基于Blending-Clustering集成学习的大坝变形预测模型

【目的】变形是反映大坝结构性态最直观的效应量,构建科学合理的变形预测模型是保障大坝安全健康运行的重要手段。针对传统大坝变形预测模型预测精度低、误报率高等问题导致的错误报警现象,【方法】选取不同预测模型和聚类算法集成,构建了一种Blending-Clustering集成学习的大坝变形预测模型,该模型以Blending对单一预测模型集成提升预测精度为核心,并通过Clustering聚类优选预测值改善模型稳定性。以新疆某面板堆石坝变形监测数据为实例分析,通过多模型预测性能比较,对所提出模型的预测精度和稳定性进行全面评估。【结果】结果显示:Blending-Clustering模型将预测模型和聚类算法集成,均方根误差(RMSE)和归一化平均百分比误差(nMAPE)明显降低,模型的预测精度得到显著提高;回归相关系数(R~2)得到提升,模型具备更强的拟合能力;在面板堆石坝上22个测点变形数据集上的预测评价指标波动范围更小,模型的泛化性和稳定性得到有效增强。【结论】结果表明:Blending-Clustering集成预测模型对于预测精度、泛化性和稳定性均有明显提升,在实际工程具有一定的应用价值。

A deep reinforcement learning approach to gasoline blending real-time optimization under uncertainty

The gasoline inline blending process has widely used real-time optimization techniques to achieve optimization objectives,such as minimizing the cost of production.However,the effectiveness of real-time optimization in gasoline blending relies on accurate blending models and is challenged by stochastic disturbances.Thus,we propose a real-time optimization algorithm based on the soft actor-critic(SAC)deep reinforcement learning strategy to optimize gasoline blending without relying on a single blending model and to be robust against disturbances.Our approach constructs the environment using nonlinear blending models and feedstocks with disturbances.The algorithm incorporates the Lagrange multiplier and path constraints in reward design to manage sparse product constraints.Carefully abstracted states facilitate algorithm convergence,and the normalized action vector in each optimization period allows the agent to generalize to some extent across different target production scenarios.Through these well-designed components,the algorithm based on the SAC outperforms real-time optimization methods based on either nonlinear or linear programming.It even demonstrates comparable performance with the time-horizon based real-time optimization method,which requires knowledge of uncertainty models,confirming its capability to handle uncertainty without accurate models.Our simulation illustrates a promising approach to free real-time optimization of the gasoline blending process from uncertainty models that are difficult to acquire in practice.

基于Prophet算法和Blending集成学习的实时负荷中期预测

目前的中期负荷预测一般未考虑负荷实时状态,而负荷数据的非线性、季节性、随机性、时序性特征将影响实时负荷的中期预测。构建一个实时负荷中期预测的框架,采用Prophet算法提取负荷数据的季节性部分,采用Blending集成学习对负荷数据的非季节部分进行滚动预测,将季节性部分和非季节性部分合成中期负荷实时数据。爱尔兰电力系统的算例结果验证了模型的有效性和稳定性。

Multi-objective Design of Blending Fuel by Intelligent Optimization Algorithms

Fuel design is a complex multi-objective optimization problem in which facile and robust methods are urgently demanded.Herein,a complete workflow for designing a fuel blending scheme is presented,which is theoretically supported,efficient,and reliable.Based on the data distribution of the composition and properties of the blending fuels,a model of polynomial regression with appropriate hypothesis space was established.The parameters of the model were further optimized by different intelligence algorithms to achieve high-precision regression.Then,the design of a blending fuel was described as a multi-objective optimization problem,which was solved using a Nelder–Mead algorithm based on the concept of Pareto domination.Finally,the design of a target fuel was fully validated by experiments.This study provides new avenues for designing various blending fuels to meet the needs of next-generation engines.

