Monitoring the Quality of Water Production Process in Surabaya Using Max-Mcusum Control Chart Based on Residual Deep Learning LSTM Model

EasyChair Preprint 14541

Abstract

This study examines the water quality at the Surya Sembada water treatment plant in Surabaya, Indonesia, by analyzing turbidity, pH, permanganate index, and chlorine residual. Recognizing the inherent autocorrelation within these parameters, a Long Short-Term Memory (LSTM) neural network was implemented to model their temporal dependencies. Optimal LSTM hyperparameters were determined through rigorous experimentation using MSE, RMSE, and MAE as evaluation metrics. Residuals from the LSTM model was subsequently analyzed using a Maximum Multivariate Cumulative Sum (MCUSUM) control chart. Phase I analysis indicated a statistically nonconforming process, suggesting a significant process shift. Subsequent Phase II monitoring confirmed ongoing process instability. The application of LSTM modeling and Max-MCUSUM control charting in this study provides a robust framework for early detection of anomalies and process deviations in water treatment operations, facilitating timely corrective actions and improvements in water quality management

Keyphrases: LSTM, Max-MCUSUM, PDAM, Statistical Quality Control, control chart

@booklet{EasyChair:14541,
  author    = {Veneza Rafa Aliyah and Muhammad Ahsan},
  title     = {Monitoring the Quality of Water Production Process in Surabaya Using Max-Mcusum Control Chart Based on Residual Deep Learning LSTM Model},
  howpublished = {EasyChair Preprint 14541},
  year      = {EasyChair, 2024}}