新的鲁棒滤波算法及在微机电系统-惯性导航系统／全球定位系统中应用

摘要

为了提高系统对不确定干扰的鲁棒性并保持较高的精度，提出一种新的鲁棒滤波算法．当系统满足可检测和可镇定的条件时，通过引入适当的滤波增益矩阵，得到指数稳定的滤波器，从而将不确定干扰对估计误差的影响限制在给定的范围内，实现鲁棒性和精度的要求．用微机电系统一惯性导航系统／全球定位系统组合导航系统中得到的试验数据对该算法进行离线验证，试验结果表明，新算法的计算量较小，并且在有不确定噪声干扰的情况下能使系统保持较高的精度．%A new robust filtering algorithm is proposed for enhancing the system robustness to uncertain disturbances and keeping a high accuracy. When the system is detectable and stabilizable, by introducing suitable filter gain matrix, we develop an exponentially stable filter to restrain the impact from uncertain disturbances to estimation errors. The new algorithm is verified offline with experimental data from an micro-electro-mechanical system-based inertial navigation system/global positioning system integrated navigation system; results show that the system operates in a low calculation complexity, but keeps a high accuracy in the presence of uncertain disturbances.