A TRUST REGION APPROACH WITH MULTIVARIATE PADÉ MODEL FOR THE OPTIMIZATION OF REGULAR AND FRACTIONAL ORDER CIRCUITS
Shaimaa Ebid Kamel Ebid;
Abstract
Optimization is very important to find optimal nominal values of the designable system parameters. The system is required to satisfy the design specifications as well as not to be too sensitive to parameter variations. Parameter variations can be due to noise or unavoidable statistical fluctuations in the fabrication process. Usually the objective function defined by the system specifications is computationally expensive. Since the optimization process requires a significant number of function evaluations, it is recommended to represent the objective function by building up a model that approximates the objective function within a certain trust region. Many models are used among them linear and quadratic models. In this thesis, the objective function is approximated by building rational models called multivariate Padé model over a sequence of trust regions. The multivariate Padé model is constructed by using data points of O(n), where n is the number of design parameters. The proposed approach is tested by applying it to several bench mark problems.
In case of the fractional order circuits, the sensitivities are derived based on a matrix approach. An adjoint matrix approach is used in the derivation of the sensitivities in this thesis. The sensitivity with respect to the fractional derivative orders α and β are also derived. The use of fractional order elements instead of regular integer elements enhances a better circuit performance. Optimal design using the derived sensitivity can be obtained using efficient gradient optimization techniques.
The yield is defined as the probability that a design satisfies the specifications. It is difficult process to calculate as yield is represented as a statistical function. Therefore, it is desired to obtain a good starting point for yield optimization process. The proposed method with gradient optimization technique is used to obtain this starting po
In case of the fractional order circuits, the sensitivities are derived based on a matrix approach. An adjoint matrix approach is used in the derivation of the sensitivities in this thesis. The sensitivity with respect to the fractional derivative orders α and β are also derived. The use of fractional order elements instead of regular integer elements enhances a better circuit performance. Optimal design using the derived sensitivity can be obtained using efficient gradient optimization techniques.
The yield is defined as the probability that a design satisfies the specifications. It is difficult process to calculate as yield is represented as a statistical function. Therefore, it is desired to obtain a good starting point for yield optimization process. The proposed method with gradient optimization technique is used to obtain this starting po
Other data
| Title | A TRUST REGION APPROACH WITH MULTIVARIATE PADÉ MODEL FOR THE OPTIMIZATION OF REGULAR AND FRACTIONAL ORDER CIRCUITS | Authors | Shaimaa Ebid Kamel Ebid | Issue Date | 2016 |
Recommend this item
Similar Items from Core Recommender Database
Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.