustrate the usefulness of the results.     

On the use of discontinuous nonlinear bistable dynamics to increase the responsiveness of energy harvesting devices

There is promise in the use of bistable devices to transduce ambient vibrations into electrical power. However, it is critical to sustain the relatively large amplitude snap-through motion, or interwell motion, to significantly improve the responsiveness of bistable devices as compared to linear resonance-based approaches. This work posits that relatively stiff structural elements can be placed in the vicinity of the equilibria of bistable devices such that the discontinuous change in dynamics will tend to eject an otherwise small amplitude motion into the large amplitude interwell orbit that is to be preferred for energy harvesting applications. The discontinuous nonlinear dynamic equations of motion are derived and a proxy system parametrically studied. These numerical studies demonstrate that discontinuous nonlinear bistable devices have a significantly broadened frequency range that elicits the large amplitude snap through behavior. It is also seen that interwell motion is achievable at significantly reduced excitation amplitudes through these discontinuous structural elements.     

An exact nonlinear hybrid-coordinate formulation for flexible multibody systems

The previous low-order approximate nonlinear formulations succeeded in capturing the stiffening terms, but failed in simulation of mechanical systems with large deformation due to the neglect of the high-order deformation terms. In this paper, a new hybrid-coordinate formulation is proposed, which is suitable for flexible multibody systems with large deformation. On the basis of exact strain–displacement relation, equations of motion for flexible multibody system are derived by using virtual work principle. A matrix separation method is put forward to improve the efficiency of the calculation. Agreement of the present results with those obtained by absolute nodal coordinate formulation (ANCF) verifies the correctness of the proposed formulation. Furthermore, the present results are compared with those obtained by use of the linear model and the low-order approximate nonlinear model to show the suitability of the proposed models. The project supported by the National Natural Science Foundation of China (10472066, 50475021).     

Normal modes for piecewise linear vibratory systems

A method to construct the normal modes for a class of piecewise linear vibratory systems is developed in this study. The approach utilizes the concepts of Poincaré maps and invariant manifolds from the theory of dynamical systems. In contrast to conventional methods for smooth systems, which expand normal modes in a series form around an equilibrium point of interest, the present method expands the normal modes in a series form of polar coordinates in a neighborhood of an invariant disk of the system. It is found that the normal modes, modal dynamics and frequency-amplitude dependence relationship are all of piecewise type. A two degree of freedom example is used to demonstrate the method.     

Rainfall frequency and seasonality identification through artificial neural networks

The artificial neural network technique is experimented to cope with the study of the sub-annual seasonal non-stationarity of the rainfall process. The homogeneity of the climatic signals inside each of the natural 12 monthly classes is analyzed, adopting a multilayer feed-forward network with error back-propagation. The possibility of identifying monthly based seasons from only daily rainfall data is found to be quite limited. The coupling of rainfall and temperature statistics is instead confirmed to be a fundamental climatic indicator. Contrary to what is commonly expected, the season uncertainty appears higher in summer and in winter than in spring or autumn. The hypothesis of defining any monthly based pluviometric regime is however demonstrated to be generally difficult to sustain, revealing the necessity of adopting an unsupervised criterion to identify any seasonal filter of the rainfall process.

---

Sommario Viene sperimentata la tecnica delle reti neurali artificiali per affrontare lo studio della non-stazionarità stagionale del processo di precipitazione. Viene analizzata l'omogeneità dei segnali climatici all'interno di ciascuna delle 12 naturali classi mensili, adottando un rete di tipo multistrato feed-forward con retro-propagazione dell'errore. La possibilità di indentificare le stagioni su base mensile con i soli dati di precipitazione si dimostra essere piuttosto limitata. Viceversa, l'accoppiamento fra le statistiche di pioggia e temperatura si rivela essere un fondamentale indicatore climatico. Al contrario di quanto comunemente reputato, l'incertezza nella stagionalità appare essere più alta in estate ed in inverno che in primavera e in autunno. L'ipotesi sulla definizione su base mensile dei regimi pluviometrici appare comunque difficile da sostenere, rivelando la necessità di adottare un criterio senza supervisione per l'identificazione del filtro stagionale relativo al processo di precipitazione.

     

Output frequency properties of nonlinear systems

Nonlinear systems usually have complicated output frequencies. For the class of Volterra systems, some interesting properties of the output frequencies are studied in this paper. These properties show theoretically the periodicity of the output super-harmonic and inter-modulation frequencies and clearly demonstrate the mechanism of the interaction between different output harmonics incurred by different input nonlinearities in system output spectrum. These new results have significance in the analysis and design of nonlinear systems and nonlinear filters in order to achieve a specific output spectrum in a desired frequency band by taking advantage of nonlinearities. Examples and discussions are given to illustrate these new results.     

