摘要
根据时间序列预测的特点和要求,分析了传统时间序列预测方法的不足,提出了将卡尔曼滤波应用于时间序列预测。推导了基于卡尔曼滤波的ARMA模型参数实时更新算法,并采用功率谱密度分析方法确定预测模型的形式与阶数。最后,通过对光纤陀螺随机漂移建模进行了实证研究。
Considering the characteristics and requirements of fault forecasting,the deficiencies of traditional forecast methods were pointed out.The method that applies Kalman Filter in time series forecasting was put forward.The method for estimating parameters of ARMA（autoregressive moving average） based on Kalman Filter was introduced.Components of the model were analyzed with PSD（power spectrum density）.The theory above was demonstrated through the modeling of random drift for FOG（fiber optic gyro）.
出处
《仪表技术》
2010年第7期37-39,共3页
Instrumentation Technology
关键词
时间序列
卡尔曼滤波
预测
time series
Kalman filtering
forecast
