Stochastic Reconstruction of an Oolitic Limestone by Generative Adversarial Networks

Stochastic image reconstruction is a key part of modern digital rock physics and material analysis that aims to create representative samples of microstructures for upsampling, upscaling and uncertainty quantification. We present new results of a method of three-dimensional stochastic image reconstruction based on generative adversarial neural networks (GANs). GANs are a family of unsupervised learning methods that require no a priori inference of the probability distribution associated with the training data. Thanks to the use of two convolutional neural networks, the discriminator and the generator, in the training phase, and only the generator in the simulation phase, GANs allow the sampling of large and realistic volumetric images. We apply a GAN-based workflow of training and image generation to an oolitic Ketton limestone micro-CT unsegmented gray-level dataset. Minkowski functionals calculated as a function of the segmentation threshold are compared between simulated and acquired images. Flow simulations are run on the segmented images, and effective permeability and velocity distributions of simulated flow are also compared. Results show that GANs allow a fast and accurate reconstruction of the evaluated image dataset. We discuss the performance of GANs in relation to other simulation techniques and stress the benefits resulting from the use of convolutional neural networks . We address a number of challenges involved in GANs, in particular the representation of the probability distribution associated with the training data.     

Recurrent and convolutional neural networks for deep terrain classification by autonomous robots

The future challenge for field robots is to increase the level of autonomy towards long distance (>1 km) and duration (>1h) applications. One of the key technologies is the ability to accurately estimate the properties of the traversed terrain to optimize onboard control strategies and energy efficient path-planning, ensuring safety and avoiding possible immobilization conditions that would lead to mission failure. Two main hypotheses are put forward in this research. The first hypothesis is that terrain can be effectively detected by relying exclusively on the measurement of quantities that pertain to the robot-ground interaction, i.e., on proprioceptive signals. Therefore, no visual or depth information is required. Then, artificial deep neural networks can provide an accurate and robust solution to the classification problem of different terrain types. Under these hypotheses, sensory signals are classified as time series directly by a Recurrent Neural Network or by a Convolutional Neural Network in the form of higher-level features or spectrograms resulting from additional processing. In both cases, results obtained from real experiments show comparable or better performance when contrasted with standard Support Vector Machine with the additional advantage of not requiring an a priori definition of the feature space.     

基于并行卷积神经网络的地磁方向适配性分析

针对地磁方向适配性分析时人工特征提取主观性较强、所取特征难以表达深层的结构性特征的问题,并为了进一步提高方向适配性分析的准确率,提出了一种基于并行卷积神经网络的地磁方向适配性分析方法。首先,从不同角度建立了地磁场在6个代表方向上的适配性分析图;然后,从同一磁场的不同角度出发,利用卷积神经网络自动完成了特征学习,得到了更为全面的方向适配性特征描述;最后,在并行卷积神经网络所得特征的基础上,利用BP网络建立了地磁方向适配性的分析模型。仿真结果证明,该方法可以有效避免人工特征提取和计算等复杂步骤,实现了地磁方向适配性分析的自动化,而且可以获得优于传统网络和单路卷积神经网络的准确率。     

基于Fuzzy ARTMAP神经网络的景像匹配实时图选取方法

提出一种新的基于Fuzzy ARTMAP神经网络并利用图像直方图特征的快速景像匹配实时图选取方法。与已有的方法相比,该方法充分考虑了图像的边缘、亮度、对比度、信噪比等特征对影像实时图质量的影响,具有自适应聚类、收敛导速,实时性好,分类准确率高和通用性强等优点。将该方法应用于SMGS（景像匹配制导系统）进行实时图像的自动选取,可大大提高SMGS的智能性,可靠性和实时性。     

