A supervisory control system for ATIS/ATMS integration. Part 2: neuro-fuzzy calibration and analyses

Abstract

This paper addresses the problem of integrating advanced traveller information systems (ATIS) and advanced traffic management systems (ATMS). A fuzzy-logic structure that allows the integration of existing successful ATIS/ATMS control implementations is presented. Part 1 of this paper addressed the formulation of the three operational schemes: ATIS stand-alone system, ATMS stand-alone control system, and a supervisory dual non-cooperative ATIS/ATMS control. Approximate simulation-based optimisation algorithms are devised as representations of the existing control logic operating these schemes. In Part 2 of the paper, the supervisory dual non-cooperative scheme is represented by a fuzzy-logic system. The structures of the fuzzy logics for the stand-alone systems are presented. A neural nets algorithm is then utilised to develop the knowledge base of the supervisory fuzzy system and to calibrate its parameters. The neural nets’ algorithm utilises data replicates generated by the approximate optimisation algorithms (presented in Part 1). The details of the fuzzy system's training are briefly presented. The results of the fuzzy system's calibration and effectiveness are discussed.