Rafał Biedrzycki

Reliability and Efficiency of Differential Evolution Based Method of Determination of Jiles-Atherton Model Parameters for X30CR13 Corrosion Resisting Martensitic Steel

Paper presents new possibilities of Jiles-Atherton model of magnetic hysteresis parameters determination. The main problem connected with this model is the fact that its parameters have to be determined during the optimization process. However, due to the local minima on the target function, the gradient optimization methods are not effective, whereas evolutionary strategies, such as (μ+λ) strategy, are very time consuming. Results of calculation presented in the paper indicate that differential strategies create possibility of reliable and fast determination of Jiles-Atherton model parameters. Paper also presents guidelines for practical determination of model’s parameters, which is very important from practical point of view.

Accuracy and Bandwidth Optimization of the Over-Determined Offset-Short Reflectometer Calibration

We present a new method for calculating line lengths of offset-short standards so as to provide a broadband and accurate one-port vector-network-analyzer (VNA) calibration. Our method is based on an approximate uncertainty analysis of corrected VNA measurements. We estimate the maximum value of the total variance (i.e., the trace of the covariance matrix) of errors in those measurements at a single frequency, and then quantify the quality of the offset-short calibration in a given frequency range with two metrics: the upper bound for the total variance and the lower calibration frequency. We then apply a global bi-objective optimization to these metrics in order to determine the optimal lengths. In order to verify our approach we first validate assumptions made in the approximate uncertainty analysis through a Monte Carlo simulation. We further perform an experiment in which we compare measurements of verification devices after calibrating the VNA with different offset-short sets designed with our method and with a set of reference airlines. Finally, we perform an uncertainty analysis for these measurements, which demonstrates the tradeoff between the offset-short calibration bandwidth and accuracy, and shows that our approach for selecting the offset-short line lengths provides the broadest bandwidth for a given calibration accuracy.

Determination of Jiles-Atherton Model Parameters Using Differential Evolution

Effective and robust method of determination of Jiles-Atherton model’s parameters is one of the most significant problem connected with magnetic hysteresis loop modelling. Parameters of this model are determined during the optimisation process targeting experimental results of hysteresis loop measurements. However, due to appearance of local minima, the cognitive methods have to be applied. One of the most common method are evolutionary strategies. On the other hand, typical evolutionary strategies, such as μ + λ are expensive from the point of view of calculation time. To overcome this problem, differential evolution was applied. As a result, the calculation time for determination of Jiles-Atherton model’s parameters was significantly reduced.

KIS: An automated attribute induction method for classification of DNA sequences

Abstract This paper presents an application of methods from the machine learning domain to solving the task of DNA sequence recognition. We present an algorithm that learns to recognize groups of DNA sequences sharing common features such as sequence functionality. We demonstrate application of the algorithm to find splice sites, i.e., to properly detect donor and acceptor sequences. We compare the results with those of reference methods that have been designed and tuned to detect splice sites. We also show how to use the algorithm to find a human readable model of the IRE (Iron-Responsive Element) and to find IRE sequences. The method, although universal, yields results which are of quality comparable to those obtained by reference methods. In contrast to reference methods, this approach uses models that operate on sequence patterns, which facilitates interpretation of the results by humans.

A version of IPOP-CMA-ES algorithm with midpoint for CEC 2017 single objective bound constrained problems

This paper presents the RB-IPOP-CMA-ES algorithm which is an enhanced version of IPOP-CMA-ES. The algorithm uses midpoint of the population as an approximation of the optimum. The midpoint fitness is also used to introduce a new restart trigger for IPOP. Other IPOP restart triggers and parameters are also corrected. The performance of the proposed approach is evaluated on 30 problems from the CEC 2017 benchmark for 10, 30, 50 and 100 dimensions. The results confirm that RB-IPOP-CMA-ES achieves better results than its version that does not utilize midpoint and is a considerable improvement over a plain IPOP-CMA-ES.

Quasi-Stability of Real Coded Finite Populations

This contribution analyzes dynamics of mean and variance of real chromosomes in consecutive populations of an Evolutionary Algorithm with selection and mutation. Quasi-stable state is characterized with an area in which population mean and variance will remain roughly unchanged for many generations. Size of the area can be indirectly estimated from the infinite population analysis and is influenced by the population size, selection type and parameter, and the mutation variance. The paper gives formulas that define this influence and illustrates them with numerical examples.

Machine learning methods in data fusion systems

In heterogeneous, multisensor and multitarget data fusion systems the notion of “levels” is used in order to divide the complex problem of discovering relationships between objects into parts which are easier to understand. In presented paper we consider classifiers as general feature generators, these algorithms are able to connect data from different sensors and different observations. The classifier increases the level of data abstraction, which simplifies the architecture of following system components in data fusion chain. A data fusion engine named DAFNE uses the presented paradigm in its classifier module. The module was implemented in Python and C++, the Naïve Bayesian and decision tree classifiers were used. The tests on simulated data shows improvement of data quality via fusion. The system design allowed to attain real-time processing with limited data volume.