The optimization of crop harvesting processes for commonly cultivated crops is of great importance in the aim of agricultural industrialization. Nowadays, the utilization of machine vision has enabled the automated id...
详细信息
Large volumes of distributed energy resources (DERs), such as solar photovoltaic (PV) plants are integrated into the power distribution system due to increased awareness of climate change. These DERs introduce variabl...
Large volumes of distributed energy resources (DERs), such as solar photovoltaic (PV) plants are integrated into the power distribution system due to increased awareness of climate change. These DERs introduce variable and uncertain generation sources due to changing weather conditions. This makes operations and controls challenging and complex. To better understand and manage the dynamic nature of solar PV power plants, digital twins (DTs) will be needed. DTs based on artificial intelligence (AI) methods can be applied to replicate the dynamics of PV plants. This study utilizes a popular paradigm of AI - neural networks to create a variety of data-driven DT (DD-DT) prediction models for a 1 MW solar PV plant located at Clemson University in South Carolina, USA. State-of-the-art internet of things (IoT) based real-time measurements are used to develop the DD-DTs. Typical results for short-term PV power prediction for DTs implemented using multilayer perceptron neural networks (MLPNNs) and Elman recurrent neural networks (ERNNs) are presented in this paper.
Mobile edge computing (MEC) mitigates the energy and computation burdens on mobile users (MUs) by offloading tasks to the network edge. To optimize MEC server utilization through effective resource allocation, a well-...
详细信息
The lungs are integral to facilitating the respiratory processes crucial for human survival. However, lung infections can pose severe threats to human health. In medical diagnostics, numerous tools and techniques exis...
The lungs are integral to facilitating the respiratory processes crucial for human survival. However, lung infections can pose severe threats to human health. In medical diagnostics, numerous tools and techniques exist for assessing and diagnosing lung-related issues. Among these, Chest X-ray images have emerged as a widely favoured choice due to their distinct advantages, such as accessibility and the ability to provide valuable insights into pulmonary conditions. This work is dedicated to developing a robust multiclass category system for Chest X-ray images. A notable feature of this work is its emphasis on the efficient execution of the categorization work on a low-power heterogeneous embedded device, demonstrating the potential for practical applications in resource-constrained environments. One of this study's key highlights lies in exploring different optimization algorithms and their impact on category accuracy. The study conducts comprehensive experiments utilizing four prominent optimizers: Adam, Adamax, RMSprop, and SGD. Through these experiments, we observed that the proposed modified design attained the highest rate of accuracy, 97.16% when the Adam optimizer was employed. This outcome underscores the significance of optimizer selection in developing accurate and reliable diagnostic models for Chest X-ray image categorization, ultimately contributing to advancements in medical imaging and healthcare.
Direct policy search has achieved great empirical success in reinforcement learning. Recently, there has been increasing interest in studying its theoretical properties for continuous control, and fruitful results hav...
详细信息
This paper presents Visual Evaluative AI, a decision aid that provides positive and negative evidence from image data for a given hypothesis. This tool finds high-level human concepts in an image and generates the Wei...
详细信息
This paper presents a powerline energy harvesting circuit to power wireless sensor nodes for powerline safety monitoring. The magnetic harvester is a ring-shaped nano-crystalline magnetic core, which transforms the ac...
This paper presents a powerline energy harvesting circuit to power wireless sensor nodes for powerline safety monitoring. The magnetic harvester is a ring-shaped nano-crystalline magnetic core, which transforms the ac powerline current to ac voltage. The major building blocks of the circuit are a buck-boost converter operating in discontinuous conduction mode (DCM) and a microcontroller unit (MCU) for maximum power point tracking (MPPT). The MPPT algorithm based on the perturb and observe senses the current flowing into the load and adjusts the duty cycle of the buck-boost converter to match the source impedance. The magnetic core delivers 6.98 W to an optimal $200\ \Omega$ resistor directly attached to the core under the powerline current of 30 A. The output power of the proposed circuit is 4.86 W with the optimal load resistance of $R_{L}=250\ \Omega$ , resulting in the conversion efficiency of 70%.
In this paper, we present a dilated convolutional neural network-based deep learning system for the challenging problem of building accurate roof pixel detection from city aerial multispectral images in the visible an...
In this paper, we present a dilated convolutional neural network-based deep learning system for the challenging problem of building accurate roof pixel detection from city aerial multispectral images in the visible and infrared spectral bands. The proposed system uses an automatic spectral band selection strategy and an automatic histogram-based threshold-finding method to improve the network's prediction accuracy performance. Current existing systems train a single model on all the cities of interest data at once. These systems also usually use only RGB bands and avoid post-processing of the predicted images for increased prediction accuracies. In this paper, we present the following solutions to these object detection challenges, 1) we propose and implement a separate model for each city dataset, 2) we propose an optimal band selection strategy to choose the most informative visible-infrared bands from multispectral images, 3) we propose an adaptive and automatic histogram-based optimal threshold finding technique for building roof object segmentation. Our results show that having a separate model for each city rather than only one model for all the combined cities increases the system's performance by 20.77 % in building roof pixel prediction in any of the cities considered. We show that our system also has a 35.6% increase in F1 score compared to its state-of-the-art counterpart, the winning implementation of the SpaceNet building detection national challenge, which took place in 2017.
Ease of calibration and high-accuracy task-space state-estimation purely based on onboard sensors is a key requirement for enabling easily deployable cable robots in real-world applications. In this work, we incorpora...
Ease of calibration and high-accuracy task-space state-estimation purely based on onboard sensors is a key requirement for enabling easily deployable cable robots in real-world applications. In this work, we incorporate the onboard camera and kinematic sensors to drive a statistical fusion framework that presents a unified localization and calibration system which requires no initial values for the kinematic parameters. This is achieved by formulating a Monte-Carlo algorithm that initializes a factor-graph representation of the calibration and localization problem. With this, we are able to jointly identify both the kinematic parameters and the visual odometry scale alongside their corresponding uncertainties. We demonstrate the practical applicability of the framework using our state-estimation dataset recorded with the ARAS-CAM suspended cable driven parallel robot, and published as part of this manuscript.
Empirical studies have shown that real-life queueing systems, such as contact centers, exhibit non-Markovian and nonstationary behaviors. Consequently, analyzing their performance poses significant challenges. In this...
详细信息
ISBN:
(数字)9798331534202
ISBN:
(纸本)9798331534219
Empirical studies have shown that real-life queueing systems, such as contact centers, exhibit non-Markovian and nonstationary behaviors. Consequently, analyzing their performance poses significant challenges. In this paper, we propose a simulation-based autoregressive deep learning algorithm (SADLA) for predicting service levels in non-Markovian, nonstationary queueing systems. Our method leverages modern recurrent neural networks, which are trained on synthetic data to capture the intrinsic spatio-temporal characteristics of queueing systems. Our findings demonstrate that SADLA achieves high prediction accuracy while reducing computational complexity by six orders of magnitude compared to traditional simulation methods. The implications of our research extend beyond accurate queue performance analysis; by embracing the learning capabilities of neural networks, our approach contributes to the advancement of the overall performance and resilience of real-life service systems.
暂无评论