Detecting Time-Delayed Causations in Coal Mill based on Convergent Cross Mapping *

Coal mill has many factors influencing the grinding quality, which will eventually affect the combustion performance of the furnace. Therefore, it is necessary to identify the interactions among these process variables, which can be used for many purposes like identifying the disturbances propagation. This paper presents the application using the extended convergent cross mapping technique to exploit time-delayed causal relationships between any two process variables. Hence, all the pairwise results can be collected and summarized in a final causal map, including unidirectional or bidirectional, direct or indirect links in transport chains. The techniques of parameter initialization and indirect causations separated from direct ones are discussed and analyzed. And the application results show various causations with strong, moderate or weak effects. For example, the primary air flowrate and the differential mill pressure have great impacts on most of the rest variables, whereas the coal-air mixture temperature only responds to others, which is consistent with the physical characters as the input and output of the coal mill system.