This project deals with the analysis and development of an innovative sensing system that combines conventional remote sensors and signal processing techniques that extract the corrupted portion of signals. Typically, in diagnostic systems the sensor output is corrupted by the machine itself since the sensor cannot be generally placed at the site where the signal to be measured is generated. For example, in machine tools, some cutting force sensors depend upon the measurement of strain in the outer race of the machine tools, others depend upon the load measured at the base of the tool turret, and still others depend on the current consumed by the feed or spindle motor. A monitorability index (MI) is developed in this research which not only improves the sensor reliability and performance, but also provides a means of systematically optimizing the design of a particular sensing system. The design procedure yields a simple rational method to determine the most appropriate signals to measure and the type of transducers to use. The technique of using a model to correct the corrupted signal belongs to the class of techniques known as model-based estimators. This sensing system, which is based on MI and is called Model Based Sensing (MBS) is implemented for tool force and spindle bearing monitoring on existing machine tools and its performance is compared with traditional model-based estimation strategies.*** //
