Prediction of purified water quality parameters for use in the pharmaceutical industry
Purified water, Pharmaceutical industry, Prediciton, Time series
Purified water plays a crucial role in the pharmaceutical industry, impacting the quality of processes and end products. Therefore, strict parameters must be followed to certify the quality of the purified water produced. Thus, monitoring the purified water production process is essential to ensure that quality parameters are maintained and complied with, and anticipating behavior and possible quality deviations can significantly help with this task. This study proposes a machine learning model based on LSTM networks to predict multiple steps for the conductivity of purified water in a treatment plant intended for industrial pharmaceutical use. For the development of the research, a pipeline was proposed that involved the stages of data collection and understanding, analysis and pre-processing, modeling and evaluation of the results. Real data from a pharmaceutical laboratory's water treatment plant, NUPLAM, was used. LSTM networks were chosen as the basis for prediction due to their efficiency in predictive tasks, thanks to their ability to memorize short- and long-term data. The results obtained showed that the choice of appropriate parameters and hyperparameters was essential for the efficiency of the proposed model. Furthermore, LSTM-based models are able to predict water conductivity, especially for short prediction horizons, while they lose accuracy as the horizon increases. This work contributes to the improvement of monitoring and control processes in pharmaceutical industries, including NUPLAM, as well as making an academic contribution to the study of time series prediction for water treatment systems.