Title :
A Particle-Swarm-Optimization-Based Decentralized Nonlinear Active Noise Control System
Author :
George, Nithin V. ; Panda, Ganapati
Author_Institution :
Sch. of Electr. Sci., Indian Inst. of Technol. Bhubaneswar, Bhubaneswar, India
Abstract :
This paper proposes a functional-link-artificial-neural-network-based (FLANN) multichannel nonlinear active noise control (ANC) system trained using a particle swarm optimization (PSO) algorithm suitable for nonlinear noise processes. The use of PSO algorithm in a multichannel ANC environment not only reduces the local minima problem but also removes the requirement of computationally expensive modeling of the secondary-path transfer functions. A decentralized version of a multichannel nonlinear ANC is also developed, which facilitates scaling up of an existing ANC setup without rederiving the learning rules. This is possible as the controller module of each channel is independent of others. Simulation study of the two new multichannel ANC systems demonstrates comparable mitigation performance. However, the decentralized one is preferred to as it possesses the added advantage of scalability.
Keywords :
active noise control; decentralised control; learning systems; neurocontrollers; nonlinear control systems; particle swarm optimisation; transfer functions; FLANN; PSO algorithm; controller module; decentralized nonlinear active noise control system; functional-link-artificial-neural-network; learning rules; local minima problem; multichannel ANC environment; multichannel nonlinear ANC system; multichannel nonlinear active noise control system; nonlinear noise processes; particle-swarm-optimization algorithm; secondary-path transfer functions; Active noise reduction; Adaptive filters; Artificial neural networks; Distributed control; Particle swarm optimization; Transfer functions; Adaptive filter; decentralized control; functional link artificial neural network (FLANN); nonlinear active noise control (ANC); particle swarm optimization (PSO);
Journal_Title :
Instrumentation and Measurement, IEEE Transactions on
DOI :
10.1109/TIM.2012.2205492