Model-based microburst identification using a hybridized extended Kalman filter with genetic algorithm

Abstract

Microburst (MB) wind shear is one of the most important meteorological dangers threatening the aircraft (AC) safety and the life of passengers. Though there are some ground-based 3D Lidar systems to detect low-level MB wind shears to alert the pilots, there have been fewer scientific attempts to identify model-based MB parameters via AC onboard air and position data. The latter refers to the development and identification of an acceptable MB model upon which an automatic flight control (AFC) system can be designed to control the AC through wind shear microburst. In essence, accurate knowledge of MB model is an essential prerequisite for design and analysis of AFC systems that can safely fly the AC against microbursts, especially in crucial phases of flight such as takeoff and landing. The present study focuses on online estimation of MB parameters whose results pave the way for effective MB autopilot designs for safe flights through MB. The proposed task is accomplished via a model-based approach using the AC six degrees of freedom (6 DoF) equations of motion (EOM) integrated with the latest verified model of the MB utilizing the extended Kalman filter (EKF). In addition to a sensitivity analysis to determine the key MB model parameters, the performance of the estimation process is enhanced via a hybridization of the genetic algorithm (GA) with the EKF. The results are promising and indicate that the proposed scheme can identify the MB model parameters with sufficient accuracy needed for online applications with AFC design.