Predicting Energy Budgets in Droplet Dynamics: A Recurrent Neural Network Approach