Comparison of Machine Learning Techniques for Condition Assessment of Sewer Network

Abstract

Assessment of the structural condition of sewer pipes is one of the critical steps in asset management and support investment decisions; therefore, structural condition models with high accuracy are important that can help the utility managers and other authorities correctly assess the current condition of the sewage network and effectively initiate maintenance and rehabilitation strategies. The main objective of this research is to assess the potential application of advanced machine learning (ML) for predicting the structural condition of sewer pipes with a case study in Ålesund city, Norway. Nine physical factors (i.e., age, diameter, depth, slope, length, pipe type, material, pipe form, and connection type) and ten environmental factors (i.e., rainfall, geology, landslide area, building area, population, land cover, groundwater level, traffic volume, distance from the road, and soil type) were used to assess the sewer structural condition employing seventeen ML models. After processing the sewer inspections, 1,159 of 1,449 individual pipelines were used to train the structural condition model. The performance of ML models was validated using the 290 remaining inspected sewer pipes. The area under the Receiver Operating Characteristic (AUC-ROC) curve and accuracy (ACC) showed that the Random Forest (AUC-ROC = 77.6% and ACC = 78.3%) is a sensitive model for predicting the condition of sewer pipes in the study area. Based on the RF model, maps of predicted conditions of sewers were generated that may be useful for utilities and water managers to establish future sewer system maintenance strategies.

Utilization of Augmented Reality Technique for Sewer Condition Visualization

“…The study introduces a workflow from data collection and processing to ML implementation to predict the condition of sewer pipes, along with a 3D model for sewer condition visualization. The study builds on earlier studies of the authors on sewer condition assessment and prediction using ML for the Ålesund Municipality in Norway [3][4][5]. The platform developed in this study will allow sewer infrastructure managers to visualize the condition of pipes on-site for planning and maintenance.…”

“…The best output produced from the models was selected as the input for visualizing using the application developed in this study. For more information on the application of ML models for sewer condition assessment, readers are referred to [3][4][5].…”

“…Subsequently, the outcomes were visualized on both mobile devices and HoloLens. Additional data, such as road networks or building footprints, received from the Norwegian Mapping Authority (https://www.kartverket.no/en) (accessed on 20 March 2020) [4], were used to build the 3D model. These data were converted into shapefile (*.shp) format, which is an Environmental Systems Research Institute, Inc. (ESRI) (Redlands, CA, USA) vector data storage format for storing the location, shape, and attributes of geographic features.…”

“…In this regard, 3D models of the sewer network (e.g., pipes, manholes, or pumps) were created and visualized using a high-performance computer system associated with the simulation programs [24,42]. Sewer conditions were predicted using ML and deep learning models based on input data, as described in Nguyen, Bui and Seidu [4]. The network components' output was coded based on their corresponding indexes for visualization purposes.…”

“…To plan investments, deterioration models could be used to assess the current and future status of sewer pipes by accounting for intrinsic (e.g., size, age, or material) and extrinsic (e.g., rainfall, soil type, or population) factors. Models used for the assessment and prediction of sewer conditions include regression [3], classification [4], and hybrid models [5]. The results of these models have been presented in Geographic Information System (GIS) for visualization of the geolocation of pipes that need improvement.…”