Modeling of nonstationary time-series data. Part II. Dynamic periodic trends

This paper presents deterministic models for simulating nonstationary time-series data with dynamic periodic trends. These models have been applied to monthly energy data recorded in the lunar calendar. Simulation results have shown good accuracy when compared with results obtained from established modeling techniques. The models developed in this paper are useful not only to electric utilities using the lunar calendar but also to those utilities faced with the modeling of data whose actual length of seasonality is not known precisely.