基于Blending集成学习的多源信息液压系统多类故障诊断研究

针对传统故障诊断方法准确性不高、耗时长问题,研究通过多个EfficientNet模型对传感器数据进行预训练,并使用XGBoost作为元学习器,提出了一种基于Blending集成学习的多源信息液压系统多类故障诊断方法。实验结果表明,各个子分类器在训练次数达到300次后趋于收敛,准确率均达到95%左右。该方法具有较高的准确性和鲁棒性,为液压系统故障诊断提供了一种有效的解决方案。

基于Blending集成学习模型的电力市场日前出清电价预测

精准的掌握未来电价信息对把握市场的运行状态、支撑市场参与各方进行有效决策、推动市场主体合理优化资源配置具有重要意义。因此基于Blending集成学习机制构建了一种面向于日前电价预测的综合模型。该模型充分考虑电价的高波动性特点,采用Ashin变换,减小了输入数据波动性对预测模型的影响;选取SVM、LightGBM、EWNN、SARIMAX 4种较为成熟的单一电价预测模型作为初级学习器,进而保证了基于Blending集成学习模型的预测精度。选用美国PJM电力市场实际运行数据对上述构建的电价预测模型进行验证,通过对预测结果的对比分析,表明构建的基于Blending集成学习机制的电价综合预测模型集成了多种传统预测模型的优点,具有较好的准确性与稳定性。

基于FPFF-Blending模型融合的个体工商户信用评价研究

个体工商户信用评价研究往往通过单一机器学习模型建立,其预测精确率较低,抗干扰能力较弱。基于特征金字塔的FPFF特征融合算法,应用于Blending模型融合框架,建立个体工商户信用评价异质融合模型,并赋予模型可解释性,综合解决单一模型稳定性较差、原有Blending框架融合模型过拟合、融合模型缺乏可解释性的问题。通过对个体工商户数据集进行实证实验,结果表明:融合模型较单一机器学习模型在个体工商户信用评价场景下具有更优的预测性能和泛化能力。

Closed-loop scheduling optimization strategy based on particle swarm optimization with niche technology and soft sensor method of attributes-applied to gasoline blending process

Gasoline blending scheduling optimization can bring significant economic and efficient benefits to refineries.However,the optimization model is complex and difficult to build,which is a typical mixed integer nonlinear programming(MINLP)problem.Considering the large scale of the MINLP model,in order to improve the efficiency of the solution,the mixed integer linear programming-nonlinear programming(MILP-NLP)strategy is used to solve the problem.This paper uses the linear blending rules plus the blending effect correction to build the gasoline blending model,and a relaxed MILP model is constructed on this basis.The particle swarm optimization algorithm with niche technology(NPSO)is proposed to optimize the solution,and the high-precision soft-sensor method is used to calculate the deviation of gasoline attributes,the blending effect is dynamically corrected to ensure the accuracy of the blending effect and optimization results,thus forming a prediction-verification-reprediction closed-loop scheduling optimization strategy suitable for engineering applications.The optimization result of the MILP model provides a good initial point.By fixing the integer variables to the MILPoptimal value,the approximate MINLP optimal solution can be obtained through a NLP solution.The above solution strategy has been successfully applied to the actual gasoline production case of a refinery(3.5 million tons per year),and the results show that the strategy is effective and feasible.The optimization results based on the closed-loop scheduling optimization strategy have higher reliability.Compared with the standard particle swarm optimization algorithm,NPSO algorithm improves the optimization ability and efficiency to a certain extent,effectively reduces the blending cost while ensuring the convergence speed.

The application study of dual-energy CT nonlinearblending technique in pulmonary angiography

Objective This study aimed to explore the feasibility of enhancing image quality in computed tomography(CT) pulmonary angiography (CTPA) and reducing radiation dose using the nonlinear blending (NLB)technique of dual-energy CT.Methods A total of 61 patients scheduled for CTPA were enrolled, and 30 patients underwent dual-energyscanning. Nonlinear blending images (NLB group) and three groups of linear blending images (LB group,80 kV group, and 140 kV group) were reconstructed after scanning;31 patients underwent single-energyscanning (120 kV group). The CT values and standard deviations of the pulmonary trunk, left and rightpulmonary arteries, and ipsilateral back muscle at the bifurcation level of the left and right pulmonaryarteries were measured. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of the fivegroups were calculated. The subjective image quality of the five groups was assessed. The radiation dosesof the dual- and single-energy groups were recorded and calculated.Results The CNR and SNR values of blood vessels in the NLB group were significantly higher than thosein the LB, 140 kV, and 80 kV groups (CNR of pulmonary artery trunk: t = 3.50, 4.06, 7.17, all P < 0.05;SNRof pulmonary trunk: t = 3.76, 4.71, 6.92, all P < 0.05). There were no statistical differences in the CNR andSNR values between the NLB group and 120 kV group (P > 0.05). The effective radiation dose of the dualenergygroup was lower than that of the single-energy group (t = –4.52, P < 0.05). The subjective scores ofimages in the NLB group were the highest (4.28 ± 0.74).Conclusion The NLB technique of dual-energy CT can improve the image quality of CTPA and reducethe radiation dose, providing more reliable imaging data for the clinical diagnosis of pulmonary embolism.

