Remaining Useful Life (RUL) of a component or system is defined as the length from current time to end useful life. Accurate RUL estimation plays critical role in Prognostics and Health Management(PHM). Data driven approaches for use sensor data operational estimate RUL. Traditional regression based recent Convolutional Neural Network (CNN) approach features created sliding windows build models. However, sequence information not fully considered these approaches. Sequence learning models...

While early emphasis of Infrastructure as a Service (IaaS) clouds was on providing resource elasticity to end users, providers are increasingly interested in over-committing their resources maximize the utilization and returns capital investments.In principle, hedges that users-on average-only need small portion leased resources.When such hedge fails (i.e., demand far exceeds available physical capacity), must mitigate this provider-induced overload, typically by migrating virtual machines...

Real life data often includes information from different channels. For example, in computer vision, we can describe an image using features, such as pixel intensity, color, HOG, GIST feature, SIFT etc.. These aspects of the same objects are called multi-view (or multi-modal) data. Low-rank regression model has been proved to be effective learning mechanism by exploring low-rank structure real But previous only works on single view In this paper, propose a imposing constraints model. Most...

Current client/server or multiagent based embedded database systems are hard to match the quality of service distributed industry monitoring. To address issues, an cloud (ECDBS) method is proposed. First, ECDBS framework constructed, and a dual-timing transaction control (DT <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> C) proposed increase real-time performance stableness processing. Then, computing middleware subsystem developed...

In mobile edge computing systems, the server placement problem is mainly tackled as a multi-objective optimization and solved with mixed integer programming, heuristic or meta-heuristic algorithms, etc. These methods, however, have profound defect implications such poor scalability, local optimal solutions, parameter tuning difficulties. To overcome these defects, we propose novel algorithm based on deep q-network reinforcement learning, dubbed DQN-ESPA, which can achieve placements without...

In machine learning and data mining, dimensionality reduction is one of the main tasks. Linear Discriminant Analysis (LDA) a widely used supervised algorithm it has attracted lot research interests. Classical finds subspace to minimize within-class distance maximize between-class distance, where computed using arithmetic mean all distances. However, some limitations. First, gives equal weight distances, large could dominate result. Second, does not consider pairwise thus classes may overlap...

Dynamic dispatching is one of the core problems for operation optimization in traditional industries such as mining, it about how to smartly allocate right resources place at time. Conventionally, industry relies on heuristics or even human intuitions which are often short-sighted and sub-optimal solutions. Leveraging power AI Internet Things (IoT), data-driven automation reshaping this area. However, facing its own challenges large-scale heterogenous trucks running a highly dynamic...

Data-driven Remaining Useful Life (RUL) estimation for systems with abrupt failures is a very challenging problem. In these systems, the degradation starts close to failure time and accelerates rapidly. Normal data no sign of can act as noise in training step, prevent RUL estimator model from learning patterns. This degrade performance significantly. Therefore, it critical identify mode during step. Moreover, application predicting when system normal not showing any generate inaccurate...

Linear Discriminant Analysis (LDA) is a widely-used supervised dimensionality reduction method in computer vision and pattern recognition. In null space based LDA (NLDA), well-known extension, between-class distance maximized the of within-class scatter matrix. However, there are some limitations NLDA. Firstly, for many data sets, matrix does not exist, thus NLDA applicable to those datasets. Secondly, uses arithmetic mean distances gives equal consideration all distances, which makes larger...

It is well recognized that strain and deflection data are important indexes to judge the safety of truss structures. Specifically, shape sensing technology can estimate deformation a structure by exploiting discrete without considering material property conditions. To fill gap in which most methods SHM (structural health monitoring) cannot be directly used predict displacement field, this paper proposed novel inverse finite element method (iFEM) algorithm based on equivalent stiffness...

Singular value decomposition (SVD) is the mathematical basis of principal component analysis (PCA). Together, SVD and PCA are one most widely used formalism/decomposition in machine learning, data mining, pattern recognition, artificial intelligence, computer vision, signal processing, etc. In recent applications, regularization becomes an increasing trend. this paper, we present a regularized (RSVD), efficient computational algorithm, provide several theoretical analysis. We show that...

Prognostics and Health Management (PHM) is an emerging engineering discipline which concerned with the analysis prediction of equipment health performance. One key challenges in PHM to accurately predict impending failures equipment. In recent years, solutions for failure have evolved from building complex physical models use machine learning algorithms that leverage data generated by However, problems pose a set unique make direct application traditional classification impractical. These...

Efficient dispatching rule in manufacturing industry is key to ensure product on-time delivery and minimum past-due inventory cost. Manufacturing, especially the developed world, moving towards on-demand meaning a high mix, low volume mix. This requires efficient that can work dynamic stochastic environments, it allows for quick response new orders received over disparate set of shop floor settings. In this paper we address problem manufacturing. Using reinforcement learning (RL), propose...

In equipment health classification, machines in normal, degradation and critical stages are classified based on domain experts KPI (Remaining Useful Life). Higher values indicate healthier machines. GANs can be used to generate sensor data for different stages. There challenges this type of generation. Firstly, the generated samples should well separated. For example, it is not preferred that stage have higher than stage. Secondly, equally with each other. instance, normal more like However,...

A novel nanocomposite proton-exchange membrane (PEM) was obtained by combined electrospinning and solution casting of a composite sulfophenylated poly(ether ether ketone ketone) (SP-PEEKK) maleic anhydride modified nanocellulose (MN). SP-PEEKK prepared polymerization between phenyl hydroquinone 1,4-bis(4-fluorobenzoyl)benzene followed post-sulfonation. Nanocellulose (NCC) the acid treatment MCC with sulfuric acid, MN carboxyl group modifying NCC anhydride. PEMs 2% (MN2) showed water uptake...

Support Vector Machine (SVM) is an efficient classification approach, which finds a hyperplane to separate data from different classes. This determined by support vectors. In existing SVM formulations, the objective function uses L2 norm or L1 on slack variables. The number of vectors measure generalization errors. this work, we propose Minimal SVM, L0.5 result model further reduces and increases performance.