In this brief, the compensation for both the nonlinear hysteresis and the vibrational dynamics effects of piezo actuators is studied. Piezo actuators are the enabling device in many applications such as atomic force microscopy (AFM) to provide nanoto atomic-levels precision positioning. During high-speed, largerange positioning, however, large positioning errors can be generated due to the combined hysteresis and dynamics effects of piezo actuators, making it challenging to achieve precision positioning. The main contribution of this brief is the use of an inversion-based iterative control (IIC) technique to compensate for both the hysteresis and vibrational dynamics effects of piezo actuators. The convergence of the IIC algorithm is investigated by capturing the input-output behavior of piezo actuators with a cascade model consisting of a rate-independent hysteresis at the input followed by the dynamics part of the system. The size of the hysteresis and the vibrational dynamics variations that can be compensated for (by using the IIC method) is quantified. The IIC approach is illustrated through experiments on a piezotube actuator used for positioning on an AFM system. Experimental results show that high-speed, large-range precision positioning can be achieved by using the proposed IIC technique. Furthermore, the proposed IIC algorithm is also applied to experimentally validate the cascade model and the rate-independence of the hysteresis effect of the piezo actuator.Index Terms-Atomic force microscopy, hysteresis, iterative methods, nanotechnology, piezoelectric materials.

This review covers the core concepts and design decisions of TensorFlow. TensorFlow, originally created by researchers at Google, is the most popular one among the plethora of deep learning libraries. In the field of deep learning, neural networks have achieved tremendous success and gained wide popularity in various areas. This family of models also has tremendous potential to promote data analysis and modeling for various problems in educational and behavioral sciences given its flexibility and scalability. We give the reader an overview of the basics of neural network models such as the multilayer perceptron, the convolutional neural network, and stochastic gradient descent, the most commonly used optimization method for neural network models. However, the implementation of these models and optimization algorithms is time-consuming and error-prone. Fortunately, TensorFlow greatly eases and accelerates the research and application of neural network models. We review several core concepts of TensorFlow such as graph construction functions, graph execution tools, and TensorFlow’s visualization tool, TensorBoard. Then, we apply these concepts to build and train a convolutional neural network model to classify handwritten digits. This review is concluded by a comparison of low- and high-level application programming interfaces and a discussion of graphical processing unit support, distributed training, and probabilistic modeling with TensorFlow Probability library.

Early detection of epidermal growth factor receptor (EGFR) mutation, particularly EGFR T790M mutation, is of clinical significance. The aim of the present study was to compare the performances of amplification refractory mutation system-based quantitative polymerase chain reaction (ARMS-qPCR) and droplet digital polymerase chain reaction (ddPCR) approaches in the detection of EGFR mutation and explore the feasibility of using ddPCR in the detection of samples with low mutation rates. EGFR gene mutations in plasmid samples with different T790M mutation rates (0.1–5%) and 10 clinical samples were detected using the ARMS-qPCR and ddPCR approaches. The results demonstrated that the ARMS-qPCR method stably detected the plasmid samples (6,000 copies) with 5 and 1% mutation rates, while the ddPCR approach reliably detected those with 5% (398 copies), 1% (57 copies), 0.5% (24 copies) and 0.1% (average 6 copies) mutation rates. For the 10 clinical samples, the results for nine samples by the ARMS-qPCR and ddPCR methods were consistent; however, the sample N006, indicated to be EGFR wild-type by ARMS-qPCR, was revealed to have a clear EGFR T790M mutation with seven copies of mutant alleles in a background of 6,000 wild-type copies using ddPCR technology. This study demonstrates the feasibility of applying the ddPCR system to detect EGFR mutation and identified the advantage of ddPCR in the detection of samples with a low EGFR mutation abundance, particularly the secondary EGFR T790M resistance mutation, which enables early diagnosis before acquired resistance to tyrosine kinase inhibitors becomes clinically detectable.

No abstract

This article presents an iterative-based feedforward-feedback control approach to achieve high-speed atomic force microscope (AFM) imaging. AFM-imaging requires precision positioning of the probe relative to the sample in all x-y-z axes directions. Particularly, this article is focused on the vertical z-axis positioning. Recently, a current-cycle-feedback iterative-learning-control (CCF-ILC) approach has been developed for precision tracking of a given desired trajectory (even when the desired trajectory is unknown), which can be applied to achieve precision tracking of sample profile on one scanline. In this article, we extend this CCF-ILC approach to imaging of entire sample area. The main contribution of this article is the convergence analysis and the use of the CCF-ILC approach for output tracking in the presence of desired trajectory varation between iterations—the sample topography variations between adjacent scanlines. For general case where the desired trajectory variation occurs between any two successive iterations, the convergence (stability) of the CCF-ILC system is addressed and the allowable size of desired trajectory variation is quantified. The performance improvement achieved by using the CCF-ILC approach is discussed by comparing the tracking error of using the CCF-ILC technique to that of using feedback control alone. The efficacy of the proposed CCF-ILC control approach is illustrated by implementing it to the z-axis control during AFM-imaging. Experimental results are presented to show that the AFM-imaging speed can be substantially increased.

Following the fourth industrial revolution, and with the recent advances in information and communication technologies, the digital twinning concept is attracting the attention of both academia and industry worldwide. A microgrid digital twin (MGDT) refers to the digital representation of a microgrid (MG), which mirrors the behavior of its physical counterpart by using high-fidelity models and simulation platforms as well as real-time bi-directional data exchange with the real twin. With the massive deployment of sensor networks and IoT technologies in MGs, a huge volume of data is continuously generated, which contains valuable information to enhance the performance of MGs. MGDTs provide a powerful tool to manage the huge historical data and real-time data stream in an efficient and secure manner and support MGs' operation by assisting in their design, operation management, and maintenance. In this paper, the concept of the digital twin (DT) and its key characteristics are introduced. Moreover, a workflow for establishing MGDTs is presented. The goal is to explore different applications of DTs in MGs, namely in design, control, operator training, forecasting, fault diagnosis, expansion planning, and policy-making. Besides, an up-to-date overview of studies that applied the DT concept to power systems and specifically MGs is provided. Considering the significance of situational awareness, security, and resilient operation for MGs, their potential enhancement in light of digital twinning is thoroughly analyzed and a conceptual model for resilient operation management of MGs is presented. Finally, future trends in MGDTs are discussed.

No abstract

No abstract