Parameter Identification of Hydro-Mechanical Processes Using Artificial Intelligence Systems

Abstract

The article is aimed at developing the comprehensive approach of parameter identification of complicated hydro-mechanical system by means of artificial intelligence systems (AIS) using artificial neural networks (ANN) and genetic algorithms (GA). The scientific novelty of the proposed method is inconsequent implementation of numerical analysis approach (e. g., FEM and CFD simulations), mathematical modeling of hydro-mechanical processes using the quasilinear regression procedure (QLRP), and artificial intelligence systems. Algorithms and related design schemes for implementation of the abovementioned technique are proposed for solving the interdisciplinary problems of aero-elastic interaction of gas-dispersed flow with the previously deformed flexible plates in a separation channel, static calculations of fixtures for parts manufacturing, as well as dynamic analysis of centrifugal compressors' rotors. All the presented has a significant advantage in comparison with the direct solving of the abovementioned problems. This advantage is in the ability of the comprehensive sequence "FEM / CFD-ANN / GA-QLRP" to predict the solutions of highly nonlinear mathematical models describing hydro-mechanical processes with uncertainties. As a result, unknown parameters of the mathematical models describing the complicated hydro-mechanical interactions are refined under an incompleteness of the initial data.