A FRI-FCM Based Fuzzy Monitoring System For Industrial Processes

Process monitoring is an effective way for keeping the stability and safety of industrial production processes as well as improving the product quality and enhancing the economic benefits. Recognized as a kind of qualified improvement for traditional data/diagram process monitoring, fuzzy monitoring has attracted lots of attentions ever since. However, most existing fuzzy monitoring approaches strongly rely on fuzzy rules for decision-making which are only eligible for simple industrial processes.FRI-FCM is a comprehensive inference algorithm combing the fuzzy rule inference and fuzzy cognitive map. By taking full advantage of data, FRI-FCM can decrease the probability that rules might become invalid due to increasing data, being able to accommodate large-scale data treatments. In regard to complicated industrial processes, this thesis uses FRI-FCM to formulate a fuzzy monitoring approach able to employ simplified classifying production rules to represent decision-making knowledge. As a result, a fuzzy matching based inference approach is established to predicate abnormal states during the industrial process running.Moreover, this thesis develops a fuzzy monitoring system which consists of five components responsible for displaying the supervisory data, maintaining the fuzzy rules, predicating abnormal process states, recording the process operations and managing the system, respectively.The proposed inference approach and the system developing technology are applied to monitor the DMF recovery process. Through the data collecting, fuzzy rules building and FRI-FCM based fuzzy reasoning, the process abnormality could be successfully predicted, showing the benefits of the contribution.