Within-class multimodal classification

Multimedia Tools and Applications - In many real-world classification problems there exist multiple subclasses (or clusters) within a class; in other words, the underlying data distribution is...     

Analysis of phosphorus removal and anaerobic selector performance in a full-scale activated sludge process in Singapore.

This paper analyses the performance of the anaerobic selector (A/O process) in a full-scale activated sludge process receiving mostly industrial sewage discharge (> 60%) in Singapore. In addition to the sludge settleability, enhanced biological phosphorus removal (EBPR) was studied. The sludge volume index (SVI) reduced from 200 to 80 ml g(-1) and foaming was suppressed significantly, indicating the effectiveness of the anaerobic selector in improving sludge settleability. The phosphorus removal efficiency was 66%, and 7.5 mg HAc-COD was consumed per mg PO4(3-) -P removed. In the anaerobic compartment, 31% of the SCOD and 73% of the acetic acid in the settled sewage were removed with PO4(3-) -P release of 14.1 mg PO4(3-)-P l(-1). The linear correlation between PO4(3-) -P release in the anaerobic compartment and PO4(3-) -P uptake in the aerobic compartment indicates that there is about 0.8 mg PO4(3-) -P release in the anaerobic compartment per mg PO34(3-) -P uptake in the aerobic compartment. The fates of volatile fatty acids (VFAs) and its short chain acids (SCAs) in the process were studied and discussed.     

The effects of nonlocal impact ionization on the speed of avalanche photodiodes

The nonlocal enhancement in the velocities of charge carriers to ionization is shown to outweigh the opposing effects of dead space, increasing the avalanche speed of short avalanche photodiodes (APDs) over the predictions of a conventional local model which ignores both of these effects. The trends in the measured gain-bandwidth product of two short InAlAs APDs reported in the literature support this result. Relatively large speed benefits are predicted to result from further small reductions in the lengths of short multiplication regions.     

Impact of power control and lognormal shadowing on the mean transmit power of Bluetooth devices

This letter analyzes Bluetooth's power-control algorithm with a goal to study the mean transmit power required in the presence of lognormal shadowing. The following results are found. 1) A smaller power-control step size yields a lower mean transmit power. 2) When the standard deviations of lognormal shadowing are 3 and 6 dB, respectively, a Bluetooth device needs to consume 11.1 dB and 15.0 dB more in the transmit energy than the minimum one required in the absence of shadowing. 3) The transmit energy consumption varies by around 6 dB among Bluetooth devices as a result of the /spl plusmn/6 dB tolerance in the golden receive power range, which has a nominal size of 20 dB.     

Finite element piezothermoelasticity analysis and the active control of FGM plates with integrated piezoelectric sensors and actuators

 An efficient finite element model is presented for the static and dynamic piezothermoelastic analysis and control of FGM plates under temperature gradient environments using integrated piezoelectric sensor/actuator layers. The properties of an FGM plate are functionally graded in the thickness direction according to a volume fraction power law distribution. A constant displacement-cum-velocity feedback control algorithm that couples the direct and inverse piezoelectric effects is applied to provide active feedback control of the integrated FGM plate in a closed loop system. Numerical results for the static and dynamic control are presented for the FGM plate, which consists of zirconia and aluminum. The effects of the constituent volume fractions and the influence of feedback control gain on the static and dynamic responses of the FGM plates are examined. Received: 13 March 2002 / Accepted: 5 March 2003 The work described in this paper was supported by a grant awarded by the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. CityU 1024/01E).     

PHYSICS-BASED SIMULATION OF HIGH SPEED MACHINING

Computer simulation of high speed machining processes can provide a unique insight and reduce the number of design iterations required to advance and optimize the process. Predictive modeling of high speed machining of exotic materials has been hindered by the nonlinear behavior of this type of materials at extremely high strain, strain rate, and temperatures. This paper presents a physics-based modeling technology that includes the change in the material constitutive equation and the friction characterization at cutting speeds up to 400 m min^-1. The dependence of the accuracy of the predicted parameters, such as the chip formation on cutting forces, chip/tool/workpiece interface temperature, stress and strain distributions are also discussed. The fundamentals of metal cutting were utilized to understand the effect of parameter changes in regimes that are outside current empirical knowledge databases.     

Performance of combined cobalt—nickel—zirconia and HZSM-5 catalyst systems for carbon monoxide hydrogenation

The synthesis of hydrocarbons via hydrogenation of carbon monoxide was investigated over cobalt—nickel—zirconia catalysts of various compositions in combination with zeolite HZSM-5 in “mixed bed” and “follow bed” arrangements. These combinations resulted in the formation of aromatics in amounts as high as 30-35 wt% under relatively mild operating conditions (1 atm, 250–280°C). Although the olefinicity of C₂ and C₃ fractions in the product stream was higher in the mixed bed compared to the follow bed arrangement, the selectivities to aromatics were comparable in the two bed arrangements. The aromatic selectivity was found to be sensitive to operating conditions. The formation of aromatics was favored at high HZSM-5/metal catalyst ratios, low space velocities and high reaction temperatures. The product distributions obtained using various metal/zeolite bifunctional catalysts have been discussed.     

Anatomically constrained electrical impedance tomography forthree-dimensional anisotropic bodies

As shown previously for two-dimensional geometries, anisotropy effects should not be ignored in electrical impedance tomography (EIT) and structural information is important for the reconstruction of anisotropic conductivities. Here, we describe the static reconstruction of an anisotropic conductivity distribution for the more realistic three-dimensional (3-D) case. Boundaries between different conductivity regions are anatomically constrained using magnetic resonance imaging (MRI) data. The values of the conductivities are then determined using gradient-type-algorithms in a nonlinear-indirect approach. At each iteration, the forward problem is solved by the finite element method. The approach is used to reconstruct the 3-D conductivity profile of a canine torso. Both computational performance and simulated reconstruction results are presented together with a detailed study on the sensitivity of the prediction error with respect to different parameters. In particular, the use of an intracavity catheter to better extract interior conductivities is demonstrated