Monitor the Strength Status of Buildings Using Hybrid Machine Learning Technique

Abstract

Standard inspections of buildings are not always possible because of human flaws in prediction. Hence, we need more stable, scalable, and efficient automated processes. Structure Health Monitoring (SHM) is one of the automation systems for forecasting potential losses in building structures. This article suggested how to monitor the strength status of buildings by using Hybrid Machine Learning Technique (HMLT). HMLT contains two-hybrid procedures. One for identifying the most significant features in Dataset using Hybrid Feature Selection Method (HFSM). HFSM uses the combined features of Mutual information (MI) and Rough Set Theory (RST) for feature selection. Another method is optimized classifiers such as Support Vector Machine (SVM) and Artificial Neural Networks (ANN) are used for the classification and predicting the accuracy i.e. predicting the strength status of buildings. Now the proposed method is applied on Earthquake Damage Dataset (Gorkha Earthquake in April 2015). Training and 10- fold cross-validation procedure pragmatic to features. Then the performance of proposed method has been evaluated using the F1-score and accuracy metrics and get 91% and 92% respectively. Finally, the result analysis demonstrates the importance of the proposed approach in predicting the status of the building strength.

Keywords

• KNN: K-nearest neighbors
• RF: random forest
• GBM: gradient boosted machines
• SVM: support vector machine
• ANN: artificial neural networks
• HMLT: hybrid machine learning technique