Excessive Computational Load Research Articles

Toward efficient and lightweight sea–land segmentation for remote sensing images

Sea–land segmentation is of great significance for autonomous coastline monitoring, which is fundamental research in the remote sensing community. Due to the diverse contents and easily confused sea–land boundaries contained in remote sensing images, it is always challenging to achieve precise sea–land segmentation for complex scenarios. Although existing deep learning-based methods have exhibited promising performance, excessive computational load and insufficient use of hierarchical features remain unresolved. In this paper, we contribute to addressing the problems by developing an efficient and lightweight convolutional neural network (CNN) termed E-Net. On the one hand, the proposed network adopts a novel E-shaped architecture that reforms the conventional U-codec structure to make full use of hierarchical features at different depths, so that the sea–land segmentation effect can be significantly improved without excessive computational load. On the other hand, a contextual aggregation attention mechanism (CA2M) is designed to further facilitate efficient aggregation and transmission of contextual information, so that the fuzzy and irregular sea–land boundaries can be accurately distinguished. Extensive experiments reveal that our approach not only produces superior sea–land segmentation effect but also demonstrates promising computational efficiency. Specifically, the proposed E-Net achieves state-of-the-art sea–land segmentation performance with 92.78% and 93.62% mean Intersection over Union (mIoU) on the SLSD and HRSC2016 datasets, respectively, while the frames per second (FPS) reaches 108.032 with as low as 52.287G floating point operations per second (FLOPs).

An enhanced direct position determination of non-circular sources via sparse Bayesian inference and grid refinement strategy

In this study, an off-grid sparse Bayesian inference (OGSBI) based direct position determination (DPD) algorithm is investigated. Existing SBI-based DPD algorithms are confronted with the challenge of excessive computational loads and lack of consideration for non-circular (NC) signals. To address these limitations, we present an enhanced OGSBI-based DPD algorithm for multiple non-circular sources. By utilizing the conjugate information of the NC signals, we expand the dimensionality of the data matrix to achieve a significant improvement in the localization performance. Additionally, a grid refinement strategy is developed to alleviate the computational loads, which involves an initial search to determine the approximate source locations, followed by fine localization using a denser grid. Moreover, the computational complexity and Cramér-Rao lower bound are derived to provide a comprehensive analysis of the proposed algorithm. Numerical simulations demonstrate the superiority of the proposed algorithm in terms of both localization accuracy and computational efficiency.

A Novel Symbol Detection Strategy based on Discrete Elephant Herding Optimization Algorithm for MIMO-FBMC/OQAM System

The rapidly growing technological enhancements in wireless communication have pushed the researchers to develop various transmission schemes in recent decades. Multiple-input multiple-output - filter bank multicarrier/offset quadrate amplitude modulation (MIMO-FBMC/OQAM) is one of these schemes with exclusive features. However, unless an efficient symbol detector is employed, we cannot fully benefit from the superior features of MIMO-FBMC/OQAM system as the symbol vectors cannot be detected accurately at the receiver side. An ideal symbol detector is expected to have the capability of recovering the symbol vectors in the most accurate way without causing too much enhancement in the computational complexity. To meet this expectation, we have modified the conventional maximum likelihood (ML) strategy. Instead of using an exhaustive search procedure, which guarantees the highest detection performance but causes an excessive computational load in the ML scheme, an efficient metaheuristic algorithm called discrete elephant herding optimization (DEHO) was employed for optimizing the symbol vectors. With the integration of DEHO to the conventional ML in the aforementioned way, we have not only obtained a considerable reduction in the complexity of ML, but also reached the near-optimal symbol detection performance by outperforming the other symbol detectors considered in this paper.

Research on Forest Flame Detection Algorithm Based on a Lightweight Neural Network

To solve the problem of the poor performance of a flame detection algorithm in a complex forest background, such as poor detection performance, insensitivity to small targets, and excessive computational load, there is an urgent need for a lightweight, high-accuracy, real-time detection system. This paper introduces a lightweight object-detection algorithm called GS-YOLOv5s, which is based on the YOLOv5s baseline model and incorporates a multi-scale feature fusion knowledge distillation architecture. Firstly, the ghost shuffle convolution bottleneck is applied to obtain richer gradient information through branching. Secondly, the WIoU loss function is used to address the issues of GIoU related to model optimization, slow convergence, and inaccurate regression. Finally, a knowledge distillation algorithm based on feature fusion is employed to further improve its accuracy. Experimental results based on the dataset show that compared to the YOLOv5s baseline model, the proposed algorithm reduces the number of parameters and floating-point operations by approximately 26% and 36%, respectively. Moreover, it achieved a 3.1% improvement in mAP0.5 compared to YOLOv5s. The experiments demonstrate that GS-YOLOv5s, based on multi-scale feature fusion, not only enhances detection accuracy but also meets the requirements of lightweight and real-time detection in forest fire detection, commendably improving the practicality of flame-detection algorithms.

