Coordinated measurements and estimation using quantized particle swarm optimization
MetadataShow full item record
The field of optimization is very vast and is growing by the hour, every day we use hundreds of optimization algorithms in various fields of sciences to solve and optimize problems with large variances in both degree of complexity and time requirements, and there is a constant need for better and faster optimization algorithms. This thesis presents one such optimization algorithm, named as Quantized Particle Swarm Optimization (QPSO) algorithm, which is an improvement of standard Particle Swarm Optimization (PSO) algorithm and retains most of its properties as PSO has been empirically shown to have a very good performance. In this new algorithm the particles can additionally communicate with each other with via a new variable called “Quantizer” which exploits the group effect to improve algorithm performance. Simulation results suggesting advantages of QPSO algorithm over PSO are presented for eight benchmark functions used for testing, Moreover the convergence analysis for deterministic version of QPSO is provided, and an application of this algorithm to solve threat surface parameter estimation problem is also presented. Initial work on a heuristic approach to localize and estimate the parameters of unknown threat sources by coordinated measurements taken by kinematically constrained sensing robots is described. This approach is tested with simulations for one source and two sources case and results are presented. Our goal was to most efficiently recover true parameters of the threat sources within the fastest time.