实时监控下基于二进制蝴蝶优化算法与特征选择的堆石坝压实质量评估研究

Compaction quality evaluation of rockfill dam based on feature selection algorithm with b-BOA under real-time monitoring

传统压实质量检测手段具有取点有限、效率低、反馈不及时等不足,实时压实质量评估已发展为堆石坝压实质量监控的重要内容。基于碾轮振动加速度信号分析的CMV、CCV指标获取简单快速、难度低,但这些指标与压实质量之间的关系容易受到碾压参数和料源参数的影响。针对上述问题,综合考虑碾压参数、料源参数以及CMV、CCV指标,提出了一种基于二进制蝴蝶优化算法(binary Butterfly Optimization Algorithm,b-BOA)和XGBoost特征选择的堆石坝压实质量评估模型。结合现场碾压试验,将该模型应用于某工程全强风化料区的压实质量评估。结果表明,本文提出的模型考虑碾压参数和料源参数对CMV、CCV指标的影响,可有效地进行特征选择,同时与常用压实质量评估模型相比,具有更佳的评估性能。

The conventional inspection methods for compaction quality exist issues of limited spot testing,low efficiency,and untimely feedback.Real-time compaction quality evaluation is an important process of compaction quality monitoring for rockfill dam.The CMV and CCV indicators based on the analysis of the roller vibration acceleration signal can be obtained easily and timely,but the relationship between these indicators and the compaction quality is affected by the rolling parameters and the material parameters.In response to this problem,considering rolling parameters,material parameters,CMV and CCV indicators,this paper proposes a compaction quality evaluation model for rockfill dams based on feature selection algorithm with binary butterfly optimization algorithm(b-BOA)and XGBoost.Combined with the on-site rolling test,the model was applied to the evaluation of the compaction quality of the full-strength weathered material area of a project.The results show that the model,which considers the impact of rolling parameters and material parameters on CMV and CCV indicators,can selects features effectively and outperform common models for compaction quality evaluation.