Abstract Oligonucleotide microarray has become one of the most powerful tools in the areas such as genotyping, gene expression, and SNPs detection. However quality control is critical for rapid development of this technology. Several approaches have been attempted to monitor the fabrication steps from oligonucleotide probe printing to data analysis. Yet, quality control methods for presynthesized oligonucleotide probes have not been available. Here we propose a novel approach for the control of the quality of oligonucleotide probes using the technology of denaturing high-performance liquid chromatography (DHPLC). In this work, all possibly existing types of poorly synthesized oligonucleotide probes prepared by artificial method were used to validate the discriminating ability of DHPLC. It was found that DHPLC can not only provide the purity ratio of the oligonucleotide probes accurately, but also detect misincorporation of chemicals in the oligonucleotides such as one base substitution, one base deletion, or failed amino modification which is difficult to be distinguished through routine methods such as PAGE. For 277 oligonucleotide probes, 31% of the probes had a purity ratio of 70% or less. The results indicate that DHPLC can be a practical approach to control the quality of oligonucleotide probes before they are printed to the microarray slides.
A vision inspection system is introduced to detect drainage pipelines, identify the inner wall side damage situation such as crack, hole etc.. Pipeline image is preprocessed using image processing technique and segmented by threshold to find out external contour of the damage region. Then, damage feature is extracted to identify the damage type using both statistics recognition and BP neural network (BP NN) recognition. In the end, recognition difference between statistic and BPNN is discussed.
Abstract Compared with traditional motors, ultrasonic motors have the advantages of small size and no noise and are widely used in modern technological fields. This article proposes a neural network-based ultrasonic motor control method and optimizes it to address the difficulty of controlling ultrasonic motors. Firstly, the principle and equivalent circuit of ultrasonic motors are studied, and a phase-shifted PWM method suitable for driving ultrasonic motors is proposed. Secondly, to meet the nonlinear characteristics of ultrasonic motors, a NARX neural network that can be used for nonlinear systems is adopted for optimization control. Finally, an experimental platform was built for the experiment, and the results showed that the proposed H-bridge driving circuit has a wide input voltage range. After using the NARX neural network, the output voltage of the H-bridge circuit is more stable, which can provide a stable driving voltage for the ultrasonic motor. It can provide a certain reference value for the application of ultrasonic motors in modern technology.
This paper presents a high-speed communication and real-time control system for AC servo system. A FPGA-based current control system for AC servo machine is developed to achieve high-performance control. In the communication system, sigma-delta modulation is employed to compress data and to transmit the signal over the transmission channels between the controller and the controlled plant. The current control loop is based on the deadbeat control method, and the space vector method is employed to fire the switches of a PWM inverter. Simulation results show that the proposed system can compensate the possible noise in the transmission channels. In addition, the current controller is implemented in field-programmable gate array (FPGA). The experimental results of current control loop with FPGA-based solutions are also given in this paper.
Deep neural networks (DNNs) have gained a strong momentum among various applications. The enormous matrix-multiplication exhibited in the above DNNs is computation and memory intensive. Resistive random-access memory crossbar (RRAM-crossbar) consisting of memristor cells can naturally carry out the matrix-vector multiplication. RRAM-crossbar-based accelerator, therefore, has two orders of magnitude of higher energy-efficiency than conventional accelerators. The imperfect fabrication process of RRAM-crossbars, however, causes various defects and process variations. These fabrication imperfections not only result in significant yield loss but also degrade the accuracy of DNNs executed on the RRAM-crossbars. In this article, we first propose an accelerator-friendly neural-network training method, by leveraging the inherent self-healing capability of the neural network, to prevent the large-weight synapses from being mapped to the imperfect memristors. Next, we propose a dynamic adjustment mechanism to extend the above method for DNNs, such as multilayer perceptrons (MLPs), wherein the imperfect-memristor induced errors can accumulate and magnify through multiple layers. Such off-device training method is a pure software solution, and it is unable to provide enough accuracy for convolutional neural networks (CNNs). Several works propose error-tolerable hardware design by allowing the retraining of CNNs on the RRAM-crossbar. Although this hardware-based on-device training method is effective, the frequent write operation on RRAM-crossbar hurt the endurance of RRAM-crossbars. Consequently, we propose a software and hardware co-design methodology to effectively preserve the classification accuracy of CNN with few on-device training iterations. The experimental results show that the proposed method can guarantee ≤1.1% loss of accuracy for resistance variations in MLP and CNN. Moreover, the proposed method can guarantee ≤1% loss of accuracy even when stuck-at-faults (SAFs) rate = 20%.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
