基于优化的萤火虫算法的INS/GNSS/视觉融合导航系统参数优化研究

Parameter Optimization Study for an Integrated INS/GNSS/Vision Navigation System Using an Enhanced Firefly Algorithm

摘要:

在惯性导航系统、全球导航卫星系统和视觉传感器的组合导航系统中,基准卡尔曼滤波器存在参数选择依赖经验、收敛性能差以及对噪声统计特性变化敏感等缺点,为解决这些问题,提出一种基于改进萤火虫算法的自优化方法。为攻克多传感器融合导航系统中参数优化的难题,在萤火虫算法中引入 Lévy飞行原理,利用其重尾分布特性,增强种群多样性,增强整体搜寻能力；通过制定自适应参数调整策略,实现算法探索与开发之间的动态平衡；实施精英存储机制,有效防止高质量解的退化；基于新息序列的统计特性构建适应度函数,联合在线优化过程噪声协方差矩阵Q和观测噪声协方差矩阵R,使整体搜索效率和算法收敛性能得到显著提升。仿真结果表明,优化后的算法在位置和姿态估计精度上优于基准卡尔曼滤波算法,收敛速度更快,系统稳定性更高,鲁棒性更强,为参数自适应优化多传感器融合导航系统提供了有效方法。

Abstract:

In order to solve the challenges of standard Kalman filter in INS/GNSS/ visual embedded navigation system, such as parameter adjustment based on experience, suboptimal convergence performance and sensitivity to noise statistical changes, an automatic parameter optimization method based on improved glowworm algorithm is proposed. In order to solve the parameter optimization problem in multi-sensor fusion navigation system, Lévy flight mechanism is introduced into glowworm algorithm, and the characteristics of long tail distribution are used to increase population diversity and improve the overall search ability. In addition, by designing an adaptive parameter adjustment strategy, the dynamic balance between the process of algorithm exploration and utilization can be realized, and a better individual retention mechanism can be introduced to effectively inhibit the degradation of the optimal solution.Based on this and utilizing the statistical properties of innovative sequences, fitness functions for the covariance matrix of the combined real-time optimization process noise Q and the covariance matrix of the observed noise R have been developed. This will significantly improve the efficiency of global navigation and the convergence performance of the algorithm. Simulation results show that the optimization algorithm not only outperforms traditional Kalman filters in terms of position and attitude estimation accuracy but also enhances convergence speed and further improves system stability and robustness, paving the way for automatic parameter adaptation in a multi-sensor fusion navigation system.