Daniel Schütz

Development of PLC-Based Software for Increasing the Dependability of Production Automation Systems

This paper presents the elaboration of a concept to develop and implement real-time capable industrial automation software that increases the dependability of production automation systems by means of soft sensors. An application example with continuous behavior as it is a typical character treat of process automation is used to illustrate the initial requirements. Accordingly, the modeling concept is presented which supports application development and which is supplemented by an implementation approach for standard automation devices, e.g., programmable logic controllers. The paper further comprises an evaluation which adapts the concept for two use cases with discrete behavior (typical character treat of manufacturing automation) and validates the initially imposed requirements.

Real-time capable software agents on IEC 61131 systems - Developing a tool supported method

Abstract In this paper an approach that utilizes the paradigm of agent-orientation for the software development of automation systems, is presented. We discuss different methods for agent-oriented software development in general and motivate a novel approach that meets the special requirements for control software in production automation. We show how this novel approach can enhance the dependability of control software by integrating concepts for diagnosis and failure compensation in real-time. Furthermore, we present a modelling approach, that will be integrated into a market leading IEC 61131 software development tool and by that, allow application engineers to implement agent control software.

On Modelling the State-Space of Manufacturing Systems using UML

Abstract Motivated by the idea of implementing all degrees of freedom into the field-level automation software of a manufacturing system, this paper presents an approach to develop the space of action of manufacturing systems by means of modelling their state-space. The meta-model of the Unified Modelling Language (UML) is extended for an object-oriented description of machine components. This description contains the operations a machine component is able to execute as well as the preconditions and effects of the operation in relation to the sensors of a component. From the class descriptions of machine components, models of manufacturing plants can be composed. Since the preconditions and effects of a components operation reference the machine states of the manufacturing system, from these models the state-space of a manufacturing system can be derived.

Model-based testing of PLC software: test of plants' reliability by using fault injection on component level

Abstract In this paper, the current situation of how PLC software is tested in industry is analyzed and the challenges on new testing approaches are identified using real industry code and a survey conducted within industry. The different possible and most relevant faults that may occur and must be dealt with are identified and requirements for testing approaches concerning component failures are derived. Further on, an approach to generate tests for error handling routines, which test the reliability of plants by injecting the corresponding faults is presented. The test cases are generated from timing sequence diagrams in combination with failure mode and effects analysis. In order to inject the faults at relevant points during the execution of the control software, IEC 61131-3 code is analyzed for the derivation of the test cases.

Interface Behavior Modeling for Automatic Verification of Industrial Automation Systems' Functional Conformance

Abstract To facilitate engineering and evolution of automation systems, ensuring the correctness of the design models is an important topic. Industrial automation systems are composed of various heterogeneous elements designed by different disciplines such as mechanical, electrical/electronic and software engineering. In this contribution, an approach for modeling industrial automation systems is presented which is based on interface behavior modeling of design artifacts and which supports automatic verification of their functional conformance while considering information from various disciplines.

Highly reconfigurable production systems controlled by real-time agents

Flexible plant components can significantly increase the flexibility of manufacturing systems and enable concepts like mass-customized production. This paper presents an approach for production planning and execution for flexible manufacturing system components, based on software agents. The agents are implemented directly on a PLC, making them capable of real-time operation. Additionally, a service-interface contributes to the vertical integration of the approach into the higher level planning of a flexible production system. Using the presented software agents, flexible plant components can be fully automated and integrated in modern FMS, leading to a higher degree of flexibility and dependability of the overall system. The approach is evaluated on a flexible fixture device as a FMS component, capable of automatic reconfiguration.

Design, implementation and evaluation of a hybrid approach for software agents in automation

Multi-agent systems (MAS) are used to increase flexibility and reliability of automated production plants, by reducing their control structures rigidness through the use of autonomous, interacting software agents. This paper comprises a concept for MAS, which are implemented partly on the Manufacturing Execution Systems (MES)-layer and partly on the field device layer, trying to combine the advantages found in other works that used MAS on only one of these layers. The concept contains a software architecture, a communication structure between different runtime platforms (MES and field devices) and processes, including an adaptive path finding algorithm. An evaluation showed that the proposed MAS is comparable in production performance, while increasing fault tolerance compared to a static legacy implementation.

Software changes in factory automation: Towards automatic change based regression testing

Changes of software in factory automation are frequent and resource-consuming in quality assurance when proving compliance with functional specifications. Automated testing can help minimizing required resources for software engineering. However, changes in the software cause a need for re-evaluating functionality through tests. To reduce resource consumption, existing relevant tests can be re-executed after ensuring their compatibility with the software after the changes. In this paper, an approach for enabling the automatic identification and classification of changes is presented, which can be used to efficiently select existing and help adapting incompatible test cases, leading to an overall increase in test efficiency. The approach is based on a detailed Programmable Logic Controller (PLC) program library and code analysis (both in IEC 61131-3) for several types of machines of industrial companies from the domain of factory automation.

Evaluation of a newly developed model-driven PLC programming approach for machine and plant automation

In machine and plant automation model-driven programming of Programmable Logic Controller (PLC) software is becoming an alternative to the state-of-the-art programming. As part of the related work the authors first present the results of a previously finished project on model-driven approaches for Multi Agent System (MAS) PLC programming, in particular using the Systems Modeling Language (SysML). A tool supported procedure model was developed that assists automation engineers implementing MAS in order to enhance reliability of the plants as well as reusability of implemented modules. Subsequently, related work on the evaluation of PLC programming approaches is discussed and the qualitative evaluation study for our newly developed procedure model is presented. In the last section, we present the study results and discuss our findings and their meaning for future developments and experiments.

MDE of manufacturing automation software — Integrating SysML and standard development tools

In the paper, a SysML-based approach that supports the model-driven engineering (MDE) of manufacturing automation software is presented. The approach adapts the Systems Modeling Language (SysML) and defines a specialized language profile, the SysML-AT, to support automated code generation for IEC 61131-3 runtime environments. For adapted SysML Parametric Diagrams (PD), a prototypical tool support was realized inside a commonly used automation software development environment. Inside the prototypical tool, the linking between the realized PD model editor and online data from the IEC 61131-3 runtime environment enables direct debugging within a SysML-AT model. The concept was evaluated by case studies and by usability experiments, which proved the usability of the SysML-based MDE approach for future users.