Abstract:

To monitor the daily living of the old people living alone, a study of their activity amount based on time series analysis and fuzzy pattern recognition is proposed, in which the daily activity of the old people living alone is used as the effective index to monitor their living state. The predictive model for the activity amount time series is established by using random time series analysis. Using the fuzzy pattern recognition method, the difference between the amount of monitoring activity and the amount of forecast activity is identified. When the gap is more than the normal range, the warning information will be reflected to the guardian of the elderly. The result shows the accuracy rate of abnormal behavior recognition was 96. 97%. The study provides an alternative to activity amount monitoring of the old people.

Key words: time series analysis, fuzzy pattern recognition, autoregressive integrated moving average(ARIMA) model, membership degree