Continuous-time bilinear system identification using single experiment with multiple pulses

A novel method is presented for the identification of a continuous-time bilinear system from the input?Coutput data generated by a single experiment with multiple pulses. In contrast to the conventional approach utilizing multiple experiments, the current work documents the advantage of using a single experiment and sets up a procedure to obtain bilinear system models. The special pulse inputs employed by earlier research can be avoided and accurate identification of the continuous-time system model is possible by performing a single experiment incorporating a class of control input sequences combining pulses with free-decay response. The algorithm presented herein is more attractive in practice for the identification of bilinear systems. Numerical examples presented demonstrate the methodology developed in the paper.     

A general method for obtaining shear stress and normal stress functions from parallel disk rheometry data

The problems of converting the torque and normal force versus rim shear rate data generated by parallel disk rheometers into shear stress and normal stress difference as functions of shear rate are formulated as two independent integral equations of the first kind. Tikhonov regularization is used to obtain approximate solutions of these equations. This way of handling parallel disk rheometer data has the advantage that it is independent of the rheological constitutive equation and noise amplification is kept under control by the user-specified parameter in Tikhonov regularization. If the fluid under test exhibits a yield stress, Tikhonov regularization computation will simultaneously give an estimate of the yield stress. The performance of this method is demonstrated by applying it to a number of data sets taken from the published literature and to laboratory measurements conducted specifically for this investigation.     

A fast harmonic balance technique for periodic oscillations of an aeroelastic airfoil

The harmonic balance (HB) method is utilized to obtain the periodic solutions for the two-dimensional airfoil with cubic nonlinearity in pitch undergoing subsonic flow. In the course of formulating the HB algebraic system, the manipulation software Mathematica is employed to deal with the complex Fourier coefficients involved with the nonlinear term. In general, to solve the HB algebraic system, either a symbolic calculation or a numerical approximation of the Jacobian matrix is required in each iteration, which is computationally expensive. To remedy this drawback, the Jacobian matrix is explicitly derived in this paper. The effects of exploiting the explicit Jacobian matrix on the accuracy and efficiency of the HB method are investigated, through comparing with the case using a numerical Jacobian matrix calculated by a three-point difference technique. Moreover, the spectral analysis is applied to the periodic motions by the numerical method to provide insight into the distribution of the dominant frequencies, so as to provide a reasonable suggestion for the truncation of the Fourier series expansion in the HB method. In addition, a frequency modulation phenomenon is identified in the pitch motions via spectral analysis, whose effect on the accuracy of the HB method is examined both numerically and analytically. Finally, illustrative examples validate that the HB method with as many harmonics as the spectral analysis suggests can yield sufficiently accurate solutions.     

Adaptive neural DSC for stochastic nonlinear constrained systems under arbitrary switchings

This paper focuses on the primary resonance analysis of a dual-rotor system having two rotor unbalance excitations of different rotating speeds and being connected by an inter-shaft ball bearing. Due to the complex excitation condition and the complicated nonlinear bearing forces of the inter-shaft bearing, the general analytical methods, e.g., the multiple scales method or the harmonic balance method, are failed to give the theoretical solutions. Thus, the harmonic balance–alternating frequency/time domain (HB–AFT) method is formulated to deal with this problem. The basic idea of the method is using the inverse discrete Fourier transform and the discrete Fourier transform, instead of the direct analytical relationship between the supposed solutions of the system and the nonlinear forces, to construct the harmonic expressions of the nonlinear forces, which is the so-called alternating frequency/time domain technique. By using the HB–AFT method, therefore, a Newton– Raphson iteration procedure can be performed to demonstrate the approximate solutions of the system. Accordingly, the frequency responses of the system affected by some critical parameters, such as rotating speed ratio, unbalances of both the inner and outer rotors, and clearance of the inter-shaft bearing, are analyzed, respectively. A variety of phenomena including double resonance peaks, biperiodic and quasi-periodic behaviors, and resonance hysteresis phenomenon are obtained, which are discussed in detail through diagrams for separated frequency responses with different frequency components and rotors’ orbits for different combinations of system parameters. Most prominently, for a relatively small unbalance of rotor as well as a relatively large clearance of the inter-shaft bearing, the resonance hysteresis phenomena are more obvious. The results obtained are also compared with the direct numerical simulation results, and the comparisons show good agreements. In addition, the methodology formulated in this paper is a general approach, which can be applied to other engineering systems with multi-frequency excitations.     

