Uncertainty quantification of a deep learning model for failure rate prediction of water distribution networks. (August 2023)

• Record Type: Journal Article
• Authors: Fan, Xudong; Zhang, Xijin; Yu, Xiong Bill
• Is Part Of: Reliability engineering & system safety. Volume 236(2023)
• Publication Date: 2023-08
• Subjects: Water distribution network -- Failure rate -- Uncertainty quantifications -- Deep learning -- Long short term memory (LSTM); Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2023.109088
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: Developed a framework to quantify the uncertainties of deep learning-based failure rate prediction of water distribution networks. Introduced a probability-based LSTM model that allows to simultaneously determine the mean value and standard deviations of model predictions. Validated the framework with data from a large in-service water supply network. Identified the influence of different sources of uncertainties on the reliability of ML model prediction. Recommended methods to reduce the uncertainties associated with the LSTM model prediction. Abstract: Predicting the time-dependent pipe failure rate of the water distribution networks (WDNs) is important for planning its renewal budget but also challenging due to the complex factors involved. The recent development of machine learning techniques provides a novel approach for accurate failure rate prediction based on historical data. However, the inherited randomness and uncertainty of water pipe failures and machine learning algorithms are often ignored in the training and prediction process. This article develops an uncertainty quantification framework for deep learning-based WDN system-wide failure rate prediction. The framework integrates a probabilistic long short-term memory (LSTM) model with a Monte Carlo method. The historical climate data and WDN pipe maintenance data over the past 35 years for Cuyahoga County, USA, are used to illustrate this deep-learning model-based uncertainty quantification framework in this study. A statistical time series regression model, ARIMAX, is used as a comparison benchmark. The results show that the LSTM model outperforms the ARIMAX model in prediction accuracy in most years by considering the uncertainties. Besides, the uncertainty range of the LSTM model prediction is 50% of that of the ARIMAX model. The results also identified the major contributing factors to the uncertainties of LSTM machine learning model prediction. The proposed uncertainty framework features excellent extensibility and can be adapted to quantify uncertainties with other types of machine learning models.