Health Monitoring and Safety Early Warning System for Foundation Pit Support Structure Using Support Vector Machine and Ensemble Learning Algorithm

This paper proposes a health monitoring and safety early warning system for foundation pit support structure based on SVM and ensemble learning algorithm. This system can achieve real time supervision and hazard forecast based on the SVM classification capability and the merits of ensemble learning. Firstly, the system adopts the transducer to gather all kinds of monitoring data, then obtains the valid information by preprocess and characteristic extraction. Secondly, SVM is applied to classify the condition, and then the method of ensemble learning is adopted to make the model more accurate and robust. Experiments indicate that the precision of this system is up to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 2. 5 \%}$</tex>, and it is more than <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{10 \%}$</tex> better than the conventional one. As for real time supervision, the system's response time can be controlled in 5s, so that it can respond quickly to unexpected security incidents. By comparing with other traditional monitoring methods, the algorithm proposed in this paper has obvious advantages in multi-category classification, data noise processing and model stability. The performance of ensemble learning algorithm in multi-source data fusion and optimization prediction is particularly outstanding, which can effectively improve the performance of early warning system.