Artificial intelligence application for drinking water treatment plants using julia and deep learning techniques

Julia is a dynamic programming language that compiles on the fly. It was designed by MIT for mathematical calculations and high-level algorithms required by artificial intelligence studies. With this thesis, 21 parameters taken from the Drinking Water Treatment Plant, the analysis sequence number and the average temperature value of the day to which the data belong, 23 parameters belonging to 826 days and 6 SCADA parameters were analyzed. Data normalization was performed before analysis. With 11 regression models, including Multi-Output Regression, the dataset was evaluated by cross-validation results. Data augmentation technique was used to increase accuracy. In this way, the number of data was increased to 5598. K-fold and repeated k-fold methods were used for test and training sets. The results were shared through tables. With this thesis, it has contributed to the elimination of the damages that may occur in the short or long term as a result of human intervention in the parameters of a substance such as drinking water, which is of vital importance for the society.