A Fortune 500 OEM needed to find a way of accurately determining the temperature of a gas flow entering a component in a large engine system. A test system was able to record the temperature but the production systems lacked the same sensors. High temperatures in the production systems seemed likely to be causing premature failure, but more information was required.
Using measurements from the sensors available in both the test and production systems, develop a predictive statistical model of the gas temperatures based on the test system measurements. This predictive model can then be used to estimate the temperatures in the production systems.
An accurate prediction model of the temperature in the test system has been developed. Paths identified for expansion of model to include more data and to cover different classes of engine operating conditions including stall and partial shutdown. Application of these virtual sensors is to control recommendation and predictive maintenance.
