基于SOFM和BP短期负荷预测方法

基于以自组织特征映射神经网络(Self-organizing feature map,SOFM)先聚类、神经网络再预测的模型以往多用在对疾病、天气方面的预测,由此提出了一种以SOFM与误差反向传播算法的神经网络(Back Propagation,BP)相组合应用为基本原理的短期电力负荷预测的组合模型。该模型主要基于SOFM网络的主要特性聚类,预先将训练样本集采取聚类分析,对其分为具有相似数据的若干子类,再根据每一子类构造一个BP网络模型。利用内蒙古自治区某市的实际日平均负荷数据进行仿真,证明了本文方法的有效性。     

Targeting the Mitochondrial Protein VDAC1 as a Potential Therapeutic Strategy in ALS

Impaired mitochondrial function has been proposed as a causative factor in neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS), caused by motor neuron degeneration. Mutations in superoxide dismutase (SOD1) cause ALS and SOD1 mutants were shown to interact with the voltage-dependent anion channel 1 (VDAC1), affecting its normal function. VDAC1 is a multi-functional channel located at the outer mitochondrial membrane that serves as a mitochondrial gatekeeper controlling metabolic and energetic crosstalk between mitochondria and the rest of the cell and it is a key player in mitochondria-mediated apoptosis. Previously, we showed that VDAC1 interacts with SOD1 and that the VDAC1-N-terminal-derived peptide prevented mutant SOD1 cytotoxic effects. In this study, using a peptide array, we identified the SOD1 sequence that interacts with VDAC1. Synthetic peptides generated from the identified VDAC1-binding sequences in SOD1 directly interacted with purified VDAC1. We also show that VDAC1 oligomerization increased in spinal cord mitochondria isolated from mutant SOD1^G93A mice and rats. Thus, we used the novel VDAC1-specific small molecules, VBIT-4 and VBIT-12, inhibiting VDAC1 oligomerization and subsequently apoptosis and associated processes such as ROS production, and increased cytosolic Ca²⁺. VBIT-12 was able to rescue cell death induced by mutant SOD1 in neuronal cultures. Finally, although survival was not affected, VBIT-12 administration significantly improved muscle endurance in mutant SOD1^G93A mice. Therefore, VBIT-12 may represent an attractive therapy for maintaining muscle function during the progression of ALS.     

Interlocked MXene/rGO aerogel with excellent mechanical stability for a health-monitoring device

Two-dimensional (2D) materials have attracted considerable interest thanks to their unique electronic/physical–chemical characteristics and their potential for use in a large variety of sensing applications. However, few-layered nanosheets tend to agglomerate owing to van der Waals forces, which obstruct internal nanoscale transport channels, resulting in low electrochemical activity and restricting their use for sensing purposes. Here, a hybrid MXene/rGO aerogel with a three-dimensional (3D) interlocked network was fabricated via a freeze-drying method. The porous MXene/rGO aerogel has a lightweight and hierarchical porous architecture, which can be compressed and expanded several times without breaking. Additionally, a flexible pressure sensor that uses the aerogel as the sensitive layer has a wide response range of approximately 0–40 kPa and a considerable response within this range, averaging approximately 61.49 kPa^–1. The excellent sensing performance endows it with a broad range of applications, including human-computer interfaces and human health monitoring.     

Ultra wideband CMOS digital T-type attenuator with low phase errors

A proposed inductive-phase-compensation ultra wideband CMOS digital T-type attenuator design based on an analysis of minimising phase errors is presented in this letter. In a standard CMOS technology, the proposed attenuator is analytically demonstrated to have low phase errors due to the inductive-phase-compensation network. A design equation is inferred and a wide-band 4dB attenuation bit digital attenuator with low phase errors is designed as a test vehicle for the proposed approach.     

Microanalysis of supersaturated gas release based on wall-attached bubbles

Supersaturation of dissolved gases in natural water, due to spillage from high dams and other factors, may cause fish mortality. In previous experiments, the dissipation coefficient has been used to denote the degassing process of total dissolved gas(TDG) saturation. These experiments mainly analyzed supersaturated TDG dissipation from a macroscopic view. To precisely clarify the mechanism of supersaturated TDG release, this study investigated bubble adsorption at a wall surface from a microscopic view. The experiment was conducted in a Plexiglas-wall container filled with supersaturated TDG water. A model that calculates the adsorption flux of supersaturated TDG by a solid wall, and helps describe construction for a contact angle at a three-phase intersection, was developed according to Young's equation. This model was used to investigate the formation process of bubbles adsorbed on a solid polymethyl methacrylate(PMMA) surface in supersaturated TDG water. The adsorption effect of a solid wall on TDG release was analyzed based on the experimental data. The modeling results were compared with observations under different wall area conditions, and it was found that TDG release tended to increase with wall area. This study helps improve our understanding of the mechanisms of supersaturated TDG release and provides an important theoretical method for accurate calculation of the release process. The adsorption flux model of the solid wall provides mitigation measures to combat the adverse effects of TDG supersaturation,which will be beneficial to the protection of aquatic organisms in hydropower-regulated rivers.     

