Physical Representation Learning and Parameter Identification from Video Using Differentiable Physics

International Journal of Computer Vision - Representation learning for video is increasingly gaining attention in the field of computer vision. For instance, video prediction models enable activity...     

Modeling of bubble‐size distribution in water and freon co‐blown free rise polyurethane foams

A model has been developed to simulate the foam characteristics obtained, when chemical (water) and physical (Freon) blowing agents are used together for the formation of polyurethane foams. The model considers the rate of reaction, the consequent rise in temperature of the reaction mixture, nucleation of bubbles, and mass transfer of CO₂ and Freon to them till the time of gelation. The model is able to explain the experimental results available in literature. It further predicts that the nucleation period gets reduced with increase in water (at constant Freon content), whereas with increase in Freon (at constant water) concentration nucleation period decreases marginally leading to narrower bubble‐size distribution. By the use of uniform sized nuclei added initially, the model predicts that the bubble‐size distribution can be made independent of the rate of homogeneous nucleation and can, thus, offer an extra parameter for its control. © 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2014 , 131, 40745.     

Towards farm-level health management of offshore wind farms for maintenance improvements

This paper studies a conceptual architecture for health management of offshore wind farms. To this aim, various necessary enablers of a health management system are presented to improve reliability and availability while optimizing maintenance costs. The main focus lies on improving existing condition monitoring systems based on concepts of condition-based maintenance and reliability centered maintenance. A brief review of the relevant state-of-the-art is presented and gaps to be filled towards realization of such health management system are discussed.     

Optimal power control in interference-limited fading wirelesschannels with outage-probability specifications

We propose a new method of power control for interference-limited wireless networks with Rayleigh fading of both the desired and interference signals. Our method explicitly takes into account the statistical variation of both the received signal and interference power and optimally allocates power subject to constraints on the probability of fading induced outage for each transmitter/receiver pair. We establish several results for this type of problem. We establish tight bounds that relate the outage probability caused by channel fading to the signal-to-interference margin calculated when the statistical variation of the signal and interference powers is ignored. This allows us to show that well-known methods for allocating power, based on Perron-Frobenius eigenvalue theory, can be used to determine power allocations that are provably close to achieving optimal (i.e., minimal) outage probability. We show that the problems of minimizing the transmitter power subject to constraints on outage probability and minimizing outage probability subject to power constraints can be posed as a geometric program (GP). A GP is a special type of optimization problem that can be transformed to a nonlinear convex optimization problem by a change of variables and therefore solved globally and efficiently by interior-point methods. We also give a fast iterative method for finding the optimal power allocation to minimize the outage probability     

Globally Constrained Power Control Across Multiple Channels in Wireless Data Networks

We investigate multi-channel transmission schemes for packetized wireless data networks. The transmitting unit transmits concurrently in several orthogonal channels (for example, distinct FDMA bands or CDMA codes) with randomly fluctuating interference and there is a global constraint on the total power transmitted across all channels at any time slot. Incoming packets to the transmitter are queued up in separate buffers, depending on the channel they are to be transmitted in. In each time slot, one packet can be transmitted in each channel from its corresponding queue. The issue is how much power to transmit in each channel, given the interference in it and the packet backlog, so as to optimize various power and delay costs associated with the system. We formulate the general problem taking a dynamic programming approach. Through structural decompositions of the problem, we design practical novel algorithms for allocating power to various channels under the global power constraint.     

Content-based image retrieval system for cutting tools

Content based image retrieval has recently been investigated extensively. Images and video are queried to identify particular scenes or objects. The method can serve as an alternative to keyword type retrieval or complement such a system; depending on the application, it may be the only viable method of search. In this paper we consider the retrieval of cutting tools based on image content. Three methods are used for image retrieval which are based on matching of colour histograms, angle count histograms and image intensity histograms. The developed system is able to retrieve images that are reasonably similar. Additional work is needed to detect objects that are part of an image and also to improve the accuracy of the search.