Research on multi-data fusion fire detection based on GWO-LSTM regression prediction model
Year: 2024
Type: conference-paper
Abstract: Aiming at the current situation of unsatisfactory fire prediction index and accuracy, this study obtains the optimal network hyperparameter values for LSTM by introducing the GWO algorithm, obtaining the optimal solution of model parameters through fitness calculation, compensating for the drawback of traditional LSTM easily converging to local optimal solutions, accelerating the convergence speed of LSTM neural networks, and reducing training time. The MATLAB simulation experiments show that the accuracy of the improved LSTM model is 92.54%, 97.04%, and 88.17% respectively, with prediction accuracy improvements of 10.6%, 4.33%, and 22.3%. This significant improvement demonstrates the effectiveness of the improvement method and provides a reference for the development of fire detection technology.
Authors: Haoyang Li, Jianxiang Liu, Zhenkun Cao, Yujun Liu, Ruju Ni
Institutions: Qilu University of Technology, Shandong Academy of Sciences
Language: English
Subfield: Safety, Risk, Reliability and Quality
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4401518957