Search

It is well known that any kind of time series algorithm requires past information to model the inherent temporal relationship between past and future. This temporal dependency (i.e. number of past samples required for a good prediction) is generally addressed by feeding a number of past instances to the model in an empirical manner. Conventional approaches mostly rely on offline model, making them impractical to be adopted in the online or streaming context. Hence, a novel method of online temporality analysis is proposed in this paper. The estimated temporality is then employed to form an Adaptive Temporal Neural Network (ATNN) with an elastic memory capable of automatically selecting number of past samples to be used. Temporality change or drift can be a common occurrence in data streams. Hence a drift detection mechanism is also proposed. Once such drift is detected, a drift handling mechanism kicks in which utilizes the rate of drift, making our solution truly autonomous. The entire mechanism is termed as LEMON: Learning Elastic Memory Online. LEMON although not a time series model in itself, can work with any predictive models to improve their performance. Synthetic datasets are used here as proof of correct temporality estimation and drift detection whereas real world datasets are employed to demonstrate how LEMON improves the predictive performance and speed of an existing model with the knowledge of temporality and drift. Nanyang Technological University We would like to sincerely thank Energy Research Institute at Nanyang Technological University (ERI@N), Singapore for their continued support. The work was funded by their SMES project.

A collocation method has been recently developed as a powerful alternative to Galerkin's method in the context of isogeometric analysis, characterized by significantly reduced computational cost, but still guaranteeing higher‐order convergence rates. In this work, we propose a novel adaptive isogeometric analysis meshfree collocation (IGAM‐C) for the two‐dimensional (2D) elasticity and frictional contact problems. The concept of the IGAM‐C method is based upon the correspondence between the isogeometric collocation and reproducing kernel meshfree approach, which facilitates the robust mesh adaptivity in isogeometric collocation. The proposed method reconciles collocation at the Greville points via the discretization of the strong form of the equilibrium equations. The adaptive refinement in collocation is guided by the gradient‐based error estimator. Moreover, the resolution of the nonlinear equations governing the contact problem is derived from a strong form to avoid the disadvantages of numerical integration. Numerical examples are presented to demonstrate the effectiveness, robustness, and straightforward implementation of the present method for adaptive analysis. NRF (Natl Research Foundation, S’pore) Accepted version

While significant efforts for online learning have been devoted to arrive at reliable predictions of crisp values, the problem of prediction interval (PI) in practical data is one of the underexplored areas in the existing literature. PI aims to produce upper and lower bound predictions which capture possible domain solution. This paper aims to extend a prominent meta-cognitive learning algorithm, namely meta-cognitive interval type-2 fuzzy inference system (McIT2FIS), to cope with the problem of prediction interval in real-time. McIT2FIS is constructed under interval type-2 fuzzy inference system and realizes the meta-cognitive learning theory featuring the basic three elements of human learning: what-to-learn, how-to-learn, when-to-learn. Unlike existing works in PI, McIT2FIS-PI works fully in the online mode and is capable of performing automatic knowledge acquisition from data streams. The efficacy of McIT2FIS-PI has been experimentally validated in a real-world wave characteristics prediction in Semakau Island, Singapore, where it is capable of delivering accurate short-term prediction intervals of wave parameters. The performance of McIT2FIS-PI is also compared with existing state-of-the-art fuzzy inference systems in benchmark problems where it attains competitive accuracy while retaining comparable complexity. Nanyang Technological University The present work was supported by Energy Research Institute @Nanyang Technological University, Interdisciplinary Graduate School, Singapore.

In assessment of the response of floating wind turbines under extreme wave conditions, structure stability and survivability is of utmost importance in the design and implementation. The experimental upwind horizontal axis floating wind turbine on a tension leg platform (TLP) setup was validated with the strongly coupled partitioned six degree-of-freedom rigid body motion solver (Chow and Ng 2016). After tuning of the unknown variables such as the tendon stiffness and damping coefficients with the decay tests, the system ran with the coupled fluid-motion numerical solver resulted in accurate estimations of the natural frequencies and damping ratios. Together with a modified restrain system to model the tendons, the response of the floating wind turbine under regular and focused waves simulations were found to be well-predicted. A stability analysis was performed to determine the iterations that should be ran every time step. The median of the time steps converged within 8.7 iterations. EDB (Economic Devt. Board, S’pore) Published version

Tremendous research efforts have been made to develop advanced lithium-ion battery (LIB) anode materials with high capacity, long cycle life and excellent rate capability. Metal–organic frameworks (MOFs) are regarded as appealing candidates to address the issues such as huge volume change, pulverization and sluggish ion transport kinetics. In this work, a 2D few-layer black phosphorus/NiCo MOF (BP/NiCo MOF) hybrid is rationally designed and prepared through a facile route in a solution of Ni2+, Co2+ and benzenedicarboxylic acid (BDC) at room temperature. The carboxylate groups in BDC2− can not only chelate with metal ions but also bond with BP, resulting in a stable 2D hybrid structure. The BP/NiCo MOF electrode can provide abundant redox-active sites and ensure high Li+ storage capability. The 2D nanostructure and porous structure of BP/NiCo MOF can not only provide favorable pathways for charge transport but also buffer the volume change upon cycling, leading to the enhanced electrochemical performance with good cycling stability and excellent rate capability. The rationally designed 2D BP/NiCo MOF demonstrates high reversible capacity (853 mA h g−1 at 0.5 A g−1), long cycle life and excellent high rate capability (398 mA h g−1 at 5 A g−1 after 1000 cycles).

