Mewma Control Chart for Monitoring Clean Water Quality by PDAM Production in Surabaya City Based on Residual of Generative Adversarial Network

EasyChair Preprint 14642

Abstract

This study investigates the water quality characteristics of pH, turbidity, and KMnO4 at the Ngagel II water treatment plant operated by Surya Sembada water treatment in Surabaya, Indonesia. Phase I and II analyses revealed that while the water quality parameters met established standards, the presence of autocorrelation compromised data reliability. To address this, a Generative Adversarial Network (GAN) model was developed and optimized to generate residual values capable of reducing autocorrelation. The performance of the GAN was evaluated using Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE) metrics. The residual series was subsequently monitored using a Moving Average Exponential Weighted Moving Average (MEWMA) control chart with a smoothing parameter λ of 0.4. Phase I analysis indicated a statistically controlled process after outlier removal. However, Phase II monitoring detected out-of-control signals, suggesting process instability. The findings demonstrate the potential of GAN-based residual analysis in mitigating autocorrelation in water quality data. Nevertheless, the complexity of GAN training and the computational resources required for optimal model development pose significant challenges.

Keyphrases: Forecasting, Generative Adversarial Network, MEWMA, PDAM, control chart

@booklet{EasyChair:14642,
  author    = {Raditya Widi Indarsanto and Muhammad Ahsan},
  title     = {Mewma Control Chart for Monitoring Clean Water Quality by PDAM Production in Surabaya City Based on Residual of Generative Adversarial Network},
  howpublished = {EasyChair Preprint 14642},
  year      = {EasyChair, 2024}}