Iryna Turchenko

Approach to neural-based identification of multisensor conversion characteristic

The neural network based method of individual conversion characteristic identification of multisensor using reduced number of its calibration/testing results is proposed in this paper. The proposed method is based on reconstruction of surface points of multisensor conversion characteristic by modular neural network. Each neural network module reconstructs separate point of the surface. The simulation results show high reconstruction accuracy of the first approximation phase of the method.

Improvement of identification accuracy of multisensor conversion characteristic using SVM

A method of individual conversion characteristic identification of multisensor using reduced number of its calibration/testing results is described in this paper. The proposed method is based on the neural-based reconstruction (approximation or prediction) of surface points of multisensor conversion characteristic. Each neural network module reconstructs separate point of the surface. Our results show that the use of a Support Vector Machine (SVM) model allows improving the reconstruction accuracy of multisensor conversion characteristic. The reconstruction results obtained by SVM are compared with the results obtained by a multi-layer perceptron (MLP).

Recognition of Multi-Sensor Output Signal Using Modular Neural Networks Approach

An approach of modular neural networks using for recognition of an output signal of multi-parameter sensor (MPS) is described in this paper. A simple mathematical model of MPS is presented. The structure schemes of single neural network and modular neural networks are designed. The experimental researches of both methods are fulfilled using MATLAB software, the comparison results are given in the end of the paper.

Simulation Modeling of Neural-Based Method of Multi-Sensor Output Signal Recognition

The possibility of artificial neural network usage for recognition of a signal of a multiparameter sensor (MPS), described by different mathematical models, is described in this paper. These mathematical models are developed for the cases, when MPS conversion characteristics have positive derivatives, negative derivatives and derivatives of different sign at similar and opposite increasing of MPS output signal. The model of neural network, used for recognition, as well as achieved results of simulation modeling of a multiparameter sensor signal recognition using MATLAB software are presented in the end of this paper

A Low-Cost Network Controller for Security Systems

This paper describes the design of a low-cost network controller that can be used to interface sensors to security systems. This controller can use either a two-wired network interface with reconfigurable structure or an open optical channel for data communication with the network server. This design reduces the cost of data communication channels between sensors and the controller.

Predetermined movement of mobile robot using neural networks

We describe the experimental results of neural networks application for mobile robot control on predetermined trajectory of the road. Considered is the formation process of training sets for neural network, their structure and simulating features. Researches have showed robust mobile robot movement on different pans of the road

Optimized design of the universal ternary gates for quantum/reversible computing

Multiple-valued logic is a promising choice for future computer technologies, which provides a set of advantages comparing to binary circuits. In this paper, we have developed a genetic algorithm-based synthesis of ternary reversible circuits using Muthukrishnan-Stroud gates. The method for chromosomes coding, as well as a reasonable choice of algorithm parameters, allowed obtaining circuits for ternary Toffoli and modified Fredkin gates, which are better than other published methods in terms of quantum cost, delay times and amount of input ancillary and output garbage qutrits.

Method of adaptive control structure learning based on model of test's complexity

Method of adaptive control structure learning, which provides significant cost savings in time without loss of learning effectiveness are proposed in this article.

Recognition of MPS Output Signal Described by Different Mathematical Models

The possibility of artificial neural network usage for recognition of a signal of a multi-parameter sensor (MPS), described by different mathematical models, is described in this paper. These mathematical models are developed for the cases, when MPS conversion characteristics have positive derivatives, negative derivatives and derivatives of different sign at similar and opposite increasing of MPS output signal. The model of neural network, used for recognition, as well as achieved results of simulation modeling of a multi-parameter sensor signal recognition using MATLAB software are presented in the end of this paper.

Tasks distribution for students testing based on fuzzy logic

The article suggests the implementation of the tasks distribution system for students testing considering grade point average and current progress. This system is based on fuzzy logic. Fuzzy system rules were constructed and the system was checked for correctness of functioning.