CRJ 700 Longitudinal Aerodynamic Coefficients Identification using Support Vector Machine

Abstract

This paper presents a novel methodology developed by the Laboratory of Applied Research in Actives Controls, Avionics, and AeroServoElasticity (LARCASE) for predicting the longitudinal aerodynamic coefficients of an aircraft. The proposed technique utilizes Support Vector Regression (SVR) and has been successfully applied to the Bombardier CRJ 700 regional jet. Flight test data was collected during a longitudinal stick-fixed motion flight test procedure utilizing a level D CRJ-700 Virtual Research Simulator (VRESIM), which has been developed by CAE Inc. and Bombardier and is recognized by the Federal Aviation Administration (FAA) as the highest qualification for flight dynamics and propulsion models. The collected data was used to train and develop a multidimensional Support Vector Regression model capable of predicting longitudinal aerodynamic coefficients variation at any point on the flight envelope, defined by altitude, speed, weight, and center of gravity position. The choice of SVR hyperparameters was determined using the Bayesian Optimization (BO) method and is detailed in the paper. Validation of the resulting models was conducted by comparing predicted flight data with experimental data obtained from the Level D Bombardier CRJ 700 VRESIM, while considering the same pilot inputs. The results demonstrate that the SVR methodology is highly accurate, with average relative errors smaller than 1% for lift and drag aerodynamic coefficients, and an average error of 5.74% for pitching moment aerodynamic coefficients. Overall, the presented methodology provides a promising approach for predicting longitudinal aerodynamic coefficients in the field of aerospace research and could have significant implications for the design and optimization of aircraft.