A perturbation method for evaluating nonlinear normal modes of a piecewise linear two-degrees-of-freedom system

The classical Lindstedt–Poincaré method is adapted to analyze the nonlinear normal modes of a piecewise linear system. A simple two degrees-of-freedom, representing a beam with a breathing crack is considered. The fundamental branches of the two modes and their stability are drawn by varying the severity of the crack, i.e., the level of nonlinearity. Results furnished by the asymptotic method give insight into the mechanical behavior of the system and agree well with numerical results; the existence of superabundant modes is proven. The unstable regions and the bifurcated branches are followed by a numerical procedure based on the Poincarè map.     

Interaction of free and forced nonlinear normal modes in two-DOF dissipative systems under resonance conditions

The resonance dynamics of a dissipative spring-mass and of a dissipative spring-pendulum system is studied. Internal resonance case is considered for the first system; both external resonances and simultaneous external and internal resonance are studied for the second one. Analysis of the systems resonance behavior is made on the base of the concept of nonlinear normal vibration modes (NNMs) by Kauderer and Rosenberg, which is generalized for dissipative systems. The multiple time scales method under resonance conditions is applied. The resulting equations are reduced to a system with respect to the system energy, arctangent of the amplitudes ratio and the difference of phases of required solution in the resonance vicinity. Equilibrium positions of the reduced system correspond to nonlinear normal modes; in energy dissipation case they are quasi-equilibriums. Analysis of the equilibrium states of the reduced system permits to investigate stability of nonlinear normal modes in the resonance vicinity and to describe transfer from unstable vibration mode to stable one. New vibration regimes, which are called transient nonlinear normal modes (TNNMs) are obtained. These regimes take place only for some particular levels of the system energy. In the vicinity of values of time, corresponding to these energy levels, the TTNM attract other system motions. Then, when the energy decreases, the transient modes vanish, and the system motions tend to another nonlinear normal mode, which is stable in the resonance vicinity. The reliability of the obtained analytical results is confirmed by numerical and numerical-analytical simulations.     

Analysis of time-periodic nonlinear dynamical systems undergoing bifurcations

In this study a new procedure for analysis of nonlinear dynamical systems with periodically varying parameters under critical conditions is presented through an application of the Liapunov-Floquet (L-F) transformation. The L-F transformation is obtained by computing the state transition matrix associated with the linear part of the problem. The elements of the state transition matrix are expressed in terms of Chebyshev polynomials in timet which is suitable for algebraic manipulations. Application of Floquet theory and the eigen-analysis of the state transition matrix at the end of one principal period provides the L-F transformation matrix in terms of the Chebyshev polynomials. Since this is a periodic matrix, the L-F transformation matrix has a Fourier representation. It is well known that such a transformation converts a linear periodic system into a linear time-invariant one. When applied to quasi-linear equations with periodic coefficients, a dynamically similar system is obtained whose linear part is time-invariant and the nonlinear part consists of coefficients which are periodic. Due to this property of the L-F transformation, a periodic orbit in original coordinates will have a fixed point representation in the transformed coordinates. In this study, the bifurcation analysis of the transformed equations, obtained after the application of the L-F transformation, is conducted by employingtime-dependent center manifold reduction andtime-dependent normal form theory. The above procedures are analogous to existing methods that are employed in the study of bifurcations of autonomous systems. For the two physical examples considered, the three generic codimension one bifurcations namely, Hopf, flip and fold bifurcations are analyzed. In the first example, the primary bifurcations of a parametrically excited single degree of freedom pendulum is studied. As a second example, a double inverted pendulum subjected to a periodic loading which undergoes Hopf or flip bifurcation is analyzed. The methodology is semi-analytic in nature and provides quantitative measure of stability when compared to point mappings method. Furthermore, the technique is applicable also to those systems where the periodic term of the linear part does not contain a small parameter which is certainly not the case with perturbation or averaging methods. The conclusions of the study are substantiated by numerical simulations. It is believed that analysis of this nature has been reported for the first time for this class of systems.     

Parametric identification of nonlinear systems using multiple trials

It is observed that the harmonic balance (HB) method of parametric identification of nonlinear system may not give right identification results for a single test data. A multiple-trial HB scheme is suggested to obtain improved results in the identification, compared with a single sample test. Several independent tests are conducted by subjecting the system to a range of harmonic excitations. The individual data sets are combined to obtain the matrix for inversion. This leads to the mean square error minimization of the entire set of periodic orbits. It is shown that the combination of independent test data gives correct results even in the case where the individual data sets give wrong results.     

Design of state estimator for uncertain neural networks via the integral-inequality method

The issue of state estimation is studied for a class of neural networks with norm-bounded parameter uncertainties and time-varying delay. Some new linear matrix inequality (LMI) representations of delay-dependent stability criteria are presented for the existence of the desired estimator for all admissible parametric uncertainties. The proposed method is based on the S-procedure and an extended integral inequality which can be deduced from the well-known Leibniz–Newton formula and Moon’s inequality. The results extend some models reported in the literature and improve conservativeness of those in the case that the derivative of the time-varying delay is assumed to be less than one. Two numerical examples are given to show the effectiveness and superiority of the results.     

