Stress wave based material identification of buried pipelines using signal processing and deep learning

K I M Iqbal

doi:10.17918/00011183

Lead-based water pipelines pose a significant public health risk in the US. The challenge lies in locating these pipelines, as current identification technologies have limitations and utility records are often inaccurate. This study introduces a technology based on stress wave propagation, a non-invasive method for identifying service line materials on both the public (utility) and private (customer) sides. Stress waves are generated at the curb-stop valve of the service line using an impact hammer (input signal) and an extension rod. Piezoelectric accelerometer sensors placed on the soil surface then detect the pipe's responses (output signal). Since buried service lines are surrounded by soil and contain water, the stress wave propagation is non trivial. This study outlines several steps to effectively utilize the proposed technology for distinguishing service line materials. Firstly, the study performed numerical simulations to investigate the applicability of the proposed method by examining wave propagation properties that could be used in a stress wave approach to identify buried lead-based pipelines. For instance, the dispersion curves differ significantly for steel, copper, lead, and PVC pipes filled with water. While the soil surrounding pipes causes a decrease in wave propagation energy due to the energy leakage into the soil medium, this phenomenon can enable the detection of leaked waves with sufficiently sensitive sensors installed near the soil surface. The received signals vary for different types of pipe materials, allowing to differentiate among service line materials. Building on insights from the numerical simulations, the study conducted controlled lab experiments, which suggest that stress wave propagation could become a valuable tool for identifying buried water SL materials. Subsequently, the study field-tested the proposed technology on nearly 419 service lines across 20 cities in the United States. The recorded signals were processed to extract frequency response functions (FRF) and wavelet transform (WT) features to capture both spectral and time-frequency information. However, the geographic diversity of the dataset introduced significant variation in soil depth, surface conditions, and material properties, which complicated the development of reliable physics-based models. As a result, approaches relying solely on time-of-arrival to extract speed features, FRF, or WT were insufficient to consistently distinguish between lead and non-lead materials. The study then explored various deep learning algorithms, including MLP and 1D-CNN using FRF data, and 2D-CNN with transfer learning, CNN-BiLSTM, and transformer-based models (TimeSformer) using wavelet-transformed features. Prior to model training, all data underwent necessary preprocessing steps to extract robust features (FRF and WT matrices) from the raw stress-wave signals. This thesis presents detailed discussions on data preparation, development of model architectures, and performance evaluations. The proposed technology, integrated with a deep learning framework, demonstrated promising performance on both test datasets and blind prediction of unknown service line materials.

Stress wave based material identification of buried pipelines using signal processing and deep learning

Files and links (1)

Abstract

Metrics

Details

Stress wave based material identification of buried pipelines using signal processing and deep learning

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media