An Intelligent Recording Method for Field Geological Survey Data in Hydraulic Engineering Based on Speech Recognition
Journal Title: Journal of Civil and Hydraulic Engineering - Year 2024, Vol 2, Issue 4
Abstract
Field data collection is a crucial component of geological surveys in hydraulic engineering. Traditional methods, such as manual handwriting and data entry, are cumbersome and inefficient, failing to meet the demands of digital and intelligent recording processes. This study develops an intelligent speech recognition and recording method tailored for hydraulic engineering geology, leveraging specialized terminology and speech recognition technology. Initially, field geological work documents are collected and processed to create audio data through manual recording and speech synthesis, forming a speech recognition training dataset. This dataset is used to train and construct a speech-to-text recognition model specific to hydraulic engineering geology, including fine-tuning a Conformer acoustic model and building an N-gram language model to achieve accurate mapping between speech and specialized vocabulary. The model's effectiveness and superiority are validated in practical engineering applications through comparative experiments focusing on decoding speed and character error rate (CER). The results demonstrate that the proposed method achieves a word error rate of only 2.6% on the hydraulic engineering geology dataset, with a single character decoding time of 15.5ms. This performance surpasses that of typical speech recognition methods and mainstream commercial software for mobile devices, significantly improving the accuracy and efficiency of field geological data collection. The method provides a novel technological approach for data collection and recording in hydraulic engineering geology.
Authors and Affiliations
Zuguang Zhang, Qiubing Ren, Wenchao Zhao, Mingchao Li, Leping Liu, Yuangeng Lyu
Keywords
Related Articles
- EP ID EP752503
- DOI https://doi.org/10.56578/jche020403
- Views 39
- Downloads 0
Comparative Analysis of 1D and 2D Modeling Approaches for Scour Depth Estimation: A Case Study of the Kelanisiri Bridge, Sri Lanka
The scouring process, characterised by the erosion of sediment around bridge piers due to fluid flow, poses a significant risk to the structural integrity of bridges. Scour depth, defined as the vertical distance from th...
Evaluation of Rainwater Harvesting and Bio-pore Infiltration Holes for Flood Mitigation and Soil Conservation
Rainwater harvesting (RH) techniques, specifically the implementation of Bio-pore Infiltration Holes (BIH), have been investigated as cost-effective and practical methods for managing surface runoff and mitigating flood...
Numerical Simulation of Resistivity Response Characteristics in Seepage Detection of Cutoff Walls Using Cross-Hole Resistivity Tomography
Cutoff walls are an essential method for seepage prevention in dams. During the construction and operation of reservoirs, factors such as construction techniques, variations in groundwater conditions within the dam body,...
Risk Assessment of High-grade Highway Construction Based on Combined Weighting and Fuzzy Mathematics
High-grade highways are an important part of the modern comprehensive transportation system. However, due to frequent natural disasters, harsh meteorological conditions, and fragile geological environments, high-grade hi...
Evaluating Flood Hazard Mitigation through Sustainable Urban Drainage Systems in Bor, Jonglei State, South Sudan
In response to the escalating pressures of urbanization and population growth on the ecosystems and flood risks in Bor County, Jonglei State, South Sudan, this study proposes the implementation of Sustainable Urban Drain...