Ambient temperature learning algorithm for automotive vehicles

Abstract

An ambient temperature learning algorithm provides a series of ambient temperature estimates which are calculated as a function of the induction air temperature as measured by an induction temperature sensor and this series of estimates is then filtered to provide an accurate reading of the actual ambient temperature. To increase the accuracy of the ambient temperature following an engine-off soak period, the initial estimate of the ambient can be made based upon several criteria. If the engine has not run for several hours, the induction system measurement from which the estimates are made is a good starting value. How long the engine has been off can be determined by a soak timer in the vehicle or by comparing the engine coolant temperature with the induction system temperature. If the coolant and induction system temperatures are very close, the soak period can be assumed long enough for the induction temperature to be a good starting point for the learned ambient temperature. If the soak has not been long, as indicated by either the soak timer or the temperature comparison methods, then the learned ambient from the last drive cycle is a better initial value for the ambient temperature. The algorithm eliminates the need for a separate temperature sensor for measuring the ambient temperature.