A REAL-TIME HYBRID DEEP LEARNING MODEL FOR PREDICTING FIRE OUTBREAKS CAUSED BY ELECTRICAL FAILURE

In residential, commercial, and industrial settings, fire outbreaks continue to pose a serious and ongoing threat, frequently leading to irreparable damage, financial loss, and fatalities.Although fire protection technology has advanced, the majority of traditional systems are reactive rather than proactive.Because these systems are made to identify the existence of a fire only after it has started, there is little or no chance of taking preventative or early intervention measures.The researchers in this study created a hybrid deep learning model that combines feedforward neural networks (FFNN) and long short-term memory (LSTM) networks.This model has shown great promise in predicting fire outbreaks brought on by electrical failures.Utilizing FFNN's capacity to extract intricate patterns and LSTM's proficiency with sequential timeseries data, the suggested model produced exceptional performance measures, such as a robust F1-score of 97.2%, accuracy of 98.2%, precision of 96.8%, and recall of 97.6%.With a false positive rate of less than 2.3%, these findings demonstrate the model's high capacity to accurately identify both typical and potentially dangerous electrical activities.The model's short inference latency (~50 ms) further validated its Igbudu Kingsley Ezebunwo, Nlerum Promise Anebo https://iaeme.com/Home/