Breeding Estimated Particle Filter

As the normal particle filter has an expensive computation and degeneracy problem, a propagation-prediction particle filter is proposed. In this scheme, particles after transfer are propagated under the distribution of state noise, and then the produced filial particles are used to predict the corresponding parent particle referring to measurement, in which step the newest measure information is added into estimation. Therefore predicted particle would be closer to the true state, which improves the precision of particle filter. Experiment results have proved the efficiency of the algorithm and the great predominance in little particles case.

Study on Techniques for Optimization of Fault Diagnosis and Tests Based on PSO

This paper analyzes the optional flow chart for different kinds of tests based on hybrid diagnostic modeling technique. Furthermore, combining with varied diagnostic tests, we use the modified particle swarm algorithm (PSO) to improve the disadvantages of the traditional fault diagnosis and tests (FDT). According to the result of the filter experiment, the proposed method performs better and requires for lower ability of computation and takes less time for testing. The method can effectively optimize FDT.