Semantic Query Optimization (SQO) dates back to the eighties and basically relies on integrity constraints (mainly functional dependencies) to achieve logical query optimization. Applied to temporal databases with integrity constraints given as temporal functional dependencies (TFDs), SQO turns out to be very difficult in the general case due to the time dimension. Motivated by an application on Intelligent buildings, we have defined a simple class of temporal queries for which SQO techniques apply. Given a temporal database, a set of TFD and a logical temporal query belonging to the class, we propose a new rewriting technique allowing to produce a more efficient query, equivalent to the initial query. Interestingly, the time dimension opens new opportunities for SQO. We apply our proposition to a scenario of Intelligent building (IBs). We point out that temporal databases and TFDs are well-suited for modeling sensor data in IB. We show how IB inhabitants' preferences on their life condition can be naturally represented with TFDs.