Sensor Systems for PHM
“Garbage in, garbage out” is a saying familiar to many of us. In other words, when the quality of input data is poor, the low quality of outputs is unavoidable. Prognostics and health management (PHM) techniques can be developed through four sequential steps: (i) data collection, (ii) feature extraction, (iii) diagnostics and prognostics, and (iv) health management. If the first step of data collection is not designed properly, the final outcome of the PHM techniques will be unreliable, regardless of the soundness of the remaining three steps. As a result, the PHM techniques starting with poor data collection can be easily identified and abandoned by the user. Therefore, data sensing is the first and most critical step in the implementation of PHM techniques for electronic products.
Hyunseok Oh,Michael H. Azarian, Shunfeng Cheng, and Michael G. Pecht, “Sensor Systems for PHM”, IEEE 2018.