Low pressure environments multi-sensor fire detection algorithm ensemble model of quadratic optimization choice
An ensemble classification model based on Quadratic Optimization Choice(QOC)is proposed to target the problem of insensitive smoke detection in low-pressure aircraft cargo compartment environments.The model leverages a comprehensive evaluation of multiple fire feature parameters,including CO concentration,temperature,humidity,dual-wavelength PTR,Sauter mean diameter and growth rates.Through the Weka attribute evaluator,highly correlated attributes are identified and selected as input features for the subsequent steps.In the feature engineering stage,the candidate classifiers are ranked based on their performance and suitability for the given problem.To combine the strengths of different classifiers effectively,a soft voting mechanism is adopted.This mechanism allows the secondary classifiers to contribute their predictions to the final decision-making process based on their respective confidences.Therefore,the ensemble model can leverage the diverse perspectives of its component classifiers,resulting in improved accuracy and robustness in real-flight environments.The meta-classifier chosen for the model integration is Multilayer Perceptron(MLP)whose flexibility allows it to adapt and generalize well to various scenarios.To assess the performance of the ensemble model,precision,recall,F1,F2,and F3 scores are used for comparison with KNN and MLP.Experimental results demonstrate that the proposed method outperforms both K Nearest Neighbor(KNN)and MLP when applied to smoke detection in a 60 kPa low-pressure environment.Overall,the ensemble classification model based on QOC,combined with soft voting and MLP as a meta-classifier,offers a solution for enhancing the accuracy and robustness of smoke detection in low-pressure aircraft cargo compartments.This ensemble classification model has 0.972 4 and 0.960 1 classification accuracies,which are 0.087 2 and 0.062 6 higher than KNN and 0.036 8 and 0.182 2 higher than MLP.The comprehensive evaluation of fire feature parameters and the thoughtful selection of classifiers contribute to the model's ability to handle complexities and variations,making it a valuable tool for ensuring flight safety and mitigating potential fire hazards.
safety engineeringfire detectionaircraft cargo compartmentmulti-parameterensemble model
万方数据
维普
