支持向量回归算法在土体含水率AHFO测试中的应用研究

APPLICATION RESEARCH OF SUPPORT VECTOR REGRESSION ALGORITHM IN AHFO TEST OF SOIL WATER CONTENT

主动加热光纤法(AHFO)可实现土体原位含水率的分布式测量,但受场地复杂环境的影响,现有含水率计算模型解译获得的测试结果容易发生较大误差。本文在试验研究的基础上,探究了土体干密度对AHFO法含水率计算的影响,并将支持向量回归(SVR)算法引入含水率的计算模型中,与传统计算模型进行了对比。研究结果表明:土体干密度是影响含水率计算模型精度的重要因素,干密度波动越大则传统计算模型的计算误差也越大,土体较高含水率下会降低计算模型的敏感性;与传统的含水率计算模型相比,考虑干密度影响的径向基核支持向量回归方法具有更高的计算精度,建议在含水率AHFO法的测试中推广应用。

Active heating optical fiber method(AHFO)can realize the distributed measurement of soil moisture content in situ. But due to the complex environment of the site, the test results obtained by the interpretation of the existing moisture content calculation model are prone to large errors. Based on the experimental research, this paper explored the effect of soil dry density on the calculation of water content of AHFO method, and introduced support vector regression(SVR)algorithm into the calculation model of water content. The numerical result was compared with the traditional calculation model. The research results show that the dry density of the soil is an important factor that affects the accuracy of the water content calculation model. The larger the dry density fluctuation, the larger the calculation error of the traditional calculation model. The higher the soil content, the lower the sensitivity of the calculation model. Compared with the traditional water content calculation model, the radial basis kernel support vector regression method considering the influence of dry density has a higher calculation accuracy, and it is recommended to be applied in the test of the water content AHFO method.