基于膜计算和蚁群算法的融合算法在云计算资源调度中的研究

针对云计算下的资源调度的问题,提出将蚁群算法的个体与云计算中的可行性资源调度进行对应,首先对云计算资源调度进行描述,其次针对蚁群算法的路径选择引入了平衡因子,对信息素进行了局部研究和全局研究,将蚁群个体引入到膜计算中,通过膜内运算和膜间运算,提高了算法的局部和全局收敛的能力,最后在云计算资源分配中,引入匹配表概念,将云计算任务和资源进行匹配,融合后的算法提高了算法的整体性能.仿真实验说明在网络消耗,成本消耗,能量消耗上有了明显的降低,提高了资源分配效率。     

基于改进的布谷鸟算法在云计算资源的研究

如何进行更好地资源调度一直都是云计算研究的热点,本文在云计算资源算法中引入布谷鸟算法,针对布谷鸟算法中出现的收敛速度快,容易局部震荡等现象,本文首先引入高斯变异算子来处理每一个阶段中的鸟窝最佳位置的选择,然后通过自适应动态因子来调整不同阶段中的鸟窝位置的选择,使得改进后的算法收敛精度提高,通过适应度函数的平衡以及遗传算法中的三种操作,使得本文算法能够有效的提高云计算环境下的资源分配效率,降低了网络消耗。在Cloudsim平台仿真实验中,通过三个方面的比较,本文算法在性能上、资源调度效率和任务调度方面都有很大改进,有效提高了云计算系统的资源调度能力。     

基于关键路径链和PACGA的网络总线资源调度方法研究

针对传统工业控制网络总线资源调度算法在节点数量逐渐增加时收敛速度慢和搜索精度不高,且准确度及效率低等问题, 提出了一种基于关键路径链和多态蚁群遗传算法(PACGA)的资源调度方法,采用关键路径链的调度算法获取需求调度的节点,不同节点间采用多态蚁群遗传算法进行资源的调度,依据照工业控制网络资源调度的特征,用自适应调整挥发系数增强节点的全局搜索性能,通过候选节点集方法缩小搜索区域提高算法的搜索效率,完成工业控制网络总线资源的高效调度。仿真实验说明,该种方法在工业控制过程中任务数量较多的情况下仍然具备较高的运行效率和精度,并且具有较低的运行时间,具有较强的应用价值。     

Autonomous DRL-based energy efficient VM consolidation for cloud data centers

The exponential data growth and demands for high computing resources lead to excessive resource use in cloud data centers, which cause an increase in energy consumption and high carbon emissions in the environment. So, the high energy consumption, inefficient resource usage, and quality of service assurance (QoS) are major challenges for cloud data centers. The dynamic consolidation of Virtual Machines (VMs) is proven to be an efficient way to tackle these issues while reducing energy consumption and improving resource utilization in data centers. It reduces the number of active hosts for energy efficiency by switching under-utilized and idle hosts into lower power mode. So, several heuristics and Artificial Intelligence (AI) based VM consolidation approaches have been published in papers. Most existing approaches rely on the aggressive consolidation of VMs for energy efficiency, thus causing performance degradation and high SLA violation. However, an automated solution is needed to reduce energy consumption and SLA violation by ensuring efficient resource usage in the cloud data center environment. Therefore, this article proposes an energy-efficient autonomous VM consolidation (AVMC) mechanism that has Deep Reinforcement Learning (DRL) based agent for performing VM consolidation decisions. The DRL agent learns the optimal distribution of VMs in the data center, considering energy efficiency and QoS assurance. The real-time workload traces from PlanetLab have been used to validate the proposed mechanism. Experimental results reveal the superiority of the proposed AVMC system over the existing models. AVMC reduced the energy consumption and SLA violation rate significantly.     

基于深度残差网络的脱机手写汉字识别研究

摘 要: 手写汉字识别是模式识别与机器学习的重要研究方向和应用领域。近年来,随着深度学习理论方法的完善、新技术的层出不穷,深度神经网络在图像识别分类、图像生成等典型应用中取得了突破性的进展,其中,深度残差网络作为最新的研究成果,已成功应用于手写数字识别、图片识别分类等多个领域。本文将研究深度残差网络在脱机孤立手写汉字识别中的应用方法,通过改进残差学习模块的单元结构,优化深度残差网络性能,同时通过对训练集的预处理,从数据层面实现训练生成模型性能的提升,最后设计实验,验证深度残差网络、End-to-End模式在脱机手写汉字识别中的可行性,分析、总结存在的问题及今后的研究方向。     