“新工科”背景下应用型大学给排水专业智能化课程群建设探索

智能建造技术的发展及新工科建设都要求培养适应产业升级需求的跨学科复合型人才,目前已有部分高校开展了针对服务于土木工程建设的智能建造专业的探索,但是针对给排水专业智能化人才培养的研究寥寥无几。在院校和企业调研、文献调研、毕业生调研的基础上,掌握了给排水专业智能化人才培养的知识和技能需求,并在此基础上设计了给排水专业智能化培养方向课程群,采用新建课程与融入课程相结合的方式,设置了BIM应用技术及软件实训、Python语言程序设计、物联网技术、数据库技术、网络爬虫与数据分析等新建课程并将新兴信息技术融入现有专业课程,以培养智能化水务人才,服务国家智慧水务系统建设。     

重磁异常的快速均衡边界识别方法

边界识别是位场数据解释中一项重要任务,可用于资源探测等领域,是目前研究的热点问题之一.对于深部异常体目标,采用传统的边界识别方法易出现模糊、发散、变形等问题,而改进的方法大多采用高阶导数,虽然能够很好地识别出深部异常体边界,但是计算复杂,易受噪声干扰.为了改善以上问题,提出了一种稳定的快速边界识别方法,仅利用符号函数和...     

浅谈金属材料磨损失效及防护措施

为了进一步深入研究铁基非晶涂层的制备和性能,利用人工神经网络和多因素多水平设计对超音速火焰喷涂技术制备的铁基非晶纳米晶涂层的工艺参数和性能进行了优化,以期对铁基非晶涂层的实际应用提供参考. 多因素多水平试验通过设计分析方法和设计实验参数,建立神经网络模型,研究了煤油量、氧气量、送粉率和喷涂距离四项工艺参数对涂层孔隙率、硬度、结合强度和沉积效率的影响规律. 采用扫描电镜、透射电镜等手段表征了粉末及涂层的显微结构和内部微观组织形貌;采用X射线衍射仪、同步热分析仪等设备对粉末和涂层成分、相组成和非晶程度进行了观察分析. 验证了工艺参数优化过程的计算机模拟结果,确定了最佳喷涂工艺参数范围,进一步提升了涂层的性能. 讨论了喷涂过程中孔隙形成的微观过程和非晶纳米晶对涂层性能的影响机理. 研究表明各工艺参数对涂层性能是多因素互相影响的,理论上最佳喷涂工艺参数为煤油量23 L·h⁻¹、氧气量51 L·h⁻¹、送粉率72 g·min⁻¹、喷涂距离280 mm,制备出的铁基非晶涂层厚度约为270 μm、孔隙率约为1.3%、结合强度约为84 MPa、硬度约为1110 HV_0.3. 涂层非晶程度在80%左右,纳米晶尺寸为3～5 nm, 涂层在600 ℃以下不会发生晶化过程.

     

螺旋藻多糖超声波提取方案的优化

研究螺旋藻多糖热碱液的提取、过氧化氢法脱色的条件并进行工艺优化。在单因素实验结果的基础上,分别以多糖得率和多糖脱色率为指标,通过响应面法优化热碱法提取螺旋藻多糖和过氧化氢法脱色的工艺条件,并分别得到热碱液提取螺旋藻多糖的最优提取工艺条件和过氧化氢法最优脱色条件。结果表明,热碱液提取法最优条件：NaOH质量分数1.5%、提取温度为90 ℃、提取时间4 h及料液比1:20 g/mL,在此条件下,得到螺旋藻多糖得率为5.18%±0.23%;过氧化氢法的最佳脱色条件：过氧化氢溶液添加量为17%、脱色时间为55 min、脱色温度为50 ℃,在此条件下得到的实际脱色率为82.54%±0.03%,多糖保留率为83.45%±0.13%。本研究所得结果对于螺旋藻多糖的开发利用具有一定参考价值。

     

用自产生和自组织神经网络对超声医学图像进行自动分割

本文研究用自产生和自组织神经网络方法进行超声心脏图像的自动分割。这种无监督的聚类方法能够自动搜索最佳的网络输出节点数而获取图像中的目标数,从而完成对图像的自动分割。实验结果表明,与自组织特征映射方法相比,本文的方法具有许多重要的优点。