(369o) Wiener Dynamic Modeling Under Inputs with Continuous-Time Stochastic Process Noise

Many chemical process input variables have a continuous-time stochastic (CTS) behavior. The nature of these variables is a persistent time correlated variation that manifest as process noise as it deviates in time from the nominal level. This work introduces methodologies in process identification for inputs with this behavior with application for measured and unmeasured inputs with CTS process noise components.

Two parameter estimation techniques are proposed. The first one, Method 1, is a derivative-free approach that uses sample moments and analytical expressions for population moments to estimate the CTS model parameters. The second, Method 2, approximates derivatives using a finite difference equation, and requires much less data to achieve a desired level of accuracy.

Three studies are presented. The first one evaluates the statistical properties of the estimators of Method 1 in a Monte Carlo simulation study. The second one provides an example of using Method 1 for process identification. The third one provides an example that highlights the strengths of Method 2 over Method 1 and illustrates it use in process identification.