邻域搜索蚁群算法在无线传感器网络数据融合中的应用

为了减少网络中的数据传输量,提高数据融合率,降低网络延时,针对无线传感器网络数据融合问题的研究,提出了一种邻域搜索蚁群算法。首先利用蚁群算法寻找最短路径的优势,构造最短路径。为了避免蚁群算法的早熟收敛和收敛速度慢的问题,当达到一定的迭代次数后,运用具有可变邻域搜索的变异算子对搜索结果进行优化。算法不但考虑了无线传感器网络节点能量消耗也考虑了数据传输的网络延时问题。实验结果表明,该算法减少了网络能耗,降低了网络延时,稳定性更好,性能更优。     

云计算环境下资源调度系统设计与实现

在云计算环境下,对开放的网络大数据库信息系统中的数据进行优化调度,提高数据资源的利用效率和配置优化能力。传统的资源调度算法采用资源信息的自相关匹配方法进行资源调度,当数据传输信道中的干扰较大及资源信息流的先验数据缺乏时,资源调度的均衡性不好,准确配准度不高。提出一种基于云计算资源负载均衡控制和信道自适应均衡的资源调度算法,并进行调度系统的软件开发和设计。首先构建了云计算环境下开放网络大数据库信息资源流的时间序列分析模型,采用自适应级联滤波算法对拟合的资源信息流进行滤波降噪预处理,提取滤波输出的资源信息流的关联维特征,通过资源负载均衡控制和信道自适应均衡算法实现资源调度改进。仿真结果表明,采用资源调度算法进行资源调度系统的软件设计,提高了资源调度的信息配准能力和抗干扰能力,计算开销较小,技术指标具有优越性。     

一种采用种群多样性监控和实时更新策略的粒子群优化算法

针对粒子群算法优化后期容易出现早熟收敛问题,建立一种具有种群多样性监测和实时更新策略的改进方法.首先建立种群健康度指标用来评价粒子群进化状态;其次提出随机扰动策略和离心搜索策略用于丰富粒子群的种群多样性,增强算法的全局搜索能力,并提出梯度搜索策略用于精确、高效地搜寻当前邻域内的局部极值点,提高算法的计算效率.最后建立种群健康度反馈机制,使粒子可以实时感知种群的健康程度,并自适应地采用不同的粒子更新策略,保证粒子群处于健康进化水平.将新方法应用于优化实例,并与其它改进方法进行性能比较,结果验证了新方法的有效性.     

Investigation of acoustic and geometric effects on the sonoreactor performance

In this work, three design configurations of a sonoreactor are considered under various operating conditions, and the acoustic characteristics during water sonication are investigated while using an immersed-type ultrasonic flat transducer probe in a sonoreactor model. Numerical models are also developed to simulate the sonication process, and they are successfully validated and compared with available data in the literature. Several sets of numerical investigations are conducted using the finite-element method and solved by the computational acoustics module in the COMSOL Multiphysics. The effects of the acoustical and geometrical parameters are investigated, analyzed, and reported, including the ultrasonic frequency, acoustic intensity, and scaling-up the reactor. The present study includes a parametric investigation examining the change of the ultrasonic frequency, intensity, and probe immersion depth on the performance. The results of the parametric study show that the highest cavitation energy corresponds to the maximum magnitude of negative pressure that takes place in the range of 60–80 kHz. The cavitation energy analyses are conducted under the conditions of 20 kHz of frequency and at 36 W input power. It is found that the cavitation energy of 15.87 W could produce 2.98 × 10⁻¹⁰ mol/J of sonochemical efficiency. In addition, the effect of altering the transducer probe depth changes the acoustic pressure field insignificantly. Furthermore, a recommendation is made to improve the sonochemical efficiency by introducing more considerable ultrasound input power while operating the sonoreactor at an ultrasonic frequency lower than 60 kHz. The results presented in this paper provide a comprehensive assessment of different sonoreactors and the feasibility of scaling-up their production rate.     

