In this thesis, two novel approaches aiming at increasing the effectiveness and efficiency during the model-based testing of programmable controllers in automation systems are presented: design-to-test (DTT) and plant features (PFs).
These two approaches deal with black-box conformance testing, where the specifications and implementations can be modeled as finite state machines (FSMs). Given an automation system, the testing objective is to validate whether the implemented controller conforms to expected input-output behavior with regard to their specification models. However, existing testing methods suffer from various issues and are therefore not well applicable for current industrial applications.
On the one hand, the DTT approach aims to improve the effectiveness of complete testing, which is indispensable for critical systems. The specification models are automatically checked and modified with limited design overhead in order to improve the testability of their physical implementation, namely its controllability, observability, and single-input-change testability. This approach also guarantees, by design, that the behavior of the implementation remains unchanged during its normal execution, i.e., when disconnected from a test bench.
On the other hand, the PF approach attempts to enhance the efficiency of testing for large scale systems where complete testing is hardly realistic. Plant features are manually modeled using simple templates (which also limits the modeling overhead), and then automatically fed into test generation. As a result, the input space of a system under test and the number of meaningful test cases can be significantly reduced, and consequently, the length of an executable test sequence can also be significantly shortened. It is worth mentioning that the obtained shortened test sequence guarantees full coverage of the whole nominal behavior of a system under test.
Based on case studies, these two approaches outperform the current methods and advance the model-based testing of programmable controllers.
«
In this thesis, two novel approaches aiming at increasing the effectiveness and efficiency during the model-based testing of programmable controllers in automation systems are presented: design-to-test (DTT) and plant features (PFs).
These two approaches deal with black-box conformance testing, where the specifications and implementations can be modeled as finite state machines (FSMs). Given an automation system, the testing objective is to validate whether the implemented controller conforms...
»