S10  Fluidodinâmica de Escoamento Multifásico    Titulo: HORIZONTAL TWO PHASE FLOW PATTERN IDENTIFICATION BY NEURAL NETWORKS   Resumo : A MULTIPHASE FLUID CAN FLOW ACCORDING TO SEVERAL FLOW REGIMES. THE PROBLEM ASSOCIATED WITH MULTIPHASE SYSTEMS ARE BASICALLY RELATED TO THE BEHAVIOR OF MACROSCOPIC PARAMETERS, SUCH AS PRESSURE DROP, THERMAL EXCHANGES AND SO ON, AND THEIR STRONG CORRELATION TO THE FLOW REGIME. FROM THE INDUSTRIAL APPLICATIONS POINT OF VIEW, THE SAFETY AND LONGEVITY OF EQUIPMENT AND SYSTEMS CAN ONLY BE ASSURED WHEN THEY WORK ACCORDING TO THE FLOW REGIMES FOR WHICH THEY WERE DESIGED TO. THIS IMPLIES IN THE NEED TO DIAGNOSE FLOW REGIMES IN REAL TIME. THE AUTOMATIC DIAGNOSIS OF FLOW REGIMES REPRESENTS AN OBJECTIVE OF EXTREME IMPORTANCE, MAINLY FOR APPLICATIONS ON NUCLEAR AND PETROCHEMICOL INDUSTRIES. IN THIS WORK, A NEURAL NETWORK IS USED IN ASSOCIATION TO A PROBE OF DIRECT VISUALIZATION FOR THE IDENTIFICATION OF A GAS-LIQUID FLOW HORIZONTAL REGIMES, DEVELOPED IN AN EXPERIMENTAL CIRCUIT. MORE SPECIFICALLY, THE SIGNALS PRODUCED BY THE PROBE ARE USED TO COMPOSE A QUALITATIVE IMAGE OF THE FLOW, WICH IS PROMPTLY SENT TO THE NETWORK FOR THE RECOGNITION OF THE REGIMES. RESULTS ARE PRESENTED FOR DIFFERENT TRANSITIONS AMONG THE FLOW REGIMES, WICH DEMONSTRATE THE EXTREMELY SATISFACTORY PERFORMANCE OF THE DIAGNOSIS SYSTEM. KEYWORDS: NEURAL NETWORK, MULTIPHASE FLOW, FLOW REGIMES, DIAGNOSIS     Autores : Crivelaro, Kelen O. Seleghim Jr., Paulo 0     Trabalho Completo :   COBEM99 | Comissão  Organizadora | Palestras | Sessões Técnicas | Autor | Simpósios e Sessões Especiais | Revisores |  Título  dos  Trabalhos | Local