An Innovative Method for Gas Leakage Detection Device Based on XGBoost-A-BiGRU Based Approach

Recently, there has been an uptick in both the incidence of house fires and the danger the system pose to people's lives and possessions. Even if a safe distance away from the leak, liquid petroleum gas (LPG) can still ignite. The regulator being left on or a defective rubber hose are the two most common causes of fires. Making a system to alert people of gas leaks is, thus, extremely important. To that end, this research presents a gas leakage alarm system to keep an eye out for emergencies like these. The proposed procedure begins with three stages: data preprocessing, feature selection, and model training. The wavelet de-noising technique is used in preprocessing data. Time domain features are utilized during the feature selection process. Once features have been extracted, the data is used to train XGBoost-A-BiGRU models. The proposed strategy outperforms XGBoost and BiGRU, two well-known rivals. The greatest accuracy of 98.70% was reached using the proposed strategy.