Time series analysis has been extensively used in many fields, such as system identification, modeling and data predication, and played an important role in system design, planning and performance analysis. The focus of time series application study is how to improve the accuracy and computation speed of the parameter estimation. Many researchers have carried out system modeling study by applying time series analysis and have gained their research results. The traditional methods such as maximum likelihood estimation, moment estimate and least square estimate which exit the defect of low precision, poor convergence and parameter estimation white noises coupling, are mostly utilized in parameter estimation for model. Taking this as basis the data forecasting and anomaly detection are conducted, which is hard to ensure the system’s stability. Different from the traditional algorithm, this paper proposes a new weighted iterative stage parameter estimation algorithm which avoids the coupling with white noise estimation of ARMA model parameter and improves the accuracy of parameter estimation. In theory, this algorithm tends to provide a good convergence performance. The experimental results based on ARIMA model show that the algorithm can improve the accuracy of parameter estimation and provide a good convergence performance.