Kalman filter is one of the most common ways to deal with dynamic data and has been widely used in project fields. However, the accuracy of Kalman filter for discrete dynamic system is poor when the observation matrix is ill-conditioned. Therefore, the method for overcoming the harmful effect caused by ill-conditioned observation matrix in discrete dynamic system is studied in this paper.The causes of the ill-conditioned observation matrix and its effect on Kalman filter are analyzed. Biased Kalman filter and its algorithm are proposed by combining the biased estimation and Kalman filter in the sense of mean square error (MSE). The methods of choosing biased parameter in the new algorithm is proposed. By separately exerting some disturbance on the observation matrix and observation vector, two simulations are carried out. The experimental results show that the traditional Kalman filter is inaccurate when the observation matrix is ill-conditioned, and the biased Kalman filter is more accurate than the traditional one in terms of MSE.