Forward kinematics analysis of the 6-3 SPM by using neural networks

A feed-forward neural network has been trained using backpropagation algorithm to solve the forward kinematics problem of the 6-3 Stewart Platform Mechanism (SPM). The forward kinematics problem of the SPM does not have a unique solution since it involves solving a polynomial of order 16. Purely translational, purely rotational and general spatial data sets have been used in training and testing, and then an optimization procedure has been applied to fine-tune the solution. The method yields results fast and accurate enough such that it can be used instead of a gyro and a position sensor for real time control of the mechanism.     

On a computational approach for the approximate dynamics of averaged variables in nonlinear ODE systems: Toward the derivation of constitutive laws of the rate type

A non-perturbative approach to the time-averaging of nonlinear, autonomous ordinary differential equations is developed based on invariant manifold methodology. The method is implemented computationally and applied to model problems arising in the mechanics of solids.     

Globally convergent homotopy algorithms for nonlinear systems of equations

Probability-one homotopy methods are a class of algorithms for solving nonlinear systems of equations that are accurate, robust, and converge from an arbitrary starting point almost surely. These new globally convergent homotopy techniques have been successfully applied to solve Brouwer fixed point problems, polynomial systems of equations, constrained and unconstrained optimization problems, discretizations of nonlinear two-point boundary value problems based on shooting, finite differences, collocation, and finite elements, and finite difference, collocation, and Galerkin approximations to nonlinear partial differential equations. This paper introduces, in a tutorial fashion, the theory of globally convergent homotopy algorithms, deseribes some computer algorithms and mathematical software, and presents several nontrivial engineering applications.This work was supported in part by DOE Grant DE-FG05-88ER25068, NASA Grant NAG-1-1079, and AFOSR Grant 89-0497.     

Vibration response of rotating mechanical systems using experimental techniques and artificial neural networks

A neural network predictor investigation is presented for analyzing vibration parameters of a rotating system. The vibration parameters of the system, such as amplitude, velocity, and accelertion in the vertical direction, were measured at the bearing points. The system's vibration and noise were analyzed for different working conditions. The designed neural predictor has three layers, which are input, hidden, and output layers. In the hidden layer, 10 neurons were used for this approximation. The results show that the network can be used as an analyzer of such systems in experimental applications.     

Analytical procedure for determining the linear and nonlinear effective properties of the elastic composite cylinder

In this work we consider a cylindrical structure composed of a nonlinear core (inhomogeneity) surrounded by a different nonlinear shell (matrix). We elaborate a technique for determining its linear elastic moduli (second order elastic constants) and the nonlinear elastic moduli, which are called Landau coefficients (third order elastic constants). Firstly, we develop a nonlinear perturbation method which is able to turn the initial nonlinear elastic problem into a couple of linear problems. Then, we prove that only the solution of the first linear problem is necessary to calculate the linear and nonlinear effective properties of the heterogeneous structure. The following step consists in the exact solution of such a linear problem by means of the complex elastic potentials. As result we obtain the exact closed forms for the linear and nonlinear effective elastic moduli, which are valid for any volume fraction of the core embedded in the external shell.     

An experimental study of nonlinear dynamic system identification

A technique for robust identification of nonlinear dynamic systems is developed and illustrated using both digital simulations and analog experiments. The technique is based on the Minimum Model Error optimal estimation approach. A detailed literature review is included in which fundamental differences between the current approach and previous work is described. The most significant feature of the current work is the ability to identify nonlinear dynamic systems without prior assumptions regarding the form of the nonlinearities, in contrast to existing nonlinear identification approaches which usually require detailed assumptions of the nonlinearities. The example illustrations indicate that the method is robust with respect to prior ignorance of the model, and with respect to measurement noise, measurement frequency, and measurement record length.     

Analysis of periodic-quasiperiodic nonlinear systems via Lyapunov-Floquet transformation and normal forms

In this paper a general technique for the analysis of nonlinear dynamical systems with periodic-quasiperiodic coefficients is developed. For such systems the coefficients of the linear terms are periodic with frequency ω while the coefficients of the nonlinear terms contain frequencies that are incommensurate with ω. No restrictions are placed on the size of the periodic terms appearing in the linear part of system equation. Application of Lyapunov-Floquet transformation produces a dynamically equivalent system in which the linear part is time-invariant and the time varying coefficients of the nonlinear terms are quasiperiodic. Then a series of quasiperiodic near-identity transformations are applied to reduce the system equation to a normal form. In the process a quasiperiodic homological equation and the corresponding ‘solvability condition’ are obtained. Various resonance conditions are discussed and examples are included to show practical significance of the method. Results obtained from the quasiperiodic time-dependent normal form theory are compared with the numerical solutions. A close agreement is found.     

The stability of limit cycles in nonlinear systems

In this paper, a necessary condition is first presented for the existence of limit cycles in nonlinear systems, then four theorems are presented for the stability, instability, and semistabilities of limit cycles in second order nonlinear systems. Necessary and sufficient conditions are given in terms of the signs of first and second derivatives of a continuously differentiable positive function at the vicinity of the limit cycle. Two examples considering nonlinear systems with familiar limit cycles are presented to illustrate the theorems.