检索结果-内蒙古大学图书馆

FPGA Implementation of Image Registration Using Accelerated CNN

SENSORS 2023年第14期23卷 6590-6590页

作者： Aydin, Seda Guzel Bilge, Hasan Sakir Bingol Univ Dept Elect & Elect Engn TR-12000 Bingol Turkiye Gazi Univ Biomed Calibrat & Res Ctr BIYOKAM TR-06560 Ankara Turkiye

Background: Accurate and fast image registration (IR) is critical during surgical interventions where the ultrasound (US) modality is used for image-guided intervention. Convolutional neural network (CNN)-based IR methods have resulted in applications that respond faster than traditional iterative IR methods. However, general-purpose processors are unable to operate at the maximum speed possible for real-time CNN algorithms. Due to its reconfigurable structure and low power consumption, the field programmable gate array (FPGA) has gained prominence for accelerating the inference phase of CNN applications. Methods: This study proposes an FPGA-based ultrasound IR CNN (FUIR-CNN) to regress three rigid registration parameters from image pairs. To speed up the estimation process, the proposed design makes use of fixed-point data and parallel operations carried out by unrolling and pipelining techniques. Experiments were performed on three US datasets in real time using the xc7z020, and the xcku5p was also used during implementation. Results: The FUIR-CNN produced results for the inference phase 139 times faster than the software-based network while retaining a negligible drop in regression performance of under 200 MHz clock frequency. Conclusions: Comprehensive experimental results demonstrate that the proposed end-to-end FPGA-based accelerated CNN achieves a negligible loss, a high speed for registration parameters, less power when compared to the CPU, and the potential for real-time medical imaging.

关键词： accelerated CNN field programmable gate array image registration ultrasound

来源：评论

学校读者我要写书评

暂无评论

Error estimation and enhanced stiffness sensitivity in contact resonance force microscopy with a multiple arbitrary frequency lock-in amplifier (MAFLIA)

引用

MEASUREMENT SCIENCE AND TECHNOLOGY 2020年第11期31卷 115009-115009页

作者： Flater, Erin E. Mugdha, Arya C. Gupta, Saurabh Hudson, William A. Fahrenkamp, Abbigail A. Killgore, Jason P. Wilson, Jesse W. Luther Coll Dept Phys Decorah IA 52101 USA Colorado State Univ Dept Elect & Comp Engn Ft Collins CO 80521 USA NIST Appl Chem & Mat Div Boulder CO 80305 USA Colorado State Univ Sch Biomed Engn Ft Collins CO 80521 USA

In contact resonance force microscopy and related dynamic atomic force microscopy methods, an accurate description of the real-time cantilever dynamics is essential to the mapping of local material properties, such as viscoelasticity, piezo response, and chemical composition. Stiffness and damping variations of the tip-sample contact result in variations in the cantilever's resonance frequency and quality factor as it scans a sample of interest. When measuring characteristics of the resonance, generally, there is a tradeoff between full spectral coverage, best obtained by sweeping the amplitude versus frequency response in the time or frequency domain, and high-speed information, obtained by observing the cantilever response at one or two discrete frequencies, that may be required to track a resonance frequency that changes spatially. Here, we introduce a new option for performing contact resonance force microscopy with a low-cost multifrequency lock-in amplifier system with up to eight simultaneous independent excitation and detection frequencies. We demonstrate how the multifrequency approach can measure contact resonance frequency, quality factor, amplitude, and phase during imaging, with high precision and error estimation, without the need for frequency-tracking feedback. We show, using a wood composite sample, that this multifrequency approach can determine resonance frequency and quality factor, and associated uncertainty. This ability to estimate uncertainty of resonance parameters is not possible with 1 and 2 frequency methods. We further utilize the multifrequency lock-in to develop a novel means of increasing the stiffness range for highly sensitive nanomechanical sensing by dividing the eight lock-in frequencies to monitor two or four simultaneous eigenmodes, each of which is optimized for sensitivity in a particular stiffness regime. Overall, we show how multifrequency lock-in amplifiers with observation frequency chosen to coincide with an expected eig

