Abstract:

The structural safety problem of rural reconstruction houses structure is prominent, especially the ' Changsha 4·29 civil housing collapse accident ' highlights the seriousness of the problem. In order to solve this problem, firstly, based on machine learning algorithm, the safety early warning index system of rural reconstruction housing structure was constructed. Secondly, the initial data was preprocessed by standardization, and the imbalance of sample categories was solved based on oversampling algorithm. The classical machine learning algorithm was used to construct the safety early warning model of rural housing structure. Then, the ensemble learning algorithm was used to optimize the original model to improve the accuracy of the model. Finally, the importance of each early warning index was sorted. The results show that the classical machine learning algorithm has a better prediction effect on support vector machine (SVM), and the ensemble algorithm has a better effect on stacking method, with an overall prediction rate of 85.3%. The more important early warning indicators are 17, such as too large construction area, long construction year, no construction sketch, non-six key investigation houses, independent foundation, self-built houses in the urban-rural fringe for renting special groups for profit, too large number of upper floors, irregular design, and unqualified design.

Key words: rural reconstruction houses, safety of housing structure, early warning model, machine learning, integrated algorithm