Wyniki wyszukiwania - BazTech

1

Combination of the Levenberg–Marquardt and diferential evolution algorithms for the ftting of postseismic GPS time series

Li Jingwei, Zhan Wei, Guo Bofeng, Li Shuiping, Guo Binghui

Acta Geophysica

|

2021

|

Vol. 69, no. 2

405--414

EN

Postseismic global positioning system (GPS) time series are of fundamental importance for investigating the physical mechanisms of postseismic deformations, as well as the construction and maintenance of terrestrial reference frames. Particularly, methods for constructing accurate ftting models for such time series are critical. Based on the physical features of postseismic deformation models, we propose a new algorithm that combines the strengths of the Levenberg–Marquardt (LM) and dif ferential evolution (DE) algorithms, that is, the LM+DE algorithm. In this algorithm, the parameters are initialised by the constrained DE algorithm; the fnal parameters of the postseismic model are then solved by the LM algorithm. To validate the proposed method, DE, LM, and LM+DE were compared using synthetic and observational data from the 2011 Tohoku Earthquake. For all tests based on synthetic data, the LM+DE algorithm consistently converged to the global solution and the residual is small, regardless of how the independent parameter was varied. In the 2011 Tohoku earthquake, the parameters calculated by the LM+DE algorithm matched consistently for the global solution with a 100% passing rate after constraints were provided for the ratios of the initial relaxation time parameters. In contrast, the LM and DE algorithms individually achieved passing rates of only 22% and 1%, respectively. These results demonstrate that the proposed LM+DE algorithm efectively solves the initial estimate problem in the ftting of nonlinear postseismic models, and also ensures that the fts are mathematically optimal and consistent with physical reality.

2

Artificial Neural Network Approach to Mobile Location Estimation in GSM Network

Ezema L. S., Ani C. I.

International Journal of Electronics and Telecommunications

|

2017

|

Vol. 63, No. 1

39--44

EN

The increase in utilisation of mobile location-based services for commercial, safety and security purposes among others are the key drivers for improving location estimation accuracy to better serve those purposes. This paper proposes the application of Levenberg Marquardt training algorithm on new robust multilayered perceptron neural network architecture for mobile positioning fitting for the urban area in the considered GSM network using received signal strength (RSS). The key performance metrics such as accuracy, cost, reliability and coverage are the major points considered in this paper. The technique was evaluated using real data from field measurement and the results obtained proved the proposed model provides a practical positioning that meet Federal Communication Commission (FCC) accuracy requirement.

3

Aircraft dynamic model identification on the basis of flight data recorder registers

Lasek L., Lichota P.

Challenges of Modern Technology

|

2013

|

Vol. 4, no. 1

16--20

EN

We investigate the problem of an aircraft dynamic model parametric identification using dimensional derivatives as an example. Identification is done in offline mode, in the time domain. Flight parameters used for identification are obtained from Flight Data Recorder, that register them during each scheduled flight. We investigate the possibility of application of Maximum Likelihood Estimation that belongs to the Output Error Methods class. The likelihood function is defined for n-dimensional multivariate normal distribution. Unknown covariance matrix is estimated with the use of measured data and output equation. Output equation is calculated with Runge–Kutta fourth order method. In order to find the cost function minimum we consider using Levenberg-Marquardt Algorithm, where derivatives are calculated with central difference formulas and small perturbations theory. Mathematical model of an aircraft is obtained through flight dynamics classical approach. Rigid body model of an aircraft is assumed. Coordinate Systems Transformations are done using Euler’s Rotation Theorem with angle order typical for flight dynamics. Equations of motion are obtained from Newtons Second Law of Motion in body fixed coordinate system Oxyz, that is located at aircraft’s center of gravity. Turbulence is modeled as a bias, and also is an object of identification. We implement this method in Matlab R2009b environment.

4

Practical inverse analysis in metal forming

Pschera R., Buchmayr B.

Computer Methods in Materials Science

|

2011

|

Vol. 11, No. 2

436-441

EN

This article is thought as a contribution to the practical use of inverse analysis in the field of metal forming. The first part of the paper gives a short overview of the field of inverse analysis and the different algorithms to solve the mathematical problems; in the second part the application to different problems in metal forming is discussed. These include the validation of Finite Element Models, i.e. cross rolling, the determination of damage parameters and the determination of heat transfer coefficients for heat treatment simulations. It is shown that inverse analysis is a comfortable tool when lots of analysis data need to be handled efficiently.

PL

Ideą artykułu jest omówienie praktycznych aspektów zastosowania rozwiązania odwrotnego w plastycznej przeróbce metali. W pierwszej części pracy przedstawiono krótkie omówienie problematyki rozwiązań odwrotnych i przedstawiono różne algorytmy dla matematycznego rozwiązania tych problemów. W drugiej części pracy przedstawiono aplikacje do różnych zadań w plastycznej przeróbce metali. Zaprezentowane przykłady to walidacja modelu MES dla walcowania poprzeczno-klinowego, przewidywania parametrów zniszczenia materiału oraz współczynnika wymiany ciepła w symulacjach procesów obróbki cieplnej. Wykazano, że analiza odwrotna jest użytecznym narzędziem kiedy duża liczba danych musi być przetwarzana w sposób efektywny.

5

Optimisation of neural network controller architecture in DC motor model

Józefowicz K., Kołota J., Stepień S.

Journal of Applied Computer Science

|

2008

|

Vol. 16, nr 2

81-88

EN

The past few years have witnessed a dynamic growth variety of neural network applications. Range of these applications is very wide especially in industrial process control. As in nature, the neural network is determined by the connections between the elements, and we can train its to perform the particular function by adjusting special values (weights between elements) [1,4,7]. This paper presents DC motor model controlled by neural network Proportional-Derivate controller in comparison with classic PD controller. The research concern different network architectures and training functions. The models are presented and the experimental results signals are shown using graphical charts.

6

Adaptation of the Regularization Parameters in the Nm-Delta Networks

Gołąbek P., Kosiński W.

International Journal of Applied Mathematics and Computer Science

|

2000

|

Vol. 10, no 4

779-790

EN

The paper describes an application of regularization techniques to an automatic choice of parameters driving the learning process in the NM-Delta neural network architecture. The heterogeneous learning algorithm is identified as very similar to the Levenberg-Marquardt method but with a considerably smaller computational cost and different justification of parameter selection. The performance of the modified algorithm proves to be comparable with that of the Levenberg-Marquardt.