Research on Particle Filter Algorithm Based on Neural Network
- DOI
- 10.2991/lemcs-14.2014.242How to use a DOI?
- Keywords
- Particle filter; Particle degeneracy BP neural network Generalized regression neural network (GRNN)
- Abstract
Aiming at the weight degeneracy phenomena in particle filter algorithm, the improved particle filtering algorithm based on neural network was presented. BP neural network and generalized regression neural network (GRNN) are adapted to improve resampling algorithm and importance probability density function. This algorithm utilizes the nonlinear mapping function of BP neural network. First of all, to sample from the importance density function of particle weight division, the weighted particle is splitted into two small weight particles. Then, abandon the weight of very small particles, and adjust the particles with smaller weight by using the neural network. This algorithm optimizes the sample from importance density function by generalized regression neural network. Through GRNN, the samples are adjusted. The samples are more nearer to the posterior probability density. Simulation results show that the particle filter algorithm based on neural network can improve the diversity of particle samples, increase the effective particle number, reduce the mean square error, and the filtering performance is improved. It is proved that this particle filter algorithm based on neural network is available and effective.
- Copyright
- © 2014, the Authors. Published by Atlantis Press.
- Open Access
- This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).
Cite this article
TY - CONF AU - Ershen Wang AU - Xingkai Li AU - Tao Pang PY - 2014/05 DA - 2014/05 TI - Research on Particle Filter Algorithm Based on Neural Network BT - Proceedings of the International Conference on Logistics, Engineering, Management and Computer Science PB - Atlantis Press SP - 1084 EP - 1088 SN - 1951-6851 UR - https://doi.org/10.2991/lemcs-14.2014.242 DO - 10.2991/lemcs-14.2014.242 ID - Wang2014/05 ER -