Design Of An Early Warning Visualization System For Urban Power Grid Underground Cable Tunnels Based On WebGL

In response to the challenges posed by low lighting conditions and limited ventilation in cable galleries,there is a need for real-time and accurate monitoring of the internal environment and timely warning of hazardous gases.This thesis presents the design of a WebGL-based visualization system for cable gallery warnings,with the following specific contents:(1)Requirement analysis and system function design were conducted.The networking design of the cable gallery warning visualization system was then implemented,encompassing the design of the cable gallery on-site data collection module,environmental parameter prediction module,environmental parameter warning module,and system visualization module.(2)To achieve the system’s prediction and warning functions,a deep learning network model combining Transformer,LSTM,and CNN was developed.The Transformer effectively captures correlations between different time points in a sequence,enhancing prediction accuracy.The LSTM captures time-dependence in the time series data,while the CNN captures local features through convolution operations,further improving prediction accuracy.Additionally,the system’s warning strategy was designed.Comparative experiments were conducted among the Transformer-LSTMCNN network model,LSTM network model,LSTM-CNN network model,and Transformer-LSTM network model.The results demonstrate that the TransformerLSTM-CNN network model outperforms the others in terms of prediction performance.(3)To achieve the goal of visualizing the system warnings,specific parameters for the monitoring camera,sensors,data collector,and streaming media server were configured.A three-dimensional scene model of the cable gallery was created,enabling3 D visualization on the web.The system also encompasses interaction functions,realtime data transmission,warning functions for abnormal cable gallery environment conditions,and monitoring video display functions.To address the issue of extended loading time for three.js 3D scenes,optimization was implemented in two aspects:reducing the number of faces and normal maps during scene model creation and compressing the scene models using Draco during loading.These optimizations effectively improve model loading time while maintaining model quality.The designed WebGL-based cable gallery warning visualization system enables accurate monitoring of the cable gallery’s internal environment and provides real-time warnings for hazardous gases.Through the system’s warning function,personnel can promptly detect and address abnormal issues in the cable gallery environment,thereby ensuring the safety and stability of the cable gallery.