Comfort, carbon emissions, and cost of building envelope and photovoltaic arrangement optimization through a two-stage model

The design of a high-performance building necessitates tradeoffs among multiple performance objectives, and many simulation-based optimization tools have been developed for this purpose. In practice, these tools are often constrained by excessive computational load and tight project deadlines. This study aimed to develop a holistic approach to rapidly identify optimal building design schemes. A novel two-stage model was developed as a surrogate to conventional physics-based models, which were then linked to multi-objective optimization algorithms, namely, NSGA-II, NSGA-III, and C-TAEA, in search of the Pareto optimal and best design schemes. The performance of the two-stage model was further enhanced by applying the least absolute shrinkage and selection operator (LASSO) and Neural Architecture Search methods and learning from the statistical layer. The above approach was tested in the design of an apartment building for seniors in Northern China such as thermal comfort, carbon, and cost. The optimal design achieved through the compromise optimum can significantly reduce thermal discomfort, life-cycle carbon emissions, and high levels of daylighting conditions. The design scheme was visualized using a geometrical remodel module. This study contributes methodologically to complex multi-objective building optimization problems with practical implications for design.

Learning-based model predictive energy management for fuel cell hybrid electric bus with health-aware control

Advanced energy management strategy (EMS) can ensure healthy, stable, and efficient operation of the on-board energy systems. Model Predictive Control (MPC) and Deep Reinforcement Learning (DRL) are two powerful control methods that have been extensively researched in the field of vehicle energy management. However, there are some problems with both approaches. On the one hand, MPC is difficult to cope with the complex systems and the excessive computational load caused by the non-linear solving over long prediction horizon, on the other hand, DRL lacks adaptability to different driving conditions and is poorly interpretable. Therefore, this paper innovatively proposes a learning-based model predictive (L-MPC) EMS for fuel cell hybrid electric bus (FCHEB) with health-aware control. This method effectively merges the advantages of control theory and machine learning. Specifically, firstly, the precise aging models for vehicular energy systems are established and incorporated into the optimization framework along with hydrogen consumption related metrics. Secondly, the principles of the learning-based MPC algorithm are thoroughly elucidated. In addition, to ensure driving details under future conditions, a speed predictor based on a double-layer Bi-directional Long Short-Term Memory (BiLSTM) is proposed at the strategy supervision layer. Finally, the superiority of the proposed strategy in prolonging the lifespan of the energy systems and reducing overall vehicle operating cost is verified by comprehensive comparisons with state-of-the-art MPC and DRL methods under real-world collected driving condition.

Stochastic mirror descent method for linear ill-posed problems in Banach spaces

Consider linear ill-posed problems governed by the system for , where each A i is a bounded linear operator from a Banach space X to a Hilbert space Y i . In case p is huge, solving the problem by an iterative regularization method using the whole information at each iteration step can be very expensive, due to the huge amount of memory and excessive computational load per iteration. To solve such large-scale ill-posed systems efficiently, we develop in this paper a stochastic mirror descent method which uses only a small portion of randomly selected equations at each iteration step and incorporates convex regularization terms into the algorithm design. Therefore, our method scales very well with the problem size and has the capability of capturing features of sought solutions. The convergence property of the method depends crucially on the choice of step-sizes. We consider various rules for choosing step-sizes and obtain convergence results under a priori early stopping rules. In particular, by incorporating the spirit of the discrepancy principle we propose a choice rule of step-sizes which can efficiently suppress the oscillations in iterates and reduce the effect of semi-convergence. Furthermore, we establish an order optimal convergence rate result when the sought solution satisfies a benchmark source condition. Various numerical simulations are reported to test the performance of the method.

