Application of differential Evolution as an Optimisation Method for Geo-Acoustic Inversion
other
In an accompanying paper ([1]) the global optimization method Differential Evolution (DE) is described and its performance assessed and optimized. DE is shown to be more efficient than a genetic algorithm (GA). Here, DE is applied for geo-acoustic inversion. The receiving system is a vertical array. The sound source was at a fixed position. From the received acoustic signals, spanning 8 hours, 41 snapshots are used for inverting for the geo-acoustic parameters. The variability in the resulting parameter estimates stems from the imperfect optimization method, variability in the water column and noise in the data. Both DE and GA were used for the optimization, again demonstrating superior performance of DE. Also DE is used for simulations quantifying the three contributions to the parameter uncertainty. These three origins are shown to almost completely account for the parameter uncertainty.
TNO Identifier
186379
Publisher
TNO
Source title
Proceedings of the Seventh European Conference on Underwater Acoustics, ECUA 2004, 5-8 July, 2004, Delft, The Netherlands
Place of publication
The Hague
Files