2D Map Representation Research Articles

A deep learning method for bearing fault diagnosis based on Cyclic Spectral Coherence and Convolutional Neural Networks

Accurate fault diagnosis is critical to ensure the safe and reliable operation of rotating machinery. Data-driven fault diagnosis techniques based on Deep Learning (DL) have recently gained increasing attention due to theirs powerful feature learning capacity. However, one of the critical challenges lies in how to embed domain diagnosis knowledge into DL to obtain suitable features that correlate well with the health conditions and to generate better predictors. In this paper, a novel DL-based fault diagnosis method, based on 2D map representations of Cyclic Spectral Coherence (CSCoh) and Convolutional Neural Networks (CNN), is proposed to improve the recognition performance of rolling element bearing faults. Firstly, the 2D CSCoh maps of vibration signals are estimated by cyclic spectral analysis to provide bearing discriminative patterns for specific type of faults. The motivation for using CSCoh-based preprocessing scheme is that the valuable health condition information can be revealed by exploiting the second-order cyclostationary behavior of bearing vibration signals. Thus, the difficulty of feature learning in deep diagnosis model is reduced by leveraging domain-related diagnosis knowledge. Secondly, a CNN model is constructed to learn high-level feature representations and conduct fault classification. More specifically, Group Normalization (GN) is employed in CNN to normalize the feature maps of network, which can reduce the internal covariant shift induced by data distribution discrepancy. The proposed method is tested and evaluated on two experimental datasets, including data category imbalances and data collected under different operating conditions. Experimental results demonstrate that the proposed method can achieve high diagnosis accuracy under different datasets and present better generalization ability, compared to state-of-the-art fault diagnosis techniques.

Location Based Services for Human Self-Localization

Abstract. Human self-localisation is an important part of everyday life. In order to determine one’s own position and orientation, the allocentric representation, usually in the form of a map, has to be aligned with one’s own egocentric representation of the real world. This requires objects (anchor points) that are present in both representations. We present two novel approaches that aim to simplify the process of alignment and thus the self-localisation. The Viewshed approach is based on visibility analysis and the Image Recognition approach identifies objects and highlights them on the map. On the basis of an empirical experiment with 30 participants in the city of Vienna, Austria, the two approaches were compared with each other as well as with a standard approach using a 2D map representation. The goal is to assess and compare aspects like efficiency, user experience, and cognitive workload. Results show that the Image Recognition method provided the best support and was also most popular among users. The Viewshed method performed well below expectations.

The Lay of the Land: Sensing and Representing Topography

Navigating, and studying spatial navigation, is difficult enough in two dimensions when maps and terrains are flat. Here we consider the capacity for human spatial navigation on sloped terrains, and how sloping terrain is depicted in 2D map representations, called topographic maps. First, we discuss research on how simple slopes are encoded and used for reorientation, and to learn spatial configurations. Next, we describe how slope is represented in topographic maps, and present an assessment (the Topographic Map Assessment), which can be administered to measure topographic map comprehension. Finally, we describe several approaches our lab has taken with the aim of improving topographic map comprehension, including gesture and analogy. The current research reveals a rich and complex picture of topographic map understanding, which likely involves perceptual expertise, strong spatial skills, and inferential logic.

Ecosystem services visualization and communication: A demand analysis approach for designing information and conceptualizing decision support systems

The concept of ecosystem services (ES) is broadly established in research and in communities of interest. The European Commission (EU) has embraced these conceptual approaches in order to provide policy makers with decision-supportive information concerning the supply of and demand for ES. It is, however, not yet clear how ES information should be represented to fulfill decision-supportive functions or even to process the data in such a manner that it is understandable. Knowledge about the ideal representation and communication of ES information integrated into decision support systems (DSSs) is particularly key for guiding users through such systems. In order to determine the correct representation type for a given situation and intended use, we developed a demand analysis, distributed through an online survey, to identify user demands for ES information. A principal component analysis depicts that requirements were highly heterogeneous among respondents of this study. Five components describing the representation type can, however, be identified, depending on the situation of application and the intended use of the ES information by the respondents: (1) 3D landscape visualizations are preferred for analyzing and exploring ES-related information; (2) texts and abstracts are preferred for communication and discussion support; (3) thematic 2D map representations are preferred to support scenario development in public applications; (4) abstract 3D landscape visualizations facilitate estimations in group applications; and (5) charts and tables, in combination with thematic 2D map representations, support analyses. However, while certain representation types are function- and/or situation-specific, no representation type can be used as a panacea. A demand analysis, as presented in this paper, can contribute to the definition of how ES information is to be integrated into DSSs and how it needs to be designed to be (decision-) supportive.

"Building block picture" of the electronic structure of aqueous cysteine derived from resonant inelastic soft X-ray scattering.

The electronic structure of the amino acid L-cysteine in an aqueous environment was studied using resonant inelastic soft X-ray scattering (RIXS) in a 2D map representation and analyzed in the framework of a "building block" approach. The element selectivity of RIXS allows a local investigation of the electronic structure of the three functional groups of cysteine, namely, the carboxyl, amino, and thiol groups, by measuring at the O K, N K, and S L2,3 edges, respectively. Variation of the pH value allows an investigation of molecules with protonated and deprotonated functional groups, which can then be compared with simple reference molecules that represent the isolated functional groups. We find that such building blocks can provide an excellent description of X-ray emission spectroscopy (XES) and RIXS spectra, but only if all nearest-neighbor atoms are included. This finding is analogous to the building block principle commonly used in X-ray absorption spectroscopy. The building blocks show a distinct spectral character (fingerprint) and allow a comprehensive interpretation of the cysteine spectra. This simple approach opens the path to investigate the electronic structure of more complex biological molecules in aqueous solutions using XES and RIXS.

Quantitative analysis of X-ray absorption spectra using a 2D map representation

A promising possibility for the quantitative analysis of X-ray absorption near edge structure (XANES) spectra of nanosized electrode materials is demonstrated. We used a 2D map representation technique, which utilizes the values of the first derivatives of the absorbance with respect to the inserted Li + content plotted over the two-dimensional space defined by the inserted Li + content (mole) versus photon energy (eV) as a single map. The technique was applied to XANES spectra of the Li y CoO system in the first Li + insertion reaction for determining the structural and electronic variations associated with the change in Li + content. The obtained show that the intensities of two peaks at 7725 and 7711 eV increased with the Li + content and the difference of intensity change of these two peaks carried out for successive couples of spectra yielded the largest changes at 1.05 and 1.98 mol of Li content. This approach for quantitative analysis of XANES without using conventional simulation techniques enable us to interpret X-ray absorption spectroscopy (XAS) as a quantitative analytical technique with greater confidence.

Web-Based Application for Telerobotics

In the present paper the architecture of Web-based telerobotics system is presented. A remote Web interface for telecontrol facilitates natural and easily distributed human-robot interaction and remote coordination in real time. A microrobot with tactile range finders for local map building is used. The user participates in control procedure through finding the optimal trajectory as a virtual hand drawing over the 2D map representation of the environment. Two proxy systems are proposed in order to over come the unreliability of the Internet connection and to ensure smooth running of the interactive process on the client side. In order to achieve these goals a reflexive module and fuzzy prompt system are used.