CHC-UNet: Cascaded Heterogeneous Convolution UNet Framework for Adaptive Image Dehazing

Haze is a phenomenon of degrading the visual quality of outdoor images by wet aerosols and particles suspended in the atmosphere under poor weather conditions. The incident light on to the object will be scattered and causes the distortion of color, contrast reduction and generate the halo artifacts. In this article, a Cascaded Heterogeneous Convolution UNet (CHC-UNet) with redefined Squeeze-and-Excitation (SE) mechanism framework is designed to reduce the adverse hazes’ effect in the images. The Heterogeneous Convolution kernels with size 3×3 and 1×1 are used in each block of UNet to extract the features at various image resolutions and to overcome excessive computation load. The adaptive residual connections preserve the original image features, prevent from overfitting problem and make the dehazing process more thorough. The SE Algorithm is re-defined to better represent the channel-wise image features with Local Feature Fusion (LFF) module. The qualitative and quantitative findings of the ablation experiments show that the suggested framework outperforms state-of-the-art methodologies in the examined literature using the benchmark I-Haze, O-Haze, and D-Hazy datasets. The proposed framework has an average improvement from 17.95 dB to 27.52 dB in Peak Signal-to-Noise Ratio (PSNR), from 0.56 to 0.87 in Structural SIMilarity (SSIM) index on retrieved images, from 4.98 to 3.12 in Naturalness Image Quality Evaluator (NIQE). Furthermore, the framework has varying degrees of improvement in the subjective visual effects of the dehazed output, and it also has the added advantage of optimum running time in comparison with the existing dehazing methods.

Test Results of Variance Reduction Techniques Applied to Deep Penetration Problem

Nowadays, there is a problem of a lack of computer power to conduct high-precision reactor analysis. There are several factors that increase the excessive computational load and make it difficult to calculate nuclear reactor full-scale models using Monte Carlo method. Among them is the large neutron-flux attenuation, which is observed in deep penetration problems. Various reduction techniques are used to increase the efficiency of Monte Carlo calculations in such tasks. It allows to reduce the statistical uncertainty of the functional evaluation without increasing the number of neutron histories. This work is devoted to study and testing of variance reduction techniques in the deep penetration problem. To demonstrate the possibility of using non-analogue Monte Carlo modeling a test problem was formulated. To quantify the efficacy of applying the variance reduction methods, FOM characteristic is considered, which is a function of the relative error in a flux estimate and the computational time of the simulation. The article considers non-analogue techniques implemented in the MCU (Monte Carlo Univarsal) and OpenMC codes. As part of the study, a module of the OpenMC code was developed that allows to automatically generate weight windows. It is shown that variance reduction techniques increase the calculation efficiency by several times, particularly, the weight windows method in OpenMC make it possible to achieve a 7-fold increase in the efficiency of neutron flux estimation with the same number of simulated histories. The formulated recommendations can be used in the calculation of innovative nuclear reactors full-scale models.

Cluster Analysis and Model Comparison Using Smart Meter Data.

Load forecasting plays a crucial role in the world of smart grids. It governs many aspects of the smart grid and smart meter, such as demand response, asset management, investment, and future direction. This paper proposes time-series forecasting for short-term load prediction to unveil the load forecast benefits through different statistical and mathematical models, such as artificial neural networks, auto-regression, and ARIMA. It targets the problem of excessive computational load when dealing with time-series data. It also presents a business case that is used to analyze different clusters to find underlying factors of load consumption and predict the behavior of customers based on different parameters. On evaluating the accuracy of the prediction models, it is observed that ARIMA models with the (P, D, Q) values as (1, 1, 1) were most accurate compared to other values.

Advanced electromagnetic modeling of large-scale high-temperature superconductor systems based on H and T-A formulations

The development of the high-temperature superconductors (HTS) has allowed the emergence of diverse superconductor devices. Some of these devices, like wind power generators and high-field magnets, are classified as large-scale HTS systems, because they are made of several hundreds or thousands of turns of conductors. The electromagnetic analysis of such systems cannot be addressed by means of the available analytical models. The finite-element method has been extensively used to solve the H formulation of the Maxwell’s equations, thus far with great success. Nevertheless, its application to large scale HTS systems is still hindered by excessive computational load. The recently proposed T-A formulation has allowed building more efficient models for systems made of HTS tapes. Both formulations have been successfully applied in conjunction with the homogenization and multi-scaling methods, these advanced methods allow reducing the required computational resources. A new advanced method, called densification, is proposed here. The most important contribution of this article is the comprehensive comparison of the strategies emerged from the combined use of the two formulations and the three advanced methods.