采用两步优化器的深度配点法与深度能量法求解薄板弯曲问题

随着计算机技术的进步以及机器学习算法的进一步发展,深度学习方法逐渐被广泛引用于各行各业中。本文发展并比较了适应于工程计算的深度配点法与深度能量法并应用于求解薄板弯曲问题。深度配点法采用物理驱动的深度神经网络来,并将物理信息（偏微分方程强形式）引入到损失函数中,最终将求解薄板弯曲问题简化为优化问题。深度能量法则是采用系统总势能驱动的神经网络。根据最小势能原理,在所有的可能位移场中,真实位移场的总势能取最小值,因此我们可以使用总势能构造损失函数,从而求解薄板弯曲问题。对于边界条件,通过罚函数法将有约束最优化问题转化为求解无约束最优化问题。深度配点法与深度能量法的适用性基于神经网络的通用近似定理。由于物理信息跟总势能的引入,增加了神经网络训练的困难,为了解决这个问题,我们发展了两步优化器方法。数值结果表明,深度配点法与深度能量法很适合求解薄板弯曲问题,并且程序实现简单,实现了真正意义上的“无网格法”。     

Classifying wakes produced by self-propelled fish-like swimmers using neural networks

We consider the classification of wake structures produced by self-propelled fish-like swimmers based on local measurements of flow variables. This problem is inspired by the extraordinary capability of animal swimmers in perceiving their hydrodynamic environments under dark condition. We train different neural networks to classify wake structures by using the streamwise velocity component, the crosswise velocity component, the vorticity and the combination of three flow variables, respectively. It is found that the neural networks trained using the two velocity components perform well in identifying the wake types, whereas the neural network trained using the vorticity suffers from a high rate of misclassification. When the neural network is trained using the combination of all three flow variables, a remarkably high accuracy in wake classification can be achieved. The results of this study can be helpful to the design of flow sensory systems in robotic underwater vehicles.     

Stability criteria for periodic neural networks with discrete and distributed delays

In this paper, the stability analysis problem is dealt with for a class of periodic neural networks with both discrete and distributed time delays. Both global asymptotic and exponential stabilities are considered. The existence of the periodic solutions of the addressed neural networks is briefly discussed. Then, by constructing different Lyapnuov--Krasovskii functionals and using some analysis techniques, several new easy-to-test sufficient conditions are derived, respectively, for checking the globally asymptotic stability and globally exponential stability of the delayed neural networks. These results are useful in the design and applications of globally exponentially stable and periodic oscillatory neural circuits for recurrent neural networks with mixed time delays. A simulation example is provided to demonstrate the effectiveness of the results obtained.     

基于Mask R-CNN网络的磨损颗粒智能识别与应用

针对设备磨损故障诊断中磨粒识别技术难度高、工作主观经验影响大等问题,采用深度学习技术开展了磨粒智能识别的研究,提出了基于Mask R-CNN卷积神经网络的磨粒数字化表征方法. 该方法利用迁移学习训练基于Mask R-CNN网络的磨粒识别模型对图像中磨粒进行识别和实例分割,然后使用Suzuki85算法、迭代算法、等比例计算方法计算出磨粒的真实尺寸,解决了磨粒分析中难定量分析的问题. 结果表明:基于Mask R-CNN网络(采用R-101-FPN骨干网络)训练的磨粒识别模型可以对图像中多个异常磨损颗粒进行识别,综合准确率和召回率达到当前图像识别领域的主流水平. 辅以上述Suzuki85等算法,成功实现磨粒图像的定量评价分析,对促进设备故障诊断技术的自动化发展和工业应用具有一定的实际应用价值.      

