Convergence and stability of numerical solutions to a class of index 1 stochastic differential algebraic equations with time delay

2010 ◽  
Vol 215 (11) ◽  
pp. 4008-4021 ◽  
Author(s):  
Xiaomei Qu ◽  
Chengming Huang ◽  
Chao Liu
Author(s):  
Hajrudin Pasic

Abstract Presented is an algorithm suitable for numerical solutions of multibody mechanics problems. When s-stage fully implicit Runge-Kutta (RK) method is used to solve these problems described by a system of n ordinary differential equations (ODE), solution of the resulting algebraic system requires 2s3 n3 / 3 operations. In this paper we present an efficient algorithm, whose formulation differs from the traditional RK method. The procedure for uncoupling the algebraic system into a block-diagonal matrix with s blocks of size n is derived for any s. In terms of number of multiplications, the algorithm is about s2 / 2 times faster than the original, nondiagonalized system, as well as s2 times in terms of number of additions/multiplications. With s = 3 the method has the same precision and stability property as the well-known RADAU5 algorithm. However, our method is applicable with any s and not only to the explicit ODEs My′ = f(x, y), where M = constant matrix, but also to the general implicit ODEs of the form f (x, y, y′) = 0. In the solution procedure y is assumed to have a form of the algebraic polynomial whose coefficients are found by using the collocation technique. A proper choice of locations of collocation points guarantees good precision/stability properties. If constructed such as to be L-stable, the method may be used for solving differential-algebraic equations (DAEs). The application is illustrated by a constrained planar manipulator problem.


2021 ◽  
pp. 147715352199159
Author(s):  
X Sun ◽  
LB Kong ◽  
PY Zhou ◽  
M Xu

In machine vision systems, the objectives to be detected, such as circuit boards, may be composed of many different materials and shapes, which can lead to highlights, shadows and error information in captured images under traditional uniform illumination. A lighting system that generates tailored irradiance in different regions in machine vision is needed. This paper proposes a design method for an LED lens and obtains special total-internal-reflection (TIR) lenses that generate tailored illumination, which can adapt to the reflectance and shape of a target. Differential-algebraic equations (DAEs) based on the conservation of flux are established to ensure the uniform illumination on the targets, nonlinear equations based on the edge-ray principle are employed to generate the spots with tailored shapes and the numerical solutions can be fitted into the proposed lenses. Six different tailored faculae are generated to verify the proposed method. The results show that the uniformity of tailored facula can exceed 87%, and the light efficiency can exceed 85%. In particular, the contrast of the irradiance among different regions can be adjusted by the boundary conditions; thus, the proposed method can satisfy the complex demand for machine vision and be applied to improve vision detection systems in production lines.


2003 ◽  
Vol 139 (2-3) ◽  
pp. 259-264 ◽  
Author(s):  
Ercan Çelik ◽  
Erdal Karaduman ◽  
Mustafa Bayram

Author(s):  
H. Ersin Erol ◽  
Altuğ İftar

This paper presents a complete approach for designing stabilizing linear time-invariant decentralized finite-dimensional or retarded time-delay output feedback controllers for linear time-invariant systems of delay differential-algebraic equations. The proposed approach is based on the sequential design of the local controllers by using a centralized controller design algorithm. In this sequential design approach, the local controller to be designed at each step is determined depending on the mobility of the rightmost modes with respect to the controllers that have not yet been designed and closed with the system. Since no predefined sequence is followed, a sequence that can target the least effort and dimension for each agent can be aimed. Also, in the proposed approach, a base controller effort can be targeted for each control agent, so that the effort required to stabilize the system can be distributed among the local controllers. In the centralized controller design algorithm used for the design of each local controller, the parameters of the controllers are changed stepwise in a quasi-continuous way to shift the targeted rightmost modes towards the stable area. For a time-delay controller, the desired mode placement can be achieved by applying small changes stepwise to the elements of the matrices and the time-delays of the controller while time-delays remain always non-negative. The effect of small perturbations on the time-delays in the open-loop system or to be added by the controller to be designed is taken into account to ensure some degree of robustness against all possible perturbations on the delays. The effectiveness of the proposed design approach is demonstrated by a numerical example.


Sign in / Sign up

Export Citation Format

Share Document