Advanced three-dimensional electromagnetic modelling using a nested integral equation approach

SUMMARY Most of the existing 3-D electromagnetic (EM) modelling solvers based on the integral equation (IE) method exploit fast Fourier transform (FFT) to accelerate the matrix–vector multiplications. This in turn requires a laterally uniform discretization of the modelling domain. However, there is often a need for multiscale modelling and inversion, for instance, to properly account for the effects of non-uniform distant structures and, at the same time, to accurately model the effects from local anomalies. In such scenarios, the usage of laterally uniform grids leads to excessive computational loads, in terms of both memory and time. To alleviate this problem, we developed an efficient 3-D EM modelling tool based on a multinested IE approach. Within this approach, the IE modelling is first performed at a large domain and on a (laterally uniform) coarse grid, and then the results are refined in the region of interest by performing modelling at a smaller domain and on a (laterally uniform) denser grid. At the latter stage, the modelling results obtained at the previous stage are exploited. The lateral uniformity of the grids at each stage allows us to keep using the FFT for the acceleration of matrix–vector multiplications. An important novelty of the paper is the development of a ‘rim domain’ concept that further improves the performance of the multinested IE approach. We verify the developed tool on both idealized and realistic 3-D conductivity models, and demonstrate its efficiency and accuracy.

Fast Modelling and Identification of Hydraulic Brake Plants for Automotive Applications

Diffusion of electric and hybrid vehicles is accelerating the development of innovative braking technologies. Calibration of accurate models of a hydraulic brake plant involves availability of large amount of data whose acquisition is expensive and time consuming. Also, for some applications, such as vehicle simulators and hardware in the loop test rig, a real-time implementation is required. To avoid excessive computational loads, usage of simplified parametric models is almost mandatory. In this work, authors propose a simplified functional approach to identify and simulate the response of a generic hydraulic plant with a limited number of experimental tests. To reproduce complex nonlinear behaviours that are difficult to be reproduced with simplified models, piecewise transfer functions with scheduled poles are proposed. This innovative solution has been successfully applied for the identification of the brake plant of an existing vehicle, a Siemens prototype of instrumented vehicle called SimRod, demonstrating the feasibility of proposed method.

Time Synchronization Between EtherCAT Network and External Processor

We employ an external processor to distribute the master computing load in the EtherCAT networked control system and develop a method for time synchronization between the Linux on the external processor and EtherCAT network. The synchronization between the Linux on the external processor and EtherCAT master is performed by using general-purpose input/output and shared memory, and the synchronization accuracy is significantly improved by compensating for the system time setting latency and general-purpose input detection latency in the EtherCAT master. The developed method is implemented as part of the master application without modifying the EtherCAT protocol or requiring an excessive computational load. By performing extensive experiments and analyzing the results, we demonstrate that the developed method achieves the synchronization error means less than 1 μs with standard deviations less than 5 μs between the external processor and EtherCAT network. This synchronization accuracy can be acceptable for various networked control applications based on Linux.

Deep learning-based video surveillance system managed by low cost hardware and panoramic cameras

The design of automated video surveillance systems often involves the detection of agents which exhibit anomalous or dangerous behavior in the scene under analysis. Models aimed to enhance the video pattern recognition abilities of the system are commonly integrated in order to increase its performance. Deep learning neural networks are found among the most popular models employed for this purpose. Nevertheless, the large computational demands of deep networks mean that exhaustive scans of the full video frame make the system perform rather poorly in terms of execution speed when implemented on low cost devices, due to the excessive computational load generated by the examination of multiple image windows. This work presents a video surveillance system aimed to detect moving objects with abnormal behavior for a panoramic 360∘ surveillance camera. The block of the video frame to be analyzed is determined on the basis of a probabilistic mixture distribution comprised by two mixture components. The first component is a uniform distribution, which is in charge of a blind window selection, while the second component is a mixture of kernel distributions. The kernel distributions generate windows within the video frame in the vicinity of the areas where anomalies were previously found. This contributes to obtain candidate windows for analysis which are close to the most relevant regions of the video frame, according to the past recorded activity. A Raspberry Pi microcontroller based board is employed to implement the system. This enables the design and implementation of a system with a low cost, which is nevertheless capable of performing the video analysis with a high video frame processing rate.

