cankobanz / understanding-supercapacitor-charge-discharge-rates-using-machine-learning

The project aims to design machine learning algorithm which is able to predict energy densities of supercapacitors by using input data that is enriched from the results came from image processing techniques.

Understanding-Supercapacitor-Charge-Discharge-Rates-Using-Machine-Learning

ABSTRACT

The project aims to design machine learning algorithm which is able to predict energy densities of supercapacitors by using input data that is enriched from the results came from image processing techniques. The created machine learning algorithm will make possible to save time and money spent during material selection process of supercapacitors. Input data has extent of 22 features and 2189 observations. By using image processing, porosity and pore volume columns are improved since 84 of the 628 predicted data is filled with more the results come from more advanced technique. Enriched data is used to train 5 different machine learning models that are namely linear regression, ridge regression, lasso regression, regression tree and artificial neural network. The RMSE test results of the trained models are 5.2343, 5.2335, 5.1288, 3.7315 and 5.010, relatively. The results are satisfactory which implies that models predict energy densities accurately.