Methodology for Dynamic Data-Driven Online Flight Capability Estimation

Abstract

This paper presents a data-driven approach for the online updating of the flight envelope of an unmanned aerial vehicle subjected to structural degradation. The main contribution of the work is a general methodology that leverages both physics-based modeling and data to decompose tasks into two phases: expensive offline simulations to build an efficient characterization of the problem and rapid data-driven classification to support online decision making. In the approach, physics-based models at the wing and vehicle level run offline to generate libraries of information covering a range of damage scenarios. These libraries are queried online to estimate vehicle capability states. The state estimation and associated quantification of uncertainty are achieved by Bayesian classification using sensed strain data. The methodology is demonstrated on a conceptual unmanned aerial vehicle executing a pullup maneuver, in which the vehicle flight envelope is updated dynamically with onboard sensor information. During vehicle operation, the maximum maneuvering load factor is estimated using structural strain sensor measurements combined with physics-based information from precomputed damage scenarios that consider structural weakness. Compared to a baseline case that uses a static as-designed flight envelope, the self-aware vehicle achieves both an increase in probability of executing a successful maneuver and an increase in overall usage of the vehicle capability. = number of observable vector data samples N s = number of sensors n = load factor n max = maximum load factor before failure n truth max = truth reference maximum load factor n op = operational load factor decided upon using dynamic capability estimate n static op = operational load factor decided upon using static capability estimate from design n util = average utilization of maximum vehicle capability p· = probability p op = threshold probability for decisions using dynamic capability estimate R = number of damage library records S = support vector machine discriminant function ("score") S = observable vector space s = observable vector s = observable vector measurement V = airspeed w c = chordwise extent of damage on wing w s = spanwise extent of damage on wing X = vehicle state space x = vehicle state α j = support vector machine weight for jth training sample β 1 , β 2 = probabilistic support vector machine regression model parameters ϵ k = kth strain gage rosette output ε k = plane strain at location of kth strain gage rosette

Simulation-Based Classification; a Model-Order-Reduction Approach for Structural Health Monitoring

“…Instead of proceeding from a law-based model, the data-based approach ( [25,26]) is based on the collection of a dataset of offline training data from all the possible healthy and damaged states of interest. The dataset can be collected (i) by performing experiments on the structure itself or on similar structures (see, e.g., [26]), or (ii) by performing synthetic experiments based on a (possibly parametrized) mathematical model of the structure of interest (see, e.g., [36,35,40]). Given the dataset, machine learning algorithms are used to train a classifier that assigns measured data from the monitoring phase to the relevant diagnostic class label.…”

“…If we rely on a high-fidelity solver based on a Finite Element (FE) discretization the construction of the offline dataset leads to an unaffordable computational burden. For this reason, most of the early literature ( [32,45]) resort to surrogate models, while more recent works focus on adaptive sampling schemes ( [3,6,5,40]), in both cases to reduce the number of FE solves. Both these approaches face the problem of curse-of-dimensionality in the number of features Q and/or in the number of parameters P , and have been mostly applied to static data.…”

Dynamic data-driven model reduction: adapting reduced models from incomplete data

“…In [28], online parameter identification from measurements is considered for DDDAS with proper generalized decomposition. The work [1,33,34,37] considers model reduction for structural health monitoring in DDDAS.…”

Methodology for Path Planning with Dynamic Data-Driven Flight Capability Estimation

“…Full implementation details for this aerostructural model are given in Ref. 9. It is important to note that the extent of damage modeling of this tool is reduction in the moduli of damaged elements in the 2D cross section of VABS via the k loss parameter.…”

“…9. In the offline phase, we compute probabilistic damage libraries characterizing capability using high-fidelity physics-based models as well as determine allowable control actions for different vehicle state distributions.…”