Multi-sensory Machine Diagnosis on Security Printing Machines with Two Layer Conflict Solving

In order to reduce time consuming and expensive flawed production in Security Printing Machines an inspection system for early recognition of consecutive errors is developed. It shall avoid printing errors by combining measuring data from several sensors with expert knowledge. The inspection quality is improved by acquiring several information sources, using different physical quantities, integrating expert knowledge and perception, extracting reasonable features, and generating intuitive results. The TLCS (Two Layer Conflict Solving) approach is based on the Evidence Theory and uses conflict solving to fuse data. The first layer applies the Conflict Modified Dempster-Shafer-Theory (CMDST). Every two sensors‘ data are combined and conflicts are solved between individuals. In the second layer the data is fused using the results from the CMDST and the sensors’ original observations by the Group Conflict Redistribution (GCR). We introduce an improvement of the TLCS approach with reference to highly complex machine conditioning applications. In this context, the sensors are grouped to attributes applying expert knowledge. The fusion of the fuzzyfied sensor’s observations that are elements of one particular attribute is accomplished by the TLCS. Subsequently, the attributes’ conditions are merged using an Ordered Weighted Averaging Operator. In security printing machines the wiping unit is the most sensible part. It is responsible for removing surplus ink around the engravings. Even small parameter manipulations cause errors during the production. Experienced machine operators recognize errors before they occur and stabilize the production by changing wiping unit parameters mainly. The fusion approach is evaluated in a wiping simulator. Current, impact sound, temperature and force are acquired and processed. Wear, parameter changes, and mechanical disturbances are detected by the introduced algorithm