论文标题
部分可观测时空混沌系统的无模型预测
Segmentation of pores within concrete-epoxy interface using synchronous chemical composition mapping and backscattered electron imaging
论文作者
论文摘要
储层计算是预测湍流的有力工具，其简单的架构具有处理大型系统的计算效率。然而，其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间，然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用，该系统模拟了湍流的若干特征。我们表明，使用径向基函数作为非线性投影器，即使只有部分观测并且不知道控制方程，也能稳健地捕捉复杂的系统非线性。最后，我们表明，当测量稀疏、不完整且带有噪声，甚至控制方程变得不准确时，我们的网络仍然可以产生相当准确的预测，从而为实际湍流系统的无模型预测铺平了道路。
The method of assessing porosity using images from scanning electron microscopy is ineffective in situations where the substrate and the coating have a significantly different average atomic number, which results in a different contrast of backscattered electrons. For this reason, the analyzes carried out to date were usually based on the manual distinction of important microstructure elements, which was followed by selective thresholding. However, this method depends on the subjective action of the researcher, and may result in errors that are difficult to estimate. The presented method is based on the simultaneous collection of both the image of backscattered electrons and elemental maps obtained using an energy dispersive X-ray analyzer. Multivariate comparative analysis of the matrix was used to characterize a technique of distinguishing epoxy resin from a hydrated cement-based substrate. The proposed algorithm significantly facilitates the detection of individual interface components, and also makes comparative analyzes repeatable and reliable.