关键词： atomic force microscopy contact resonance force microscopy lock-in amplifier field programmable gate array multifrequency AFM digital signal processing

来源：评论

学校读者我要写书评

暂无评论

A neural network accelerated optimization method for FPGA

引用

JOURNAL OF COMBINATORIAL OPTIMIZATION 2024年第5期47卷 84-84页

作者： Hu, Zhengwei Zhu, Sijie Wang, Leilei Cao, Wangbin Xie, Zhiyuan North China Elect Power Univ Dept Elect & Commun Engn Baoding 071003 Peoples R China

A neural network accelerated optimization method for FPGA hardware platform is proposed. The method realizes the optimized deployment of neural network algorithms for FPGA hardware platforms from three aspects: computational speed, flexible transplantation, and development methods. Replacing multiplication based on Mitchell algorithm not only breaks through the speed bottleneck of neural network hardware acceleration caused by long multiplication period, but also makes the parallel acceleration of neural network algorithm get rid of the dependence on the number of hardware multipliers in FPGA, which can give full play to the advantages of FPGA parallel acceleration and maximize the computing speed. Based on the configurable strategy of neural network parameters, the number of network layers and nodes within layers can be adjusted according to different logical resource of FPGA, improving the flexibility of neural network transplantation. The adoption of HLS development method overcomes the shortcomings of RTL method in designing complex neural network algorithms, such as high difficulty in development and long development cycle. Using the Cyclone V SE 5CSEBA6U23I7 FPGA as the target device, a parameter configurable BP neural network was designed based on the proposed method. The usage of logical resources such as ALUT, Flip-Flop, RAM, and DSP were 39.6%, 40%, 56.9%, and 18.3% of the pre-optimized ones, respectively. The feasibility of the proposed method was verified using MNIST digital recognition and facial recognition as application scenarios. Compare to pre-optimization, the test time of MNIST number recognition is reduced to 67.58%, and the success rate was lost 0.195%. The test time for facial recognition applications was reduced to 69.571%, and the success rate of combining LDA algorithm was lost within 4%.

关键词： field programmable gate array Neural network Hardware acceleration Mitchell algorithm High-level synthesis

来源：评论

学校读者我要写书评

暂无评论

Precise single step and multistep short-term photovoltaic parameters forecasting based on reduced deep convolutional stack autoencoder and minimum variance multikernel random vector functional network

引用

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 2024年第PartA期136卷

作者： Sahani, Mrutyunjaya Choudhury, Sasmita Siddique, Marif Daula Parida, Tanmoy Dash, Pradipta Kishore Panda, Sanjib Kumar Natl Univ Singapore Dept Elect & Comp Engn Singapore Singapore Siksha O Anusandhan Deemed Be Univ Multidisciplinary Res Cell Bhubaneswar 751030 India

Accurate prediction of photovoltaic (PV) power can significantly alleviate energy crises. However, the inherent randomness and intermittency of PV power pose challenges to the stability and safety of PV-penetrated grid systems. To address this, we have developed a novel hybrid model: a reduced deep convolutional stack autoencoder with a minimum variance multikernel random vector functional link network (RDCSAEMVMRVFLN). This model enhances grid efficacy and safety. We extract the most informative band-limited intrinsic mode functions (BLIMFs) of highly nonlinear and nonstationary solar energy parameters using an entropy, kurtosis, and correlation coefficient-based information-oriented variational mode decomposition (IOVMD). These efficient BLIMFs are concatenated and input into the RDCSAE for rich, abstract, and discriminative representation computation. A less computationally complex MVMKRVFLN regression method, incorporating these refined representations, is proposed for superior prediction accuracy and reduced computational complexity. Our method shows exceptional performance in predicting solar temperature, irradiation, and power for multi-horizon forecasts with minimal error metrics (correlation coefficients of 0 . 999 +/- 0 . 001, 0 . 992 +/- 0 . 001, 0 . 986 +/- 0 . 02 and 0 . 978 +/- 0 . 02, and RMSE of 0 . 016 +/- 0 . 001, 0 . 024 +/- 0 . 001, 0 . 034 +/- 0 . 001 and 0 . 045 +/- 0 . 001 for the interval of 10 minutes, 30 minutes, 1 hour and 3 hours respectively) in both single-step and multistep forecasting compared to conventional methods. The RDCSAE-MVMKRVFLN model is implemented on a high-speed Xilinx Virtex-5 FPGA embedded processor to validate its simplicity, robustness, and practicability. Additionally, we examine the prediction performance using real-time data from a 1 MW solar farm in Odisha, India, demonstrating the model's effectiveness and superiority.

