Research of the Reliability of the Electrical Supply System of Airports and Aerodromes Using Neural Networks

Research of the Reliability of the Electrical Supply System of Airports and Aerodromes Using Neural Networks

Serhii Mykolaiovych Boiko (Kremenchuk Flight College, Kharkiv National University of Internal Affairs, Ukraine), Yurii Shmelev (Kremenchuk Flight College, Kharkiv National University of Internal Affairs, Ukraine), Viktoriia Chorna (Kremenchuk Flight College, Kharkiv National University of Internal Affairs, Ukraine) and Marina Nozhnova (Kremenchuk Flight College, Kharkiv National University of Internal Affairs, Ukraine)
DOI: 10.4018/978-1-7998-1415-3.ch012

Abstract

The system of supplying airports and airfields is subject to high requirements for the degree of reliability. This is due to the existence of a large number of factors that affect the work of airports and airfields. In this regard, the control systems for these complexes must, as soon as possible, adopt the most optimal criteria for the reliability and quality of the solution. This complicates the structure of the electricity supply complex quite a lot and necessitates the use of modern, reliable, and high-precision technologies for the management of these complexes. One of them is artificial intelligence, which allows you to make decisions in a non-standard situation, to give recommendations to the operator to perform actions based on analysis of diagnostic data.
Chapter Preview
Top

Analysis Of Mathematical Description Of Neural Networks

ANN concerning the type of activation functions that are part of the structure of the AN is divided (Shibzukhov, 2006):

  • 1)

    Homogeneous ANNs that consist of one type neurons with a single activation function.

  • 2)

    Heterogeneous ANNs that consist of neurons with different activation functions.

ANN, depending on the state in which neutrons are, divided into analogue and binary, and depending on the number of neurons that change their status at some time, divided into synchronous ANN when only one neuron changes its state; asynchronous, when its state changes several neutrons (a group of neurons)(Barsky, 2013). Classification of the ANN can be represented as Table 1.

Table 1.
Classification of the ANN
NoClassification TypeSpecies
1By the number of layersSingle layer
Multilayer
2By way of spreading the signalDirect Signal Distribution
Reverse signal propagation
3According to the architecture of NNFull networks
Multilayer network with serial communications
Weakly connected networks
4NN with feedbackHopfield Network
Hamming's Networks
5NN with feedbackLoop-to-loop networks
Layered and fully connected networks
6Types of activation functions in the neuronHomogeneous (homogeneous)
Heterogeneous (different)
7For the state of neurons (excited, inhibited)Analogue networks
Binary networks
8By the number of neurons that simultaneously change the stateSynchronous
Asynchronous
9NN Bidirectional associative memory (BAM)Feedback Networks
Modeling of plastic-stable perception
10NN Architecture of Adaptive-Resonance Theory (ART)
11NN architectures ART-1, ART-2, ART-3
12Congnitron, neoconnitronNN modeling of the visual system of the human brain

Key Terms in this Chapter

LP: Learning Pair.

OUT: Outgoing Neural Network Signals.

AdB: Adjacent Block.

RPS: Reserve Passage of the Signal.

A&A: Airports and Aerodromes.

AcB: Activation Block.

NN: Neural Network.

MB: Matching Block.

FSP: Forward Signal Passing.

Complete Chapter List

Search this Book:
Reset