Contextual processing of ECT measurement information towards detection of process emergency states
Paper i proceeding, 2013
This papers shows attempt of using prior knowledge in order to learn about the industrial process system behaviour. The system being studied here is the pneumatic conveying of solids monitored with electrical capacitance tomography (ECT). The prior knowledge is the non-invasive measurement information derived on the basis of intentionally introduced structural change in the experimental process flow rig. The focus is on the examination of pseudo-emergency states of bulk flow. Real case pneumatic transport failure usually consists in pipeline blockage. Such blockage typically develops starting as material slug build up. In contrast, our experiments investigate artificially induced flow pre-blockage states monitored with ECT measurement system. Collected measurement data are contextually processed in order to detect the pipeline state in sense of flow regime and flow behaviour. ECT raw data analysis strives to uncover a relation between measured values and flow change tendency. This research is aimed at giving rise to conclusions about possibility of contextual data processing in order to properly adjust pneumatic conveying control and a possible role of ECT records in this aspect. Authors show here that for now it is possible to detect abnormal process state such as blockage signal and the presented methodology coupled with nowadays computational techniques are anticipated to lead to abnormal states forecasting such as blockage threat recognition.
raw measurement data
prior knowledge modelling