A methodology for fault isolation and identification in automated equipments

The paper addresses the problem of reducing the production lines downtimes through an automatic real-time fault isolation and identification system. The paper presents first a methodology, from system specification to implementation on industrial PLC's, for fault isolation based on a special timed analysis and discrete event observers, and in addition shows a real industrial application of the approach, where the whole models, methodology and on-line diagnoser have been implemented and tested on industrial PLC controlling an industrial equipment. The methodology lays on the explicit modeling of time-outs, the clear separation of nominal and non-nominal control behaviour, a suitable physical component model library, a dedicated Java interface for specification, an automatic code generation for PLC-open compliant platforms.