为实现实际应用中的非线性、非高斯系统中的状态估计,结合粒子滤波非线性估计的优势和自适应神经模糊推理系统(ANFIS)的非线性逼近功能,建立了ANFIS一粒子滤波模型。该模型首先通过ANFIS消除测量信号中有色噪声的影响,再运用粒子滤波实现对状态的最优估计,从而进一步提高估计精度。仿真结果表明ANFIS与PF的级联滤波较单一的粒子滤波均值减少了65%,方差减小了74.4%。ANFIS一粒子滤波对于强非线性系统的噪声消除效果显著,使状态估计精度得到了较大提高,证明了该级联滤波模型的有效性。
F develops th system(AN state signa demo or the practical application of nonlinear, e ANFIS- Particle filter cascaded filteri FIS) nonlinear approximation function estimation. ANFIS is used 1 is processed by the nstrate that with the respectively, systems, and the proposed partm casca ANFIS-particle the state estim model. non-Gaussian noise system s ng model based on the adap and particle filter's obvious tate estlmat paper tive neuro-fuzzy inference advantages for non-linear eliminate the bias in the colored noise of the signal, filter to realize the optimal state estimation. The s filter model the mean and variance filter model atlon accuracy has significant noise cancell has been greatly enhanced, then the filtered imulation results are reduced by 65% and 74% ation effect for strongly nonlinear which verifies the effectiveness of