关键词： Reduced deep convolutional stack autoencoder Information-oriented variational mode decomposition Minimum variance multikernel random vector functional link network Single step and multistep forecasting field programmable gate array

来源：评论

学校读者我要写书评

暂无评论

A resilient and intelligent multi-objective energy management for a hydrogen-battery hybrid energy storage system based on MFO technique

引用

RENEWABLE ENERGY 2024年 222卷

作者： Elkholy, M. H. Senjyu, Tomonobu Metwally, Hamid Farahat, M. A. Irshad, Ahmad Shah Hemeida, Ashraf M. Lotfy, Mohammed Elsayed Univ Ryukyus Dept Elect & Elect Engn Okinawa 9030213 Japan Zagazig Univ Fac Engn Elect Power & Machines POB 44519 Zagazig Egypt Kandahar Univ Engn Fac Dept Energy Engn Kandahar 3801 Afghanistan Aswan Univ Fac Energy Engn Dept Elect Engn Aswan 81528 Egypt

In this paper, a new design and flexible energy management strategy are presented for microgrids. The proposed intelligent energy management system (IEMS) achieves effective integration between the resilient microcontroller, chosen for its rapid response speed and its capability to perform multiple operations simultaneously, and the optimization techniques to enhance the power quality. The IEMS is designed using the FPGA board, chosen for its flexibility and capability to handle multiple and complex operations simultaneously. The experimental testing of the IEMS demonstrates a significant level of effectiveness in managing energy. To enhance system performance and ensure cost-effective reliability, advanced optimization techniques are employed. This study deals with a complex multi-objective optimization problem involving the limitations of energy generation, load demand, and a hydrogen-battery hybrid energy storage system. The moth-flame optimization (MFO) algorithm is chosen to solve this optimization problem due to its rapid convergence rate and accuracy. The effectiveness of the MFO algorithm is assessed by comparing it with several new algorithms. The obtained results show the robust performance of the IEMS and its high responsiveness to dynamic operational scenarios. It can observe, gather, and analyze data in real-time. It achieves a remarkable 1.287 % reduction in operating costs within a short timeframe.

关键词： Hydrogen-battery hybrid energy storage system Moth-flame Optimization Operating reserve field programmable gate array Backup energy management system

来源：评论

学校读者我要写书评

暂无评论

Giga-sample Pulse Acquisition and Digital Processing for Photomultiplier Detectors

引用

JOURNAL OF FUSION ENERGY 2022年第1期41卷 8-8页

作者： Walewski, W. Nowakowski, P. Nowak Vel Makowski, D. Lodz Univ Technol Dept Microelect & Comp Sci Lodz Poland

Digital Pulse Processing offers multiple advantages over traditional analogue processing chains. As a disadvantage, produce gigabytes of data every second. Storing and processing such data rates in real-time still remains a challenge. Analogue solutions are not riddled with this issue, however, they offer limited flexibility and modifiability. This work highlights the advantages of Digital Pulse Processing over Analogue Pulse Processing and describes a successful implementation of a digital pulse detection and acquisition system based on field programmable gate arrays. The system is tasked with processing pulses generated by a Photo Multiplier Tube nuclear detector. Incoming signals are sampled at a 1 GS/s rate, so to enable full acquisition resolution, throughput is reduced with digital detection filters and leading-edge triggering or with a derivative zero-crossing detector. Three different fast timing filters are adapted to high-speed real-time acquisition and compared in a simulated scenario. A trapezoidal filter is implemented in firmware alongside the detection channel for pulse height analysis. Thanks to the use of reprogrammable devices, the system remains versatile and can be remotely adapted to different needs with no additional hardware costs.