Effect of electron and proton irradiation on recombination centers in GaAsN grown by chemical beam epitaxy

Deep level transient spectroscopy (DLTS) was deployed to study the evolution, upon electron irradiation and hydrogenation of GaAsN grown by chemical beam epitaxy, of the main nitrogen-related nonradiative recombination center (E1), localized at 0.33 eV below the bottom edge of the conduction band of the alloy. On one hand, the electron irradiation was found to enhance the density of E1 depending on the fluence dose. On the other hand, the hydrogenation was found to passivate completely E1. Furthermore, two new lattice defects were only observed in hydrogenated GaAsN films and were suggested to be in relationship with the origin of E1. The first defect was an electron trap at average thermal activation energy of 0.41 eV below the CBM of GaAsN and was identified to be the EL5-type native defect in GaAs, originating from interstitial arsenic (As_i). The second energy level was a hole trap, newly observed at average thermal activation energy of 0.11 eV above the valence band maximum of the alloy and its origin was tentatively suggested to be in relationship with the monohydrogen–nitrogen (N–H) complex. As the possible origin of E1 was tentatively associated with the split interstitial formed from one N atom and one As atom in single V-site [(N–As)_As], we strongly suggested that the new hole trap took place after the dissociation of E1 and the formation of N–H complex.     

A highly efficient 5 kW peak power Nd:YAG laser with time-shared fiber optic beam delivery

Here, we report on the development of an efficient, high peak power lamp pumped Nd:YAG laser with time-shared fiber optic beam delivery. A maximum average output power of 270 W with 100 J maximum pulse energy and 5 kW peak power has been achieved with an electrical to laser conversion efficiency of 5.4%, which is on higher side for typical lamp pumped solid-state lasers. We have improved efficiency by spectral conversion and water flow optimization in the pump cavity, with a resulting beam quality comparable to commercial systems of similar power level. The resonator has been designed for stable operation from single-shot to 200 Hz repetition rate. A study of pulse-to-pulse laser energy stability for different resonator configurations has also been performed. The resonator was designed to achieve a good beam quality for the whole range of operation with a maximum beam parameter product of 15 mm mrad (M²45). A simple mechanism for time-shared fiber optic port selection has also been devised. Material processing applications such as cutting of stainless steel sheets up to 14 mm thickness and welding of metals such as carbon steel with weld depths up to 2 mm using the developed laser system has also been reported.     

DLTS investigation of GaP under hydrostatic pressure

Abstract

Deep Level Transient Spectroscopy (DLTS) was applied to nitrogen related deep electron trap 0.4 eV in green emitting diodes of GaP under hydrostatic pressure. The pressure coefficient of the level energy is determinated as equal -31 meV/kbar with respect to the valence band edge.     

Improved Deep Q-Network for User-Side Battery Energy Storage Charging and Discharging Strategy in Industrial Parks

Battery energy storage technology is an important part of the industrial parks to ensure the stable power supply, and its rough charging and discharging mode is difficult to meet the application requirements of energy saving, emission reduction, cost reduction, and efficiency increase. As a classic method of deep reinforcement learning, the deep Q-network is widely used to solve the problem of user-side battery energy storage charging and discharging. In some scenarios, its performance has reached the level of human expert. However, the updating of storage priority in experience memory often lags behind updating of Q-network parameters. In response to the need for lean management of battery charging and discharging, this paper proposes an improved deep Q-network to update the priority of sequence samples and the training performance of deep neural network, which reduces the cost of charging and discharging action and energy consumption in the park. The proposed method considers factors such as real-time electricity price, battery status, and time. The energy consumption state, charging and discharging behavior, reward function, and neural network structure are designed to meet the flexible scheduling of charging and discharging strategies, and can finally realize the optimization of battery energy storage benefits. The proposed method can solve the problem of priority update lag, and improve the utilization efficiency and learning performance of the experience pool samples. The paper selects electricity price data from the United States and some regions of China for simulation experiments. Experimental results show that compared with the traditional algorithm, the proposed approach can achieve better performance in both electricity price systems, thereby greatly reducing the cost of battery energy storage and providing a stronger guarantee for the safe and stable operation of battery energy storage systems in industrial parks.     

Analysis of current loss from a series-parallel combination of dye-sensitized solar cells using electrochemical impedance spectroscopy

Dye-sensitized solar cells (DSCs) have been proposed as a substitute for silicon crystalline solar cells which have a high manufacturing cost but it is still difficult to fabricate highly efficient DSC module assemblies. Therefore, in this work, an externally connected module assembly was investigated for industrial applications of DSCs. The equivalent circuit of a DSC was determined using typical electrical components and the cause of a current loss in the parallel connection was analyzed using electrochemical impedance spectroscopy. Also, an externally connected module has been constructed using 50 DSCs, where each cell has an active area of 8 cm² (4.62 cm × 1.73 cm) and a conversion efficiency of 4.21% under 1 sun illumination (P_in of 100 mW/cm²). As a result, the externally connected DSC module assembly has an output of 7.4 V and 200 mA, and shows stable performance, with an energy conversion efficiency of 4.44% under 0.45 sun illuminations.     