彩色光弹性干涉影像分析系统

自行开发的“彩色光弹性干涉影像分析系统”首先利用CCD成像和图像采集设备,将光弹图像以数字图像的形式存储到计算机,然后通过对存储的光弹图像进行处理,得到物体边界、等差线、等倾线等数据。最后根据这些数据,绘制出主应力迹线,并进行二维的和三维的应力分析。本文着重介绍了系统整体设计以及系统研制的难点问题(彩色光弹图像处理、主应力迹线的绘制等)。系统可以通过对彩色图像进行分解,应用目前已经比较成熟的灰度光弹图像处理技术,来完成彩色图像的处理;也可以直接应用彩色信息来确定条纹级数,进行相关处理。彩色图像能够比灰度图像提供更精确的图像信息,以满足高精度测量的要求。     

Recognition of partially overlapped particle images using the Kohonen neural network

A neural network is proposed for the recognition of partially overlapped particle images in the analysis of Particle Tracking Velocimetry (PTV) frames. The Kohonen neural network is an approximation to an optimum classifier. In this work it allows single particle images to be distinguished from overlapped particle images by shape analysis: it classifies 99.1% of the spots correctly (in test images). If a spot has an almost circular shape, the barycenter co-ordinates are extracted. If the spot shape is far from being circular, it is believed to be a particle overlap, and a procedure to find more centroids is activated.The particle recognizer based on the Kohonen neural network is tested on both multi-exposed and single-exposure images at high particle density, and compared to a particle recognizer that did not consider the partial overlap. The management of overlapped particles causes the neural network to produce a big improvement in the number of barycenters that can be extracted from these images. The practical consequence is that the seeding density in PTV can be increased, so as to improve the spatial resolution of the technique in the velocity field calculation.Marcello Sallusti and Paolo Monti produced the synthetic PTV images that were used in this project. Gianni Leuzzi helped clearing the statistic properties of the neural network through infinite discussions. We would like to thank them all, since we are aware that they all gave a decisive contribute to the project.     

Neural network solution to an inverse problem associated with the eigenvalues of the Stokes operator

A numerical method, based on the design of two artificial neural networks, is presented in order to approximate the viscosity and density features of fluids from the eigenvalues of the Stokes operator. The finite element method is used to solve the direct problem by training a first artificial neural network. A nonlinear map of eigenvalues of the Stokes operator as a function of the viscosity and density of the fluid under study is then obtained. This relationship is later inverted and refined by training a second artificial neural network, solving the aforementioned inverse problem. Numerical examples are presented in order to show the effectiveness and the limitations of this methodology.     

Some methods of training radial basis neural networks in solving the Navier‐Stokes equations

The purpose of this research is to analyze the application of neural networks and specific features of training radial basis functions for solving 2‐dimensional Navier‐Stokes equations. The authors developed an algorithm for solving hydrodynamic equations with representation of their solution by the method of weighted residuals upon the general neural network approximation throughout the entire computational domain. The article deals with testing of the developed algorithm through solving the 2‐dimensional Navier‐Stokes equations. Artificial neural networks are widely used for solving problems of mathematical physics; however, their use for modeling of hydrodynamic problems is very limited. At the same time, the problem of hydrodynamic modeling can be solved through neural network modeling, and our study demonstrates an example of its solution. The choice of neural networks based on radial basis functions is due to the ease of implementation and organization of the training process, the accuracy of the approximations, and smoothness of solutions. Radial basis neural networks in the solution of differential equations in partial derivatives allow obtaining a sufficiently accurate solution with a relatively small size of the neural network model. The authors propose to consider the neural network as an approximation of the unknown solution of the equation. The Gaussian distribution is used as the activation function.     

Quantized energy-to-peak state estimation for persistent dwell-time switched neural networks with packet dropouts

Nonlinear Dynamics - This paper pays close attention to the problem of energy-to-peak state estimation for a class of neural networks under switching mechanism. Persistent dwell-time switching...     

Dynamic stability recognition of cylindrical shallow shells in Kelvin–Voigt viscoelastic medium under transverse white noise excitation

Nonlinear Dynamics - This paper investigates the problem of delay-dependent dissipativity for a class of Markovian jump neural networks with a time-varying delay. A generalized integral inequality...     

