Evaluation of a UML-Based Versus an IEC 61131-3-Based Software Engineering Approach for Teaching PLC Programming

A field experiment investigated the evaluation, teaching, and application of two different approaches to automatic control in programmable logic controllers, in particular comparing the Unified Modeling Language (UML) to the classic procedural paradigm (IEC 61131-3). A total of 85 apprentices from a vocational school for production engineering with a specialization in mechatronics took part in the training and the experiment. This paper details the results of the training using both approaches, and the correlations found between the modeling and/or programming performance and cognitive abilities, interest, workload, expertise, and school grades. In general, the results show that students can be trained to carry out authentic programming tasks within one and a half days, even for beginners in programming. The data distinguish the two approaches. Function Block Diagram programming (IEC 61131-3) can be best predicted by the grade in mathematics, programming experience, and cognitive demand. For performance in UML class diagram and state chart $({\rm UML}/{\rm CD}+{\rm SC})$ modeling, the grade in mathematics plays an even more prominent role; this explains the greater variance in modeling performance in the UML group than in the 61131/Function Block Diagram group. With respect to other findings, the paper concludes that special problem-solving skills and skills for abstract thinking should be taught when teaching UML-based modeling approaches.