Blocky corrugations recognition based on graph theory for corrugated plate of plate heat exchanger

A graph theory-based method is proposed to identify two types of blocky corrugations on the corrugated plate of plate heat exchangers. A graph network is established basing on the region group constituted from a defect-free corrugated plate. Each blocky region in the group is taken as a node. And a set of unique lines between the nodes are taken as the arcs. A feature-data-template (FDT) based on a four-dimensional feature vector is defined to represent the location and the attribute of the nodes. Then, an active exploring-based graph traversal strategy is proposed based on the theory of web crawler model. It builds arcs actively from the root node to the lower-level nodes by the FDT established above. Experimental results show that the CRR of sample is 97.67% and the CRR of region is 99.57%. And the time-consuming is in millisecond level. It shows great potential for the practical application.