Search

We present an event-triggered observer design for linear time-invariant systems, where the measured output is sent to the observer only when a triggering condition is satisfied. We proceed by emulation and we first construct a continuous-time Luenberger observer. We then propose a dynamic rule to trigger transmissions, which only depends on the plant output and an auxiliary scalar state variable. The overall system is modeled as a hybrid system, for which a jump corresponds to an output transmission. We show that the proposed event-triggered observer guarantees global practical asymptotic stability for the estimation error dynamics. Moreover, under mild boundedness conditions on the plant state and its input, we prove that there exists a uniform strictly positive minimum inter-event time between any two consecutive transmissions, guaranteeing that the system does not exhibit Zeno solutions. Finally, the proposed approach is applied to a numerical case study of a lithium-ion battery.

We propose a framework for designing observers possessing global convergence properties and desired asymptotic behaviors for the state estimation of nonlinear systems. The proposed scheme consists in combining two given continuous-time observers: One, denoted as global, ensures (approximate) convergence of the estimation error for any initial condition ranging in some prescribed set, while the other, denoted as local, guarantees a desired local behavior. We make assumptions on the properties of these two observers, and not on their structures, and then explain how to unite them as a single scheme using hybrid techniques. Two case studies are provided to demonstrate the applicability of the framework. Finally, a numerical example is presented.

This work deals with the experimental analysis of wake losses fluctuations at full-scale wind turbines. The test case is a wind farm sited on a moderately complex terrain: 4 turbines are installed, having 2 MW of rated power each. The sources of information are the time-resolved data, as collected from the OPC server, and the 10-minutes averaged SCADA data. The objective is to compare the statistical distributions of wake losses for far and middle wakes, as can be observed through the "fast" lens of time-resolved data, for certain selected test-case time series, and through the "slow" lens of SCADA data, on a much longer time basis that allow to set the standards of the mean wake losses along the wind farm. Further, time-resolved data are used for an insight into the spectral properties of wake fluctuations, highlighting the role of the wind turbine as low-pass filter. Summarizing, the wind rose, the layout of the site and the structure of the data sets at disposal allow to study middle and far wake behavior, with a "slow" and "fast" perspective., QC 20170830

Generic techniques are available for the design of local observers for nonlinear continuous-time systems. By local, we mean that the state estimate is guaranteed to converge to the true state, provided its initial value is nearby the plant initial condition. The main drawback of this approach is the right initialization of the observer, which may not be easy in practice. To overcome this potential issue, we propose to combine two observers: a local one, and an other observer, which provides (approximate) estimates but in a global sense, namely for any initial condition. We explain how to combine these two observers with a hybrid scheme guaranteeing global asymptotic convergence. The hybrid observer, called uniting observer, matches the local observer when the initial conditions are nearby the initial conditions of the observed system. We illustrate the proposed approach by means of a numerical example.

We revisit the calculation of curvature corrections to the pp-wave energy of type IIA string states on \textAd\textS4 ×\mathbbCP3 {\text{Ad}}{{\text{S}}_4} \times \mathbb{C}{P^3} initiated in arXiv:0807.1527. Using the near pp-wave Hamiltonian found in arXiv:0912.2257, we compute the first non-vanishing correction to the energy of a set of bosonic string states at order 1/R 2, where R is the curvature radius of the background. The leading curvature corrections give rise to cubic, order 1/R, and quartic, order 1/R 2, terms in the Hamiltonian, for which we implement the appropriate normal ordering prescription. Including the contributions from all possible fermionic and bosonic string states, we find that there exist logarithmic divergences in the sums over mode numbers which cancel between the cubic and quartic Hamiltonian. We show that from the form of the cubic Hamiltonian it is natural to require that the cutoff for summing over heavy modes must be twice the one for light modes. With this prescription the strong-weak coupling interpolating function h(¿), entering the magnon dispersion relation, does not receive a one-loop correction, in agreement with the algebraic curve spectrum. However, the single magnon dispersion relation exhibits finite-size exponential corrections., We revisit the calculation of curvature corrections to the pp-wave energy of type IIA string states on AdS4×CP3 initiated in arXiv:0807.1527. Using the near pp-wave Hamiltonian found in arXiv:0912.2257, we compute the first non-vanishing correction to the energy of a set of bosonic string states at order 1/R^2, where R is the curvature radius of the background. The leading curvature corrections give rise to cubic, order 1/R, and quartic, order 1/R^2, terms in the Hamiltonian, for which we implement the appropriate normal ordering prescription. Including the contributions from all possible fermionic and bosonic string states, we find that there exist logarithmic divergences in the sums over mode numbers which cancel between the cubic and quartic Hamiltonian. We show that from the form of the cubic Hamiltonian it is natural to require that the cutoff for summing over heavy modes must be twice the one for light modes. With this prescription the strong-weak coupling interpolating function h(¿), entering the magnon dispersion relation, does not receive a one-loop correction, in agreement with the algebraic curve spectrum. However, the single magnon dispersion relation exhibits finite-size exponential corrections.

Today the study of biological intelligence constitutes perhaps the most challenging yet rewarding frontier of research. A most compelling notion attracting many researchers is that such complex intelligence originates from a vast interconnected architecture containing approximately 1012 relatively simple computational units or neurons. The massively parallel architecture, sporting approximately 1015 adaptable neural interconnectors, is responsible for the extraordinary pattern processing capabilities of the brain. Among the pattern processing capabilities of interest to machine intelligence researchers pursuing engineering applications are fast pattern recognition by association, apparent infinite memory capacity, generalization, and robust, fault-tolerant performance. Furthermore, these system level capabilities are obtained biologically at modest expense, for the human brain consumes energy at a rate comparable to that of a single VLSI digital signal processing chip. Such efficiency arises from the rather conservative processing units, operating with 1 ms time constants and slowly communicating information (100ms) along passive neuronal channels.