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 8
- Downloads 0
A Smoothed Particle Hydrodynamics Approach for One-Dimensional Dam Break Flow Simulation with Boussinesq Equations
The Smoothed Particle Hydrodynamics (SPH) method has been applied to solve the Boussinesq equations in order to simulate hypothetical one-dimensional dam break flows (DBFs) across varying depth ratios. Initial simulation...
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...
Optimization of Ball-Milling Parameters for Enhanced Compressive Strength of Coal Gangue Cement Paste: A Response Surface Methodology Approach
The substitution of cement with coal gangue powder (CGP) offers significant potential for energy conservation, emission reduction, and environmental sustainability. To optimize the mechanical properties of coal gangue ce...
Mechanisms of Cracking and Stress Control During the Construction Phase of Concrete Face Rockfill Dams in Cold Regions
The construction phase of concrete face rockfill dams is often marred by prominent panel cracking issues, with a lack of reliable foundations for anti-cracking design. To control tensile stresses and enhance crack resist...
Designing Induced Joints in RCC Arch Dams for Enhanced Crack Prevention: A Contact Unit Simulation and Equivalent Strength Theory Approach
Decades of engineering practice have substantiated that the implementation of construction joints is a pivotal method for mitigating dam cracking. The integration of various joint types, notably transverse and induced jo...