Finding the Direction of Disturbance Propagation in a Chemical Process Using Transfer Entropy

Abstract

Abstract-In continuous chemical processes, variations of process variables usually travel along propagation paths in the direction of the control path and process flow. This paper describes a data-driven method for identifying the direction of propagation of disturbances using historical process data. The novel concept is the application of transfer entropy, a method based on the conditional probability density functions that measures directionality of variation. It is sensitive to directionality even in the absence of an observable time delay. Its performance is studied in detail and default settings for the parameters in the algorithm are derived so that it can be applied in a large scale setting. Two industrial case studies demonstrate the method.

Linearized Transfer Entropy for Continuous Second Order Systems

“…It has been suggested that the main utility of the TE is to, given a sequence of observations, assess the direction of information flow in a coupled system. More specifically, one computes the difference T E i→j − T E j→i with a positive difference suggesting information flow from i to j (negative differences indicating the opposite) [2,4]. In the system modeled by Equation (10) one would heuristically understand the information as flowing from the drive signal to the response.…”

“…Moreover, TE is defined in such a way as to provide insight into the direction of the coupling (is component A driving component B or vice-versa?). Since its introduction, the TE has been applied to a diverse set of systems, including biological [1,3], chemical [4], economic [5], structural [6,7], and climate [8]. A number of papers in the Neurosciences also have focused on the TE as a useful way to draw inference about coupling [9][10][11].…”

A practical method for identifying the propagation path of plant-wide disturbances

“…Transfer entropy [11] adds information about time into a Bayesian network because it tests hypotheses concerning the joint and conditional probabilities of past and current values in a time series. Cause-and-effect analysis based on transfer entropy has been shown to work well in industrial case studies [12], but it does not give an explicit estimate of the time delay.…”

“…The effect of rearrangment is that whichever variable ends up in the first row is the candidate for the root cause because all other variables have time delays relative to that one. An algorithm for the rearrangement was presented in [12]. The causality matrix then has the following form:…”

Nearest neighbors method for detecting transient disturbances in process and electromechanical systems

“…Methods for persistent disturbances, however, are not appropriate for transient disturbances because they rely on the repetition of the abnormal dynamic episode [14,15,16,17].…”