Automated monitoring of pilot tube microtunneling installations through pattern recognition in time-series data of hydraulic pressure

Understanding the jacking forces and productivity of trenchless construction is critical for appropriate technology selection and proactive control of boring operations. Civil engineers can correlate construction operation parameters (e.g., equipment, forces, and timing) with contextual information (e.g., soil characteristics) for identifying issues influencing the trenchless construction productivity. Manual data collection methods for capturing the tunneling construction operations lack the capabilities of acquiring details of the construction, making detailed workflow analysis difficult. This paper presents an automated data collection and interpretation approach for supporting detailed tunneling productivity analysis. This approach uses a data logger to automatically record the hydraulic pressures of various construction equipment used during Pilot Tube Microtunneling (PTMT) installations. A pattern recognition method can then automatically classify time series patterns of the hydraulic pressure, and detect cycles of construction operations within the three phases of the PTMT process: 1) pilot tube installation; 2) casing installation; and 3) product pipe installation. This method first uses the Artificial Neural Network (ANN) algorithm to classify the time series of the hydraulic pressure for determining the phase. It then uses an anomaly detection algorithm to split the time series into boring operation cycles and individual operations (pushing, retracting, etc.) through detecting discontinuities in the time series pattern. Results from a case study indicate that this approach enables detailed productivity analysis of a typical PTMT process.