Piecewise constant line fitting on noisy ramped signals by Particle Swarm Optimization

In this study, Particle Swarm Optimization(PSO) is proposed for change point (edge) detection on noisy ramped signals. By taking moving averages between detected edges, noise on ramped signals is filtered and desired piecewise constant signals are acquired. It is required to detect edges in the immediate vicinity of actual edges. Performance of PSO is measured by the difference between estimated and actual position of edges. It is not possible to satisfy such a condition by standard PSO. Hence, in this work, two modifications to standard PSO are proposed: “PSO with uniformly distributed position vectors” and “Cascading PSO”. Throughout this work, all implementations are done on real signals which indicate generated powers by plants.