Identification of the pCOィイD22ィエD2 control system using a nonlinear autoregressive moving average model

点击翻译按钮获取中文摘要

Three mass-flow controllers operated by a personal computer made a gas mixture of NィイD22ィエD2, OィイD22ィエD2 and COィイD22ィエD2, which was given to subjects. Respiratory flow and pCOィイD22ィエD2 was recorded so that minute volume (input of the system) and end-tidal pCOィイD22ィエD2 (output of the system) could be obtained. These two variables were used for system identification. The previously proposed method for system identification was improved: the slope of sigmoid functions in the neural network was varied for each unit, the sigmoid functions were expanded around the threshold in place of zero and a model for noises was added to the neural network model.