Identification of Nonlinear and Viscoelastic Properties of Flexible Polyurethane Foam

Analysis of the steady-state response of a polyurethane foam and masssystem to harmonic excitation is presented. The foam's uni-directionaldynamic behavior is modeled by using nonlinear stiffness, linearviscoelastic and velocity proportional damping components. Therelaxation kernel for the viscoelastic model is assumed to be a sum ofexponentials. The harmonic balance method is used to develop one- andtwo-term approximations to periodic solutions, and the equationsdeveloped are utilized for system identification. The identificationprocess is based on least-squares minimization of a sub-optimal costfunction that uses response data at various excitation frequencies andamplitudes. The effects of frequency range, spacing and amplitudes ofthe harmonic input on the results of the model parameter estimation arediscussed. The identification procedure is applied to measurements ofthe steady-state response of a base-excited foam-mass system. Estimatesof the system parameters at different levels of compression and inputamplitudes are thus determined. The choice of model-order and thefeasibility of describing the system behavior at several inputamplitudes with a single set of parameters are also addressed.     

Response and bifurcation analysis of a MDOF rotor system with a strong nonlinearity

A new HB (Harmonic Balance)/AFT (Alternating Frequency Time) method is further developed to obtain synchronous and subsynchronous whirling response of nonlinear MDOF rotor systems. Using the HBM, the nonlinear differential equations of a rotor system can be transformed to algebraic equations with unknown harmonic coefficients. A technique is applied to reduce the algebraic equations to only those of the nonlinear coordinates. Stability analysis of the periodic solutions is performed via perturbation of the solutions. To further reduce the computational time for the stability analysis, the reduced system parameters (mass, damping, and stiffness) are calculated in terms of the already known harmonic coefficients. For illustration, a simple MDOF rotor system with a piecewise-linear bearing clearance is used to demonstrate the accuracy of the calculated steady-state solutions and their bifurcation boundaries. Employing ideas from modern dynamics theory, the example MDOF nonlinear rotor system is shown to exhibit subsynchronous, quasi-periodic and chaotic whirling motions.     

A polycrystal plasticity model based on the mechanical threshold

A temperature and rate-dependent viscoplastic polycrystalmodel is presented.It uses a single crystal constitutive response that is based on the isotropic Mechanical Threshold Stress continuum model. This combination gives us theability to relate the constitutive model parameters between the polycrystaland continuum models. The individual crystal response is used to obtain themacroscopic response through the extended Taylor hypothesis. A Newton-Raphsonalgorithm is used to solve the set of fully implicit nonlinear equations for each crystal. The analysis also uses a novel state variable integration method which renders the analysis time step size independent for constant strain rate simulations. Material parameter estimates are obtained through an identification study, where the error between experimental and computed stress response is minimized. The BFGS method, which is used to solve theidentification problem, requires first-order gradients. These gradients arecomputed efficiently via the direct method of design sensitivity analysis.Texture augmentation is performed in a second identification study by changing crystal weights (volume fractions).     

Mathematic modeling and characteristic analysis for dynamic system with asymmetrical hysteresis in vibratory compaction

We investigate dynamic characteristics of vibratory compaction system with asymmetrical hysteresis. An asymmetrical model derived from Bouc-Wen differential equation is employed to describe hysteretic behavior of vibration engineering. A practical polynomial expression for hysteretic restoring force is deduced to be substituted into standard equation of the system, assuming that the non-linearity of the restoring force is weak. An asymptotic method, which combines Krylov-Bogolyubov-Mitropolsky (KBM) method with harmonic balance (HB) method, is applied to analyze steady-state responses of the asymmetrical hysteretic system subjected to harmonic excitation. Dynamic responses, such as the restoring force time histories and frequency responses of the system for the first order approximate, are obtained. Furthermore, numerical solution obtained using Runge-Kutta method as well as results of experiments (asphalt compaction on the Beijing-Fuzhou highway) are compared with the asymptotic solution. These results investigated that asymmetrical hysteretic model and the methods applied in this paper are quite appropriate for engineering applications.     