The bearing faults may lead to expensive and catastrophic failures, which affect the reliability of power drivetrain in electromechanical system. To reduce impact of the bearing failures, accurate analysis and timely diagnosis are required for improved reliability of the electromechanical systems. In this work, an enhanced transient current signature analysis has been investigated for aircraft application using permanent magnet synchronous motor with bearing defects. The motor current under steady-state and transient conditions is acquired from an experimental test-rig with bearing defects at different loading and speed levels. The acquired signals are first investigated using frequency domain analysis and then compared with the time–frequency domain analysis such as wavelet analysis. The discrete wavelet transform is used to analyze on the calculated residual current for the bearing fault diagnosis. The proposed time–frequency-based technique is able to provide useful features that characterize the condition of the bearings on transient current. Further, back-propagation neural network is used over the calculated features which distinguishes the defective bearing from the healthy bearing with high accuracy.

In this study, the ability of the power-law wind model to reproduce a tropical vertical wind profile is investigated. Data from a one-year observation on an offshore island of Singapore, obtained using a lidar, are used to show that the power-law model profiles are unable to capture the pertinent features of the observed wind profiles. Two alternative profile models, one based on the free convection limit scaling and the other based on the arctangent function, are introduced and tested. The latter model, although lacking physical basis, is found to be superior in predicting the measured profiles. A systematic study of the seasonal and diurnal variations of the model parameters is presented and some practical implications of the results are also discussed.

usc Refereed/Peer-reviewed When it comes to time series forecasting, lag in the predicted sequence can be a predominant issue. Unfortunately, this is often overlooked in most of the time series literature as this does not contribute to a high prediction error (i.e. MSE). However, it leads to a rather poor forecast in terms of movement prediction in time series. In this article, we tackle this basic problem with a novel trend driven mechanism. Trend, defined as the inherent pattern of the data, is extracted here and utilized next to perform a lag-free forecasting. We propose a generic and light Dual Network Solution (DNS), where the first network predicts the trend and the second network utilizes that predicted trend along with its historical information to capture the dynamical behavior of the time series efficiently. DNS exhibits a substantially improved (≈10% better) performance compared to more complex and resource-intensive state-of-the- art algorithms in large scale regression problems. Apart from the traditional Mean Squared Error (MSE), we also propose a new Movement Prediction Metric or MPM (for detection of lag in time series) as a new complementary performance metric to evaluate the efficacy of DNS better.

Mo2C/Ti64 composites with nearly full densification and enhanced tensile strength were developed by using laser powder bed fusion (LPBF). The addition of 1–3 wt% Mo2C particles with an average particle size of ~ 1 µm enabled the LPBF-printed composites with an alternately laminated α′-Ti/β-Ti microstructure along with Mo2C and in-situ synthesized M23C6 (M was Ti, Al, V, and Mo). During LPBF, most of the Mo2C particles were incorporated into the Ti64 matrix, resulting in the α′-Ti and β-Ti solid solutions. The segregation of Mo contributed to the generation of the β-Ti phase at the melt pool boundaries of the composites, leading to the alternating hard α′-Ti/soft β-Ti phases. Remarkable improvements of 22.8% and 19.5% in the hardness and tensile strength of the Ti64 matrix were achieved, respectively. The strengthening of the LPBF-printed composites was associated with the formation of the laminated α′-Ti/β-Ti structure, solid solution behavior of Mo and C, and presence of the remaining Mo2C and precipitated M23C6. The findings are anticipated to provide insights into the selection of reinforcement materials for additively manufactured composites with nearly full densities and greater mechanical strength.

Mn+1AXn phases (MAX phases for short with M: transition metal, A: A group elements, X: C or N, and n = 1–3) have attracted considerable attention due to the unique combination of the ceramic- and m...

There exist many factors affecting fatigue behavior of metallic materials, among them the effect of grain boundary (GB) segregation of alloying elements remains largely unexplored. To exploit this factor, an atomistic simulation is performed to investigate the cyclic behavior of Al bi-crystals with GB segregation of Co or Ti during their mode I cyclic loading perpendicular to GB plane. Results are given in comparison with the case of pure Al bi-crystal. Detailed analysis of the structure evolution during cyclic loading for the considered bi-crystals is presented. It is found that during the cycling, plastic deformation of the pure Al bi-crystal occurs through perfect dislocation sliding followed by twinning, while the GB segregations inhibit the formation of both perfect dislocations and twins in an Al matrix leading to decrease in ductility of the bi-crystals and intergranular crack propagation rate. The cyclic loading of the samples with GB segregation is accompanied with a generation of plenty of Shockley partials. Both the alloying elements in GBs can improve the fatigue lifetime of the Al bi-crystals. The ability of GB segregations in bi-crystalline and polycrystalline Al to retard the dislocation emission from GBs that has been first theoretically revealed in the study could be successfully used to improve fatigue characteristics of metals, especially nanocrystalline metals with a dense GB network.