Techno-economic comparative assessment of the ultrasound,electrostatic and microwave supported coalescence of binary water droplets in crude oil

In this study, comparative assessment of the technical performance, energy usage and economic impact of ultrasound, electrostatics and microwave on the coalescence of binary water droplets in crude oil was conducted. The effect of different oil properties such as crude oil viscosity (10.6–106 mPa s) and interfacial tension (IFT) (20–250 mN/m) on the coalescence time and energy consumption was examined. In addition, operation conditions such as inlet emulsion flow velocity (10–100 mm/s), electric field type, ultrasound frequency and applied voltage amplitude (0–30 kV) were evaluated. The numerical models showed good agreement with experimental findings in the literature. Moreover, the process time of the dewatering process increased with rising inlet flow velocities. The elevation of the coalescence time with velocity can be attributed to the increasing effect of flow disturbance, and the reduction of the emulsion residence time. As regards the IFT, the coalescence time reduced as the IFT was increased. This can be associated with the improved stability of emulsions formed at lowered IFT. As the maximum droplet size is directly proportional to the IFT, lowering the IFT reduces the peak diameter of the droplets that are present in the emulsion. Moreover, the coalescence time followed the order: ultrasound < microwave < electrostatics approaches under varying IFT. The coalescence energy increased from ∼15 J, ∼90 J and ∼25 mJ to ∼61 J, ∼235 J and ∼26 mJ for microwave, electrostatics and ultrasound techniques, respectively, as the viscosity was raised from 10.6 to 106 mPa s. Ultrasound coalescence showed significant energy and economic savings in comparison to microwave and electro-coalescence. Hence, ultrasound coalescence would be a potential method for standalone or integrated demulsification over the two other techniques. However, there are indications that beyond a viscosity of 300 mPa s, the effect of ultrasound becomes weak with significant hindrance to droplet movement and accumulation. This analysis provides fundamental insights on the comparative behavior of the three emulsion separation techniques.     

Micromachining NiTi tubes for use in medical devices by using a femtosecond laser

Recent growth in medical device technology has been substantially driven by developments in laser micromachining, which is a powerful fabrication technique in which nickel–titanium (Nitinol, NiTi) alloy materials that exhibit superelastic and shape memory properties are formed (e.g., self-expanding stents). In this study a NiTi tube curve surface process is proposed, involving a femtosecond laser process and a galvano-mirror scanner. The diameter of the NiTi tube was 5.116 mm, its thickness was 0.234 mm, and its length was 100 mm. The results indicated that during the machine process the ablation mechanism of the NiTi tubes was changed by altering the machining path. The path alteration enhanced the laser ablation rate from 12.3 to 26.7 μm/J. Thus the path alteration contributed to a wide kerf line, enabling the assisted air to efficiently remove the debris deposited at the bottom of the kerf during the laser ablation process. The results indicated that the NiTi tube curve process enhanced the laser ablation rate by two times and reduced the amount of energy accumulated within the materials by 50% or more. By altering the machining path using the scanning system, this process can decrease the production of heat affected zones (the accumulation of thermal energy) in medical device applications.     

DRL-M4MR: An intelligent multicast routing approach based on DQN deep reinforcement learning in SDN

Traditional multicast routing methods have some problems in constructing a multicast tree. These problems include limited access to network state information, poor adaptability to dynamic and complex changes in the network, and inflexible data forwarding. To address these defects, the optimal multicast routing problem in software-defined networking (SDN) is tailored as a multiobjective optimization problem, and DRL-M4MR, an intelligent multicast routing algorithm based on the deep Q network (DQN) deep reinforcement learning (DRL) method is designed to construct a multicast tree in a software-defined network. First, combining the characteristics of SDN global network-aware information, the multicast tree state matrix, link bandwidth matrix, link delay matrix and link packet loss rate matrix are designed as the state space of the reinforcement learning agent to solve the problem in that the original method cannot make full use of network status information. Second, the action space of the agent is all the links in the network, and the action selection strategy is designed to add the links to the current multicast tree in four cases. Third, single-step and final reward function forms are designed to guide the agent to make decisions to construct the optimal multicast tree. The double network architectures, dueling network architectures and prioritized experience replay are adopted to improve the learning efficiency and convergence of the agent. Finally, after the DRL-M4MR agent is trained, the SDN controller installs the multicast flow entries by reversely traversing the multicast tree to the SDN switches to implement intelligent multicast routing. The experimental results show that, compared with existing algorithms, the multicast tree constructed by DRL-M4MR can obtain better bandwidth, delay, and packet loss rate performance after training, and it can make more intelligent multicast routing decisions in a dynamic network environment. Code and DRL model are available at https://github.com/GuetYe/DRL-M4MR.     

