The parameters can not strongly converge to the true values when using traditional least squares cost function with noisy input data. This problem can be solved by a novel cost function which contains error variables. 当采用最小方差型的误差成本函数进行输入含噪系统的参数学习时,参数不能收敛至真值,利用包含噪声方差的误差成本函数可解决此问题。