Optimization of a Semi-Active Shock Absorber Using a Genetic Algorithm

Abstract

We propose an optimization method for a semi-active shock absorber for use in aircraft landing gear using Carroll’s FORTRAN Genetic Algorithm (GA) Driver. This method is compared with Powell’s conjugate direction method, a nonlinear programming (NP) approach, which uses not gradients, but only function values. In these optimizations, we handle variations in the maximum vertical acceleration of an aircraft during landing caused by the variation of the aircraft mass due to variations in the number of passengers and the amounts of cargo and fuel. The maximum vertical acceleration of an aircraft is set as an objective function to be minimized. Design variables searched in the first step of this optimization are discrete orifice areas formed by the outer surface of a hollow metering pin and a hole in the semi-active shock absorber. The design variable searched in the second step is an orifice area which is controlled based on the mass variation. For the GA runs, the ratio of the total number of optimum and near-optimum solutions to the total number of runs was greater than that for the NP runs. In addition, for the total GA runs, the total number of function evaluations per total number of optimum and near-optimum solutions was greater than that for the total NP runs. The optimum semi-active shock absorber is compared to the optimum passive shock absorber with respect to the variation of the acceleration of the aircraft mass. The ratio of maximum acceleration in the semi-active shock absorber to that in the passive shock absorber is 0.79 when the mass ratio is 0.65 maximum mass and is 0.58 when the mass ratio is 0.31 maximum mass.

Optimum trade-off charts considering mass variation for the design of semi-active and passive shock absorbers for landing gear

“…8 and 9, we show five optimum results among the eight results (seven obtained for the n = 1 case and one obtained in a previous study) as examples. The effectiveness of the GA used in the present study was verified in the previous study (Maemori, 2004), in which the objective function, namely, the maximum acceleration of the aircraft mass obtained using the GA was almost identical to that obtained using Powell's conjugate direction method. Each of the eight solutions was determined to be optimum when the top of the curve of the total resisting force (solid line in Fig.…”

“…We obtained seven optimum semi-active shock absorbers and seven optimum passive shock absorbers. Furthermore, we added one optimum semi-active shock absorber and one optimum passive shock absorber obtained in the previous study for consideration herein (Maemori, et al, 2004). The results for these optimum shock absorbers are shown in Figs.…”

“…In the n = 1 case, we obtained seven optimum solutions ( * ψ = 0.846, 1.221, 1.435, 1.689, 2.000, 2.438, and 3.000), namely, seven optimum semi-active shock absorbers and seven passive shock absorbers. Furthermore, to the results obtained for the n = 1 case, we added one optimum solution ( * 1 ψ = ), obtained in a previous study (Maemori, et al, 2004), as a standard shock absorber for m 1 = 130t. This optimum solution ( * 1 ψ = ), which was obtained in a single-objective optimization, would be able to obtain in the multi-objective optimization by determining adequately weights w 1 and w 2 .…”

“…Using the piecewise linear form, we can express various shapes of the curve and cope with allowable accuracy regarding the relationship between orifice area a xp and displacement x p . If lower accuracy of orifice area a xp is allowed, then we can reduce the number I max of design variables b, so that obtaining optimum design variables becomes easier (Maemori, et al, 2004).…”

“…In Reference (Wang, et al, 1999), possible optimization strategies for semi-active landing gear at touchdown and during taxiing were discussed. In addition, problems concerning mass variation have been discussed (Kobayashi, et al, 2009, Kobayashi, et al, 2013, Maemori, et al, 2004and Maemori, et al, 2003.…”