skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Model-based Layer Estimation using a Hybrid Genetic/Gradient Search Optimization Algorithm

Conference ·
OSTI ID:919230

A particle swarm optimization (PSO) algorithm is combined with a gradient search method in a model-based approach for extracting interface positions in a one-dimensional multilayer structure from acoustic or radar reflections. The basic approach is to predict the reflection measurement using a simulation of one-dimensional wave propagation in a multi-layer, evaluate the error between prediction and measurement, and then update the simulation parameters to minimize the error. Gradient search methods alone fail due to the number of local minima in the error surface close to the desired global minimum. The PSO approach avoids this problem by randomly sampling the region of the error surface around the global minimum, but at the cost of a large number of evaluations of the simulator. The hybrid approach uses the PSO at the beginning to locate the general area around the global minimum then switches to the gradient search method to zero in on it. Examples of the algorithm applied to the detection of interior walls of a building from reflected ultra-wideband radar signals are shown. Other possible applications are optical inspection of coatings and ultrasonic measurement of multilayer structures.

Research Organization:
Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
Sponsoring Organization:
USDOE
DOE Contract Number:
W-7405-ENG-48
OSTI ID:
919230
Report Number(s):
UCRL-CONF-231266; TRN: US200822%%177
Resource Relation:
Conference: Presented at: Adaptive Sensor Array Processing Workshop, Boston, MA, United States, Jun 05 - Jun 06, 2007
Country of Publication:
United States
Language:
English