Asymmetric autocorrelation function to resolve directional ambiguity in PIV images

 Autocorrelation of a double-exposed image, unlike cross-correlation between two images, produces a correlation function that is symmetric about the origin. Thus, while it is possible to calculate the speed and direction of tracer particles in a particle image velocimetry (PIV) image using autocorrelation, it is impossible to tell whether the velocity is in the positive or negative direction. This ambiguity can be resolved by spatially shifting one exposure relative to the next or labeling exposures with color or polarization, but the complexity and limitations of these methods can be prohibitive. It is, however, possible to resolve the sign of the velocity from a triple-exposed image using unequal time intervals between exposures. Triple-exposed images, like double-exposed images, correlate symmetrically about zero. The directional ambiguity, however, can be resolved by calculating the probability that the three exposures occur in a specific temporal order; that is, by assuming that the correlation has a specific sign and testing to see if the assumption is correct. Traditional spectral and statistical correlation techniques are unable to accomplish this. Presented herein is a computationally efficient asymmetric correlation function that is able to differentiate the temporal order of triple exposed images. Included is a discussion of the limitations of this function and of difficulties in experimental implementation. Received: 28 July 1997/Accepted: 18 February 1998     

Passivity of uncertain neural networks with both leakage delay and time-varying delay

In this paper, the passivity problem is investigated for a class of uncertain neural networks with leakage delay and time-varying delay as well as generalized activation functions. By constructing appropriate Lyapunov–Krasovskii functionals, and employing Newton–Leibniz formulation and the free-weighting matrix method, several delay-dependent criteria for checking the passivity of the addressed neural networks are established in linear matrix inequality (LMI), which can be checked numerically using the effective LMI toolbox in MATLAB. Two examples with simulations are given to show the effectiveness and less conservatism of the proposed criteria.     

Exponential stability in the mean square for stochastic neural networks with mixed time-delays and Markovian jumping parameters

In this paper, the stability analysis problem is considered for a class of stochastic neural networks with mixed time-delays and Markovian jumping parameters. The mixed delays include discrete and distributed time-delays, and the jumping parameters are generated from a continuous-time discrete-state homogeneous Markov process. The aim of this paper is to establish some criteria under which the delayed stochastic neural networks are exponentially stable in the mean square. By constructing suitable Lyapunov functionals, several stability conditions are derived on the basis of inequality techniques and the stochastic analysis. An example is also provided in the end of this paper to demonstrate the usefulness of the proposed criteria.     

Sampled-data state estimation for delayed neural networks with Markovian jumping parameters

This paper is concerned with the sampled-data state estimation problem for a class of delayed neural networks with Markovian jumping parameters. Unlike the classical state estimation problem, in our state estimation scheme, the sampled measurements are adopted to estimate the concerned neuron states. The neural network under consideration is assumed to have multiple modes that switch from one to another according to a given Markovian chain. By utilizing the input delay approach, the sampling period is converted into a time-varying yet bounded delay. Then a sufficient condition is given under which the resulting error dynamics of the neural networks is exponentially stable in the mean square. Based on that, a set of sampled-data estimators is designed in terms of the solution to a set of linear matrix inequalities (LMIs) which can be solved by using the available software. Finally, a numerical example is used to show the effectiveness of the estimation approach proposed in this paper.     

图像维数计算方法及其实质研究

基于颜色单元体模型,本文提出了一种间接量测表面形貌的新方法——真彩图像维数计算方法,并通过激光表面仪扫描实验对此进行了验证。该计算方法的提出,使在表面形貌的实验研究中用拍摄数码照片方法取代激光表面仪扫描成为可能。比较研究显示,该方法具有简单、方便、直观和扫描范围大等优势,是表面形貌实验研究手段上的一大改进。本文还分别探讨了二值化图像,灰度图像,真彩图像的分形特性,剖析了三类图像维数的物理实质。比较研究结果显示,对于同一试样,真彩图像维数最大,灰度图像次之,二值化图像维数最小,其维数在区间[1,2]内。以此说明,二值化图像维数是曲线维数,灰度图像维数仅考虑了颜色明度值的影响,只有真彩图像维数充分考虑了色彩、色饱和度和明度值的"贡献",该维数才是"全局"意义上的图像维数。