关键词： Digital pulse processing field programmable gate array Data acquisition Data processing Pulse height analysis Energy resolution

来源：评论

学校读者我要写书评

暂无评论

Three-Port Bi-Directional DC-DC Converter with Solar PV System Fed BLDC Motor Drive Using FPGA

引用

ENERGIES 2023年第2期16卷 624页

作者： Udayakumar, Arun Kumar Raghavan, Raghavendra Rajan Vijaya Houran, Mohamad Abou Elavarasan, Rajvikram Madurai Kalavathy, Anushkannan Nedumaran Hossain, Eklas SRM Inst Sci & Technol Dept Elect & Elect Engn Ramapuram Campus Chennai 600089 Tamilnadu India Harman Connected Serv India Pvt Ltd Automot Dept Bengaluru 560095 Karnataka India Xi An Jiao Tong Univ Sch Elect Engn Xian 710049 Peoples R China Univ Queensland Sch Informat Technol & Elect Engn St Lucia Qld 4072 Australia Kathir Coll Engn Dept Elect & Commun Engn Coimbatore 641062 Tamilnadu India Boise State Univ Dept Elect & Comp Engn Boise ID 83725 USA

The increased need for renewable energy systems to generate power, store energy, and connect energy storage devices with applications has become a major challenge. Energy storage using batteries is most appropriate for energy sources like solar, wind, etc. A non-isolated three-port DC-DC-converter energy conversion unit is implemented feeding the brushless DCmotor drive. In this paper, a non-isolated three-port converter is designed and simulated for battery energy storage, interfaced with an output drive. Based on the requirements, the power extracted from the solar panel during the daytime is used to charge the batteries through the three-port converter. The proposed three-port converter is analyzed in terms of operating principles and power flow. An FPGA-based NI LabView PXI with SbRio interface is used to develop the suggested approach's control hardware, and prototype model results are obtained to test the proposed three-port converter control system's effectiveness and practicality. The overall efficiency of the converter's output improves as a result. The success rate is 96.5 percent while charging an ESS, 98.1 percent when discharging an ESS, and 95.7 percent overall.

关键词： BLDC motor DC-DC converter field programmable gate array MPP tracking solar PV system three-portconverter (TPC) energy storage device (ESD)

来源：评论

学校读者我要写书评

暂无评论

6 Gbps real-time optical quantum random number generator based on vacuum fluctuation

引用

REVIEW OF SCIENTIFIC INSTRUMENTS 2019年第4期90卷 043105-043105页

作者： Zheng, Ziyong Zhang, Yichen Huang, Weinan Yu, Song Guo, Hong Beijing Univ Posts & Telecommun State Key Lab Informat Photon & Opt Commun Beijing 100876 Peoples R China Peking Univ Sch Elect Engn & Comp Sci State Key Lab Adv Opt Commun Syst & Networks Beijing 100871 Peoples R China Peking Univ Ctr Quantum Informat Technol Beijing 100871 Peoples R China

We demonstrate a 6 Gbps real-time optical quantum random number generator by measuring vacuum fluctuation. To address the common problem that speed gap exists between fast randomness generation and slow randomness extraction in most high-speed real-time quantum random number generator systems, we present an optimized extraction algorithm based on parallel implementation of Toeplitz hashing to reduce the influence of classical noise due to the imperfection of devices. Notably, the real-time rate of randomness extraction we have achieved reaches the highest speed of 12 Gbps by occupying less computing resources, and the algorithm has the ability to support hundreds of Gbps randomness extraction. By assuming that the eavesdropper with complete knowledge of the classical noise, our generator has a randomness generation speed of 6.83 Gbps and this supports the generation of 6 Gbps information-theoretically provable quantum random numbers, which are output in real-time through peripheral component interconnect express interface. Published under license by AIP Publishing.