基于Wi-Fi与Web的云计算资源调度算法研究

为了缩短云计算执行时间,改善云计算性能,在一定程度上加强云计算资源节点完成任务成功率,需要对云计算资源进行调度。当前的云计算资源调度算法在进行调度时,通过选择合适的调度参数并利用CloudSim仿真工具,完成对云计算资源的调度。该算法在运行时有效地进行平衡负载,导致云计算资源调度的均衡性能较差,存在云计算资源调度结果误差大的问题。为此,提出一种基于Wi-Fi与Web的云计算资源调度算法。该算法首先利用自适应级联滤波算法对云计算资源数据流进行滤波降噪,然后以降噪结果为基础,采用本体论对云计算资源进行预处理操作,最后通过人工蜂群算法完成对云计算资源的调度。实验结果证明,所提算法可以良好地应用于云计算资源调度中,有效提高了云计算资源利用率,具有实用性以及可实践性,为该领域的后续研究发展提供了可靠支撑。     

Experimental and predictive study on the performance and energy consumption characteristics for the regeneration of activated alumina assisted by ultrasound

Activated alumina used in dehumidification should be regenerated at more than 110 °C temperature, resulting in excessive energy consumption. Comparative experiments were conducted to study the feasibility and performance of ultrasonic assisted regeneration so as to lower the regeneration temperature and raise the efficiency. The mean regeneration speed, regeneration degree, and enhanced rate were used to evaluate the contribution of ultrasound in regeneration. The effective moisture diffusivity and desorption apparent activation energy were calculated by theoretical models, revealed the enhanced mechanism caused by ultrasound. Also, we proposed some specific indexes such as unit energy consumption and energy-saving ratio to assess the energy-saving characteristics of this process. The unit energy consumption was predicted by artificial neural network (ANN), and the recovered moisture adsorption of activated alumina was measured by the dynamic adsorption test. Our analysis illustrates that the introduction of power ultrasound in the process of regeneration can reduce the unit energy consumption and improve the recovered moisture adsorption, the unit energy consumption was decreased by 68.69% and the recovered moisture adsorption was improved by 16.7% under 180 W power ultrasound compared with non-ultrasonic assisted regeneration at 70 °C when initial moisture adsorption was 30%. Meanwhile, an optimal regeneration condition around the turning point could be obtained according to the predictive results of ANN, which can minimize the unit energy consumption. Moreover, it was found that a larger specific surface area of activated alumina induced by ultrasound contributed to a better recovered moisture adsorption.     

A novel deep deterministic policy gradient model applied to intelligent transportation system security problems in 5G and 6G network scenarios

Traffic congestion has been an actual problem in large cities, causing personal inconvenience and environmental pollution. To solve this problem, new applications for Intelligent Transportation System (ITS) have been created, to monitor actual traffic conditions. Therefore, fast, reliable and safe systems are desirable for creating a real intelligent transportation environment. Deep learning algorithms have been proposed for a better understanding of traffic behavior from a security-related perspective. Thus, we aim to maximize the safety problems using a deep learning algorithm, where a novel policy gradient model is presented for detecting vehicular misuse. The proposed model uses a triple network replay algorithm, maximizing the network convergence speed. Three networks are selected to optimize the policy network variables. Finally, the replay algorithm is partitioned with the aim of obtaining a faster model. Simulations on a real urban map are performed in a scenario with the integration of 5G or 6G networks. An architectural model for the integration of a Vehicular Ad-hoc Network (VANET) and cellular networks is determined in software-defined networking (SDN). The results show that the accuracy prediction of the proposed system presents better performance compared to related studies, where the proposed model increases its convergence speed and cumulative reward. Thus, the ITS improvement by the proposed deep learning algorithm increases the prediction accuracy, and reduces the transmission delay, treating the traffic path according to the congestion.