Automatic bearing fault diagnosis based on one-class ν-SVM

Rolling-element bearings are among the most used elements in industrial machinery, thus an early detection of a defect in these components is necessary to avoid major machine failures. Vibration analysis is a widely used condition monitoring technique for high-speed rotating machinery. Using the information contained in the vibration signals, an automatic method for bearing fault detection and diagnosis is presented in this work. Initially, a one-class ν-SVM is used to discriminate between normal and faulty conditions. In order to build a model of normal operation regime, only data extracted under normal conditions is used. Band-pass filters and Hilbert Transform are then used sequentially to obtain the envelope spectrum of the original raw signal that will finally be used to identify the location of the problem. In order to check the performance of the method, two different data sets are used: (a) real data from a laboratory test-to-failure experiment and (b) data obtained from a fault-seeded bearing test. The results showed that the method was able not only to detect the failure in an incipient stage but also to identify the location of the defect and qualitatively assess its evolution over time.     

Estimating above‐ground burned biomass and CO2 emissions for tropical Africa for the year 1990 with the NOAA–NASA Pathfinder AVHRR 8 km land dataset

Biomass burning combusts Earth's vegetation (in forests, savannas and agricultural lands) and occurs over huge areas of the Earth's surface. Global estimates of biomass burning are thus required in order to provide exact figures of the gas fluxes derived from this source. In this paper we use coarse resolution images for estimating above‐ground burned biomass and CO₂ emissions for tropical Africa for the year 1990. The burned land cover areas have been derived from burn scar and land cover maps using the global daily National Oceanic and Atmospheric Administration–National Aeronautics and Space Administration (NOAA–NASA) Pathfinder AVHRR 8?km land dataset. A burned area estimation of (742±222)?Mha has been considered. Monthly maximum Normalized Difference Vegetation Index (NDVI) composites and biomass density measurements have been used for modelling the temporal behaviour of above‐ground biomass for the main seasonal vegetation classes in Africa (humid savanna, derived humid savanna, dry savanna grassland and broadleaf savanna). The amount of above‐ground burned biomass and therefore CO₂ emissions can be estimated from burned land cover area, above‐ground biomass density, burn efficiency and emission factor of trace gas by land cover class. A total of 6494 (3675–9312) Tg for CO₂ emissions was computed for tropical Africa for the year 1990.     

A statistical framework for the analysis of long image time series

Coarse spatial resolution satellites are capable of observing large swaths of the planetary surface in each overpass resulting in image time series with high temporal resolution. Many change‐detection strategies commonly used in remote sensing studies were developed in an era of image scarcity and thus focus on comparing just a few scenes. However, change analysis methods applicable to images with sparse temporal sampling are not necessarily efficient and effective when applied to long image time series. We present a statistical framework that gathers together: (1) robust methods for multiple comparisons; (2) seasonally corrected Mann–Kendall trend tests; (3) a testing sequence for quadratic models of land surface phenology. This framework can be applied to long image time series to partition sources of variation and to assess the significance of detected changes. Using a standard image time series, the Pathfinder AVHRR Land (PAL) NDVI data, we apply the framework to address the question of whether the institutional changes accompanying the collapse of the Soviet Union resulted in significant changes in land surface phenologies across the ecoregions of Kazakhstan.     

Synthesis of digital optimal reduced-order compensators for asynchronously sampled systems

The synthesis of finite-horizon digital optimal reduced-order compensators is presented, for asynchronous and aperiodically sampled continuous-time systems. The dimensions of the compensator state are a priori fixed and may be time varying. Asynchronous and aperiodic sampling refers to a deterministic sampling scheme where an arbitrary, but a priori known, number of control variables is updated, and/or an arbitrary, but a priori known, number of outputs is sampled, at arbitrary, but a priori known, time instants. This sampling scheme generalizes most deterministic sampling schemes considered in the control literature. Through the use of an integral criterion the intersample behaviour is explicitly considered in the design. As a result, frequent, synchronous and periodic sampling is no longer necessary, which can be highly relevant in practice. Also the synthesis enables comparison of the optimal performance of reduced-order compensators as a function of their dimensions and the sampling scheme. The synthesis is illustrated with a numerical example.     

