基于人工神经网络的纳米多孔金力学性能分析

Analysis for Mechanical Properties of Nanoporous Metal Based on Neural Network

多孔材料孔径可控、形貌多样,已广泛应用于催化、检测、防护、环保等工业领域。纳米多孔金属(Nano Porous Metals, NPMs)作为多孔材料的一个重要分支,内部具有纳米级三维连通孔洞结构和高比表面积,引起了国内外学术界和工业界广泛的关注。其中,尤为重要的是在许多应用中起着至关重要的作用的力学完整性和可靠性。然而,NPMs制备过程和测量过程的影响因素十分复杂多变,造成试验结果的不稳定,这些因素必然会导致研究过程中的实验数据波动大、结论准确性低、工作量大、时间长、成本高等一系列缺点。为此,人工智能研究方法引入材料科学研究领域日趋广泛,但应用于纳米多孔材料及其力学性能研究几乎未见报道。本文的主要目的是建立神经网络模型,以韧带尺寸和相对密度为输入对纳米多孔金的力学性能进行测试实验。

Porous materials with controllable pore size and diverse morphology have been widely used in catalysis, detection, protection, environmental protection and other industrial fields. As an im-portant branch of porous materials, nanoporous metal (NPMs) has nano-level three-dimensional connected pore structure and high specific surface area, which has attracted extensive attention from domestic and foreign academia and industry. Of particular importance is the mechanical in-tegrity and reliability which play a vital role in many applications. However, the influencing factors of the preparation process and measurement process of NPMs are very complex and changeable, resulting in the instability of test results. These factors will inevitably lead to a series of dis-advantages such as large fluctuation of experimental data, low accuracy of conclusion, large work-load, long time and high cost in the research process. For this reason, the research method of arti-ficial intelligence has been introduced into the field of materials science increasingly widely, but the research on nanoporous materials and their mechanical properties has hardly been reported. The main purpose of this paper is to establish a backpropagation neural network model and predict the mechanical properties of nanoporous gold nanoparticles with the input of ligament size and relative density.