Feasibility Investigation of Bitumen Properties by Blending of Coal Tar Pitch

There are numerous methods and additives available to improve the durability and quality of road bitumen. A coal tar obtained by coal coking was distilled in a laboratory into fractions of initial boiling point IBP-180℃ (gasoline-like fuel), 180℃ - 360℃ (diesel-like fuel), and >360℃ (residue or coal tar pitch). The coal tar pitch was added into road bitumen by up to 1 - 5 wt% and investigated the alteration of physical and chemical properties. The physico-mechanical properties of coal tar pitch and bitumen blends, as well as the chemical group composition, were determined using standard techniques (MNS) and the SARA method, respectively. Results of 3% coal tar pitch addition into bitumen enhanced ductility by 12.4% and softening point by 1.6℃. We found that blending with bitumen coal tar pitch as a modifier could improve bitumen properties.

Conceptual Blending and Meme Humor on the Internet:The Example of COVID-19 Publicity Posters in Chinese Microblogging

Internet memes,as multimodal cultural products disseminated through the Internet,usually take the form of short videos or images that express humor or satire.The creation and dissemination of humor in memes are both creative and complex,and the successful perception of meme humor reflects humans’universal thinking capacity.Based on the theoretical framework of conceptual blending,this paper selects a set of COVID-19 publicity posters from the official Weibo account of China Guangzhou Fabu(Guangzhou Internet Information Office),analyses the multi-level structure of Internet memes,and explores the dynamic cognitive process in the interpretation of humorous memes to reveal people’s ability to make simultaneous analogies and integration between elements in different mental spaces.

基于Blending Learning的微课设计研究

通过分析比较Blending Learning概念和微课的概念、特征,提出了基于Blending Learning的微课新概念。依据Blending Learning教学过程设计模式,阐述了微课教学设计组成和原则,微课网络平台设计的基本形式、构成要素及活动特征。根据Blending Learning中师生角色、教学过程和视频录制等进行了微课分类,着重分析了讲授类和演示类微课的设计,以及微课视频录制技术应用,并且运用多媒体学习理论阐明了微课件的设计原则。

构造代数Blending曲面的Gr bner基方法

利用代数几何中关于理想的 Gr bner基的理论 ,结合 CAGD中的研究方法 ,对代数 Blending曲面做了较为细致的研究 ,给出了用 Gr bner基构造代数 Blending曲面的新方法 .该方法能够求出所有满足要求的代数Blending曲面 ,并能给出其中次数最低的曲面 .文中还讨论了如何利用代数曲面插值、最小平方逼近的方法来选取合适的自由参数 ,以达到对代数 Blending曲面进行形状控制的目的 .最后给出了一个茶壶表面造型示例 。

JiTT——Blending Learning理念下的信息化教学模式

近年来,适时教学(JiTT)作为一种能有效促进学生自主学习的教学模式,在欧美国家本科教学中受到推崇。JiTT既包含传统的面对面教学,又包含了学生自主网络探究的环节,充分体现了Blending Learning的新理念。文章通过对JiTT内涵、国外相关成功案例进行剖析,强调了成功实践JiTT的关键因素,并得出了JiTT较之传统本科授课的优势所在,以及对目前我国本科教学的启示。以期为我国当前开展的"质量工程"提供借鉴。

从Blending Learning看教育技术理论的新发展(上)

本文介绍了 Blending L earning(或 Blended L earning)的新含义 ,指出这一新含义的提出和被广泛认同 ,表明国际教育技术界的教育思想观念正在经历又一场深刻的变革 ,也是教育技术理论进一步发展的标志。作者还从对建构主义理论的反思、对信息技术教育应用认识的深化 。

A facile strategy for tuning the density of surface-grafted biomolecules for melt extrusion-based additive manufacturing applications

