Abstract In this study, a novel artificial intelligence-based approach is presented to directly estimate the surface topography. To this aim, performance of different artificial intelligence-based techniques, including the multi-layer perceptron neural, radial basis function neural, and adaptive neural fuzzy inference system networks, in estimation of the sample topography is investigated. The results demonstrate that among the designed observers, the multi-layer perceptron method can estimate surface characteristics with higher accuracy than the other methods. In the classical imaging techniques, the scanning speed of atomic force microscope is restricted due to the time required by the oscillating tip to reach the steady state motion while the closed-loop controller tries to maintain the tip vibration amplitude at a set-point value. To address this issue, we have proposed an innovative imaging technique that not only eliminates the need to a closed-loop controller but also estimates the surface topography very quick and accurate compared to the conventional imaging method. Also, the proposed technique is capable of simultaneous estimation of the topography, Hamaker parameter, and the tip-sample interaction force.

中文翻译：

---

使用神经网络和模糊网络的原子力显微镜地形估计观测器设计

摘要 在这项研究中，提出了一种新的基于人工智能的方法来直接估计表面形貌。为此，研究了基于不同人工智能的技术（包括多层感知器神经网络、径向基函数神经网络和自适应神经模糊推理系统网络）在样本地形估计中的性能。结果表明，在设计的观察者中，多层感知器方法可以比其他方法更准确地估计表面特征。在经典成像技术中，由于振荡尖端达到稳态运动所需的时间，而闭环控制器试图将尖端振动幅度保持在设定点值，因此原子力显微镜的扫描速度受到限制。为了解决这个问题，我们提出了一种创新的成像技术，与传统的成像方法相比，它不仅不需要闭环控制器，而且可以非常快速和准确地估计表面形貌。此外，所提出的技术能够同时估计地形、Hamaker 参数和尖端样品相互作用力。