Towards a taxonomy of errors in PLC programming

Abstract

Based on previous studies on programming errors and their causes, the presented paper investigates errors that application engineers in the area of machine and plant automation make while creating either function block diagrams, plcML [an adaption of the unified modelling language (UML)] or modAT4rMS code (a newly developed modelling language that adapts and combines aspects of UML and SysML). A laboratory-based study with 52 mechatronics apprentices and electrical engineering technicians with knowhow in manufacturing system design but comparably undeveloped programming skills has been conducted, in which the subjects' errors and think-aloud statements during code creation were recorded. In a subsequent step, these data have been analysed by the cognitive causes of the coding errors applying the skill-rulesknowledge framework. As a result, a taxonomy of errors is presented. Results indicate that most of the errors in the subjects' code are due to insufficient understanding of the notation's syntax, problems with the rules of encapsulation, the creation of modules and finally with the creation of variants and aggregations, which are all located at the rulebased level. Errors at the skill-based level mainly occurred during behavioural modelling with modAT4rMS. It is argued that the provided insights can be used for improving education on programmable logic controller (PLC) languages and for the design of tools that support PLC programmers at detecting and fixing errors within their code.

Evolution of software in automated production systems: Challenges and research directions

“…Unfortunately, programming performance in the evolution task was lower than 36% even referring to an expert group showing the need for adequate support. Duschl et al (2014) attempt to reveal the reasons behind the errors especially in module creation in the above mentioned study conducting interviews after the experiments. The individual erroneous code was discussed and reasons for errors were categorized.…”

A verification-supported evolution approach to assist software application engineers in industrial factory automation

“…Formal specifications defining supposedly correct behavior can also contain faults, just as the corresponding software. Thus, a finding by [24] is of high importance. Here, an experimental study of PLC programming analyzed different kinds of faults when writing code using different types of UML/SysML languages or IEC 61131-3 Function Block Diagrams (FBD).…”

Generalized test tables: A powerful and intuitive specification language for reactive systems

“…Challenges are the state space explosion, the necessary knowledge in formal description languages for writing formal specifications, and a potentially high modelling effort. Duschl et al [16] analyse the error types which can appear in code written in other modeling languages such as UML and SysML or IEC 61131-3 Function Block Diagrams (FBD). They conducted a precise analysis of error causes and concluded the most common reason for faults to be an inadequate understanding of used notations.…”