TY - JOUR AU - Shmelova, Tatyana AU - Sikirda, Yuliya AU - Zemlyanskiy, Andriy AU - Danilenko, Olena AU - Lazorenko, Vitalii PY - 2016/11/11 Y2 - 2024/03/28 TI - ARTIFICIAL NEURAL NETWORK FOR AIR TRAFFIC CONTROLLER’S PRE-SIMULATOR TRAINING JF - Proceedings of National Aviation University JA - Proceedings of National Aviation University VL - 68 IS - 3 SE - AEROSPACE SYSTEMS FOR MONITORING AND CONTROL DO - 10.18372/2306-1472.68.10905 UR - https://jrnl.nau.edu.ua/index.php/visnik/article/view/10905 SP - 13-23 AB - <p><strong><em>Purpose:</em></strong><strong><em> </em></strong><em>to develop the </em><em>neural network model for evaluating correctness and timeliness of decision-making by specialist of air traffic services during the pre-simulator training. </em><strong><em>Methods: </em></strong><em>researchers are based on the basic concepts of threat and error management in air traffic control, for characteristic of situation complexity (threat- error-undesirable condition) the methods of expert estimation and fuzzy sets theory have been used. <strong>Results</strong>: stages of the conflict situation developing have been classified </em><em>and quantitative indicators of complexity level at each stage have been defined. Four layers neural network model for evaluating correctness and timeliness of</em><em> decision-making by air traffic controller during the pre-simulator training has been built and its parameters have been obtained. The first layer (input) is exercises that perform cadets/listeners to solve potentially conflict situation, the second layer (hidden) is physiological characteristics of learner, the third layer (hidden) is the complexity of the exercise depending on the number of potentially conflict situations, the fourth layer (output) is assessment of cadets/listeners during performance of exercise. Neural network model also has additional inputs (Bias) that including restrictions on calculating parameters. With the help of modelling complex Fusion visualisation of results of educational task implementation by air traffic controller according to specified parameters have been defined</em><em>. </em><strong><em>Discussion:</em></strong><strong><em> </em></strong><em>taking into account timeliness and correctness of instructor’s tasks performance during the pre-simulator education</em><em> </em><em>with the help of using artificial neural networks will allow determining the possibility of access of specialist of air traffic services to simulator training. Multimodal system Fusion will give the possibility to improve the process of training of cadet's/listener's – air traffic controllers through automated assessment of their actions.</em></p> ER -