Artificial neural networks as approximate procedure in wireless devices designing problems
Journal Title: Вісник НТУУ КПІ. Серія Радіотехніка, Радіоапаратобудування - Year 2012, Vol 0, Issue 51
Abstract
Purpose. Present work is dedicated to the optimal configuration selection and training method of neural network (NN). This NN is architecture's element of modified NN ensemble accepted by authors for implementation of frequency-selective microwave devices design algorithms. Optimal configuration determining of NN. Optimal configuration determining of NN was received by analyzing the results of test NN training with different number of layers and neurons in these layers. The main parameters optimal configuration determining of NN is the approximation quality and total learning time. Choosing of optimal teaching method. NN training methods comparison was carried out for 7 popular training methods: Levenberg-Marquardt backpropagation, BFGS quasi-Newton backpropagation, Bayesian regulation backpropagation, Conjugate gradient backpropagation with Powell-Beale restarts, Gradient descent backpropagation, Gradient descent with momentum backpropagation and Resilient backpropagation. Conclusions. NN using allows to approximate complex features of microwave devices, such as frequency dependencies of S – parameters etc. The approximation accuracy depends on configuration and method of NN training. Increasing the number of NN layers leads to improvement of approximate characteristics. According to our results the most effective is usage of 4 layers and the number of neurons in each layer should be over the range 10 to 20. Optimal training method for complex characteristics is Bayesian regulation backpropagation, for time training reduction can be used Levenberg-Marquardt backpropagation.
Authors and Affiliations
V. Adamenko, G. Mirskikh
Keywords
Related Articles
- EP ID EP309138
- DOI 10.20535/RADAP.2012.51.41-49
- Views 78
- Downloads 0
Two-step method of the measurement data processing of the high-sensitivity radiometric system
Introduction. The radiometry systems for registration low-intensity noise signals are widely used in science and technique. Automation of such apparatus allows to improve the precision of measurements and facilitate inte...
Signal processing domain-acoustic processor in the autocorrelation mode
It is analyzed the signal / noise ratio, which can be achieved by the signal processing domain-acoustic processor. It is shown that the quality of signal processing by this method ideally determined by the quality of the...
Microwave applicator for hyperthermia on the basis of spiral periodic structure
The use of periodic structures for creation of applicators with the homogeneous or near to him distributing of the electromagnetic field is offered. The given results of modeling of the offered structure and the numerica...
Increases of speed in systems of multistation access are with the code division of channels of СDMA on the base of height telecommunication platforms
The features of construction of systems of mobile radio communication on the basis of high-altitude platforms are given, the opportunities of increase of information speed are considered consecrate, it is shown perspecti...
Device of Radio Frequency Instability Measurements
Introduction. The instability of radio frequency is divided into two types: long-term, due to systematic frequency offset, and shortterm, which is determined by the fluctuation frequency changes. The aim of the paper is...