A STUDY OF FAULT DIAGNOSIS AND RECOVERY TECHNIQUES FOR MANUFACTURING SYSTEMS Silvio Simani * Marcello Bonf` e * Cesare Fantuzzi ** * ENDIF, Universit` a di Ferrara, Via Saragat 1, 44100 Ferrara, Italy ** DISMI, Universit` a di Modena e Reggio Emilia, Via Allegri 15, 42100 Reggio Emilia, Italy Abstract: The paper describes a framework for the development of a diagnosis methodology for industrial manufacturing systems. The aim of the project is to support technicians that supervise the manufacturing plant to identify the causes of faults and failures on the machine and, in particular, to indicate a procedure for the recovery of its working condition. The requirements described in the paper have been defined giving particular care to the peculiarities of the application domain, in order to allow the design of a powerful, but easy to use, diagnostic system for industrial technicians. Keywords: 1. INTRODUCTION Manufacturing systems for, as an example, the packaging industry are complex machines, with several physical and functional interactions among different components or sub-processing parts. The mechatronic nature of these systems makes their design, testing, installation and maintenance a very challenging set of activites. For example, it is a recognized fact that software programs running on the control system of a packaging machine are dedicated in large part, even up to 70 %, to alarms monitoring and exceptions handling, while the remaining and, most of the times, simplest part is composed of basic sequencing and motion control tasks for electric actuators. However, in many cases the control software dedi- cated to machine diagnosis is not designed to sup- port the technicians that work daily on the man- ufacturing plant, but rather to prevent damages on the machine or on its products by promptly stopping all the electric motors and putting all the mechanical parts in a safe configuration. There- fore, when the machine suddenly blocks, human operators have to search for the real cause of the failure, on the basis of (sometimes vague) messages on the control system’s Human-Machine Interface (HMI), The issues related to inadeguate diagnostic and supervisory tools for manufacturing systems can be summarized in the following: • machines can operate very far from where its mechanical designers and control engineers reside, which makes difficult a direct and immediate intervention of qualified people to identify the causes of failures and to explain detailed recovery procedures; • technicians working daily with the machine may be poorly qualified; in this case, they may have difficulties in understanding cor- rectly the reasons behind a mechanical stop, so that they may not take appropriate ac-