Machine learning has demonstrated its effectiveness in fault recognition for mechanical systems. However, sufficient data for establishing accurate and reliable fault detection methods is not always available in real-world applications. Transfer learning leverages the knowledge learned from a source domain in order to bypass limitations in data availability and facilitate effective analysis in a target domain. For mechanical fault recognition, existing transfer learning methods mainly focus on transferring knowledge between different operating conditions which require training samples corresponding to all desired fault conditions from the target domain in order to realize domain adaptation. However faulted data in real applications is usually unavailable and impractical to collect. In this paper, a transfer learning-based cross-machine bearing fault recognition method is investigated. This new method sees domain adaptation take place without faulted data being available in the target domain, and thus alleviates data availability limitations. The effectiveness of the method is demonstrated in a case study in which the bearing diagnostic method is transferred from an electric motor to a wind turbine.

In recent years, with globalization in industry, there is a concern about increase of defects caused by using products in unexpected ways. To solve this problem, a method to predict such defects in a product in advance in the design stage has been proposed, in which behavior of a product and physical phenomena caused by a way of usage are represented by Petri net. However, in order to achieve a more accurate prediction, it is necessary to consider topological information about components of a product. In this research, we aim to represent topological information about elements of an electric circuit, and to make it possible to detect failures in cooperation with the method of the previous research. By modeling each element of an electrical circuit by Petri net and connecting them each other, it is possible to build a model that represents topological information about the circuit, and it can be checked whether there is a closed circuit and which element the current flows through. In addition, by reflecting the information about the circuit to the conventional model, more accurate failure detection can be performed. The fundamental effectiveness of the method was shown by case studies using an example circuit.

In ISFA-2010, we proposed a fault diagnoser design based on finite state automaton and sequential function chart to diagnose programmable logic controller (PLC) based manufacturing systems. A deficiency for that paper is that it only laid down a theoretical framework but lacking a detailed implementation and experiment validation. This paper aims at remedying the deficiency with detailed system design to diagnose discrete event system (DES) faults. Systematic diagnosis approach including modeling the plant, mapping output states with sensor readings, and designing diagnoser, was illustrated in this paper. The proposed diagnoser was implemented using Visual Basic to diagnose typical faults a PLC controlled dual robot arm. Experiment validation illustrated that the proposed diagnoser can detect and isolate DES equipment faults with a high fault detection rate of 93%, and detection rate of 80% when including product tolerance fault. The proposed diagnoser can achieve maximum fault detection delay of 9 seconds on the equipment faults. For the future direction, we are planning to extend the proposed diagnoser design to detect probabilistic faults in PLC based automated systems.