Correlated synthetic time series generation using fourier and ARMA

The ability to synthesize any number of independent but meaningful time series allows model testing to be approached statistically in a manner not possible using historical data. Whether the time series represent power consumption or demand, fuel or electricity prices, or operational power histories, the ability to test many possible scenarios statistically enables a more thorough analysis of simulation models. We have demonstrated a synthetic time series generation method that can preserve correlation between multiple data sets, in order to provide samples that are physically as well as statistically meaningful. Since the predicating methodologies of this work have been employed in industry-laboratory research collaborations [3][8], it is expected that each enhancement to the capability to produce synthetic histories will enable increasingly complex and accurate analyses. Providing a method for correlating multiple signals provides a means for analyses with multiple variable energy sources along with demand in a more realistic manner.