基于条件生成式对抗网络的油藏单井产量预测模型

为克服机器学习方法在油藏单井产量预测中的过拟合问题,提高油田生产中的产量预测精度,提出一种基于条件生成式对抗网络(CGAN)的油藏单井产量预测模型。该模型使用长短期记忆、全连接等基础神经网络,构建生成和判别网络模型。生成网络模型以产量影响因素为条件输入,生成预测产量数据,利用对数损失函数评价预测数据与真实数据之间的偏差,通过条件生成式对抗网络的博弈训练,并结合贝叶斯超参数优化算法,优化模型结构,综合提高模型的泛化能力。基于Eclipse数值模拟软件建立同一井网条件下不同地质和生产条件下的油藏单井产量数据库,以地质与生产条件等产量影响因素作为模型的条件输入,进行油藏单井产量预测。结果表明：与全连接神经网络(FCNN)、随机森林(RF)以及长短期记忆神经网络(LSTM)模型的预测结果相比,CGAN模型在测试集上的平均绝对百分比误差分别提升了2.59%、 0.81%以及1.72%,并且过拟合比最小(1.027)。说明CGAN降低了机器学习产量预测模型的过拟合程度,提高了模型的泛化能力与预测精度,验证了所提算法的优越性,对指导油田高效开发和保障我国能源战略安全具有重要意义。     

A Hybrid Method Using HAVOK Analysis and Machine Learning for Predicting Chaotic Time Series

The prediction of chaotic time series systems has remained a challenging problem in recent decades. A hybrid method using Hankel Alternative View Of Koopman (HAVOK) analysis and machine learning (HAVOK-ML) is developed to predict chaotic time series. HAVOK-ML simulates the time series by reconstructing a closed linear model so as to achieve the purpose of prediction. It decomposes chaotic dynamics into intermittently forced linear systems by HAVOK analysis and estimates the external intermittently forcing term using machine learning. The prediction performance evaluations confirm that the proposed method has superior forecasting skills compared with existing prediction methods.     

Fluidity of hydration layers nanoconfined between mica surfaces

We perform molecular dynamics simulations to investigate the shear dynamics of hydration water nanoconfined between two mica surfaces at 1 bar pressure and 298 K. Newtonian plateaus of shear viscosity comparable to the bulk value for different hydration layers D=0.92-2.44 nm are obtained. The origin of this persistent fluidity of the confined aqueous system is found to be closely associated with the rotational dynamics of water molecules, accompanied by fast translational diffusion under this confinement.     

Mixed-mean inequalities for subsets

For let and denote the arithmetic mean and geometric mean of elements of , respectively. It is proved that if is an integer in , then

with equality if and only if . Furthermore, as a generalization of this inequality, a mixed power-mean inequality for subsets is established.

     

Oscillating viscosity in a lyotropic lamellar phase under shear flow

We report some time-dependent behavior of lyotropic lamellar phase under shear flow. At fixed stress, near a layering instability, the system presents an oscillating shear rate. We build up a new stress versus shear rate diagram that includes temporal behavior. This diagram is made of two distinct branches of stationary states which correspond, respectively, to disordered and ordered multilamellar vesicle phases. When increasing the shear stress, prior to the transition to the ordered structural state, sustained oscillations of the viscosity are recorded. They correspond to periodic structural change of the entire sample between a disordered and a ordered state of multilamellar vesicles.     

Light Irradiation through Small Particles and Its Applications for Surface Nanostructuring in Near Field

We investigate the light scattering through small particles and its applications in nanostructuring, such as nanobumping, nanopatterning and dry laser cleaning. The theoretical calculation based on Mie theory provides an exact solution for sphere cavity resonance and plasmon resonance, which are two mechanisms for dielectric and metallic particles assisted surface nanostructuring in near field. The experimental results indicate that nanobumps on glass surface and subwavelength holes array on silicon surface can be formed without cracks with the self-assembly of 1 μm silica particle mask under laser irradiation. It is also found that the scattering wave by 40 nm gold particles can propagate 200 times away in terms of particle radius as recorded by photoresist under the UV light irradiation. Meanwhile, dry laser cleaning of 40nm gold particle on silicon wafer is demonstrated at plasmonic resonance frequency. The total cleaning efficiency is estimated to be 80%.     

COD速测仪测定工业废水中CODCr

采用标准回流法和仪器法（用ＣＯＤ速测仪）对工业废水化学需氧量进行对照测定，结果发现，用ＣＯＤ速测仪测定工业废水，简便、快速、精密度好、准确度高。     

Kinetics and mechanism of tertiary amine‐catalyzed cleavage of N′‐morpholino‐N‐(2′‐methoxyphenyl)phthalamide: Kinetic evidence for the presence of a reactive intermediate on the reaction path

Pseudo‐first‐order rate constants (k_obs) for tertiary amine (DABCO and Me₃N) buffer‐catalyzed cyclization of N′‐morpholino‐N‐(2′‐methoxyphenyl)phthalamide ( 1 ) to N‐(2′‐methoxyphenyl)phthalimide ( 2 ) reveal saturation (nonlinear) plots of k_obs versus [Buf]_T (total tertiary amine buffer concentration) at a constant pH. Such plots at different pH have been attributed to the presence of a reactive intermediate (T^?) formed by tertiary amine buffer‐catalyzed intramolecular nucleophilic addition of the secondary amide nitrogen to the carbonyl carbon of the tertiary amide group of 1 . © 2010 Wiley Periodicals, Inc. Int J Chem Kinet 42: 263–272, 2010