Condition Based Monitoring of a machine refers to the checking of various parameters and signaturesof the machine and then predicting the machine’s current health status. The slight changes in the machineoperating condition are carefully analyzed to know if the machine is having any fault. For monitoring thechanges, we acquire acoustic data i.e. generated while machine is running. In order to collect acoustic data, anumber of sensors are placed near various positions of the machine surface. Acquiring data from large number ofsensor positions is not economically viable. It would always be preferable to have a monitoring system thatacquires data quickly and efficiently, without compromising on the robustness of the system. Therefore there is aneed to locate some special positions on the machine, termed as “sensitive positions”, which are expected to exhibitthe fault characteristics in a much better way than others. This paper presents a novel method for rankingsensitive positions for a machine based on statistical parameters analysis. While the final list of required numberof sensitive positions is generated, the cross-correlation amongst the positions is also taken into consideration toavoid redundancy. Furthermore, a standalone application for implementing the same has been developed onAndroid platform. The proposed scheme and application can be used for variety of other applications that work onsimilar principle of acquiring data from multiple sensors.
展开▼
