Non-stationary time series could be divided into piecewise stationary stochastic signal. However, the number and locations of breakpoints, as well as the approximation function of the respective segment signal are unk...Non-stationary time series could be divided into piecewise stationary stochastic signal. However, the number and locations of breakpoints, as well as the approximation function of the respective segment signal are unknown. To solve this problem, a novel on-line structural breaks estimation algorithm based on piecewise autoregressive processes is proposed. In order to find the "best" combination of the number, lengths, and orders of the piecewise autoregressive (AR) processes, the Akaikes Information Criterion (AIC) and Yule-Walker equations are applied to estimate an AR model fit to the data. Numerical results demonstrate that the proposed estimation algorithm is suitable for different data series. Furthermore, the algorithm is used in a clinical study of electroencephalogram (EEG) with satisfactory results, and the ability to deal with real-time data is the most outstanding characteristic of on-line structural breaks estimation algorithm proposed.展开更多
1.5MW Ion Cyclotron Wave Heating system was developed, the transmitter and the antenna both have their ground loops, which will severely perturb the system’s normal operation. To avoid perturbation, a DC break was de...1.5MW Ion Cyclotron Wave Heating system was developed, the transmitter and the antenna both have their ground loops, which will severely perturb the system’s normal operation. To avoid perturbation, a DC break was designed. The S parameter and the VSWR (voltage standing wave ratio) of incident port were calculated; the thermal effect caused by conductor loss and dielectric loss was analyzed.展开更多
基金supported by Fund of National Science & Technology monumental projects under Grants No. 2012ZX03005012, 2011ZX03005-004-03, 2009ZX03003-007
文摘Non-stationary time series could be divided into piecewise stationary stochastic signal. However, the number and locations of breakpoints, as well as the approximation function of the respective segment signal are unknown. To solve this problem, a novel on-line structural breaks estimation algorithm based on piecewise autoregressive processes is proposed. In order to find the "best" combination of the number, lengths, and orders of the piecewise autoregressive (AR) processes, the Akaikes Information Criterion (AIC) and Yule-Walker equations are applied to estimate an AR model fit to the data. Numerical results demonstrate that the proposed estimation algorithm is suitable for different data series. Furthermore, the algorithm is used in a clinical study of electroencephalogram (EEG) with satisfactory results, and the ability to deal with real-time data is the most outstanding characteristic of on-line structural breaks estimation algorithm proposed.
基金The project supported by the National Science Foundation of China (No. 10274089) National High Technology Program of China
文摘1.5MW Ion Cyclotron Wave Heating system was developed, the transmitter and the antenna both have their ground loops, which will severely perturb the system’s normal operation. To avoid perturbation, a DC break was designed. The S parameter and the VSWR (voltage standing wave ratio) of incident port were calculated; the thermal effect caused by conductor loss and dielectric loss was analyzed.
