Parameters determination of Convergent Cross Mapping for causal analysis of chemical variables

Abstract

Abstract Causal analysis of chemical variables is important for safe and efficient operation of chemical processes. Convergent cross mapping (CCM) proposed recently is suitable for nonlinear systems and can calculate the time delay and causal relationship between variables accurately. However, CCM has not been directly applied for causal analysis of chemical variables, because it does not have an algorithm for calculating phase space reconstruction parameters. In this paper, the optimal parameter algorithms are selected to improve CCM, greatly expanding CCM for analyzing causal relationship of chemical variables. Finally, using Tennessee Eastman model, it’s proved that CCM with optimized parameters can basically get the causal relationship network more efficient than the original CCM and transfer entropy method.