Predicting medical image registration error with block-matching using three orthogonal planes approach

Estimating registration error is a crucial task in medical image registration. Although successful algorithms have been proposed, non-trivial challenges still exist such as lacking available data with ground truth and excessive computational load sourcing from the processing of 3D image volume. This paper presents an algorithm that can estimate target registration error accurately with fewer computations by using block-matching on three orthogonal planes rather than on a 3D volume. Furthermore, we present a way of using stereo datasets and stereo matching algorithms for predicting medical image registration error. Both contributions alleviate two crucial problems (lacking ground truth and excessive computational load) in the field of error estimation for medical image registration.

Accuracy Improvement of Master–Slave Synchronization in EtherCAT Networks

In the study, we develop a method to improve the accuracy of master-slave synchronization in EtherCAT networks. The method involves two key compensations that are not considered in the EtherCAT protocol. First, the propagation delay between the master and reference slave is measured, and the system time of the reference slave is then compensated for the measured propagation delay. Second, the bias component of the synchronization error between the master and reference slave is periodically estimated, and system time of each slave is then compensated for the estimated bias component. The entire method is implemented as part of the master application without modifying the EtherCAT protocol or requiring excessive computation load or large memory space. Thus, the developed method is advantageous because it can be immediately applied at low cost to existing EtherCAT networks. By performing extensive experiments based on a Linux-based EtherCAT master, we demonstrate that the developed method significantly improves the accuracy of the master-slave synchronization in EtherCAT networks.

Resource Aware Hybrid Energy Harvesting for Wireless Sensor Networks

Sensors consume the resources to perform different operations, and energy of the nodes may be depleted due to excessive computational load; thus, may reduce the overall network lifespan as well as coverage area. Traditional energy harvesting schemes provides energy to the nodes in linear way but these schemes depend over a single source as well as these do not interact with the routing protocol. In this paper, a Hybrid Energy Harvester scheme for wireless sensor network is introduced which can utilize multiple energy sources for harvesting and also interact with the routing protocols to fulfill their energy requirements. Simulation based analysis using various protocols are performed under the QoS constraints.

A hierarchical coordinated demand response control for buildings with improved performances at building group

Demand response control is one of the common means used for building peak demand limiting. Most of the existing demand response controls focused on single building’s performance optimization, and thus may cause new undesirable peak demands at building group, imposing stress on the grid power balance and limiting the economic savings. A few latest studies have demonstrated the potential benefits of demand response coordination, but the proposed methods cannot be applied in large scales. The main reason is that, for demand response coordination of multiple buildings, associated computational load and coordination complexity, increasing exponentially with building number, are challenges to be solved. This study, therefore, proposes a hierarchical demand response control to optimize operations of a large scale of buildings for group-level peak demand reduction. The hierarchical control first considers the building group as a ‘virtual’ building and searches the optimal performance that can be achieved at building group using genetic algorithm. To realize such optimal performance, it then coordinates each single building’s operation using non-linear programming. For validations, the proposed method has been applied on a case building group, and the study results show that the hierarchical control can overcome the challenges of excessive computational load and complexity. Moreover, in comparison with conventional independent control, it can achieve better performances in aspects of peak demand reduction and economic savings. This study provides a coordinated control for application in large scales, which can improve the effectiveness and efficiency in relieving the grid stress, and reduce the end-users’ electricity bills.

Cooperative Network Operation Design for Mobility-Aware Cloud Radio Access Network

Ultra-dense small cells operating in a cooperative manner, such as cloud radio access network (C-RAN), have been introduced to serve the numerous mobile devices in the next generation wireless networks. One of the major challenges in efficiently operating the C-RAN in a cooperative manner is the excessive overhead signaling and computation load, such as downlink channel state information (CSI) acquisition at the cloud for cooperative downlink transmission, which scales rapidly with the size of the network. In this paper, the exploitation of mobility information of devices is proposed to address the challenge of efficiently operating the C-RAN. We introduce a mobility-assisted CSI acquisition method to complement conventional pilot-based CSI acquisition methods. This method enables the C-RAN to avoid excessive overhead signaling. A low-complexity algorithm is designed to maximize the sum rate of all devices subject to the limits of backhaul capacity and base station transmit power. An adaptive algorithm has also been proposed to improve the robustness against channel uncertainty. Both theoretical and numerical analyses show that the exploitation of mobility information provides a new dimension to improve conventional transmission schemes for next-generation massive cooperative networks.