用于基桩缺陷智能识别模型的低应变法数据分析方法研究OA
Study on low-strain method data analysis for intelligent identification model of pile defects
建立基桩缺陷智能识别模型需依赖全面且有效的样本数据集,以发挥其非线性回归和预测能力,获取用于机器学习的样本数据集是模型建立过程中的关键步骤.研究提出了一种基于核主成分分析的改进数据分析方法(Pearson-KPCA),旨在为基桩缺陷智能识别模型提供可理解且充分的样本数据.研究结果显示:Pearson-KPCA算法在降低样本数据维度方面表现出显著效果,且经过该算法优化的样本数据集在基桩缺陷智能识别模型中的应用效果明显优于采用传统数据分析方法得到的样本数据集.研究为低应变法检测结果的智能分析提供了可靠的数据处理思路与方法.
The establishment of an intelligent identification model for pile defects requires a comprehensive and effective sample data set to exert its nonlinear regression and prediction capabilities.Obtaining sample data sets for machine learning is a key step in the process of model establishment.An improved data analysis method based on kernel principal component analysis(Pearson-KPCA)was proposed to provide understandable and sufficient sample data for the intelligent identification model of pile defects.The research results showed that the Pearson-KPCA algorithm had a significant effect on reducing the dimension of sample data,and the performance of the sample data set optimized by the algorithm in the intelligent recognition model of pile defects was obviously better than that of the traditional data analysis method.The research provided a reliable data processing idea and method for the intelligent analysis of low strain method detection results.
韩剑飞
安徽省·水利部淮河水利委员会水利科学研究院,安徽 合肥 230088
信息技术与安全科学
基桩缺陷识别低应变法Pearson-KPCA模型特征提取核主成分分析数据分析
pile defect identificationlow-strain methodPearson-KPCA modelcharacter extractionkernel principal component analysisdata analysis
《江淮水利科技》 2026 (1)
31-35,5
评论