No abstract

This paper deals with an original diagnosis system for LV switchboards. It is based on a local and global diagnosis approach with local temperature and current measurements. The thermal measurements including ambient and temperatures near electrical joint, are done by wireless thermal sensors. The system is composed of three complementary stages: An internet data collection stage, and two centralized data processing stages for the local detection of failures and the global diagnosis respectively. The analyses done with the diagnosis stage lead to predictive maintenance recommendations in order to avoid LV switchboard breakdown, which although rare could be catastrophic. Each stage of the centralized data processing is presented and discussed. Some results based on experimental data and expertise's information are presented to validate the feasibility of these methods.

This paper presents a fault detection and isolation (fdi) scheme with neural networks dedicated to the faults in the mechanical part of the drive. This method is designed to be suitable for a variable speed drive fed by an electronic device.After a presentation of the models of the drive components, faults characterization is reminded and the fdi scheme is introduced, trained and tested for different speeds.

No abstract

The paper deals with an entire system of monitoring and diagnosis of LV switchboards based on the measurements of currents, ambient temperatures and local temperatures of electrical joints. This system meets the needs to prevent the breakdowns of LV switchboards, which, although rare, can involve huge financial and human loss. The thermal measurements are done by wireless thermal sensor. The measured data are transmitted via internet and collected in a server, to be centrally processed. This centralized data processing includes a local detection of failures and a global diagnosis which leads to some maintenance recommendations. This paper will focus on, the local detection by comparison with an healthy model, and the global diagnosis using Bayesian network technique. The feasibility of these methods is tested with experimental data and expert's information. I.