George A. Rovithakis PhD, Manolis A. Christodoulou PhD's Adaptive Control with Recurrent High-order Neural Networks: PDF

By George A. Rovithakis PhD, Manolis A. Christodoulou PhD (auth.)

ISBN-10: 1447107853

ISBN-13: 9781447107859

ISBN-10: 1447112016

ISBN-13: 9781447112013

The sequence Advances in business keep watch over goals to record and inspire expertise move up to the mark engineering. The fast improvement of keep an eye on expertise has an influence on all components of the keep an eye on self-discipline. New idea, new controllers, actuators, sensors, new business procedures, laptop tools, new purposes, new philosophies ... , new demanding situations. a lot of this improvement paintings is living in commercial studies, feasibility learn papers and the experiences of complex collaborative initiatives. The sequence deals a chance for researchers to give a longer exposition of such new paintings in all elements of commercial keep watch over for wider and swift dissemination. Neural networks is a kind of components the place an preliminary burst of enthusiasm and optimism ends up in an explosion of papers within the journals and lots of displays at meetings however it is simply within the final decade that major theoretical paintings on balance, convergence and robustness for using neural networks up to speed platforms has been tackled. George Rovithakis and Manolis Christodoulou were attracted to those theoretical difficulties and within the sensible points of neural community purposes to commercial difficulties. This very great addition to the Advances in commercial keep watch over sequence offers a succinct file in their learn. The neural community version on the middle in their paintings is the Recurrent excessive Order Neural community (RHONN) and an entire theoretical and simulation improvement is gifted. diversified readers will locate various features of the advance of curiosity. The final bankruptcy of the monograph discusses the matter of producing or creation procedure scheduling.

Show description

Read or Download Adaptive Control with Recurrent High-order Neural Networks: Theory and Industrial Applications PDF

Best theory books

Francis Y. K. Soo (auth.)'s Mao Tse-Tung’s Theory of Dialectic PDF

The 12 months 1979 ushered in a brand new section in China's lengthy and non-stop revolu­ tion. at the moment, this new section is being symbolically spoke of, by way of the chinese language leaders themselves, because the 'New lengthy March' (a continuation of the mythical and old lengthy March) by way of modernization, which contains the 4 Modernizations: Agriculture, undefined, technological know-how and know-how, and army security.

Additional resources for Adaptive Control with Recurrent High-order Neural Networks: Theory and Industrial Applications

Sample text

20) is linearized and reduced to dla Tadt = -la - 4>Q Tm dd~ + Va, = 4>Ia - KoQ - mL, now with 4> a constant value parameter. This method of linearization and reduction can not be used when one has to alter the field excitation to fulfil additional requirements imposed to the control system however, as for example in the loss minimization problem [66]. The regulation problem of a DC motor is translated as follows: Find a state feedback to force the angular velocity Q and the armature current la, to go to zero, while the magnetic flux 4>, varies.

E(rW dr ~ - . e(rW dr + _1_ where amin:= min{ai ; i = 1, .. 1. 46) guarantees that ei and 1>i remain bounded for all i = 1, ... n, and furthermore, the "energy" of the state error e(t) is proportional to the "energy" of the modeling error v(t). , v E £2, then e(t) converges to zero asymptotically. 1. 46) to be discontinuous; therefore standard existence and uniqueness results of solutions to differential equations are in general not applicable. Wi! Wi! Wi! 46) and, in addition, guarantees the existence of a unique solution, in the sense of Caratheodory [34].

N i = 1,2, ... 3 Robust Learning Algorithms The derivation of the learning algorithms developed in the previous section made the crucial assumption of no modeling error. Equivalently, it was assumed that there exist weight vectors wi, for i = 1, ... 2. Filtered-error RHONN identifier System Model: Parametric Model: RHONN Identifier Model: Identification Error: Weight Estimation Error: Learning Law: X = F(X,u), Xi = -aiXi + wt T z, Xi = -aiXi + wT z, ei = Xi - rPi = Wi - Xi, wt, Wi = -rizei X E iR n , u E iR m i i i = = 1,2, ...

Download PDF sample

Adaptive Control with Recurrent High-order Neural Networks: Theory and Industrial Applications by George A. Rovithakis PhD, Manolis A. Christodoulou PhD (auth.)


by Michael
4.0

Rated 4.32 of 5 – based on 43 votes