TCC(graduação) - Universidade Federal de Santa Catarina. Centro Tecnológico. Engenharia de Controle e Automação.With the necessity of the world to reduce the CO2 emission on the atmosphere, all the countries are creating policies to achieve low emissions. In the EU, after 2020, all the cars produced with emissions more than 95g/km, the manufacturer will have to pay 95 for each gram above this limit, so this is an important issue for the company. In GM Hybrids department is being developed this project to help the vehicle to emit less CO2 in the atmosphere by charging the battery based on the efficiency of the generator, charging the battery when the efficiency is high to compensate it when the efficiency is low. But to have this efficiency, it is necessary to have the current and the temperature of the generator, which are not present on the CAN bus of the vehicle, so this two signals must be estimated without any extra sensors. And this project describes estimation of one of this needed signals for the efficiency, the generator temperature: sensor calibrations, measurement plans, executed tests, results taken from the measurements, conclusions about the major influences involved on the generator temperature, the model done in MATLAB/Simulink from the results taken and the perspectives for the continuation of the project. For the estimation of the temperature was defined 6 major influences on the generator inside temperature but in this project only 3 had enough time to be tested, analysed and added to the model, so the project could not have an accurate conclusion as desired. But even with more tests to be done and added in the model, could show some good results in some intervals of measurement and provide conclusions and discoveries about the signals and the car that was not known before. So this was a good step for the desired estimation of the temperature and consequently for the efficiency dependent way of charging.