Linearized harmonic balance based derivation of slow flow for some class of autonomous single degree of freedom oscillators

In this paper we exploit the embedding of linearization in the harmonic balance method developed by Wu and its collaborators to propose an approach for deriving the slow flow for some class of damped autonomous single degree of freedom oscillators. The linearized harmonic balance method is used to compute the coefficients of the harmonics of an assumed form of the solution and to derive a system of two coupled ordinary differential equations related to the slow flow. A power series procedure is next used to decouple the coupled system and to obtain the slow flow. Two examples provided to illustrate the proposed procedure show excellent results.     

梁结构时变物理参数在线识别的谐波传播法

组合梁结构在运动状态下其连接处的物理参数不断变化,导致结构的动态特性也不断发生变化.为了研究组合梁结构在运动状态下的非平稳动态特性以达到对结构振动进行实时控制的目的,必须对结构连接处的物理参数进行在线跟踪辨识.本文在现代谐波识别技术的基础上提出时变系统谐波频率与振型的在线识别算法,并利用波传播方法计算出连接处的力与位移向量,再由连接处位移连续与力平衡条件求得连接处的动刚度,进而求得结构连接处的刚度与阻尼.由于该方法可以简化测试节点,所以可以节省测试成本.仿真算例表明该方法具有良好的跟踪能力与较高的计算精度.     

一类加权全局迭代参数卡尔曼滤波算法

结合参数卡尔曼滤波算法和全局迭代推广卡尔曼滤波算法本文提出了加权全局迭代参数卡尔曼滤波算法。参数卡尔曼滤波算法可避免系统参数和状态变量之间的非线性耦合 ,同时通过带有目标函数的全局迭代算法保证能够获取到稳定、收敛的识别结果。分别针对线性结构模型和随动强化双线性结构模型进行了仿真参数识别。结果显示 ,不加权的全局迭代参数卡尔曼滤波算法对线性系统是有效的 ,而对非线性系统必须使用加权的全局迭代参数卡尔曼滤波算法。当信噪比较大 ,迭代无法得到收敛的结果时 ,目标函数保证了较好识别结果的获得     

高超声速风洞短时气动力智能辨识算法研究

风洞测力试验是高超声速飞行器研发的重要环节,随着这项技术的发展,试验模型的大尺度化成为高超声速风洞试验的趋势.在几百毫秒的有效测试时间内,大尺度测力系统刚度减弱等问题会严重导致气动力辨识精度变差,试验模型大尺度化对短时脉冲燃烧风洞精确气动力辨识带来了挑战.对此本文提出了一种新的基于传统信号处理结合深度学习的智能气动力辨...     

Three-dimensional electric potential induced by a point singularity in a multilayered dielectric medium

A simple and effective method is proposed to derive the three-dimensional electric potential induced by a point singularity of any type in an N-phase dielectric medium composed of N-2 intermediate dielectric layers of equal thickness encased in two semi-infinite dielectric media. The point singularity can include a point charge or a point electric dipole. The original boundary value problem for the N-phase medium is reduced to the determination of a single unknown three-dimensional harmonic function through satisfaction of the continuity conditions across all of the N-1 perfect planar interfaces. The single harmonic function can be completely determined after analytically solving the resulting linear recurrence relations, which are independent of the type and the specific location of the singularity. When the singularity is a point charge, we obtain the self-energy of the point charge expressed in terms of a single function and the Coulomb force on the point charge expressed in terms of the negative derivative of this function.     

Dynamics of nonlinear ecosystems under colored noise disturbances

Stochastic ecosystems of prey-predator type subjected to colored noises with broad-band spectra are investigated. Nonlinear models are considered for two different scenarios: one is the case of possible abundant prey supply and another is the case of possible large predator population. The stochastic averaging procedure is applied to obtain stationary probability solutions of the nonlinear systems. Two types of colored noise are considered: one is the low-pass filtered noise with the spectrum peak at zero frequency, and another is the randomized harmonic process with the spectrum peak at a nonzero frequency. For either type of the noises, the band width reflecting the level of the noise color can be adjusted using a single parameter. The analytical results are substantiated by those obtained from Monte Carlo simulations. It is found that the noise color has significant effects on the stationary state of the system. A narrower band width leads to a less stable system in the sense that the prey and predator populations deviate farther from the equilibrium point of the system without noise disturbances.     

