进化粒子滤波器对比研究及其在移动机器人故障诊断中的应用

Comparative Study on Evolutionary Particle Filter and Its Application to Mobile-robot Fault Diagnosis

摘要: 通过分析进化粒子滤波器与粒群优化粒子滤波器在常态和时变噪声条件下的时耗和性能，得出进化与粒群优化机制对粒子滤波的影响．发现在状态阶变情况下，进化粒子滤波器显出良好强健性，而粒群优化粒子滤波器却失去了效果．最后，将进化粒子滤波器应用于移动机器人航迹推算系统的故障诊断．实验表明：粒群优化的粒子滤波器耗时大，在噪声时变和状态突变条件下进化粒子滤波器表现出优越的估计性能与鲁棒能力，且进化粒子滤波器能准确地诊断机器人航迹推算系统各种故障．

Abstract: The influence of evolution mechanism and particle swarm optimization on particle filter is obtained by analyzing time-consumption and performance in the normal and time-varying noise conditions for the evolutionary particle filter(EPF) and particle swarm optimization particle filter(PSOPF).The EPF shows strong robustness even when state jumping abruptly, while the PSOPF fails.At last,EPF is applied to fault diagnosis of mobile-robot dead reckoning system.The experiment shows that PSOPF costs much time,EPF showes superior estimation performance and robustness capabilities in time-varying noise and state jumping conditions,meanwhile EPF can diagnose faults availably for mobile-robot dead reckoning system.