Melt extrusion-based additive manufacturing(ME-AM)is a promising technique to fabricate porous scaffolds for tissue engi-neering applications.However,most synthetic semicrystalline polymers do not possess the intrinsic biological activity required to control cell fate.Grafting of biomolecules on polymeric surfaces of AM scaffolds enhances the bioactivity of a construct;however,there are limited strategies available to control the surface density.Here,we report a strategy to tune the surface density of bioactive groups by blending a low molecular weight poly(ε-caprolactone)5k(PCL5k)containing orthogonally reactive azide groups with an unfunctionalized high molecular weight PCL75k at different ratios.Stable porous three-dimensional(3D)scaf-folds were then fabricated using a high weight percentage(75 wt.%)of the low molecular weight PCL 5k.As a proof-of-concept test,we prepared films of three different mass ratios of low and high molecular weight polymers with a thermopress and reacted with an alkynated fluorescent model compound on the surface,yielding a density of 201-561 pmol/cm^(2).Subsequently,a bone morphogenetic protein 2(BMP-2)-derived peptide was grafted onto the films comprising different blend compositions,and the effect of peptide surface density on the osteogenic differentiation of human mesenchymal stromal cells(hMSCs)was assessed.After two weeks of culturing in a basic medium,cells expressed higher levels of BMP receptor II(BMPRII)on films with the conjugated peptide.In addition,we found that alkaline phosphatase activity was only significantly enhanced on films contain-ing the highest peptide density(i.e.,561 pmol/cm^(2)),indicating the importance of the surface density.Taken together,these results emphasize that the density of surface peptides on cell differentiation must be considered at the cell-material interface.Moreover,we have presented a viable strategy for ME-AM community that desires to tune the bulk and surface functionality via blending of(modified)polymers.Furthermore,the use of alkyne-azide“click”chemistry enables spatial control over bioconjugation of many tissue-specific moieties,making this approach a versatile strategy for tissue engineering applications.

从Blending Learning看教育技术理论的新发展(下)

本文介绍了 Blending L earning(或 Blended L earning)的新含义 ,指出这一新含义的提出和被广泛认同 ,表明国际教育技术界的教育思想观念正在经历又一场深刻的变革 ,也是教育技术理论进一步发展的标志。作者还从对建构主义理论的反思、对信息技术教育应用认识的深化 。

体现Blending Learning教育思想的几种新型教学模式解析——“双创”对教学改革提出的挑战

"双创"要求教育部门培养出来的人才不仅要具有系统丰富的科学知识,还要有创新意识、创新思维和创新能力。近年来,许多学者认识到Blending Learning教育思想的重要性,要求在教学中既要重视教师的主导作用,又要注重学生的主体地位,翻转课堂模式、泰微课+洋思模式和跨越式教学模式就是在此指导下出现的"学教并重"的新型教学模式,它们的成功应用也启发了高校教学,一些大学教师将微课、慕课等与课堂教学相结合,力图培养出"双创"所需要的创新型人才。

从Blending Learning看教育技术理论的新发展

所谓Blending Learning,就是要把传统学习方式的优势和E-Learning(即数字化或网络化学习)的优势结合起来;也就是说,既要发挥教师引导、启发、监控教学过程的主导作用,又要充分体现学生作为学习过程主体的主动性、积极性与创造性。这一新含义的提出和被广泛认同,表明国际教育技术界的教育思想观念正在经历又一场深刻的变革,也标志着教育技术理论的进一步发展,也必将对建构主义理论的反思、对信息技术教育应用认识的深化、对信息技术与课程整合理论的建构、对教学设计理论的发展等产生重大影响。

基于多元模态分解与多目标算法优化的深度集成学习模型的超短期风电功率预测

针对风电功率预测问题,提出了一种基于多元变分模态分解(multivariate variational mode decomposition,MVMD)、多目标纵横交叉优化(multi-objective crisscross optimization,MOCSO)算法和Blending集成学习的超短期风电功率预测。在数据处理阶段,为了保持各序列间的同步相关性以及分解后得到本征模态函数(intrinsic mode functions,IMF)分量个数和分量频率相匹配,使用MVMD对多通道原始数据进行同步分解。针对单一机器学习模型导致预测的全面性不足,且存在精度和鲁棒性低的问题,提出基于MOCSO算法动态加权的Blending集成学习模型。通过对递归神经网络、卷积神经网络、长短期记忆网络的预测结果进行动态加权集成,并通过MOCSO优化调整权重,以提高模型的预测准确性与稳定性。实验结果表明,所提预测模型不仅有效,且显著优于其他预测模型。