智能优化的代价评估粒子滤波算法

针对噪声未知情况下粒子滤波算法进行状态估计时准确性低的问题,对智能优化的代价评估粒子滤波算法进行研究。结合代价评估粒子滤波算法特点设计了智能优化重采样策略,利用概率质量函数评价粒子的可信度,通过交叉变异操作引导粒子向风险较小的区域移动,改善基于风险和代价进行样本更新而导致的样本贫乏问题;通过对风险较大粒子的变异,扩展了粒子的后验分布区域。仿真结果表明,提出的智能优化代价评估粒子滤波算法具有良好的粒子优化性能,能在噪声统计特性未知情况下提高状态估计的精度。

Cost reference particle filter algorithm of intelligent optimization

Aiming at the problem of low accuracy of state estimation of the particle filter (PF) algorithm with unknown noise, the intelligent optimization cost reference particle filter (IOCRPF) algorithm is investigated. The intelligent optimization resampling strategy is designed based on the characteristics of the cost reference particle filter (CRPF) algorithm. The probability mass function is used to evaluate the credibility of particles, and the crossover and mutation operations are used to guide the particles to move to the less risky areas. Therefore, it can improve the impoverishment of samples resulted from samples update based on risk and cost. And the posterior distribution area is extended through the mutation of the riskier particles. The simulation results show that the IOCRPF algorithm has superior performance of particle optimization and improves the accuracy of state estimation under the condition of unknown noise statistical characteristics.