NEURAL NETWORKS FOR AIR TRAFFIC CONTROL

POWERFUL NEW COMPUTATIONAL DEVICES AND METHODS BASED ON BIO-LOGICAL PRINCIPLES OF INFORMATION PROCESSING ARE CURRENTLY BEING APPLIED IN PROBLEMS IN PATTERN RECOGNITION COMBINATORIAL OPTIMIZATION, EVIDENTIAL KNOWLEDGE PROCESSING AND IN MANYOTHER PROBLEMS WHICH HAVE RESISTED SOLUTIONS USING CONVEN- TIONAL ALGORITHMIC APPROACHES. THE RESEARCH PROPOSED HEREININVOLVES EXPLORING THE FEASIBILITY OF APPLYING NEURAL NET- WORKS TO AIR TRAFFIC CONTROL PROBLEMS SUCH AIR TRAFFIC FLOW CONTROL OPTIMIZATION, AND CONFLICT PREDICTION AND RESOLUTIONNETROLOGIC AND ITS SUBCONTRACTORS, TRW AND BALL CORPORATION,WILL PERFORM A TRADE-OFF ANALYSIS WHICH IDENTIFIES THE CANDIDATE AIR TRAFFIC CONTROL PROBLEMS WHICH ARE MOST LIKELY TO SOLVABLE BY NEURAL NETWORK APPROACHES. LIMITED NEUROCOMPU- TER EXPERIMENTATION WILL BE PERFORMED IN PHASE I SO THAT THENETWORKS BEST SUITED VERY MORE EXTENSIVE DEMONSTRATRATIONS AND TESTS IN A PHASE II PROGRAM CAN BE CONFIGURED.