11:00
Data driven methods 1
Chair: Eric GARRIGUES
11:00
30 mins
|
IMU-based deformation reconstruction for highly flexible wings
Yuhui Zhang, Changchuan Xie, Yang Meng
Abstract: The wings of high aspect ratio are typically highly flexible, resulting in significant elastic deformations during flight. The aerodynamic shape of the aircraft undergoes substantial changes, reflecting pronounced geometric nonlinearity in deformation characteristics. Such extensive deformations render traditional aeroelastic analysis methods, based on the assumption of small deformations, inapplicable. Therefore, there is a need for nonlinear analysis methods to enable real-time monitoring of large wing deformations.
This paper proposes a wing deformation reconstruction method based on distributed inertial measurement units. By establishing the coordinate transformation relationship between the local coordinate system and the reference coordinate system at any arbitrary measurement point before and after the deformation of highly flexible wings, a correlation between wing deformation and the output information from distributed inertial measurement units is established. Position and orientation information is obtained through coordinate transformation and numerical integration of the output from inertial measurement units:
where and are acceleration and angular velocity from IMU, the subscript b and g represent the IMU at B and G, the superscript represents the skewed symmetric matrix , is the transformation matrix from B to G and is the position vector from B to G.
The overall configuration of the wing at any given moment can be interpolated by synthesizing the position and orientation information from various measurement points. Time-domain simulations and numerical calculations are conducted on a highly flexible wing model (Fig.(a)), validating the proposed method. The results of wingtip deformation (Z-direction) reconstruction are presented in Fig.(b).
Fig. (a) Highly flexible wing model and (b) wingtip z-direction deformation reconstruction
The findings indicate that the deformation recognition accuracy is satisfactory within a certain time frame. However, beyond a specific duration, the accumulation of errors, such as noncommutativity rotation errors and numerical integration errors, causes the algorithm results to gradually deviate from the actual deformation values.
|
11:30
30 mins
|
System Identification and Control Techniques Applied to Nonlinear Aeroservoelastic Models
Marcus V. G. Muniz, Adriano Argiolas, João L. F. Azevedo
Abstract: The present work employs linear system techniques to the identification of aerodynamic loads over an airfoil with a control surface subject to unsteady airflow. The response of the aeroelastic system to different inputs in pitch, plunge and control surface degrees of freedom are computed by solving the unsteady Euler equations. The flow solver is based on a cell centred, finite volume scheme. The time marching scheme is a second-order accurate, 5-stage, explicit scheme. The mesh displacement simulating the airfoil movement is imposed using Radial Basis Functions (RBFs) with compact support. The impact of different RBFs is analysed. Discrete steps are used to establish the amplitude for each degree of freedom in the study of the SIMO system identification methodology. The MIMO methodology employs Walsh functions, step like functions, to set the mesh displacements. Spectral density functions (PSD and Cross-PSD), used to compute the aerodynamic transfer functions, are obtained through the classical Welch algorithm. Window size and type follow the lessons learned from previous work in the research group. Open and closed loop flutter stability analyses are performed using an optimal linear quadratic regulator (LQR). The results obtained so far show that the aerodynamic lift and pitch moment coefficient transfer functions, using the SIMO methodology for pitch and plunge degrees of freedom, match the rigid mesh displacement approach. Control surface displacements transfer functions are being validated against potential theory for low Mach numbers. The differences of each approach are highlighted on open and closed loop stability boundaries presented using the root-locus method. Validation will also be performed with the PK-method results applied to a large span finite wing modelled in MSC NASTRAN. The results obtained herein will be subsequently used in a new framework being developed for aeroservoelastic analysis. The aerodynamic transfer functions will be combined with a multibody code (MSC Adams) which can be coupled with a nonlinear control law. This is a step further on the common practice which employs linear aerodynamics obtained from DLM tuned to higher fidelity data employing linearized control laws.
|
12:00
30 mins
|
Identification of Flexible Aircraft Parameters with Varying Flexibility Configurations: Integrating Elastic Effects and Sensor Data
Abraão Ferreira de Sousa Neto, Kaique Silveira Viana Costa, Flávio Luiz Cardoso-Ribeiro
Abstract: The necessity to mitigate pollutant emissions highlights the importance of research into flexible aircraft. Identifying models that accurately represent these aircraft is essential for the validation of early-stage design models and control design. This study focuses on performing a parametric system identification in the time domain for aircraft with varying levels of flexibility. The approach employs a simplified longitudinal stability and control model for short-period dynamics, rooted in the Quad-M methodology (Maneuver, Measurements, Model, and Method). The system identification technique used is the output error method, applied to a flexible model aircraft in three different flexibility configurations. Data were collected through nonlinear simulation of the flexible aircraft. Comparison of identification results across the different flexible configurations indicates an improvement in parametric values by incorporating elastic effects into the identification models. The study also explores the feasibility of various sensors to more closely simulate flight test procedures. Identifications are analyzed by comparing deflection measurements and accelerometers as observational variables, with acceleration measurements providing more accurate parameter estimations. Future work should extend the analysis presented to system identification using flight test data.
|
|