In this research, the process of monitoring of the electric variable on a 14 Ah prismatic LiFePO4 battery has been carried out. The variables monitored include electric current, voltage, energy and internal resistance to be analysed for its effect on the temperature variable on the battery. An analysis of the relationship between the increase of temperature and the efficiency of energy has also been done. This process succeeded in getting the electrothermal value or heat arising from the electric variable in the battery. Electro thermal in the battery cell obtained the highest value 19.5 KJ and in the module obtained a value of 25.04 KJ, while the rate of electrothermal addition varies from 2.5 J/s to 22.5 J/s in a single cell and 20 J/s to 180 J/s on the battery module. Monitoring has also been implemented in the process of releasing battery energy both cells and modules. Monitoring of variable voltage, current, battery capacity, time and temperature has been done thus found that T of the battery was 20 0C when emptied with a discharge rate of 2.1 C and the temperature change of at least 3 0C at 0.7 C. While at 1.4 C, the temperature rises around 12 0C. In the battery module, the temperature rises around 6 0C when the battery module emptied at a rate of discharge 0.7 C, 15 0C at 1.4 C and around 20 0C at 2.1C. Machine learning can be used to estimate the increase of the temperature in a battery based on changes in voltage and electric current. This is done in order to determine the maximum electric current that can be supplied to the battery thus the thermal conditions of the battery can be maintained. The accuracy of estimating temperature value by using SVR on a single battery cell was 91.2% with RMSE was 1.107 0C while for the modules obtained 82.37% with RMSE is 1.18 0C. The accuracy value used RF for single cells was 97.28% with RMSE was 0 .625 0C and 98% with RMSE was 0.3 0C for battery modules.
Keywords: Prismatic LiFePO4 Battery, Support Vector Machine, Electrothermal, Random Forest