【摘要】 为了得到鲁棒的自适应预测算法,对噪声的影响做了系统研究。在噪声能量较小时,推导出噪声对自适应预测算法影响的近似表达式,发现基函数的导数是影响预测精度的主要原因,根据该式选择鲁棒基函数可以改进算法的抗噪声性能。同时,分析了鲁棒的误差函数对算法精度的影响,并提出了进一步减小噪声影响的方法。数值仿真表明,改进的算法具有更好的抗噪声性能。更多还原

【Abstract】 To achieve robust prediction algorithm,this paper studied how Gaussian noise deteriorates the performance of a predict algorithm.If the noise is low,its influence can be derived using a first-order approximation,and the expression implies that the differentials of the base functions is the determinant to the robust property of an algorithm.The influence of robust error function was analyzed,and another way to decrease noise was put forward.The improved prediction method has been applied to two chaotic systems and shows a robust performance to Gaussian noise.更多还原