Electricity demand management mechanisms are expected to play a key role in smart grid infrastructures to reduce buildings power demand at peak hours, by means of dynamic pricing strategies. Unfortunately these kinds of mechanisms require the users to manually set a lot of configuration parameters, thereby reducing the usability of these solutions. In this paper we propose a system, developed within the BEE Project, for predicting the usage of household appliances in order to automatically provide inputs to electricity management mechanism, exactly in the same way a user could do. In our architecture we use a wireless power meter sensor network to monitor home appliances consumption. Data provided by sensors are then processed every 24 hours to forecast which devices will be used on the next day, at what time and for how long. This information represents just the input parameters required by load demand management systems, hence avoiding complex manual settings by the user.