In-Flight Weight and Balance Identification Using Neural Networks

New hybrid adaptive control approach for aircraft with centre of gravity variation

“…Theorem 1: If the assumptions (i) and (ii) are satisfied, then the following adaptation rules forα i (i = 1, 2) and the adaptive control laws described by (12) and (17) can guarantee that the states achieve sliding surfaces S i = 0 (i = 1, 2), and the state errors e i (i = 1, 2) converge to zero.…”

“…h(Ma, α, γ , δ e ) represents the corresponding non-linear function from the above variables to r N . As the outstanding model approximators, NNs are selected to approximate the non-linear function (12). Based on the flight test data, standard feed-forward NNs with variables (Ma, α, γ , δ e ) as inputs and r N as output are trained.…”

Preliminary development of an onboard weight and balance estimator for commercial aircraft

“…W&B affects, in a substantial way, the static and dynamic flight characteristics of an aircraft making this information desirable for automatic flight control, fault detection (failure to deploy stores, presence of fuel leak, etc.) and identification systems [12]. Current methods to update W&B information in-flight are based on fuel flow measurements; so, the development of new inflight W&B systems can also provide an alternative/backup system to such methods.…”

“…Further developments in in-flight CG position estimation, using Kalman filters, accelerometers and the ADMIRE aircraft model, were made by Stanley [61,62]. Similarly, to Idan et al [12], the technique was developed and applied to a simulated aircraft, rather than a real one.…”

“…In steady symmetric flight, there is equilibrium between lift L, drag D, thrust T and weight W and the pitching moment M y on the airplane is zero. Per [12], transforming these forces and moments into standard dimensionless quantities (coefficients), where the product of dynamic pressure q  and wing area S are used as a unit of force and the wing mean aerodynamic chord (c) is used as a unit of length, we get the dimensionless total pitching moment C M and weight C W defined by:…”