关键词： Data acquisition Electronic noise Quantum fluctuations field programmable gate array Entropy Quantum random number generators Quantum vacuum states Beam splitter Detectors Cryptography

来源：评论

学校读者我要写书评

暂无评论

A high-performance, hardware-based deep learning system for disease diagnosis

引用

PEERJ COMPUTER SCIENCE 2022年 8卷 e1034页

作者： Siddique, Ali Iqbal, Muhammad Azhar Aleem, Muhammad Lin, Jerry Chun-Wei Natl Univ Comp & Emerging Sci Lahore Campus Lahore Pakistan Univ Macau Taipa Macau Peoples R China Univ Lancaster Lancaster England Natl Univ Comp & Emerging Sci Islamabad Pakistan Western Norway Univ Appl Sci Bergen Norway

Modern deep learning schemes have shown human-level performance in the area of medical science. However, the implementation of deep learning algorithms on dedicated hardware remains a challenging task because modern algorithms and neuronal activation functions are generally not hardware-friendly and require a lot of resources. Recently, researchers have come up with some hardware-friendly activation functions that can yield high throughput and high accuracy at the same time. In this context, we propose a hardware-based neural network that can predict the presence of cancer in humans with 98.23% accuracy. This is done by making use of cost-efficient, highly accurate activation functions, Sqish and LogSQNL. Due to its inherently parallel components, the system can classify a given sample in just one clock cycle, i.e., 15.75 nanoseconds. Though this system is dedicated to cancer diagnosis, it can predict the presence of many other diseases such as those of the heart. This is because the system is reconfigurable and can be programmed to classify any sample into one of two classes. The proposed hardware system requires about 983 slice registers, 2,655 slice lookup tables, and only 1.1 kilobits of on-chip memory. The system can predict about 63.5 million cancer samples in a second and can perform about 20 giga-operations per second. The proposed system is about 5-16 times cheaper and at least four times speedier than other dedicated hardware systems using neural networks for classification tasks.

关键词： Activation function Cancer diagnosis Deep learning field programmable gate array Hardware friendly Neural networks Swish

来源：评论

学校读者我要写书评

暂无评论

Feedback compensated 10 kW solid-state pulsed power amplifier at 352 MHz for particle accelerators

引用

REVIEW OF SCIENTIFIC INSTRUMENTS 2019年第10期90卷 104707-104707页

作者： Duc, L. Hoang Jobs, M. Lofnes, T. Ruber, R. Olsson, J. Dancila, D. Angstrom Lab Div Solid State Elect S-75121 Uppsala Sweden Angstrom Lab Dept Phys & Astron S-75120 Uppsala Sweden

This paper presents the first results of an in-house developed low-level radio frequency (LLRF) system and a 10 kW solid state power amplifier (SSPA). The design approach for the SSPA is based on eight resonant single-ended kilowatt modules combined using a planar Gysel combiner. Each of the single-ended modules is based on a two-stepped impedance resonant matching, allowing for harmonic suppression, simple design for massive production, and high-performance design. A design methodology to tune SSPA modules for optimum combining efficiency is presented thoroughly in the time domain. We characterize the power droop due to capacitor banks in the time domain. In open loop of compensation, it is about 1 kW within the pulse of peak value 10 kW and a duration of 3.5 ms. This may lead to the beam instability of the accelerator as particles are not provided with the same energy during the pulse. By incorporating our LLRF system, it is demonstrated that the objective of amplitude and phase stability is satisfied, as required in the European Spallation Source proton accelerator. The presented design also offers the advantages of compact form factor, low complexity, and better performance. In closed loop compensation, the variation of amplitude (pulse droop) is measured on the order of 20 W, which is equivalent to 0.2% at 10 kW peak output power. (C) 2019 Author(s).

关键词： Particle accelerators Synchrotrons field programmable gate array Coaxial connectors Vacuum tubes Amplifiers Pulse generators Radiowave and microwave technology field effect transistors Transformer

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：