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

Rafa Aliyah, Veneza (2024) Monitoring The Quality of Water Production Process in Surabaya Using Max-MCUSUM Control Chart Based on Residual Deep Learning LSTM Model. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003201168_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (4MB) | Request a copy

Abstract

Perusahaan Daerah Air Minum (PDAM) Surya Sembada Surabaya is a government-owned company responsible for providing clean water services in Surabaya, East Java, Indonesia. The company ensures that residents and businesses have access to safe and affordable drinking water. This study analyzed four key water quality characteristics which are turbidity, pH, KMnO4, and chlorine residual. These characteristics are interrelated and exhibit autocorrelation, which can lead to false alarms and incorrect decision-making. To address autocorrelation in time series data, both conventional and machine learning methods can be utilized. This research employs Deep Learning Long Short-Term Memory (LSTM) to capture sequential patterns in the data, making it effective in handling autocorrelation. The optimal architecture and hyperparameters for the LSTM model were determined through repeated experiments aimed at minimizing Mean Squared Error, Root Mean Squared Error, and Mean Absolute Error. The residual data based on the LSTM model were independent and suitable for monitoring using control charts. Monitoring of PDAM's clean water characteristics was performed using the Maximum Multivariate Cumulative Sum (MCUSUM) method based on LSTM residuals. Phase I analysis revealed that the control chart was not statistically controlled due to a shift in the process mean, necessitating the handling of out-of-control observations. In the Phase II control chart, many observations were still out of control, indicating that the process remained statistically uncontrolled. Univariate process capability analysis showed that turbidity and chlorine residual did not meet the company's specifications with sufficient precision. Furthermore, three variables—turbidity, KMnO4, and chlorine residual—performed inadequately in terms of accuracy. Turbidity and pH are identified as the primary factors contributing to the statistical instability of PDAM's water quality. Multivariate measurements indicated that the overall process was not yet capable of meeting the company's specifications with sufficient accuracy.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Control Chart, LSTM, Max-MCUSUM, PDAM, Statistical Quality Control
Subjects:	H Social Sciences > HA Statistics Q Science > Q Science (General) Q Science > QA Mathematics
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Veneza Rafa Aliyah
Date Deposited:	09 Aug 2024 07:29
Last Modified:	09 Aug 2024 07:29
URI:	http://repository.its.ac.id/id/eprint/115116

Actions (login required)

View Item