WSEAS Transactions on Business and Economics
Print ISSN: 1109-9526, E-ISSN: 2224-2899
Volume 14, 2017
Regression Detection in Software Controlling Industrial Systems
Authors: Sebastien Salva, William Durand
Abstract: In the industry, testing production systems, i.e., systems transforming raw materials into products, is usually performed manually and is a long and error-prone task. The automation of the testing process could be initiated with the use of models. But, for this kind of system that has a long life span, models, when they exist, are seldom up-to-date. In this paper, we do not presume the availability of any model and we propose a method to automatically test the CIM2 level of production systems i.e., the software controlling them, by combining two approaches: model inference and passive testing. Using a set of events collected from a production system, our approach combines the notions of expert system, formal models and machine learning to infer symbolic models while preventing over-generalisation (i.e., the models should not capture more behaviours than those possible in the real system). These models are then used as specifications to passively test the conformance of other production systems. We define conformance with two complementary implementation relations. The first relation is based upon the classical trace-preorder relation. The second one is a weaker relation, less strict on the parameter values found in the traces of the system under test. With the collaboration of the manufacturer Michelin, we evaluated our approach on a real production system and show that it can be used in practice to quickly generate models and to test new production systems.
Search Articles
Keywords: Software engineering, Maintenance, Model generation, Passive testing, Industrial systems
Pages: 330-353
WSEAS Transactions on Business and Economics, ISSN / E-ISSN: 1109-9526 / 2224-2899, Volume 14, 2017, Art. #35