AUTOMATION OF GENERALIZED ADDITIVE NEURAL NETWORKS FOR PREDICTIVE DATA MINING

For a new technology to make the step from experimental technology to mainstream technology, tools need to be created to facilitate the use of the developed technology in the envisaged application area. Generalized additive neural networks provide an attractive framework that shows promise in the field of predictive data mining. However, the construction of such networks is very time consuming and subjective, because it depends on the user to interpret partial residual plots and to make changes in the neural network architecture. For this technology to be accepted as a serious modeling option in the field of predictive data mining the construction process needs to be automated and the benefits of using the technique must be clearly illuminated. This article shows how intelligent search may be used to replace subjective human judgment with objective criteria and make generalized additive neural networks an attractive option for the modeler.     

GD&T-Based Characterization of Short-Range Non-contact 3D Imaging Systems

We present a series of test metrics, artifacts, and procedures for characterizing and verifying the operating limits of a short-range non-contact three-dimensional imaging system. These metrics have been designed to correspond to dimensioning and tolerancing metrics that are widely used in industry (e.g. automotive, aerospace, etc.). We introduce operating limit metrics that correspond with the geometric dimensioning and tolerancing (GD&T) metrics of Form (Flatness and Circularity), Orientation (Angularity), Location (Sphere, Corner, and Hole Position Errors), and Size (Diameter, Sphere-spacing, Plane-spacing and Angle Errors). An example is presented to illustrate how these metrics, artifacts, and associated test procedures can be used in practice.     

Analyzing probabilistic pushdown automata

The paper gives a summary of the existing results about algorithmic analysis of probabilistic pushdown automata and their subclasses.     

Automatic laser-based identification for UF6 cylinders

We present a laser item identification system (L2IS), installed in a real facility on trial, that is capable of automatically monitoring all transfers of uranium hexafluoride (UF₆) cylinders between different process areas. Each cylinder has a unique ‘fingerprint’ represented by the 3D micro-structure of its surface which remains intact even under extreme environmental conditions. L2IS automatically identifies each cylinder through exploring the 3D frontal surface acquired by the laser scanner. The system is composed of a portable unit (operated in attended mode) and a fixed installed unit (operated without inspector presence). The efficiency and accuracy of our identification system were evaluated on a large dataset acquired during several years including a full year of field testing.     

To be human is to be creative

We will assert than in the era of Ubiquitous Technology to be human is to be creative. Small children are experimental and creative actors. The socialisation process in modern societies, both at home and at educational institutions, does not enhance and develop their creativity. On the contrary, their creativity is discouraged in many ways. We conceptualise creativity developmentally: It is possible to use activities, teaching methods, motivation and procedures to enhance and develop creativity, even in older people. This paper gives some guides that can be used both at home and at work to explore, enhance and develop ones own creativity and the creativity of others. Each suggestion is presented from a practical viewpoint and then related to some of the tools and concepts that scientists and artists use in their creative endeavours.     

Ordinal and nominal classification of wind speed from synoptic pressurepatterns

Wind speed reconstruction is a challenging problem in areas (mainly wind farms) where there are not direct wind measures available. Different approaches have been applied to this reconstruction, such as measure-correlate-predict algorithms, approaches based on physical models such as reanalysis methods, or more recently, indirect measures such as pressure, and its relation to wind speed. This paper adopts the latter method, and deals with wind speed estimation in wind farms from pressure measures, but including different novelties in the problem treatment. Existing synoptic pressure-based indirect approaches for wind speed estimation are based on considering the wind speed as a continuous target variable, estimating then the corresponding wind series of continuous values. However, the exact wind speed is not always needed by wind farm managers, and a general idea of the level of speed is, in the majority of cases, enough to set functional operations for the farm (such as wind turbines stop, for example). Moreover, the accuracy of the models obtained is usually improved for the classification task, given that the problem is simplified. Thus, this paper tackles the problem of wind speed prediction from synoptic pressure patterns by considering wind speed as a discrete variable and, consequently, wind speed prediction as a classification problem, with four wind level categories: low, moderate, high or very high. Moreover, taking into account that these four different classes are associated to four values in an ordinal scale, the problem can be considered as an ordinal regression problem. The performance of several ordinal and nominal classifiers and the improvement achieved by considering the ordering information are evaluated. The results obtained in this paper present the support vector machine as the best tested classifier for this task. In addition, the use of the intrinsic ordering information of the problem is shown to significantly improve ranks with respect to nominal classification, although differences in accuracy aresmall.