Parametric identification of a chaotic base-excited double pendulum experiment

The parametric identification of a chaotic system was investigated for a double pendulum. From recorded experimental response data, the unstable periodic orbits (UPOs) were extracted and then used in a harmonic balance identification process. By applying digital filtering, digital differentiation and linear regression techniques for optimization, the results were improved. Verification of the related simulation system and linearized system also corroborated the success of the identification algorithm.     

Basic study of probabilistic approach to prediction of soil trafficability — Statistical characteristics of cone index

The principal purpose of this study is to present some of the fundamental data which might be used in the prediction of soil trafficability. These data are based on some sets of random cone index measurements which were taken from the areas of interest. In all cases, it was assumed that the individual measurements of the cone index being investigated would be independent. It is shown that the cone indices at critical layer depth can be regarded as normal random variables. The discussion in this study may be valid for the probabilistic approach of soil trafficability     

Development of data-based model-free representation of non-conservative dissipative systems

A general procedure is presented for developing data-based, non-parametric models of non-linear multi-degree-of-freedom, non-conservative, dissipative systems. Two broad classes of methods are discussed: one relying on the representation of the system restoring forces in a polynomial-basis format, and the other using artificial neural networks to map the complex transformations relating the system state variables to the needed system outputs. A non-linear two-degree-of-freedom system is used to formulate the approach under discussion and to generate synthetic data for calibrating the efficiency of the two methods in capturing complex non-linear phenomena (such as dry friction, hysteresis, dead-space non-linearities, and polynomial-type non-linearities) that are widely encountered in the applied mechanics field. Subsequently, a reconfigurable test apparatus was used to generate experimental measurements from a physical non-linear “joint” involving two-dimensional motion (translation and rotation) and complicated interaction forces between the different motion axes, among its internal elements. Both the polynomial-basis approach and the neural network method were used to develop high-fidelity, non-parametric models of the physical test article. The ability of the identified models to accurately “generalize” the essential features of the non-linear system was verified by comparing the predictions of the models with experimental measurements from data sets corresponding to different excitations than those used for identification purposes. It is shown that the identification techniques under discussion can be useful tools for developing accurate simulation models of complex multi-dimensional non-linear systems under broadband excitation.     

Time-delay identification for linear controlled systems

A new approach for time-delay identification is proposed in linear controlled systems. The delay is derived from the control loop in the system. The frequency-response function of the system is presented in terms of the impedance matrix. It is proved that the inverse form of the function may be expressed in the harmonic function, which is used to fit those data from the experiment. As an example, an isolator with the delayed feedback control is schemed to acquire such data. Using least square algorithm yields that the identified delay can reach any required accuracy.     

Control of a Nonlinear System Using the Saturation Phenomenon

In this paper, a theoretical investigation of nonlinear vibrations of a 2 degrees of freedom system when subjected to saturation is studied. The method has been especially applied to a system that consists of a DC motor with a nonlinear controller and a harmonic forcing voltage. Approximate solutions are sought using the method of multiple scales. It is shown that the closed-loop system exhibits different response regimes. The nature and stability of these regimes are studied and the stability boundaries are obtained. The effects of the initial conditions on the response of the system have also been investigated. Furthermore, the second-order solution is presented and the corresponding results are compared with those of the first-order solution. It is shown that by increasing the amplitude of the excitation voltage, the higher-order term in the solution becomes significant and causes a drift in the response. In order to verify the obtained theoretical results, they are compared with the corresponding numerical results. Good agreement between the two sets of results is observed.     

飞行器气动参数辨识进展

飞行器气动参数辨识研究的主题,是应用系统辨识技术从飞行试验数据求取气动力,从而建立飞行器动力学系统的数学模型.它研究的对象是飞行器;解决的是空气动力学问题;采用的基本方程是飞行动力学的运动方程组;应用的研究手段是现代控制论的滤波、预测和估计理论.它是处于空气动力学、飞行力学、弹性力学和现代控制论之间的应用性研究课题. 本文综述了国内外公开发表的飞行器参数辨识研究的理论结果和实践经验,分八个专题——模型辨识,参数估计,数据预处理和相容性检验,试验设计与最佳输入,弹性与非定常效应,频域辨识,闭环辨识,辨识准度与系统验证——评述其研究进展和现状.