Abstract
Industrial measuring systems nowadays are frequently based on eddy current sensors. These sensors are highly accurate, with high resolution, have good bandwidth and they are very robust against contamination in an industrial working environment. A major drawback of this technology is sensitivity of eddy current sensors to electromagnetic anisotropy of target material. This problem becomes critical in non stationary target applications, where measuring location is moving in the plane orthogonal to the sensor main axis. Compensation of induced error by lookup table is impractical due to non-stationarity of electromagnetic properties of target material. The other possibility is to smooth electromagnetic anisotropy by mechanical alteration of target surface. Unfortunately, this approach is very delicate and frequently leads to deterioration of initial situation. In this paper a new approach is presented. It is based on multiresolution signal decomposition using discrete wavelet transform, recognition of the component which is generated by electromagnetic anisotropy, and removing this component from eddy current sensor readings. This approach is dynamical in its essence and therefore it is capable of handling the non-stationary properties of electromagnetic anisotropy. The proposed method is experimentally verified. Achieved results show its applicability in real industrial conditions.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have