Compressible Diagnosis of Membrane Fouling Based on Transfer Entropy

Abstract

Membrane fouling caused by many direct and indirect triggering factors has become an obstacle to the application of membrane bioreactors (MBRs). The nonlinear relationship between those factors is subject to complex causality or affiliation, which is difficult to clarify for the diagnosis of membrane fouling. To solve this problem, this paper proposes a compressible diagnosis model (CDM) based on transfer entropy to facilitate the fault diagnosis of the root cause for membrane fouling. The novelty of this model includes the following points: Firstly, a framework of a CDM between membrane fouling and causal variables is built based on a feature extraction algorithm and mechanism analysis. The framework can identify fault transfer scenarios following the changes in operating conditions. Secondly, the fault transfer topology of a CDM based on transfer entropy is constructed to describe the causal relationship between variables dynamically. Thirdly, an information compressible strategy is designed to simplify the fault transfer topology. This strategy can eliminate the repetitious affiliation relationship, which contributes to diagnosing the root causal variables speedily and accurately. Finally, the effectiveness of the proposed CDM is verified by the measured data from an actual MBR. The results of experiments demonstrate that the proposed CDM fulfills the diagnosis of membrane fouling.