Subset Sensor Selection Optimization: A Genetic Algorithm Approach with
Innovative Set Encoding Methods
Abstract
The study presents a new approach for solving the sensor subset
selection problem using set encoding and genetic algorithms, aiming to
minimize the number of sensors while maintaining accurate spatial
estimation. The proposed method, tested on groundwater data from the
Savannah River Site, introduces novel crossover and mutation methods,
outperforming a previous greedy method with an R2
higher than 0.98 for reduced sensor counts.