Software quality measures to determine the diagnosability of PLC applications

In case of failures in a PLC controlled process the problem of diagnosis arises. Today this is often a manual task. Whenever the failure in the controlled system is not directly related to an actuator, its sources have to be tracked back through the PLC program to the corresponding sensors. This process can be very time consuming. The presented work aims at providing a means for determining this effort and to assist the diagnosis process. To do so, software quality measures are studied. The evaluation of these measures on a program shows the most complex parts and hence gives hints on possible improvements. During the calculation of the measures graphical representations of some dependency relations are derived. These can be used as an aid for the engineer in the actual diagnosis process. Two case studies illustrate the presented approach.