A posterior probabilistic criterion on the Fast look-ahead unscented Rao-Blackwellised Particle Filter

A new finding for further improvement on a computational cost reduction technique for the Fast look-ahead unscented Rao-Blackwellised Particle Filtering (Fast la-URBPF) algorithm is proposed in this paper. By means of a novel posterior probabilistic criterion, we show that the computational cost of the existing look-ahead unscented Rao-Blackwellised Particle Filtering (la-URBPF) algorithm can be significantly reduced without sacrificing its superb performance in terms of diagnosis error rates. Simulation results show how the proposed method, namely Post Fast la-URBPF, can equally achieve the performance of the la-URBPF algorithm in two different fault-related environments described by the sets of transition prior probabilities. The time usage consumption of Post Fast la-URBPF is shown to be equal or less than the previously proposed Fast la-URBPF algorithm and is also substantially lower than all existing unscented Kalman filtering-based algorithms at convergence of diagnosis error rates.